From f43f627d2f17e95c78647eeddf968d12f5c286b1 Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Mon, 4 Feb 2013 13:37:20 +0000
Subject: PM: make VT switching to the suspend console optional v3

KMS drivers can potentially restore the display configuration without
userspace help.  Such drivers can can call a new funciton,
pm_vt_switch_required(false) if they support this feature.  In that
case, the PM layer won't VT switch to the suspend console at suspend
time and then back to the original VT on resume, but rather leave things
alone for a nicer looking suspend and resume sequence.

v2: make a function so we can handle multiple drivers (Alan)
v3: use a list to track device requests (Rafael)
v4: Squash in build fix from Jesse for CONFIG_VT_CONSOLE_SLEEP=n
v5: Squash in patch from Wu Fengguang to add a few missing static
qualifiers.
v6: Add missing EXPORT_SYMBOL.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> (v3)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 include/linux/pm.h | 13 +++++++++++++
 1 file changed, 13 insertions(+)

(limited to 'include')

diff --git a/include/linux/pm.h b/include/linux/pm.h
index 03d7bb145311..e5da2f353e8f 100644
--- a/include/linux/pm.h
+++ b/include/linux/pm.h
@@ -35,6 +35,19 @@ extern void (*pm_idle)(void);
 extern void (*pm_power_off)(void);
 extern void (*pm_power_off_prepare)(void);
 
+struct device; /* we have a circular dep with device.h */
+#ifdef CONFIG_VT_CONSOLE_SLEEP
+extern void pm_vt_switch_required(struct device *dev, bool required);
+extern void pm_vt_switch_unregister(struct device *dev);
+#else
+static inline void pm_vt_switch_required(struct device *dev, bool required)
+{
+}
+static inline void pm_vt_switch_unregister(struct device *dev)
+{
+}
+#endif /* CONFIG_VT_CONSOLE_SLEEP */
+
 /*
  * Device power management
  */
-- 
cgit v1.2.3


From 3cf2667b9f8b2c2fe298a427deb399e52321da6b Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Mon, 4 Feb 2013 13:37:21 +0000
Subject: fb: add support for drivers not needing VT switch at suspend/resume
 time

Use the new PM routines to indicate whether we need to VT switch at suspend
and resume time.  When a new driver is bound, set its flag accordingly,
and when unbound, remove it from the PM's console tracking list.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/video/fbmem.c | 7 +++++++
 include/linux/fb.h    | 2 ++
 2 files changed, 9 insertions(+)

(limited to 'include')

diff --git a/drivers/video/fbmem.c b/drivers/video/fbmem.c
index dc61c12ecf8c..2af7153da2e4 100644
--- a/drivers/video/fbmem.c
+++ b/drivers/video/fbmem.c
@@ -1645,6 +1645,11 @@ static int do_register_framebuffer(struct fb_info *fb_info)
 	if (!fb_info->modelist.prev || !fb_info->modelist.next)
 		INIT_LIST_HEAD(&fb_info->modelist);
 
+	if (fb_info->skip_vt_switch)
+		pm_vt_switch_required(fb_info->dev, false);
+	else
+		pm_vt_switch_required(fb_info->dev, true);
+
 	fb_var_to_videomode(&mode, &fb_info->var);
 	fb_add_videomode(&mode, &fb_info->modelist);
 	registered_fb[i] = fb_info;
@@ -1679,6 +1684,8 @@ static int do_unregister_framebuffer(struct fb_info *fb_info)
 	if (ret)
 		return -EINVAL;
 
+	pm_vt_switch_unregister(fb_info->dev);
+
 	unlink_framebuffer(fb_info);
 	if (fb_info->pixmap.addr &&
 	    (fb_info->pixmap.flags & FB_PIXMAP_DEFAULT))
diff --git a/include/linux/fb.h b/include/linux/fb.h
index 58b98606ac26..d49c60f5aa4c 100644
--- a/include/linux/fb.h
+++ b/include/linux/fb.h
@@ -501,6 +501,8 @@ struct fb_info {
 			resource_size_t size;
 		} ranges[0];
 	} *apertures;
+
+	bool skip_vt_switch; /* no VT switch on suspend/resume required */
 };
 
 static inline struct apertures_struct *alloc_apertures(unsigned int max_num) {
-- 
cgit v1.2.3


From 11e17a0873aba66b251e77ff246590b9390cdc68 Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Tue, 19 Feb 2013 13:31:39 -0800
Subject: drm: add initial_config function to fb helper

Rather than building a config which may or may not work, let the driver
build an initial fb config.  This allows the driver to use the BIOS boot
configuration for example, displaying kernel messages and the initial fb
console on the same outputs the BIOS lit up at boot time.  If that
fails, the driver can still fall back the same way as the core.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: dri-devel@lists.freedesktop.org
Acked-by: Dave Airlie <airlied@gmail.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/drm_fb_helper.c | 23 +++++++++++++++--------
 include/drm/drm_fb_helper.h     |  4 ++++
 2 files changed, 19 insertions(+), 8 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c
index 59d6b9bf204b..6764dce44e84 100644
--- a/drivers/gpu/drm/drm_fb_helper.c
+++ b/drivers/gpu/drm/drm_fb_helper.c
@@ -1398,7 +1398,7 @@ static void drm_setup_crtcs(struct drm_fb_helper *fb_helper)
 	struct drm_mode_set *modeset;
 	bool *enabled;
 	int width, height;
-	int i, ret;
+	int i;
 
 	DRM_DEBUG_KMS("\n");
 
@@ -1419,16 +1419,23 @@ static void drm_setup_crtcs(struct drm_fb_helper *fb_helper)
 
 	drm_enable_connectors(fb_helper, enabled);
 
-	ret = drm_target_cloned(fb_helper, modes, enabled, width, height);
-	if (!ret) {
-		ret = drm_target_preferred(fb_helper, modes, enabled, width, height);
-		if (!ret)
+	if (!(fb_helper->funcs->initial_config &&
+	      fb_helper->funcs->initial_config(fb_helper, crtcs, modes,
+					       enabled, width, height))) {
+		memset(modes, 0, dev->mode_config.num_connector*sizeof(modes[0]));
+		memset(crtcs, 0, dev->mode_config.num_connector*sizeof(crtcs[0]));
+
+		if (!drm_target_cloned(fb_helper,
+				       modes, enabled, width, height) &&
+		    !drm_target_preferred(fb_helper,
+					  modes, enabled, width, height))
 			DRM_ERROR("Unable to find initial modes\n");
-	}
 
-	DRM_DEBUG_KMS("picking CRTCs for %dx%d config\n", width, height);
+		DRM_DEBUG_KMS("picking CRTCs for %dx%d config\n",
+			      width, height);
 
-	drm_pick_crtcs(fb_helper, crtcs, modes, 0, width, height);
+		drm_pick_crtcs(fb_helper, crtcs, modes, 0, width, height);
+	}
 
 	/* need to set the modesets up here for use later */
 	/* fill out the connector<->crtc mappings into the modesets */
diff --git a/include/drm/drm_fb_helper.h b/include/drm/drm_fb_helper.h
index c09511625a11..f97a8ef1d728 100644
--- a/include/drm/drm_fb_helper.h
+++ b/include/drm/drm_fb_helper.h
@@ -68,6 +68,10 @@ struct drm_fb_helper_funcs {
 
 	int (*fb_probe)(struct drm_fb_helper *helper,
 			struct drm_fb_helper_surface_size *sizes);
+	bool (*initial_config)(struct drm_fb_helper *fb_helper,
+			       struct drm_fb_helper_crtc **crtcs,
+			       struct drm_display_mode **modes,
+			       bool *enabled, int width, int height);
 };
 
 struct drm_fb_helper_connector {
-- 
cgit v1.2.3


From 2db76d7c3c6db93058f983c8240f7c7c25e87ee6 Mon Sep 17 00:00:00 2001
From: Imre Deak <imre.deak@intel.com>
Date: Tue, 26 Mar 2013 15:14:18 +0200
Subject: lib/scatterlist: sg_page_iter: support sg lists w/o backing pages

The i915 driver uses sg lists for memory without backing 'struct page'
pages, similarly to other IO memory regions, setting only the DMA
address for these. It does this, so that it can program the HW MMU
tables in a uniform way both for sg lists with and without backing pages.

Without a valid page pointer we can't call nth_page to get the current
page in __sg_page_iter_next, so add a helper that relevant users can
call separately. Also add a helper to get the DMA address of the current
page (idea from Daniel).

Convert all places in i915, to use the new API.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/drm_cache.c            |  2 +-
 drivers/gpu/drm/i915/i915_drv.h        |  2 +-
 drivers/gpu/drm/i915/i915_gem.c        |  8 ++++----
 drivers/gpu/drm/i915/i915_gem_dmabuf.c |  2 +-
 drivers/gpu/drm/i915/i915_gem_gtt.c    |  6 ++----
 drivers/gpu/drm/i915/i915_gem_tiling.c |  4 ++--
 include/linux/scatterlist.h            | 28 +++++++++++++++++++++++-----
 lib/scatterlist.c                      |  4 +---
 8 files changed, 35 insertions(+), 21 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_cache.c b/drivers/gpu/drm/drm_cache.c
index bc8edbeca3fd..bb8f58012189 100644
--- a/drivers/gpu/drm/drm_cache.c
+++ b/drivers/gpu/drm/drm_cache.c
@@ -109,7 +109,7 @@ drm_clflush_sg(struct sg_table *st)
 
 		mb();
 		for_each_sg_page(st->sgl, &sg_iter, st->nents, 0)
-			drm_clflush_page(sg_iter.page);
+			drm_clflush_page(sg_page_iter_page(&sg_iter));
 		mb();
 
 		return;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 1d091ea12fad..f69538508d8c 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1543,7 +1543,7 @@ static inline struct page *i915_gem_object_get_page(struct drm_i915_gem_object *
 	struct sg_page_iter sg_iter;
 
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, n)
-		return sg_iter.page;
+		return sg_page_iter_page(&sg_iter);
 
 	return NULL;
 }
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a1123a32dc27..911bd40ef513 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -442,7 +442,7 @@ i915_gem_shmem_pread(struct drm_device *dev,
 
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents,
 			 offset >> PAGE_SHIFT) {
-		struct page *page = sg_iter.page;
+		struct page *page = sg_page_iter_page(&sg_iter);
 
 		if (remain <= 0)
 			break;
@@ -765,7 +765,7 @@ i915_gem_shmem_pwrite(struct drm_device *dev,
 
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents,
 			 offset >> PAGE_SHIFT) {
-		struct page *page = sg_iter.page;
+		struct page *page = sg_page_iter_page(&sg_iter);
 		int partial_cacheline_write;
 
 		if (remain <= 0)
@@ -1647,7 +1647,7 @@ i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj)
 		obj->dirty = 0;
 
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, 0) {
-		struct page *page = sg_iter.page;
+		struct page *page = sg_page_iter_page(&sg_iter);
 
 		if (obj->dirty)
 			set_page_dirty(page);
@@ -1827,7 +1827,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
 err_pages:
 	sg_mark_end(sg);
 	for_each_sg_page(st->sgl, &sg_iter, st->nents, 0)
-		page_cache_release(sg_iter.page);
+		page_cache_release(sg_page_iter_page(&sg_iter));
 	sg_free_table(st);
 	kfree(st);
 	return PTR_ERR(page);
diff --git a/drivers/gpu/drm/i915/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/i915_gem_dmabuf.c
index 898615d2d5e2..c6dfc1466e3a 100644
--- a/drivers/gpu/drm/i915/i915_gem_dmabuf.c
+++ b/drivers/gpu/drm/i915/i915_gem_dmabuf.c
@@ -130,7 +130,7 @@ static void *i915_gem_dmabuf_vmap(struct dma_buf *dma_buf)
 
 	i = 0;
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, 0);
-		pages[i++] = sg_iter.page;
+		pages[i++] = sg_page_iter_page(&sg_iter);
 
 	obj->dma_buf_vmapping = vmap(pages, i, 0, PAGE_KERNEL);
 	drm_free_large(pages);
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 4cbae7bbb833..24a23b31b55f 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -123,8 +123,7 @@ static void gen6_ppgtt_insert_entries(struct i915_hw_ppgtt *ppgtt,
 	for_each_sg_page(pages->sgl, &sg_iter, pages->nents, 0) {
 		dma_addr_t page_addr;
 
-		page_addr = sg_dma_address(sg_iter.sg) +
-				(sg_iter.sg_pgoffset << PAGE_SHIFT);
+		page_addr = sg_page_iter_dma_address(&sg_iter);
 		pt_vaddr[act_pte] = gen6_pte_encode(ppgtt->dev, page_addr,
 						    cache_level);
 		if (++act_pte == I915_PPGTT_PT_ENTRIES) {
@@ -424,8 +423,7 @@ static void gen6_ggtt_insert_entries(struct drm_device *dev,
 	dma_addr_t addr;
 
 	for_each_sg_page(st->sgl, &sg_iter, st->nents, 0) {
-		addr = sg_dma_address(sg_iter.sg) +
-			(sg_iter.sg_pgoffset << PAGE_SHIFT);
+		addr = sg_page_iter_dma_address(&sg_iter);
 		iowrite32(gen6_pte_encode(dev, addr, level), &gtt_entries[i]);
 		i++;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
index f799708bcb85..c807eb93755b 100644
--- a/drivers/gpu/drm/i915/i915_gem_tiling.c
+++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
@@ -481,7 +481,7 @@ i915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj)
 
 	i = 0;
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, 0) {
-		struct page *page = sg_iter.page;
+		struct page *page = sg_page_iter_page(&sg_iter);
 		char new_bit_17 = page_to_phys(page) >> 17;
 		if ((new_bit_17 & 0x1) !=
 		    (test_bit(i, obj->bit_17) != 0)) {
@@ -511,7 +511,7 @@ i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj)
 
 	i = 0;
 	for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, 0) {
-		if (page_to_phys(sg_iter.page) & (1 << 17))
+		if (page_to_phys(sg_page_iter_page(&sg_iter)) & (1 << 17))
 			__set_bit(i, obj->bit_17);
 		else
 			__clear_bit(i, obj->bit_17);
diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h
index 2d8bdaef9611..e96b9546c4c6 100644
--- a/include/linux/scatterlist.h
+++ b/include/linux/scatterlist.h
@@ -235,13 +235,13 @@ size_t sg_copy_to_buffer(struct scatterlist *sgl, unsigned int nents,
  * sg page iterator
  *
  * Iterates over sg entries page-by-page.  On each successful iteration,
- * @piter->page points to the current page, @piter->sg to the sg holding this
- * page and @piter->sg_pgoffset to the page's page offset within the sg. The
- * iteration will stop either when a maximum number of sg entries was reached
- * or a terminating sg (sg_last(sg) == true) was reached.
+ * you can call sg_page_iter_page(@piter) and sg_page_iter_dma_address(@piter)
+ * to get the current page and its dma address. @piter->sg will point to the
+ * sg holding this page and @piter->sg_pgoffset to the page's page offset
+ * within the sg. The iteration will stop either when a maximum number of sg
+ * entries was reached or a terminating sg (sg_last(sg) == true) was reached.
  */
 struct sg_page_iter {
-	struct page		*page;		/* current page */
 	struct scatterlist	*sg;		/* sg holding the page */
 	unsigned int		sg_pgoffset;	/* page offset within the sg */
 
@@ -255,6 +255,24 @@ bool __sg_page_iter_next(struct sg_page_iter *piter);
 void __sg_page_iter_start(struct sg_page_iter *piter,
 			  struct scatterlist *sglist, unsigned int nents,
 			  unsigned long pgoffset);
+/**
+ * sg_page_iter_page - get the current page held by the page iterator
+ * @piter:	page iterator holding the page
+ */
+static inline struct page *sg_page_iter_page(struct sg_page_iter *piter)
+{
+	return nth_page(sg_page(piter->sg), piter->sg_pgoffset);
+}
+
+/**
+ * sg_page_iter_dma_address - get the dma address of the current page held by
+ * the page iterator.
+ * @piter:	page iterator holding the page
+ */
+static inline dma_addr_t sg_page_iter_dma_address(struct sg_page_iter *piter)
+{
+	return sg_dma_address(piter->sg) + (piter->sg_pgoffset << PAGE_SHIFT);
+}
 
 /**
  * for_each_sg_page - iterate over the pages of the given sg list
diff --git a/lib/scatterlist.c b/lib/scatterlist.c
index b83c144d731f..a1cf8cae60e7 100644
--- a/lib/scatterlist.c
+++ b/lib/scatterlist.c
@@ -401,7 +401,6 @@ void __sg_page_iter_start(struct sg_page_iter *piter,
 	piter->__pg_advance = 0;
 	piter->__nents = nents;
 
-	piter->page = NULL;
 	piter->sg = sglist;
 	piter->sg_pgoffset = pgoffset;
 }
@@ -426,7 +425,6 @@ bool __sg_page_iter_next(struct sg_page_iter *piter)
 		if (!--piter->__nents || !piter->sg)
 			return false;
 	}
-	piter->page = nth_page(sg_page(piter->sg), piter->sg_pgoffset);
 
 	return true;
 }
@@ -496,7 +494,7 @@ bool sg_miter_next(struct sg_mapping_iter *miter)
 		miter->__remaining = min_t(unsigned long, miter->__remaining,
 					   PAGE_SIZE - miter->__offset);
 	}
-	miter->page = miter->piter.page;
+	miter->page = sg_page_iter_page(&miter->piter);
 	miter->consumed = miter->length = miter->__remaining;
 
 	if (miter->__flags & SG_MITER_ATOMIC)
-- 
cgit v1.2.3


From a0d8dde9cbde712fad65822152d2721eec1ca656 Mon Sep 17 00:00:00 2001
From: Archit Taneja <archit@ti.com>
Date: Tue, 12 Feb 2013 16:46:46 +0530
Subject: OMAPDSS: panels: keep platform data of all panels in a single header

Structs for platform data of omapdss panels are found in headers in the
'include/video/' path. Board files populate these structs with platform
specific values, and the panel driver uses these to configure the panel.

Currently, each panel has it's own header in the above path. Move all the
omapdss panel platform data structs to a single header omap-panel-data.h.
This is useful because:

- All other omapdss panel drivers will be modified to use platform data. This
  would lead to a lot of panel headers usable only by omapdss. A lot of these
  platform data structs are trivial, and don't really need a separate header.
- Platform data would be eventually removed, and platform information would be
  passed via device tree. Therefore, omapdss panel platform data structs are
  temporary, and will be easier to remove if they are all in the same header.
- All board files will have to include the same header to configure a panel's
  platform data, that makes the board files more consistent.

Signed-off-by: Archit Taneja <archit@ti.com>
---
 arch/arm/mach-omap2/board-2430sdp.c              |   2 +-
 arch/arm/mach-omap2/board-3430sdp.c              |   2 +-
 arch/arm/mach-omap2/board-am3517evm.c            |   3 +-
 arch/arm/mach-omap2/board-cm-t35.c               |   3 +-
 arch/arm/mach-omap2/board-devkit8000.c           |   3 +-
 arch/arm/mach-omap2/board-h4.c                   |   2 +-
 arch/arm/mach-omap2/board-igep0020.c             |   2 +-
 arch/arm/mach-omap2/board-ldp.c                  |   2 +-
 arch/arm/mach-omap2/board-omap3beagle.c          |   2 +-
 arch/arm/mach-omap2/board-omap3evm.c             |   2 +-
 arch/arm/mach-omap2/board-omap3stalker.c         |   3 +-
 arch/arm/mach-omap2/board-overo.c                |   3 +-
 arch/arm/mach-omap2/dss-common.c                 |   4 +-
 drivers/video/omap2/displays/panel-generic-dpi.c |   2 +-
 drivers/video/omap2/displays/panel-n8x0.c        |   2 +-
 drivers/video/omap2/displays/panel-picodlp.c     |   2 +-
 drivers/video/omap2/displays/panel-taal.c        |   2 +-
 drivers/video/omap2/displays/panel-tfp410.c      |   2 +-
 include/video/omap-panel-data.h                  | 103 +++++++++++++++++++++++
 include/video/omap-panel-generic-dpi.h           |  37 --------
 include/video/omap-panel-n8x0.h                  |  15 ----
 include/video/omap-panel-nokia-dsi.h             |  32 -------
 include/video/omap-panel-picodlp.h               |  23 -----
 include/video/omap-panel-tfp410.h                |  35 --------
 24 files changed, 121 insertions(+), 167 deletions(-)
 create mode 100644 include/video/omap-panel-data.h
 delete mode 100644 include/video/omap-panel-generic-dpi.h
 delete mode 100644 include/video/omap-panel-n8x0.h
 delete mode 100644 include/video/omap-panel-nokia-dsi.h
 delete mode 100644 include/video/omap-panel-picodlp.h
 delete mode 100644 include/video/omap-panel-tfp410.h

(limited to 'include')

diff --git a/arch/arm/mach-omap2/board-2430sdp.c b/arch/arm/mach-omap2/board-2430sdp.c
index a3e0aaa4886b..1322ed707b30 100644
--- a/arch/arm/mach-omap2/board-2430sdp.c
+++ b/arch/arm/mach-omap2/board-2430sdp.c
@@ -38,7 +38,7 @@
 #include "gpmc-smc91x.h"
 
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
+#include <video/omap-panel-data.h>
 
 #include "mux.h"
 #include "hsmmc.h"
diff --git a/arch/arm/mach-omap2/board-3430sdp.c b/arch/arm/mach-omap2/board-3430sdp.c
index ce812decfaca..2612eeaa5889 100644
--- a/arch/arm/mach-omap2/board-3430sdp.c
+++ b/arch/arm/mach-omap2/board-3430sdp.c
@@ -35,7 +35,7 @@
 #include "common.h"
 #include <linux/omap-dma.h>
 #include <video/omapdss.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include "gpmc.h"
 #include "gpmc-smc91x.h"
diff --git a/arch/arm/mach-omap2/board-am3517evm.c b/arch/arm/mach-omap2/board-am3517evm.c
index 9fb85908a61e..1d6c28872505 100644
--- a/arch/arm/mach-omap2/board-am3517evm.c
+++ b/arch/arm/mach-omap2/board-am3517evm.c
@@ -35,8 +35,7 @@
 
 #include "common.h"
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include "am35xx-emac.h"
 #include "mux.h"
diff --git a/arch/arm/mach-omap2/board-cm-t35.c b/arch/arm/mach-omap2/board-cm-t35.c
index af2bb219e214..bccd3e51fecb 100644
--- a/arch/arm/mach-omap2/board-cm-t35.c
+++ b/arch/arm/mach-omap2/board-cm-t35.c
@@ -41,8 +41,7 @@
 
 #include <linux/platform_data/mtd-nand-omap2.h>
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 #include <linux/platform_data/spi-omap2-mcspi.h>
 
 #include "common.h"
diff --git a/arch/arm/mach-omap2/board-devkit8000.c b/arch/arm/mach-omap2/board-devkit8000.c
index 53056c3b0836..12d2126a2382 100644
--- a/arch/arm/mach-omap2/board-devkit8000.c
+++ b/arch/arm/mach-omap2/board-devkit8000.c
@@ -43,8 +43,7 @@
 #include "gpmc.h"
 #include <linux/platform_data/mtd-nand-omap2.h>
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include <linux/platform_data/spi-omap2-mcspi.h>
 #include <linux/input/matrix_keypad.h>
diff --git a/arch/arm/mach-omap2/board-h4.c b/arch/arm/mach-omap2/board-h4.c
index 812c829fa46f..0c1bdd7ee32d 100644
--- a/arch/arm/mach-omap2/board-h4.c
+++ b/arch/arm/mach-omap2/board-h4.c
@@ -34,7 +34,7 @@
 #include <asm/mach/map.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
+#include <video/omap-panel-data.h>
 
 #include "common.h"
 #include "mux.h"
diff --git a/arch/arm/mach-omap2/board-igep0020.c b/arch/arm/mach-omap2/board-igep0020.c
index bf92678a01d0..e979d48270c9 100644
--- a/arch/arm/mach-omap2/board-igep0020.c
+++ b/arch/arm/mach-omap2/board-igep0020.c
@@ -31,7 +31,7 @@
 #include <asm/mach/arch.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 #include <linux/platform_data/mtd-onenand-omap2.h>
 
 #include "common.h"
diff --git a/arch/arm/mach-omap2/board-ldp.c b/arch/arm/mach-omap2/board-ldp.c
index b12fe966a7b9..8a8e505a0e90 100644
--- a/arch/arm/mach-omap2/board-ldp.c
+++ b/arch/arm/mach-omap2/board-ldp.c
@@ -41,7 +41,7 @@
 #include "gpmc-smsc911x.h"
 
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
+#include <video/omap-panel-data.h>
 
 #include "board-flash.h"
 #include "mux.h"
diff --git a/arch/arm/mach-omap2/board-omap3beagle.c b/arch/arm/mach-omap2/board-omap3beagle.c
index c3558f93d42c..0ce91af753fa 100644
--- a/arch/arm/mach-omap2/board-omap3beagle.c
+++ b/arch/arm/mach-omap2/board-omap3beagle.c
@@ -43,7 +43,7 @@
 #include <asm/mach/flash.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 #include <linux/platform_data/mtd-nand-omap2.h>
 
 #include "common.h"
diff --git a/arch/arm/mach-omap2/board-omap3evm.c b/arch/arm/mach-omap2/board-omap3evm.c
index 48789e0bb915..233a0d528fcf 100644
--- a/arch/arm/mach-omap2/board-omap3evm.c
+++ b/arch/arm/mach-omap2/board-omap3evm.c
@@ -51,7 +51,7 @@
 #include "common.h"
 #include <linux/platform_data/spi-omap2-mcspi.h>
 #include <video/omapdss.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include "soc.h"
 #include "mux.h"
diff --git a/arch/arm/mach-omap2/board-omap3stalker.c b/arch/arm/mach-omap2/board-omap3stalker.c
index 95c10b3aa678..495b989f9040 100644
--- a/arch/arm/mach-omap2/board-omap3stalker.c
+++ b/arch/arm/mach-omap2/board-omap3stalker.c
@@ -44,8 +44,7 @@
 #include "gpmc.h"
 #include <linux/platform_data/mtd-nand-omap2.h>
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include <linux/platform_data/spi-omap2-mcspi.h>
 
diff --git a/arch/arm/mach-omap2/board-overo.c b/arch/arm/mach-omap2/board-overo.c
index 86bab51154ee..630833235cbc 100644
--- a/arch/arm/mach-omap2/board-overo.c
+++ b/arch/arm/mach-omap2/board-overo.c
@@ -47,8 +47,7 @@
 #include <asm/mach/map.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-generic-dpi.h>
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 #include "common.h"
 #include "mux.h"
diff --git a/arch/arm/mach-omap2/dss-common.c b/arch/arm/mach-omap2/dss-common.c
index 4be5cfc81ab8..9c49bbe825f7 100644
--- a/arch/arm/mach-omap2/dss-common.c
+++ b/arch/arm/mach-omap2/dss-common.c
@@ -27,9 +27,7 @@
 #include <linux/gpio.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-tfp410.h>
-#include <video/omap-panel-nokia-dsi.h>
-#include <video/omap-panel-picodlp.h>
+#include <video/omap-panel-data.h>
 
 #include "soc.h"
 #include "dss-common.h"
diff --git a/drivers/video/omap2/displays/panel-generic-dpi.c b/drivers/video/omap2/displays/panel-generic-dpi.c
index c904f42d81c1..a0c9396ca43f 100644
--- a/drivers/video/omap2/displays/panel-generic-dpi.c
+++ b/drivers/video/omap2/displays/panel-generic-dpi.c
@@ -35,7 +35,7 @@
 #include <linux/slab.h>
 #include <video/omapdss.h>
 
-#include <video/omap-panel-generic-dpi.h>
+#include <video/omap-panel-data.h>
 
 struct panel_config {
 	struct omap_video_timings timings;
diff --git a/drivers/video/omap2/displays/panel-n8x0.c b/drivers/video/omap2/displays/panel-n8x0.c
index dd1294750802..9c6b5fafeb2e 100644
--- a/drivers/video/omap2/displays/panel-n8x0.c
+++ b/drivers/video/omap2/displays/panel-n8x0.c
@@ -9,7 +9,7 @@
 #include <linux/fb.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-n8x0.h>
+#include <video/omap-panel-data.h>
 
 #define BLIZZARD_REV_CODE                      0x00
 #define BLIZZARD_CONFIG                        0x02
diff --git a/drivers/video/omap2/displays/panel-picodlp.c b/drivers/video/omap2/displays/panel-picodlp.c
index 1b94018aac3e..974ac29236aa 100644
--- a/drivers/video/omap2/displays/panel-picodlp.c
+++ b/drivers/video/omap2/displays/panel-picodlp.c
@@ -31,7 +31,7 @@
 #include <linux/gpio.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-picodlp.h>
+#include <video/omap-panel-data.h>
 
 #include "panel-picodlp.h"
 
diff --git a/drivers/video/omap2/displays/panel-taal.c b/drivers/video/omap2/displays/panel-taal.c
index a32407a5735a..031d4069f332 100644
--- a/drivers/video/omap2/displays/panel-taal.c
+++ b/drivers/video/omap2/displays/panel-taal.c
@@ -33,7 +33,7 @@
 #include <linux/mutex.h>
 
 #include <video/omapdss.h>
-#include <video/omap-panel-nokia-dsi.h>
+#include <video/omap-panel-data.h>
 #include <video/mipi_display.h>
 
 /* DSI Virtual channel. Hardcoded for now. */
diff --git a/drivers/video/omap2/displays/panel-tfp410.c b/drivers/video/omap2/displays/panel-tfp410.c
index 8281baafe1ef..a1dba868cef1 100644
--- a/drivers/video/omap2/displays/panel-tfp410.c
+++ b/drivers/video/omap2/displays/panel-tfp410.c
@@ -24,7 +24,7 @@
 #include <linux/gpio.h>
 #include <drm/drm_edid.h>
 
-#include <video/omap-panel-tfp410.h>
+#include <video/omap-panel-data.h>
 
 static const struct omap_video_timings tfp410_default_timings = {
 	.x_res		= 640,
diff --git a/include/video/omap-panel-data.h b/include/video/omap-panel-data.h
new file mode 100644
index 000000000000..40a06b5b82de
--- /dev/null
+++ b/include/video/omap-panel-data.h
@@ -0,0 +1,103 @@
+/*
+ * Header containing platform_data structs for omap panels
+ *
+ * Copyright (C) 2013 Texas Instruments
+ * Author: Tomi Valkeinen <tomi.valkeinen@ti.com>
+ *	   Archit Taneja <archit@ti.com>
+ *
+ * Copyright (C) 2011 Texas Instruments
+ * Author: Mayuresh Janorkar <mayur@ti.com>
+ *
+ * Copyright (C) 2010 Canonical Ltd.
+ * Author: Bryan Wu <bryan.wu@canonical.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __OMAP_PANEL_DATA_H
+#define __OMAP_PANEL_DATA_H
+
+struct omap_dss_device;
+
+/**
+ * struct panel_generic_dpi_data - panel driver configuration data
+ * @name: panel name
+ * @platform_enable: platform specific panel enable function
+ * @platform_disable: platform specific panel disable function
+ */
+struct panel_generic_dpi_data {
+	const char *name;
+	int (*platform_enable)(struct omap_dss_device *dssdev);
+	void (*platform_disable)(struct omap_dss_device *dssdev);
+};
+
+/**
+ * struct panel_n8x0_data - N800 panel driver configuration data
+ */
+struct panel_n8x0_data {
+	int (*platform_enable)(struct omap_dss_device *dssdev);
+	void (*platform_disable)(struct omap_dss_device *dssdev);
+	int panel_reset;
+	int ctrl_pwrdown;
+
+	int (*set_backlight)(struct omap_dss_device *dssdev, int level);
+};
+
+/**
+ * struct nokia_dsi_panel_data - Nokia DSI panel driver configuration data
+ * @name: panel name
+ * @use_ext_te: use external TE
+ * @ext_te_gpio: external TE GPIO
+ * @esd_interval: interval of ESD checks, 0 = disabled (ms)
+ * @ulps_timeout: time to wait before entering ULPS, 0 = disabled (ms)
+ * @use_dsi_backlight: true if panel uses DSI command to control backlight
+ * @pin_config: DSI pin configuration
+ */
+
+struct nokia_dsi_panel_data {
+	const char *name;
+
+	int reset_gpio;
+
+	bool use_ext_te;
+	int ext_te_gpio;
+
+	unsigned esd_interval;
+	unsigned ulps_timeout;
+
+	bool use_dsi_backlight;
+
+	struct omap_dsi_pin_config pin_config;
+};
+
+/**
+ * struct picodlp_panel_data - picodlp panel driver configuration data
+ * @picodlp_adapter_id:	i2c_adapter number for picodlp
+ */
+struct picodlp_panel_data {
+	int picodlp_adapter_id;
+	int emu_done_gpio;
+	int pwrgood_gpio;
+};
+
+/**
+ * struct tfp410_platform_data - tfp410 panel driver configuration data
+ * @i2c_bus_num: i2c bus id for the panel
+ * @power_down_gpio: gpio number for PD pin (or -1 if not available)
+ */
+struct tfp410_platform_data {
+	int i2c_bus_num;
+	int power_down_gpio;
+};
+
+#endif /* __OMAP_PANEL_DATA_H */
diff --git a/include/video/omap-panel-generic-dpi.h b/include/video/omap-panel-generic-dpi.h
deleted file mode 100644
index 127e3f20328e..000000000000
--- a/include/video/omap-panel-generic-dpi.h
+++ /dev/null
@@ -1,37 +0,0 @@
-/*
- * Header for generic DPI panel driver
- *
- * Copyright (C) 2010 Canonical Ltd.
- * Author: Bryan Wu <bryan.wu@canonical.com>
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 as published by
- * the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program.  If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef __OMAP_PANEL_GENERIC_DPI_H
-#define __OMAP_PANEL_GENERIC_DPI_H
-
-struct omap_dss_device;
-
-/**
- * struct panel_generic_dpi_data - panel driver configuration data
- * @name: panel name
- * @platform_enable: platform specific panel enable function
- * @platform_disable: platform specific panel disable function
- */
-struct panel_generic_dpi_data {
-	const char *name;
-	int (*platform_enable)(struct omap_dss_device *dssdev);
-	void (*platform_disable)(struct omap_dss_device *dssdev);
-};
-
-#endif /* __OMAP_PANEL_GENERIC_DPI_H */
diff --git a/include/video/omap-panel-n8x0.h b/include/video/omap-panel-n8x0.h
deleted file mode 100644
index 50a1302e2cfd..000000000000
--- a/include/video/omap-panel-n8x0.h
+++ /dev/null
@@ -1,15 +0,0 @@
-#ifndef __OMAP_PANEL_N8X0_H
-#define __OMAP_PANEL_N8X0_H
-
-struct omap_dss_device;
-
-struct panel_n8x0_data {
-	int (*platform_enable)(struct omap_dss_device *dssdev);
-	void (*platform_disable)(struct omap_dss_device *dssdev);
-	int panel_reset;
-	int ctrl_pwrdown;
-
-	int (*set_backlight)(struct omap_dss_device *dssdev, int level);
-};
-
-#endif
diff --git a/include/video/omap-panel-nokia-dsi.h b/include/video/omap-panel-nokia-dsi.h
deleted file mode 100644
index 04219a295539..000000000000
--- a/include/video/omap-panel-nokia-dsi.h
+++ /dev/null
@@ -1,32 +0,0 @@
-#ifndef __OMAP_NOKIA_DSI_PANEL_H
-#define __OMAP_NOKIA_DSI_PANEL_H
-
-struct omap_dss_device;
-
-/**
- * struct nokia_dsi_panel_data - Nokia DSI panel driver configuration
- * @name: panel name
- * @use_ext_te: use external TE
- * @ext_te_gpio: external TE GPIO
- * @esd_interval: interval of ESD checks, 0 = disabled (ms)
- * @ulps_timeout: time to wait before entering ULPS, 0 = disabled (ms)
- * @use_dsi_backlight: true if panel uses DSI command to control backlight
- * @pin_config: DSI pin configuration
- */
-struct nokia_dsi_panel_data {
-	const char *name;
-
-	int reset_gpio;
-
-	bool use_ext_te;
-	int ext_te_gpio;
-
-	unsigned esd_interval;
-	unsigned ulps_timeout;
-
-	bool use_dsi_backlight;
-
-	struct omap_dsi_pin_config pin_config;
-};
-
-#endif /* __OMAP_NOKIA_DSI_PANEL_H */
diff --git a/include/video/omap-panel-picodlp.h b/include/video/omap-panel-picodlp.h
deleted file mode 100644
index 1c342ef6f3a1..000000000000
--- a/include/video/omap-panel-picodlp.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * panel data for picodlp panel
- *
- * Copyright (C) 2011 Texas Instruments
- *
- * Author: Mayuresh Janorkar <mayur@ti.com>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- */
-#ifndef __PANEL_PICODLP_H
-#define __PANEL_PICODLP_H
-/**
- * struct : picodlp panel data
- * picodlp_adapter_id:	i2c_adapter number for picodlp
- */
-struct picodlp_panel_data {
-	int picodlp_adapter_id;
-	int emu_done_gpio;
-	int pwrgood_gpio;
-};
-#endif /* __PANEL_PICODLP_H */
diff --git a/include/video/omap-panel-tfp410.h b/include/video/omap-panel-tfp410.h
deleted file mode 100644
index aef35e48bc7e..000000000000
--- a/include/video/omap-panel-tfp410.h
+++ /dev/null
@@ -1,35 +0,0 @@
-/*
- * Header for TFP410 chip driver
- *
- * Copyright (C) 2011 Texas Instruments Inc
- * Author: Tomi Valkeinen <tomi.valkeinen@ti.com>
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 as published by
- * the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program.  If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef __OMAP_PANEL_TFP410_H
-#define __OMAP_PANEL_TFP410_H
-
-struct omap_dss_device;
-
-/**
- * struct tfp410_platform_data - panel driver configuration data
- * @i2c_bus_num: i2c bus id for the panel
- * @power_down_gpio: gpio number for PD pin (or -1 if not available)
- */
-struct tfp410_platform_data {
-	int i2c_bus_num;
-	int power_down_gpio;
-};
-
-#endif /* __OMAP_PANEL_TFP410_H */
-- 
cgit v1.2.3


From c3f80cbcac28dee4503a229bf1cc50c43425d857 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 12 Feb 2013 14:59:17 +0530
Subject: OMAPDSS: add fields to panels' platform data

Forthcoming panel patches will change the panel drivers to use platform
data to pass panel's gpios to the panel driver. This patch adds the
required fields and platform data structs to the omap-panel-data.h file,
so that the board files can be changed independently of the panel driver
changes.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Archit Taneja <archit@ti.com>
---
 include/video/omap-panel-data.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)

(limited to 'include')

diff --git a/include/video/omap-panel-data.h b/include/video/omap-panel-data.h
index 40a06b5b82de..6b55839b73fc 100644
--- a/include/video/omap-panel-data.h
+++ b/include/video/omap-panel-data.h
@@ -34,11 +34,18 @@ struct omap_dss_device;
  * @name: panel name
  * @platform_enable: platform specific panel enable function
  * @platform_disable: platform specific panel disable function
+ * @num_gpios: number of gpios connected to panel
+ * @gpios: gpio numbers on the platform
+ * @gpio_invert: configure gpio as active high or low
  */
 struct panel_generic_dpi_data {
 	const char *name;
 	int (*platform_enable)(struct omap_dss_device *dssdev);
 	void (*platform_disable)(struct omap_dss_device *dssdev);
+
+	int num_gpios;
+	int gpios[10];
+	bool gpio_invert[10];
 };
 
 /**
@@ -100,4 +107,46 @@ struct tfp410_platform_data {
 	int power_down_gpio;
 };
 
+/**
+ * sharp ls panel driver configuration data
+ * @resb_gpio: reset signal
+ * @ini_gpio: power on control
+ * @mo_gpio: selection for resolution(VGA/QVGA)
+ * @lr_gpio: selection for horizontal scanning direction
+ * @ud_gpio: selection for vertical scanning direction
+ */
+struct panel_sharp_ls037v7dw01_data {
+	int resb_gpio;
+	int ini_gpio;
+	int mo_gpio;
+	int lr_gpio;
+	int ud_gpio;
+};
+
+/**
+ * acx565akm panel driver configuration data
+ * @reset_gpio: reset signal
+ */
+struct panel_acx565akm_data {
+	int reset_gpio;
+};
+
+/**
+ * nec nl8048 panel driver configuration data
+ * @res_gpio: reset signal
+ * @qvga_gpio: selection for resolution(QVGA/WVGA)
+ */
+struct panel_nec_nl8048_data {
+	int res_gpio;
+	int qvga_gpio;
+};
+
+/**
+ * tpo td043 panel driver configuration data
+ * @nreset_gpio: reset signal
+ */
+struct panel_tpo_td043_data {
+	int nreset_gpio;
+};
+
 #endif /* __OMAP_PANEL_DATA_H */
-- 
cgit v1.2.3


From a0d269ec0bfab89f7b025bf6e1e596e9a6f9d999 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 27 Nov 2012 17:05:54 +0200
Subject: OMAPDSS: DSI: remove DSI & DISPC clk divisors from dssdev

struct omap_dss_device contains DSS clock divisors. The idea is that the
board file can pass precalculated divisors to the display driver.
However, these divsors are no longer needed, as the omapdss driver can
calculate the divisors during runtime.

This patch removes the divisors from omap_dss_device, and their uses
from the dsi driver.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Archit Taneja <archit@ti.com>
---
 drivers/video/omap2/dss/dsi.c | 47 ++++++++++++++++++++++++-------------------
 include/video/omapdss.h       | 21 -------------------
 2 files changed, 26 insertions(+), 42 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index 28d41d16b7be..9eb78455b7df 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -261,6 +261,13 @@ struct dsi_data {
 	struct clk *dss_clk;
 	struct clk *sys_clk;
 
+	struct dispc_clock_info user_dispc_cinfo;
+	struct dsi_clock_info user_dsi_cinfo;
+
+	enum omap_dss_clk_source user_dispc_fclk_src;
+	enum omap_dss_clk_source user_lcd_clk_src;
+	enum omap_dss_clk_source user_dsi_fclk_src;
+
 	struct dsi_clock_info current_cinfo;
 
 	bool vdds_dsi_enabled;
@@ -1200,7 +1207,7 @@ static int dsi_set_lp_clk_divisor(struct omap_dss_device *dssdev)
 	unsigned lp_clk_div;
 	unsigned long lp_clk;
 
-	lp_clk_div = dssdev->clocks.dsi.lp_clk_div;
+	lp_clk_div = dsi->user_dsi_cinfo.lp_clk_div;
 
 	if (lp_clk_div == 0 || lp_clk_div > dsi->lpdiv_max)
 		return -EINVAL;
@@ -3910,7 +3917,7 @@ static void dsi_config_cmd_mode_interleaving(struct omap_dss_device *dssdev)
 	struct omap_video_timings *timings = &dsi->timings;
 	int bpp = dsi_get_pixel_size(dsi->pix_fmt);
 	int ndl = dsi->num_lanes_used - 1;
-	int dsi_fclk_hsdiv = dssdev->clocks.dsi.regm_dsi + 1;
+	int dsi_fclk_hsdiv = dsi->user_dsi_cinfo.regm_dsi + 1;
 	int hsa_interleave_hs = 0, hsa_interleave_lp = 0;
 	int hfp_interleave_hs = 0, hfp_interleave_lp = 0;
 	int hbp_interleave_hs = 0, hbp_interleave_lp = 0;
@@ -4302,24 +4309,24 @@ int omapdss_dsi_set_clocks(struct omap_dss_device *dssdev,
 	dsi_fclk = cinfo.dsi_pll_hsdiv_dsi_clk;
 	lp_clk_div = DIV_ROUND_UP(dsi_fclk, lp_clk * 2);
 
-	dssdev->clocks.dsi.regn = cinfo.regn;
-	dssdev->clocks.dsi.regm = cinfo.regm;
-	dssdev->clocks.dsi.regm_dispc = cinfo.regm_dispc;
-	dssdev->clocks.dsi.regm_dsi = cinfo.regm_dsi;
+	dsi->user_dsi_cinfo.regn = cinfo.regn;
+	dsi->user_dsi_cinfo.regm = cinfo.regm;
+	dsi->user_dsi_cinfo.regm_dispc = cinfo.regm_dispc;
+	dsi->user_dsi_cinfo.regm_dsi = cinfo.regm_dsi;
 
-	dssdev->clocks.dsi.lp_clk_div = lp_clk_div;
+	dsi->user_dsi_cinfo.lp_clk_div = lp_clk_div;
 
-	dssdev->clocks.dispc.channel.lck_div = dispc_cinfo.lck_div;
-	dssdev->clocks.dispc.channel.pck_div = dispc_cinfo.pck_div;
+	dsi->user_dispc_cinfo.lck_div = dispc_cinfo.lck_div;
+	dsi->user_dispc_cinfo.pck_div = dispc_cinfo.pck_div;
 
-	dssdev->clocks.dispc.dispc_fclk_src = OMAP_DSS_CLK_SRC_FCK;
+	dsi->user_dispc_fclk_src = OMAP_DSS_CLK_SRC_FCK;
 
-	dssdev->clocks.dispc.channel.lcd_clk_src =
+	dsi->user_lcd_clk_src =
 		dsi->module_id == 0 ?
 		OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC :
 		OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC;
 
-	dssdev->clocks.dsi.dsi_fclk_src =
+	dsi->user_dsi_fclk_src =
 		dsi->module_id == 0 ?
 		OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DSI :
 		OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DSI;
@@ -4589,8 +4596,8 @@ static int dsi_configure_dispc_clocks(struct omap_dss_device *dssdev)
 
 	fck = dsi_get_pll_hsdiv_dispc_rate(dsidev);
 
-	dispc_cinfo.lck_div = dssdev->clocks.dispc.channel.lck_div;
-	dispc_cinfo.pck_div = dssdev->clocks.dispc.channel.pck_div;
+	dispc_cinfo.lck_div = dsi->user_dispc_cinfo.lck_div;
+	dispc_cinfo.pck_div = dsi->user_dispc_cinfo.pck_div;
 
 	r = dispc_calc_clock_rates(fck, &dispc_cinfo);
 	if (r) {
@@ -4679,13 +4686,12 @@ static void dsi_display_uninit_dispc(struct omap_dss_device *dssdev)
 static int dsi_configure_dsi_clocks(struct omap_dss_device *dssdev)
 {
 	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
+	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
 	struct dsi_clock_info cinfo;
 	int r;
 
-	cinfo.regn  = dssdev->clocks.dsi.regn;
-	cinfo.regm  = dssdev->clocks.dsi.regm;
-	cinfo.regm_dispc = dssdev->clocks.dsi.regm_dispc;
-	cinfo.regm_dsi = dssdev->clocks.dsi.regm_dsi;
+	cinfo = dsi->user_dsi_cinfo;
+
 	r = dsi_calc_clock_rates(dsidev, &cinfo);
 	if (r) {
 		DSSERR("Failed to calc dsi clocks\n");
@@ -4716,9 +4722,8 @@ static int dsi_display_init_dsi(struct omap_dss_device *dssdev)
 	if (r)
 		goto err1;
 
-	dss_select_dsi_clk_source(dsi->module_id, dssdev->clocks.dsi.dsi_fclk_src);
-	dss_select_lcd_clk_source(mgr->id,
-			dssdev->clocks.dispc.channel.lcd_clk_src);
+	dss_select_dsi_clk_source(dsi->module_id, dsi->user_dsi_fclk_src);
+	dss_select_lcd_clk_source(mgr->id, dsi->user_lcd_clk_src);
 
 	DSSDBG("PLL OK\n");
 
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index caefa093337d..255bcf558f89 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -591,27 +591,6 @@ struct omap_dss_device {
 	} phy;
 
 	struct {
-		struct {
-			struct {
-				u16 lck_div;
-				u16 pck_div;
-				enum omap_dss_clk_source lcd_clk_src;
-			} channel;
-
-			enum omap_dss_clk_source dispc_fclk_src;
-		} dispc;
-
-		struct {
-			/* regn is one greater than TRM's REGN value */
-			u16 regn;
-			u16 regm;
-			u16 regm_dispc;
-			u16 regm_dsi;
-
-			u16 lp_clk_div;
-			enum omap_dss_clk_source dsi_fclk_src;
-		} dsi;
-
 		struct {
 			/* regn is one greater than TRM's REGN value */
 			u16 regn;
-- 
cgit v1.2.3


From 4fdfdf062d896c0afa18aa2d977a933e38c7c96c Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 12 Feb 2013 15:15:21 +0200
Subject: OMAPDSS: HDMI: remove HDMI clk divisors from dssdev

struct omap_dss_device contains HDMI clock divisors. The idea is that the
board file can pass precalculated divisors to the display driver.
However, these divsors are no longer needed, as the omapdss driver can
calculate the divisors during runtime.

This patch removes the divisors from omap_dss_device, and their uses
from the hdmi driver.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Archit Taneja <archit@ti.com>
---
 drivers/video/omap2/dss/hdmi.c | 11 +++--------
 include/video/omapdss.h        |  8 --------
 2 files changed, 3 insertions(+), 16 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/dss/hdmi.c b/drivers/video/omap2/dss/hdmi.c
index 72923645dcce..b53de043c03f 100644
--- a/drivers/video/omap2/dss/hdmi.c
+++ b/drivers/video/omap2/dss/hdmi.c
@@ -472,17 +472,12 @@ static void hdmi_compute_pll(struct omap_dss_device *dssdev, int phy,
 	 * Input clock is predivided by N + 1
 	 * out put of which is reference clk
 	 */
-	if (dssdev->clocks.hdmi.regn == 0)
-		pi->regn = HDMI_DEFAULT_REGN;
-	else
-		pi->regn = dssdev->clocks.hdmi.regn;
+
+	pi->regn = HDMI_DEFAULT_REGN;
 
 	refclk = clkin / pi->regn;
 
-	if (dssdev->clocks.hdmi.regm2 == 0)
-		pi->regm2 = HDMI_DEFAULT_REGM2;
-	else
-		pi->regm2 = dssdev->clocks.hdmi.regm2;
+	pi->regm2 = HDMI_DEFAULT_REGM2;
 
 	/*
 	 * multiplier is pixel_clk/ref_clk
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 255bcf558f89..2cb2b0e812af 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -590,14 +590,6 @@ struct omap_dss_device {
 		} venc;
 	} phy;
 
-	struct {
-		struct {
-			/* regn is one greater than TRM's REGN value */
-			u16 regn;
-			u16 regm2;
-		} hdmi;
-	} clocks;
-
 	struct {
 		struct omap_video_timings timings;
 
-- 
cgit v1.2.3


From 7286a08fb5607dbcf1a47639609d53d76b60e957 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Mon, 18 Feb 2013 13:06:01 +0200
Subject: OMAPDSS: add output->name

Add name field to omapdss's outputs so that in the following patches
panels refer to the output by their name. The name also helps debugging.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Archit Taneja <archit@ti.com>
---
 drivers/video/omap2/dss/dpi.c  | 1 +
 drivers/video/omap2/dss/dsi.c  | 1 +
 drivers/video/omap2/dss/hdmi.c | 1 +
 drivers/video/omap2/dss/rfbi.c | 1 +
 drivers/video/omap2/dss/sdi.c  | 1 +
 drivers/video/omap2/dss/venc.c | 1 +
 include/video/omapdss.h        | 2 ++
 7 files changed, 8 insertions(+)

(limited to 'include')

diff --git a/drivers/video/omap2/dss/dpi.c b/drivers/video/omap2/dss/dpi.c
index cb6b280124f1..e393b6ab60db 100644
--- a/drivers/video/omap2/dss/dpi.c
+++ b/drivers/video/omap2/dss/dpi.c
@@ -512,6 +512,7 @@ static void __init dpi_init_output(struct platform_device *pdev)
 	out->pdev = pdev;
 	out->id = OMAP_DSS_OUTPUT_DPI;
 	out->type = OMAP_DISPLAY_TYPE_DPI;
+	out->name = "dpi.0";
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index 815c93047151..294c83299cea 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -5183,6 +5183,7 @@ static void __init dsi_init_output(struct platform_device *dsidev)
 			OMAP_DSS_OUTPUT_DSI1 : OMAP_DSS_OUTPUT_DSI2;
 
 	out->type = OMAP_DISPLAY_TYPE_DSI;
+	out->name = dsi->module_id == 0 ? "dsi.0" : "dsi.1";
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/hdmi.c b/drivers/video/omap2/dss/hdmi.c
index 04edf325429b..9cbca03eff43 100644
--- a/drivers/video/omap2/dss/hdmi.c
+++ b/drivers/video/omap2/dss/hdmi.c
@@ -1046,6 +1046,7 @@ static void __init hdmi_init_output(struct platform_device *pdev)
 	out->pdev = pdev;
 	out->id = OMAP_DSS_OUTPUT_HDMI;
 	out->type = OMAP_DISPLAY_TYPE_HDMI;
+	out->name = "hdmi.0";
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/rfbi.c b/drivers/video/omap2/dss/rfbi.c
index e903dd3f54d9..d9d7f3511a7b 100644
--- a/drivers/video/omap2/dss/rfbi.c
+++ b/drivers/video/omap2/dss/rfbi.c
@@ -1025,6 +1025,7 @@ static void __init rfbi_init_output(struct platform_device *pdev)
 	out->pdev = pdev;
 	out->id = OMAP_DSS_OUTPUT_DBI;
 	out->type = OMAP_DISPLAY_TYPE_DBI;
+	out->name = "rfbi.0";
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/sdi.c b/drivers/video/omap2/dss/sdi.c
index 62b5374ce438..c27b58ad8e4d 100644
--- a/drivers/video/omap2/dss/sdi.c
+++ b/drivers/video/omap2/dss/sdi.c
@@ -278,6 +278,7 @@ static void __init sdi_init_output(struct platform_device *pdev)
 	out->pdev = pdev;
 	out->id = OMAP_DSS_OUTPUT_SDI;
 	out->type = OMAP_DISPLAY_TYPE_SDI;
+	out->name = "sdi.0";
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/venc.c b/drivers/video/omap2/dss/venc.c
index 006caf3cb509..4c07a25d3d9b 100644
--- a/drivers/video/omap2/dss/venc.c
+++ b/drivers/video/omap2/dss/venc.c
@@ -819,6 +819,7 @@ static void __init venc_init_output(struct platform_device *pdev)
 	out->pdev = pdev;
 	out->id = OMAP_DSS_OUTPUT_VENC;
 	out->type = OMAP_DISPLAY_TYPE_VENC;
+	out->name = "venc.0";
 
 	dss_register_output(out);
 }
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 2cb2b0e812af..fa800b7ef832 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -541,6 +541,8 @@ struct omap_dss_writeback_info {
 struct omap_dss_output {
 	struct list_head list;
 
+	const char *name;
+
 	/* display type supported by the output */
 	enum omap_display_type type;
 
-- 
cgit v1.2.3


From 2eea5ae6c102a5088e39733115ff7762a4674887 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Wed, 13 Feb 2013 11:23:54 +0200
Subject: OMAPDSS: add output->dispc_channel

The DISPC channel used for each output is currently passed in panel
platform data from the board files.

To simplify this, and to make the panel drivers less dependent on OMAP,
this patch changes omapdss to resolve the channel independently. The
channel is resolved based on the OMAP version and, in case of DSI, the
DSI module id. This resolved channel is stored into a new field in
output, dispc_channel.

The few places where dssdev->channel was used are changed to use
output->recommended_channel. After this patch, dssdev->channel is
obsolete.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Archit Taneja <archit@ti.com>
---
 drivers/video/omap2/dss/dpi.c            | 38 ++++++++++++++++++++----
 drivers/video/omap2/dss/dsi.c            | 50 ++++++++++++++++++++++++++++++++
 drivers/video/omap2/dss/hdmi.c           |  3 +-
 drivers/video/omap2/dss/rfbi.c           |  1 +
 drivers/video/omap2/dss/sdi.c            |  1 +
 drivers/video/omap2/dss/venc.c           |  3 +-
 drivers/video/omap2/omapfb/omapfb-main.c |  2 +-
 include/video/omapdss.h                  |  4 +++
 8 files changed, 91 insertions(+), 11 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/dss/dpi.c b/drivers/video/omap2/dss/dpi.c
index e393b6ab60db..ab6d8b0e94c7 100644
--- a/drivers/video/omap2/dss/dpi.c
+++ b/drivers/video/omap2/dss/dpi.c
@@ -396,6 +396,36 @@ static int __init dpi_verify_dsi_pll(struct platform_device *dsidev)
 	return 0;
 }
 
+/*
+ * Return a hardcoded channel for the DPI output. This should work for
+ * current use cases, but this can be later expanded to either resolve
+ * the channel in some more dynamic manner, or get the channel as a user
+ * parameter.
+ */
+static enum omap_channel dpi_get_channel(void)
+{
+	switch (omapdss_get_version()) {
+	case OMAPDSS_VER_OMAP24xx:
+	case OMAPDSS_VER_OMAP34xx_ES1:
+	case OMAPDSS_VER_OMAP34xx_ES3:
+	case OMAPDSS_VER_OMAP3630:
+	case OMAPDSS_VER_AM35xx:
+		return OMAP_DSS_CHANNEL_LCD;
+
+	case OMAPDSS_VER_OMAP4430_ES1:
+	case OMAPDSS_VER_OMAP4430_ES2:
+	case OMAPDSS_VER_OMAP4:
+		return OMAP_DSS_CHANNEL_LCD2;
+
+	case OMAPDSS_VER_OMAP5:
+		return OMAP_DSS_CHANNEL_LCD3;
+
+	default:
+		DSSWARN("unsupported DSS version\n");
+		return OMAP_DSS_CHANNEL_LCD;
+	}
+}
+
 static int __init dpi_init_display(struct omap_dss_device *dssdev)
 {
 	struct platform_device *dsidev;
@@ -416,12 +446,7 @@ static int __init dpi_init_display(struct omap_dss_device *dssdev)
 		dpi.vdds_dsi_reg = vdds_dsi;
 	}
 
-	/*
-	 * XXX We shouldn't need dssdev->channel for this. The dsi pll clock
-	 * source for DPI is SoC integration detail, not something that should
-	 * be configured in the dssdev
-	 */
-	dsidev = dpi_get_dsidev(dssdev->channel);
+	dsidev = dpi_get_dsidev(dpi.output.dispc_channel);
 
 	if (dsidev && dpi_verify_dsi_pll(dsidev)) {
 		dsidev = NULL;
@@ -513,6 +538,7 @@ static void __init dpi_init_output(struct platform_device *pdev)
 	out->id = OMAP_DSS_OUTPUT_DPI;
 	out->type = OMAP_DISPLAY_TYPE_DPI;
 	out->name = "dpi.0";
+	out->dispc_channel = dpi_get_channel();
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index 294c83299cea..d13e77c01d7e 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -4946,6 +4946,55 @@ void omapdss_dsi_set_videomode_timings(struct omap_dss_device *dssdev,
 }
 EXPORT_SYMBOL(omapdss_dsi_set_videomode_timings);
 
+/*
+ * Return a hardcoded channel for the DSI output. This should work for
+ * current use cases, but this can be later expanded to either resolve
+ * the channel in some more dynamic manner, or get the channel as a user
+ * parameter.
+ */
+static enum omap_channel dsi_get_channel(int module_id)
+{
+	switch (omapdss_get_version()) {
+	case OMAPDSS_VER_OMAP24xx:
+		DSSWARN("DSI not supported\n");
+		return OMAP_DSS_CHANNEL_LCD;
+
+	case OMAPDSS_VER_OMAP34xx_ES1:
+	case OMAPDSS_VER_OMAP34xx_ES3:
+	case OMAPDSS_VER_OMAP3630:
+	case OMAPDSS_VER_AM35xx:
+		return OMAP_DSS_CHANNEL_LCD;
+
+	case OMAPDSS_VER_OMAP4430_ES1:
+	case OMAPDSS_VER_OMAP4430_ES2:
+	case OMAPDSS_VER_OMAP4:
+		switch (module_id) {
+		case 0:
+			return OMAP_DSS_CHANNEL_LCD;
+		case 1:
+			return OMAP_DSS_CHANNEL_LCD2;
+		default:
+			DSSWARN("unsupported module id\n");
+			return OMAP_DSS_CHANNEL_LCD;
+		}
+
+	case OMAPDSS_VER_OMAP5:
+		switch (module_id) {
+		case 0:
+			return OMAP_DSS_CHANNEL_LCD;
+		case 1:
+			return OMAP_DSS_CHANNEL_LCD3;
+		default:
+			DSSWARN("unsupported module id\n");
+			return OMAP_DSS_CHANNEL_LCD;
+		}
+
+	default:
+		DSSWARN("unsupported DSS version\n");
+		return OMAP_DSS_CHANNEL_LCD;
+	}
+}
+
 static int __init dsi_init_display(struct omap_dss_device *dssdev)
 {
 	struct platform_device *dsidev =
@@ -5184,6 +5233,7 @@ static void __init dsi_init_output(struct platform_device *dsidev)
 
 	out->type = OMAP_DISPLAY_TYPE_DSI;
 	out->name = dsi->module_id == 0 ? "dsi.0" : "dsi.1";
+	out->dispc_channel = dsi_get_channel(dsi->module_id);
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/hdmi.c b/drivers/video/omap2/dss/hdmi.c
index 9cbca03eff43..a6f953898d94 100644
--- a/drivers/video/omap2/dss/hdmi.c
+++ b/drivers/video/omap2/dss/hdmi.c
@@ -1012,8 +1012,6 @@ static void __init hdmi_probe_pdata(struct platform_device *pdev)
 	hdmi.ls_oe_gpio = priv->ls_oe_gpio;
 	hdmi.hpd_gpio = priv->hpd_gpio;
 
-	dssdev->channel = OMAP_DSS_CHANNEL_DIGIT;
-
 	r = hdmi_init_display(dssdev);
 	if (r) {
 		DSSERR("device %s init failed: %d\n", dssdev->name, r);
@@ -1047,6 +1045,7 @@ static void __init hdmi_init_output(struct platform_device *pdev)
 	out->id = OMAP_DSS_OUTPUT_HDMI;
 	out->type = OMAP_DISPLAY_TYPE_HDMI;
 	out->name = "hdmi.0";
+	out->dispc_channel = OMAP_DSS_CHANNEL_DIGIT;
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/rfbi.c b/drivers/video/omap2/dss/rfbi.c
index d9d7f3511a7b..1a691bb27547 100644
--- a/drivers/video/omap2/dss/rfbi.c
+++ b/drivers/video/omap2/dss/rfbi.c
@@ -1026,6 +1026,7 @@ static void __init rfbi_init_output(struct platform_device *pdev)
 	out->id = OMAP_DSS_OUTPUT_DBI;
 	out->type = OMAP_DISPLAY_TYPE_DBI;
 	out->name = "rfbi.0";
+	out->dispc_channel = OMAP_DSS_CHANNEL_LCD;
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/sdi.c b/drivers/video/omap2/dss/sdi.c
index c27b58ad8e4d..3888cfa0881c 100644
--- a/drivers/video/omap2/dss/sdi.c
+++ b/drivers/video/omap2/dss/sdi.c
@@ -279,6 +279,7 @@ static void __init sdi_init_output(struct platform_device *pdev)
 	out->id = OMAP_DSS_OUTPUT_SDI;
 	out->type = OMAP_DISPLAY_TYPE_SDI;
 	out->name = "sdi.0";
+	out->dispc_channel = OMAP_DSS_CHANNEL_LCD;
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/dss/venc.c b/drivers/video/omap2/dss/venc.c
index 4c07a25d3d9b..5cb983e2f170 100644
--- a/drivers/video/omap2/dss/venc.c
+++ b/drivers/video/omap2/dss/venc.c
@@ -786,8 +786,6 @@ static void __init venc_probe_pdata(struct platform_device *vencdev)
 
 	dss_copy_device_pdata(dssdev, plat_dssdev);
 
-	dssdev->channel = OMAP_DSS_CHANNEL_DIGIT;
-
 	r = venc_init_display(dssdev);
 	if (r) {
 		DSSERR("device %s init failed: %d\n", dssdev->name, r);
@@ -820,6 +818,7 @@ static void __init venc_init_output(struct platform_device *pdev)
 	out->id = OMAP_DSS_OUTPUT_VENC;
 	out->type = OMAP_DISPLAY_TYPE_VENC;
 	out->name = "venc.0";
+	out->dispc_channel = OMAP_DSS_CHANNEL_DIGIT;
 
 	dss_register_output(out);
 }
diff --git a/drivers/video/omap2/omapfb/omapfb-main.c b/drivers/video/omap2/omapfb/omapfb-main.c
index ca585ef37f25..f38348ea3375 100644
--- a/drivers/video/omap2/omapfb/omapfb-main.c
+++ b/drivers/video/omap2/omapfb/omapfb-main.c
@@ -2388,7 +2388,7 @@ static int omapfb_init_connections(struct omapfb2_device *fbdev,
 		struct omap_dss_device *dssdev = fbdev->displays[i].dssdev;
 		struct omap_dss_output *out = dssdev->output;
 
-		mgr = omap_dss_get_overlay_manager(dssdev->channel);
+		mgr = omap_dss_get_overlay_manager(out->dispc_channel);
 
 		if (!mgr || !out)
 			continue;
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index fa800b7ef832..41fb434093e9 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -546,6 +546,9 @@ struct omap_dss_output {
 	/* display type supported by the output */
 	enum omap_display_type type;
 
+	/* DISPC channel for this output */
+	enum omap_channel dispc_channel;
+
 	/* output instance */
 	enum omap_dss_output_id id;
 
@@ -563,6 +566,7 @@ struct omap_dss_device {
 
 	enum omap_display_type type;
 
+	/* obsolete, to be removed */
 	enum omap_channel channel;
 
 	union {
-- 
cgit v1.2.3


From 777f05cc79892c1a5c25a72cfb3cc24554378a45 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Wed, 6 Mar 2013 11:10:29 +0200
Subject: OMAPDSS: DSI: simplify dsi configuration

We have a bunch of dsi functions that are used to do the basic
configuration for DSI. To simplify things, and to make sure we have all
the necessary information, create a single dsi config function, which
does the basic configuration.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
---
 drivers/video/omap2/displays/panel-taal.c | 16 ++++---
 drivers/video/omap2/dss/dsi.c             | 74 +++++--------------------------
 include/video/omapdss.h                   | 23 +++++-----
 3 files changed, 31 insertions(+), 82 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/displays/panel-taal.c b/drivers/video/omap2/displays/panel-taal.c
index bc4c95ee41cc..eb86cbae8c2a 100644
--- a/drivers/video/omap2/displays/panel-taal.c
+++ b/drivers/video/omap2/displays/panel-taal.c
@@ -919,6 +919,13 @@ static int taal_power_on(struct omap_dss_device *dssdev)
 	struct taal_data *td = dev_get_drvdata(&dssdev->dev);
 	u8 id1, id2, id3;
 	int r;
+	struct omap_dss_dsi_config dsi_config = {
+		.mode = OMAP_DSS_DSI_CMD_MODE,
+		.pixel_format = OMAP_DSS_DSI_FMT_RGB888,
+		.timings = &dssdev->panel.timings,
+		.hs_clk = 216000000,
+		.lp_clk = 10000000,
+	};
 
 	r = omapdss_dsi_configure_pins(dssdev, &td->pin_config);
 	if (r) {
@@ -926,14 +933,9 @@ static int taal_power_on(struct omap_dss_device *dssdev)
 		goto err0;
 	};
 
-	omapdss_dsi_set_size(dssdev, dssdev->panel.timings.x_res,
-		dssdev->panel.timings.y_res);
-	omapdss_dsi_set_pixel_format(dssdev, OMAP_DSS_DSI_FMT_RGB888);
-	omapdss_dsi_set_operation_mode(dssdev, OMAP_DSS_DSI_CMD_MODE);
-
-	r = omapdss_dsi_set_clocks(dssdev, 216000000, 10000000);
+	r = omapdss_dsi_set_config(dssdev, &dsi_config);
 	if (r) {
-		dev_err(&dssdev->dev, "failed to set HS and LP clocks\n");
+		dev_err(&dssdev->dev, "failed to configure DSI\n");
 		goto err0;
 	}
 
diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index dd13bd9138fc..8cb43af60fe1 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -4278,7 +4278,7 @@ int omapdss_dsi_configure_pins(struct omap_dss_device *dssdev,
 }
 EXPORT_SYMBOL(omapdss_dsi_configure_pins);
 
-int omapdss_dsi_set_clocks(struct omap_dss_device *dssdev,
+static int dsi_set_clocks(struct omap_dss_device *dssdev,
 		unsigned long ddr_clk, unsigned long lp_clk)
 {
 	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
@@ -4293,8 +4293,6 @@ int omapdss_dsi_set_clocks(struct omap_dss_device *dssdev,
 
 	DSSDBG("Setting DSI clocks: ddr_clk %lu, lp_clk %lu", ddr_clk, lp_clk);
 
-	mutex_lock(&dsi->lock);
-
 	/* Calculate PLL output clock */
 	r = dsi_pll_calc_ddrfreq(dsidev, ddr_clk * 4, &cinfo);
 	if (r)
@@ -4336,13 +4334,10 @@ int omapdss_dsi_set_clocks(struct omap_dss_device *dssdev,
 		OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DSI :
 		OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DSI;
 
-	mutex_unlock(&dsi->lock);
 	return 0;
 err:
-	mutex_unlock(&dsi->lock);
 	return r;
 }
-EXPORT_SYMBOL(omapdss_dsi_set_clocks);
 
 int dsi_enable_video_output(struct omap_dss_device *dssdev, int channel)
 {
@@ -4884,75 +4879,26 @@ int omapdss_dsi_enable_te(struct omap_dss_device *dssdev, bool enable)
 }
 EXPORT_SYMBOL(omapdss_dsi_enable_te);
 
-void omapdss_dsi_set_timings(struct omap_dss_device *dssdev,
-		struct omap_video_timings *timings)
+int omapdss_dsi_set_config(struct omap_dss_device *dssdev,
+		const struct omap_dss_dsi_config *config)
 {
 	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
 	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
 
 	mutex_lock(&dsi->lock);
 
-	dsi->timings = *timings;
-
-	mutex_unlock(&dsi->lock);
-}
-EXPORT_SYMBOL(omapdss_dsi_set_timings);
-
-void omapdss_dsi_set_size(struct omap_dss_device *dssdev, u16 w, u16 h)
-{
-	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
-	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
-
-	mutex_lock(&dsi->lock);
+	dsi->timings = *config->timings;
+	dsi->vm_timings = *config->vm_timings;
+	dsi->pix_fmt = config->pixel_format;
+	dsi->mode = config->mode;
 
-	dsi->timings.x_res = w;
-	dsi->timings.y_res = h;
+	dsi_set_clocks(dssdev, config->hs_clk, config->lp_clk);
 
 	mutex_unlock(&dsi->lock);
-}
-EXPORT_SYMBOL(omapdss_dsi_set_size);
 
-void omapdss_dsi_set_pixel_format(struct omap_dss_device *dssdev,
-		enum omap_dss_dsi_pixel_format fmt)
-{
-	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
-	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
-
-	mutex_lock(&dsi->lock);
-
-	dsi->pix_fmt = fmt;
-
-	mutex_unlock(&dsi->lock);
-}
-EXPORT_SYMBOL(omapdss_dsi_set_pixel_format);
-
-void omapdss_dsi_set_operation_mode(struct omap_dss_device *dssdev,
-		enum omap_dss_dsi_mode mode)
-{
-	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
-	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
-
-	mutex_lock(&dsi->lock);
-
-	dsi->mode = mode;
-
-	mutex_unlock(&dsi->lock);
-}
-EXPORT_SYMBOL(omapdss_dsi_set_operation_mode);
-
-void omapdss_dsi_set_videomode_timings(struct omap_dss_device *dssdev,
-		struct omap_dss_dsi_videomode_timings *timings)
-{
-	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
-	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
-
-	mutex_lock(&dsi->lock);
-
-	dsi->vm_timings = *timings;
-
-	mutex_unlock(&dsi->lock);
+	return 0;
 }
-EXPORT_SYMBOL(omapdss_dsi_set_videomode_timings);
+EXPORT_SYMBOL(omapdss_dsi_set_config);
 
 /*
  * Return a hardcoded channel for the DSI output. This should work for
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 41fb434093e9..4dd4cea41bd4 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -282,6 +282,16 @@ struct omap_dss_dsi_videomode_timings {
 	int window_sync;
 };
 
+struct omap_dss_dsi_config {
+	enum omap_dss_dsi_mode mode;
+	enum omap_dss_dsi_pixel_format pixel_format;
+	const struct omap_video_timings *timings;
+	const struct omap_dss_dsi_videomode_timings *vm_timings;
+
+	unsigned long hs_clk;
+	unsigned long lp_clk;
+};
+
 void dsi_bus_lock(struct omap_dss_device *dssdev);
 void dsi_bus_unlock(struct omap_dss_device *dssdev);
 int dsi_vc_dcs_write(struct omap_dss_device *dssdev, int channel, u8 *data,
@@ -806,15 +816,8 @@ int dispc_ovl_setup(enum omap_plane plane, const struct omap_overlay_info *oi,
 void omapdss_dsi_vc_enable_hs(struct omap_dss_device *dssdev, int channel,
 		bool enable);
 int omapdss_dsi_enable_te(struct omap_dss_device *dssdev, bool enable);
-void omapdss_dsi_set_timings(struct omap_dss_device *dssdev,
-		struct omap_video_timings *timings);
-void omapdss_dsi_set_size(struct omap_dss_device *dssdev, u16 w, u16 h);
-void omapdss_dsi_set_pixel_format(struct omap_dss_device *dssdev,
-		enum omap_dss_dsi_pixel_format fmt);
-void omapdss_dsi_set_operation_mode(struct omap_dss_device *dssdev,
-		enum omap_dss_dsi_mode mode);
-void omapdss_dsi_set_videomode_timings(struct omap_dss_device *dssdev,
-		struct omap_dss_dsi_videomode_timings *timings);
+int omapdss_dsi_set_config(struct omap_dss_device *dssdev,
+		const struct omap_dss_dsi_config *config);
 
 int omap_dsi_update(struct omap_dss_device *dssdev, int channel,
 		void (*callback)(int, void *), void *data);
@@ -823,8 +826,6 @@ int omap_dsi_set_vc_id(struct omap_dss_device *dssdev, int channel, int vc_id);
 void omap_dsi_release_vc(struct omap_dss_device *dssdev, int channel);
 int omapdss_dsi_configure_pins(struct omap_dss_device *dssdev,
 		const struct omap_dsi_pin_config *pin_cfg);
-int omapdss_dsi_set_clocks(struct omap_dss_device *dssdev,
-		unsigned long ddr_clk, unsigned long lp_clk);
 
 int omapdss_dsi_display_enable(struct omap_dss_device *dssdev);
 void omapdss_dsi_display_disable(struct omap_dss_device *dssdev,
-- 
cgit v1.2.3


From 478d7df8af89f449bacc4e67ff35dc630400c0ca Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 5 Mar 2013 16:29:36 +0200
Subject: OMAPDSS: DSI: add enum omap_dss_dsi_trans_mode

Instead of managing DSI sync ends with booleans, add an enum for the DSI
transfer mode. This is much cleaner way to handle the DSI syncs.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
---
 drivers/video/omap2/dss/dsi.c | 15 ++++++++++-----
 include/video/omapdss.h       | 13 ++++++++++---
 2 files changed, 20 insertions(+), 8 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index 590da06e6a82..e8f5bdb2796c 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -3813,18 +3813,22 @@ static void dsi_config_vp_num_line_buffers(struct platform_device *dsidev)
 static void dsi_config_vp_sync_events(struct platform_device *dsidev)
 {
 	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
-	bool vsync_end = dsi->vm_timings.vp_vsync_end;
-	bool hsync_end = dsi->vm_timings.vp_hsync_end;
+	bool sync_end;
 	u32 r;
 
+	if (dsi->vm_timings.trans_mode == OMAP_DSS_DSI_PULSE_MODE)
+		sync_end = true;
+	else
+		sync_end = false;
+
 	r = dsi_read_reg(dsidev, DSI_CTRL);
 	r = FLD_MOD(r, 1, 9, 9);		/* VP_DE_POL */
 	r = FLD_MOD(r, 1, 10, 10);		/* VP_HSYNC_POL */
 	r = FLD_MOD(r, 1, 11, 11);		/* VP_VSYNC_POL */
 	r = FLD_MOD(r, 1, 15, 15);		/* VP_VSYNC_START */
-	r = FLD_MOD(r, vsync_end, 16, 16);	/* VP_VSYNC_END */
+	r = FLD_MOD(r, sync_end, 16, 16);	/* VP_VSYNC_END */
 	r = FLD_MOD(r, 1, 17, 17);		/* VP_HSYNC_START */
-	r = FLD_MOD(r, hsync_end, 18, 18);	/* VP_HSYNC_END */
+	r = FLD_MOD(r, sync_end, 18, 18);	/* VP_HSYNC_END */
 	dsi_write_reg(dsidev, DSI_CTRL, r);
 }
 
@@ -4171,11 +4175,12 @@ static void dsi_proto_timings(struct platform_device *dsidev)
 		int vfp = dsi->vm_timings.vfp;
 		int vbp = dsi->vm_timings.vbp;
 		int window_sync = dsi->vm_timings.window_sync;
-		bool hsync_end = dsi->vm_timings.vp_hsync_end;
+		bool hsync_end;
 		struct omap_video_timings *timings = &dsi->timings;
 		int bpp = dsi_get_pixel_size(dsi->pix_fmt);
 		int tl, t_he, width_bytes;
 
+		hsync_end = dsi->vm_timings.trans_mode == OMAP_DSS_DSI_PULSE_MODE;
 		t_he = hsync_end ?
 			((hsa == 0 && ndl == 3) ? 1 : DIV_ROUND_UP(4, ndl)) : 0;
 
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 4dd4cea41bd4..7f774c5f8b6b 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -257,6 +257,15 @@ void rfbi_bus_unlock(void);
 
 /* DSI */
 
+enum omap_dss_dsi_trans_mode {
+	/* Sync Pulses: both sync start and end packets sent */
+	OMAP_DSS_DSI_PULSE_MODE,
+	/* Sync Events: only sync start packets sent */
+	OMAP_DSS_DSI_EVENT_MODE,
+	/* Burst: only sync start packets sent, pixels are time compressed */
+	OMAP_DSS_DSI_BURST_MODE,
+};
+
 struct omap_dss_dsi_videomode_timings {
 	/* DSI video mode blanking data */
 	/* Unit: byte clock cycles */
@@ -274,9 +283,7 @@ struct omap_dss_dsi_videomode_timings {
 	int hbp_blanking_mode;
 	int hfp_blanking_mode;
 
-	/* Video port sync events */
-	bool vp_vsync_end;
-	bool vp_hsync_end;
+	enum omap_dss_dsi_trans_mode trans_mode;
 
 	bool ddr_clk_always_on;
 	int window_sync;
-- 
cgit v1.2.3


From f1e0001f973cfbec62d702cbd3d56f6b703cc90e Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 5 Mar 2013 17:21:35 +0200
Subject: OMAPDSS: DSI: use new clock calculation code

Use the new clock calculation code in the DSI driver.

The new code does not need DSI video mode parameters from the panel
driver, like the old code does. Instead the new code is given the normal
video timings, and a few DSI parameters, which are used to create DSI
video timings.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
---
 drivers/video/omap2/displays/panel-taal.c |   7 +-
 drivers/video/omap2/dss/dsi.c             | 557 +++++++++++++++++++++++++++++-
 include/video/omapdss.h                   |  20 +-
 3 files changed, 566 insertions(+), 18 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/displays/panel-taal.c b/drivers/video/omap2/displays/panel-taal.c
index eb86cbae8c2a..2fc923de23cc 100644
--- a/drivers/video/omap2/displays/panel-taal.c
+++ b/drivers/video/omap2/displays/panel-taal.c
@@ -787,6 +787,7 @@ static int taal_probe(struct omap_dss_device *dssdev)
 
 	dssdev->panel.timings.x_res = 864;
 	dssdev->panel.timings.y_res = 480;
+	dssdev->panel.timings.pixel_clock = DIV_ROUND_UP(864 * 480 * 60, 1000);
 	dssdev->panel.dsi_pix_fmt = OMAP_DSS_DSI_FMT_RGB888;
 	dssdev->caps = OMAP_DSS_DISPLAY_CAP_MANUAL_UPDATE |
 		OMAP_DSS_DISPLAY_CAP_TEAR_ELIM;
@@ -923,8 +924,10 @@ static int taal_power_on(struct omap_dss_device *dssdev)
 		.mode = OMAP_DSS_DSI_CMD_MODE,
 		.pixel_format = OMAP_DSS_DSI_FMT_RGB888,
 		.timings = &dssdev->panel.timings,
-		.hs_clk = 216000000,
-		.lp_clk = 10000000,
+		.hs_clk_min = 150000000,
+		.hs_clk_max = 300000000,
+		.lp_clk_min = 7000000,
+		.lp_clk_max = 10000000,
 	};
 
 	r = omapdss_dsi_configure_pins(dssdev, &td->pin_config);
diff --git a/drivers/video/omap2/dss/dsi.c b/drivers/video/omap2/dss/dsi.c
index a4e0560c2799..de88e69ef0a2 100644
--- a/drivers/video/omap2/dss/dsi.c
+++ b/drivers/video/omap2/dss/dsi.c
@@ -255,6 +255,24 @@ struct dsi_isr_tables {
 	struct dsi_isr_data isr_table_cio[DSI_MAX_NR_ISRS];
 };
 
+struct dsi_clk_calc_ctx {
+	struct platform_device *dsidev;
+
+	/* inputs */
+
+	const struct omap_dss_dsi_config *config;
+
+	unsigned long req_pck_min, req_pck_nom, req_pck_max;
+
+	/* outputs */
+
+	struct dsi_clock_info dsi_cinfo;
+	struct dispc_clock_info dispc_cinfo;
+
+	struct omap_video_timings dispc_vm;
+	struct omap_dss_dsi_videomode_timings dsi_vm;
+};
+
 struct dsi_data {
 	struct platform_device *pdev;
 	void __iomem	*base;
@@ -1201,6 +1219,25 @@ static unsigned long dsi_fclk_rate(struct platform_device *dsidev)
 	return r;
 }
 
+static int dsi_lp_clock_calc(struct dsi_clock_info *cinfo,
+		unsigned long lp_clk_min, unsigned long lp_clk_max)
+{
+	unsigned long dsi_fclk = cinfo->dsi_pll_hsdiv_dsi_clk;
+	unsigned lp_clk_div;
+	unsigned long lp_clk;
+
+	lp_clk_div = DIV_ROUND_UP(dsi_fclk, lp_clk_max * 2);
+	lp_clk = dsi_fclk / 2 / lp_clk_div;
+
+	if (lp_clk < lp_clk_min || lp_clk > lp_clk_max)
+		return -EINVAL;
+
+	cinfo->lp_clk_div = lp_clk_div;
+	cinfo->lp_clk = lp_clk;
+
+	return 0;
+}
+
 static int dsi_set_lp_clk_divisor(struct platform_device *dsidev)
 {
 	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
@@ -1577,8 +1614,7 @@ found:
 	return 0;
 }
 
-static void dsi_pll_calc_dsi_fck(struct platform_device *dsidev,
-		struct dsi_clock_info *cinfo)
+static void dsi_pll_calc_dsi_fck(struct dsi_clock_info *cinfo)
 {
 	unsigned long max_dsi_fck;
 
@@ -4369,7 +4405,7 @@ static int dsi_set_clocks(struct omap_dss_device *dssdev,
 		goto err;
 
 	/* Calculate PLL's DSI clock */
-	dsi_pll_calc_dsi_fck(dsidev, &cinfo);
+	dsi_pll_calc_dsi_fck(&cinfo);
 
 	/* Calculate PLL's DISPC clock and pck & lck divs */
 	pck = cinfo.clkin4ddr / 16 * (dsi->num_lanes_used - 1) * 8 / bpp;
@@ -4693,13 +4729,6 @@ static int dsi_display_init_dispc(struct platform_device *dsidev,
 			OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC);
 
 	if (dsi->mode == OMAP_DSS_DSI_CMD_MODE) {
-		dsi->timings.hsw = 1;
-		dsi->timings.hfp = 1;
-		dsi->timings.hbp = 1;
-		dsi->timings.vsw = 1;
-		dsi->timings.vfp = 0;
-		dsi->timings.vbp = 0;
-
 		r = dss_mgr_register_framedone_handler(mgr,
 				dsi_framedone_irq_callback, dsidev);
 		if (r) {
@@ -4941,24 +4970,526 @@ int omapdss_dsi_enable_te(struct omap_dss_device *dssdev, bool enable)
 }
 EXPORT_SYMBOL(omapdss_dsi_enable_te);
 
+#ifdef PRINT_VERBOSE_VM_TIMINGS
+static void print_dsi_vm(const char *str,
+		const struct omap_dss_dsi_videomode_timings *t)
+{
+	unsigned long byteclk = t->hsclk / 4;
+	int bl, wc, pps, tot;
+
+	wc = DIV_ROUND_UP(t->hact * t->bitspp, 8);
+	pps = DIV_ROUND_UP(wc + 6, t->ndl); /* pixel packet size */
+	bl = t->hss + t->hsa + t->hse + t->hbp + t->hfp;
+	tot = bl + pps;
+
+#define TO_DSI_T(x) ((u32)div64_u64((u64)x * 1000000000llu, byteclk))
+
+	pr_debug("%s bck %lu, %u/%u/%u/%u/%u/%u = %u+%u = %u, "
+			"%u/%u/%u/%u/%u/%u = %u + %u = %u\n",
+			str,
+			byteclk,
+			t->hss, t->hsa, t->hse, t->hbp, pps, t->hfp,
+			bl, pps, tot,
+			TO_DSI_T(t->hss),
+			TO_DSI_T(t->hsa),
+			TO_DSI_T(t->hse),
+			TO_DSI_T(t->hbp),
+			TO_DSI_T(pps),
+			TO_DSI_T(t->hfp),
+
+			TO_DSI_T(bl),
+			TO_DSI_T(pps),
+
+			TO_DSI_T(tot));
+#undef TO_DSI_T
+}
+
+static void print_dispc_vm(const char *str, const struct omap_video_timings *t)
+{
+	unsigned long pck = t->pixel_clock * 1000;
+	int hact, bl, tot;
+
+	hact = t->x_res;
+	bl = t->hsw + t->hbp + t->hfp;
+	tot = hact + bl;
+
+#define TO_DISPC_T(x) ((u32)div64_u64((u64)x * 1000000000llu, pck))
+
+	pr_debug("%s pck %lu, %u/%u/%u/%u = %u+%u = %u, "
+			"%u/%u/%u/%u = %u + %u = %u\n",
+			str,
+			pck,
+			t->hsw, t->hbp, hact, t->hfp,
+			bl, hact, tot,
+			TO_DISPC_T(t->hsw),
+			TO_DISPC_T(t->hbp),
+			TO_DISPC_T(hact),
+			TO_DISPC_T(t->hfp),
+			TO_DISPC_T(bl),
+			TO_DISPC_T(hact),
+			TO_DISPC_T(tot));
+#undef TO_DISPC_T
+}
+
+/* note: this is not quite accurate */
+static void print_dsi_dispc_vm(const char *str,
+		const struct omap_dss_dsi_videomode_timings *t)
+{
+	struct omap_video_timings vm = { 0 };
+	unsigned long byteclk = t->hsclk / 4;
+	unsigned long pck;
+	u64 dsi_tput;
+	int dsi_hact, dsi_htot;
+
+	dsi_tput = (u64)byteclk * t->ndl * 8;
+	pck = (u32)div64_u64(dsi_tput, t->bitspp);
+	dsi_hact = DIV_ROUND_UP(DIV_ROUND_UP(t->hact * t->bitspp, 8) + 6, t->ndl);
+	dsi_htot = t->hss + t->hsa + t->hse + t->hbp + dsi_hact + t->hfp;
+
+	vm.pixel_clock = pck / 1000;
+	vm.hsw = div64_u64((u64)(t->hsa + t->hse) * pck, byteclk);
+	vm.hbp = div64_u64((u64)t->hbp * pck, byteclk);
+	vm.hfp = div64_u64((u64)t->hfp * pck, byteclk);
+	vm.x_res = t->hact;
+
+	print_dispc_vm(str, &vm);
+}
+#endif /* PRINT_VERBOSE_VM_TIMINGS */
+
+static bool dsi_cm_calc_dispc_cb(int lckd, int pckd, unsigned long lck,
+		unsigned long pck, void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+	struct omap_video_timings *t = &ctx->dispc_vm;
+
+	ctx->dispc_cinfo.lck_div = lckd;
+	ctx->dispc_cinfo.pck_div = pckd;
+	ctx->dispc_cinfo.lck = lck;
+	ctx->dispc_cinfo.pck = pck;
+
+	*t = *ctx->config->timings;
+	t->pixel_clock = pck / 1000;
+	t->x_res = ctx->config->timings->x_res;
+	t->y_res = ctx->config->timings->y_res;
+	t->hsw = t->hfp = t->hbp = t->vsw = 1;
+	t->vfp = t->vbp = 0;
+
+	return true;
+}
+
+static bool dsi_cm_calc_hsdiv_cb(int regm_dispc, unsigned long dispc,
+		void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+
+	ctx->dsi_cinfo.regm_dispc = regm_dispc;
+	ctx->dsi_cinfo.dsi_pll_hsdiv_dispc_clk = dispc;
+
+	return dispc_div_calc(dispc, ctx->req_pck_min, ctx->req_pck_max,
+			dsi_cm_calc_dispc_cb, ctx);
+}
+
+static bool dsi_cm_calc_pll_cb(int regn, int regm, unsigned long fint,
+		unsigned long pll, void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+
+	ctx->dsi_cinfo.regn = regn;
+	ctx->dsi_cinfo.regm = regm;
+	ctx->dsi_cinfo.fint = fint;
+	ctx->dsi_cinfo.clkin4ddr = pll;
+
+	return dsi_hsdiv_calc(ctx->dsidev, pll, ctx->req_pck_min,
+			dsi_cm_calc_hsdiv_cb, ctx);
+}
+
+static bool dsi_cm_calc(struct dsi_data *dsi,
+		const struct omap_dss_dsi_config *cfg,
+		struct dsi_clk_calc_ctx *ctx)
+{
+	unsigned long clkin;
+	int bitspp, ndl;
+	unsigned long pll_min, pll_max;
+	unsigned long pck, txbyteclk;
+
+	clkin = clk_get_rate(dsi->sys_clk);
+	bitspp = dsi_get_pixel_size(cfg->pixel_format);
+	ndl = dsi->num_lanes_used - 1;
+
+	/*
+	 * Here we should calculate minimum txbyteclk to be able to send the
+	 * frame in time, and also to handle TE. That's not very simple, though,
+	 * especially as we go to LP between each pixel packet due to HW
+	 * "feature". So let's just estimate very roughly and multiply by 1.5.
+	 */
+	pck = cfg->timings->pixel_clock * 1000;
+	pck = pck * 3 / 2;
+	txbyteclk = pck * bitspp / 8 / ndl;
+
+	memset(ctx, 0, sizeof(*ctx));
+	ctx->dsidev = dsi->pdev;
+	ctx->config = cfg;
+	ctx->req_pck_min = pck;
+	ctx->req_pck_nom = pck;
+	ctx->req_pck_max = pck * 3 / 2;
+	ctx->dsi_cinfo.clkin = clkin;
+
+	pll_min = max(cfg->hs_clk_min * 4, txbyteclk * 4 * 4);
+	pll_max = cfg->hs_clk_max * 4;
+
+	return dsi_pll_calc(dsi->pdev, clkin,
+			pll_min, pll_max,
+			dsi_cm_calc_pll_cb, ctx);
+}
+
+static bool dsi_vm_calc_blanking(struct dsi_clk_calc_ctx *ctx)
+{
+	struct dsi_data *dsi = dsi_get_dsidrv_data(ctx->dsidev);
+	const struct omap_dss_dsi_config *cfg = ctx->config;
+	int bitspp = dsi_get_pixel_size(cfg->pixel_format);
+	int ndl = dsi->num_lanes_used - 1;
+	unsigned long hsclk = ctx->dsi_cinfo.clkin4ddr / 4;
+	unsigned long byteclk = hsclk / 4;
+
+	unsigned long dispc_pck, req_pck_min, req_pck_nom, req_pck_max;
+	int xres;
+	int panel_htot, panel_hbl; /* pixels */
+	int dispc_htot, dispc_hbl; /* pixels */
+	int dsi_htot, dsi_hact, dsi_hbl, hss, hse; /* byteclks */
+	int hfp, hsa, hbp;
+	const struct omap_video_timings *req_vm;
+	struct omap_video_timings *dispc_vm;
+	struct omap_dss_dsi_videomode_timings *dsi_vm;
+	u64 dsi_tput, dispc_tput;
+
+	dsi_tput = (u64)byteclk * ndl * 8;
+
+	req_vm = cfg->timings;
+	req_pck_min = ctx->req_pck_min;
+	req_pck_max = ctx->req_pck_max;
+	req_pck_nom = ctx->req_pck_nom;
+
+	dispc_pck = ctx->dispc_cinfo.pck;
+	dispc_tput = (u64)dispc_pck * bitspp;
+
+	xres = req_vm->x_res;
+
+	panel_hbl = req_vm->hfp + req_vm->hbp + req_vm->hsw;
+	panel_htot = xres + panel_hbl;
+
+	dsi_hact = DIV_ROUND_UP(DIV_ROUND_UP(xres * bitspp, 8) + 6, ndl);
+
+	/*
+	 * When there are no line buffers, DISPC and DSI must have the
+	 * same tput. Otherwise DISPC tput needs to be higher than DSI's.
+	 */
+	if (dsi->line_buffer_size < xres * bitspp / 8) {
+		if (dispc_tput != dsi_tput)
+			return false;
+	} else {
+		if (dispc_tput < dsi_tput)
+			return false;
+	}
+
+	/* DSI tput must be over the min requirement */
+	if (dsi_tput < (u64)bitspp * req_pck_min)
+		return false;
+
+	/* When non-burst mode, DSI tput must be below max requirement. */
+	if (cfg->trans_mode != OMAP_DSS_DSI_BURST_MODE) {
+		if (dsi_tput > (u64)bitspp * req_pck_max)
+			return false;
+	}
+
+	hss = DIV_ROUND_UP(4, ndl);
+
+	if (cfg->trans_mode == OMAP_DSS_DSI_PULSE_MODE) {
+		if (ndl == 3 && req_vm->hsw == 0)
+			hse = 1;
+		else
+			hse = DIV_ROUND_UP(4, ndl);
+	} else {
+		hse = 0;
+	}
+
+	/* DSI htot to match the panel's nominal pck */
+	dsi_htot = div64_u64((u64)panel_htot * byteclk, req_pck_nom);
+
+	/* fail if there would be no time for blanking */
+	if (dsi_htot < hss + hse + dsi_hact)
+		return false;
+
+	/* total DSI blanking needed to achieve panel's TL */
+	dsi_hbl = dsi_htot - dsi_hact;
+
+	/* DISPC htot to match the DSI TL */
+	dispc_htot = div64_u64((u64)dsi_htot * dispc_pck, byteclk);
+
+	/* verify that the DSI and DISPC TLs are the same */
+	if ((u64)dsi_htot * dispc_pck != (u64)dispc_htot * byteclk)
+		return false;
+
+	dispc_hbl = dispc_htot - xres;
+
+	/* setup DSI videomode */
+
+	dsi_vm = &ctx->dsi_vm;
+	memset(dsi_vm, 0, sizeof(*dsi_vm));
+
+	dsi_vm->hsclk = hsclk;
+
+	dsi_vm->ndl = ndl;
+	dsi_vm->bitspp = bitspp;
+
+	if (cfg->trans_mode != OMAP_DSS_DSI_PULSE_MODE) {
+		hsa = 0;
+	} else if (ndl == 3 && req_vm->hsw == 0) {
+		hsa = 0;
+	} else {
+		hsa = div64_u64((u64)req_vm->hsw * byteclk, req_pck_nom);
+		hsa = max(hsa - hse, 1);
+	}
+
+	hbp = div64_u64((u64)req_vm->hbp * byteclk, req_pck_nom);
+	hbp = max(hbp, 1);
+
+	hfp = dsi_hbl - (hss + hsa + hse + hbp);
+	if (hfp < 1) {
+		int t;
+		/* we need to take cycles from hbp */
+
+		t = 1 - hfp;
+		hbp = max(hbp - t, 1);
+		hfp = dsi_hbl - (hss + hsa + hse + hbp);
+
+		if (hfp < 1 && hsa > 0) {
+			/* we need to take cycles from hsa */
+			t = 1 - hfp;
+			hsa = max(hsa - t, 1);
+			hfp = dsi_hbl - (hss + hsa + hse + hbp);
+		}
+	}
+
+	if (hfp < 1)
+		return false;
+
+	dsi_vm->hss = hss;
+	dsi_vm->hsa = hsa;
+	dsi_vm->hse = hse;
+	dsi_vm->hbp = hbp;
+	dsi_vm->hact = xres;
+	dsi_vm->hfp = hfp;
+
+	dsi_vm->vsa = req_vm->vsw;
+	dsi_vm->vbp = req_vm->vbp;
+	dsi_vm->vact = req_vm->y_res;
+	dsi_vm->vfp = req_vm->vfp;
+
+	dsi_vm->trans_mode = cfg->trans_mode;
+
+	dsi_vm->blanking_mode = 0;
+	dsi_vm->hsa_blanking_mode = 1;
+	dsi_vm->hfp_blanking_mode = 1;
+	dsi_vm->hbp_blanking_mode = 1;
+
+	dsi_vm->ddr_clk_always_on = cfg->ddr_clk_always_on;
+	dsi_vm->window_sync = 4;
+
+	/* setup DISPC videomode */
+
+	dispc_vm = &ctx->dispc_vm;
+	*dispc_vm = *req_vm;
+	dispc_vm->pixel_clock = dispc_pck / 1000;
+
+	if (cfg->trans_mode == OMAP_DSS_DSI_PULSE_MODE) {
+		hsa = div64_u64((u64)req_vm->hsw * dispc_pck,
+				req_pck_nom);
+		hsa = max(hsa, 1);
+	} else {
+		hsa = 1;
+	}
+
+	hbp = div64_u64((u64)req_vm->hbp * dispc_pck, req_pck_nom);
+	hbp = max(hbp, 1);
+
+	hfp = dispc_hbl - hsa - hbp;
+	if (hfp < 1) {
+		int t;
+		/* we need to take cycles from hbp */
+
+		t = 1 - hfp;
+		hbp = max(hbp - t, 1);
+		hfp = dispc_hbl - hsa - hbp;
+
+		if (hfp < 1) {
+			/* we need to take cycles from hsa */
+			t = 1 - hfp;
+			hsa = max(hsa - t, 1);
+			hfp = dispc_hbl - hsa - hbp;
+		}
+	}
+
+	if (hfp < 1)
+		return false;
+
+	dispc_vm->hfp = hfp;
+	dispc_vm->hsw = hsa;
+	dispc_vm->hbp = hbp;
+
+	return true;
+}
+
+
+static bool dsi_vm_calc_dispc_cb(int lckd, int pckd, unsigned long lck,
+		unsigned long pck, void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+
+	ctx->dispc_cinfo.lck_div = lckd;
+	ctx->dispc_cinfo.pck_div = pckd;
+	ctx->dispc_cinfo.lck = lck;
+	ctx->dispc_cinfo.pck = pck;
+
+	if (dsi_vm_calc_blanking(ctx) == false)
+		return false;
+
+#ifdef PRINT_VERBOSE_VM_TIMINGS
+	print_dispc_vm("dispc", &ctx->dispc_vm);
+	print_dsi_vm("dsi  ", &ctx->dsi_vm);
+	print_dispc_vm("req  ", ctx->config->timings);
+	print_dsi_dispc_vm("act  ", &ctx->dsi_vm);
+#endif
+
+	return true;
+}
+
+static bool dsi_vm_calc_hsdiv_cb(int regm_dispc, unsigned long dispc,
+		void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+	unsigned long pck_max;
+
+	ctx->dsi_cinfo.regm_dispc = regm_dispc;
+	ctx->dsi_cinfo.dsi_pll_hsdiv_dispc_clk = dispc;
+
+	/*
+	 * In burst mode we can let the dispc pck be arbitrarily high, but it
+	 * limits our scaling abilities. So for now, don't aim too high.
+	 */
+
+	if (ctx->config->trans_mode == OMAP_DSS_DSI_BURST_MODE)
+		pck_max = ctx->req_pck_max + 10000000;
+	else
+		pck_max = ctx->req_pck_max;
+
+	return dispc_div_calc(dispc, ctx->req_pck_min, pck_max,
+			dsi_vm_calc_dispc_cb, ctx);
+}
+
+static bool dsi_vm_calc_pll_cb(int regn, int regm, unsigned long fint,
+		unsigned long pll, void *data)
+{
+	struct dsi_clk_calc_ctx *ctx = data;
+
+	ctx->dsi_cinfo.regn = regn;
+	ctx->dsi_cinfo.regm = regm;
+	ctx->dsi_cinfo.fint = fint;
+	ctx->dsi_cinfo.clkin4ddr = pll;
+
+	return dsi_hsdiv_calc(ctx->dsidev, pll, ctx->req_pck_min,
+			dsi_vm_calc_hsdiv_cb, ctx);
+}
+
+static bool dsi_vm_calc(struct dsi_data *dsi,
+		const struct omap_dss_dsi_config *cfg,
+		struct dsi_clk_calc_ctx *ctx)
+{
+	const struct omap_video_timings *t = cfg->timings;
+	unsigned long clkin;
+	unsigned long pll_min;
+	unsigned long pll_max;
+	int ndl = dsi->num_lanes_used - 1;
+	int bitspp = dsi_get_pixel_size(cfg->pixel_format);
+	unsigned long byteclk_min;
+
+	clkin = clk_get_rate(dsi->sys_clk);
+
+	memset(ctx, 0, sizeof(*ctx));
+	ctx->dsidev = dsi->pdev;
+	ctx->config = cfg;
+
+	ctx->dsi_cinfo.clkin = clkin;
+
+	/* these limits should come from the panel driver */
+	ctx->req_pck_min = t->pixel_clock * 1000 - 1000;
+	ctx->req_pck_nom = t->pixel_clock * 1000;
+	ctx->req_pck_max = t->pixel_clock * 1000 + 1000;
+
+	byteclk_min = div64_u64((u64)ctx->req_pck_min * bitspp, ndl * 8);
+	pll_min = max(cfg->hs_clk_min * 4, byteclk_min * 4 * 4);
+
+	if (cfg->trans_mode == OMAP_DSS_DSI_BURST_MODE) {
+		pll_max = cfg->hs_clk_max * 4;
+	} else {
+		unsigned long byteclk_max;
+		byteclk_max = div64_u64((u64)ctx->req_pck_max * bitspp,
+				ndl * 8);
+
+		pll_max = byteclk_max * 4 * 4;
+	}
+
+	return dsi_pll_calc(dsi->pdev, clkin,
+			pll_min, pll_max,
+			dsi_vm_calc_pll_cb, ctx);
+}
+
 int omapdss_dsi_set_config(struct omap_dss_device *dssdev,
 		const struct omap_dss_dsi_config *config)
 {
 	struct platform_device *dsidev = dsi_get_dsidev_from_dssdev(dssdev);
 	struct dsi_data *dsi = dsi_get_dsidrv_data(dsidev);
+	struct dsi_clk_calc_ctx ctx;
+	bool ok;
+	int r;
 
 	mutex_lock(&dsi->lock);
 
-	dsi->timings = *config->timings;
-	dsi->vm_timings = *config->vm_timings;
 	dsi->pix_fmt = config->pixel_format;
 	dsi->mode = config->mode;
 
-	dsi_set_clocks(dssdev, config->hs_clk, config->lp_clk);
+	if (config->mode == OMAP_DSS_DSI_VIDEO_MODE)
+		ok = dsi_vm_calc(dsi, config, &ctx);
+	else
+		ok = dsi_cm_calc(dsi, config, &ctx);
+
+	if (!ok) {
+		DSSERR("failed to find suitable DSI clock settings\n");
+		r = -EINVAL;
+		goto err;
+	}
+
+	dsi_pll_calc_dsi_fck(&ctx.dsi_cinfo);
+
+	r = dsi_lp_clock_calc(&ctx.dsi_cinfo, config->lp_clk_min,
+			config->lp_clk_max);
+	if (r) {
+		DSSERR("failed to find suitable DSI LP clock settings\n");
+		goto err;
+	}
+
+	dsi->user_dsi_cinfo = ctx.dsi_cinfo;
+	dsi->user_dispc_cinfo = ctx.dispc_cinfo;
+
+	dsi->timings = ctx.dispc_vm;
+	dsi->vm_timings = ctx.dsi_vm;
 
 	mutex_unlock(&dsi->lock);
 
 	return 0;
+err:
+	mutex_unlock(&dsi->lock);
+
+	return r;
 }
 EXPORT_SYMBOL(omapdss_dsi_set_config);
 
diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 7f774c5f8b6b..8821d933d9fd 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -267,9 +267,21 @@ enum omap_dss_dsi_trans_mode {
 };
 
 struct omap_dss_dsi_videomode_timings {
+	unsigned long hsclk;
+
+	unsigned ndl;
+	unsigned bitspp;
+
+	/* pixels */
+	u16 hact;
+	/* lines */
+	u16 vact;
+
 	/* DSI video mode blanking data */
 	/* Unit: byte clock cycles */
+	u16 hss;
 	u16 hsa;
+	u16 hse;
 	u16 hfp;
 	u16 hbp;
 	/* Unit: line clocks */
@@ -293,10 +305,12 @@ struct omap_dss_dsi_config {
 	enum omap_dss_dsi_mode mode;
 	enum omap_dss_dsi_pixel_format pixel_format;
 	const struct omap_video_timings *timings;
-	const struct omap_dss_dsi_videomode_timings *vm_timings;
 
-	unsigned long hs_clk;
-	unsigned long lp_clk;
+	unsigned long hs_clk_min, hs_clk_max;
+	unsigned long lp_clk_min, lp_clk_max;
+
+	bool ddr_clk_always_on;
+	enum omap_dss_dsi_trans_mode trans_mode;
 };
 
 void dsi_bus_lock(struct omap_dss_device *dssdev);
-- 
cgit v1.2.3


From 77d3595b29e2289cb6f9e4cece5fca1335554d56 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Wed, 6 Mar 2013 17:17:26 +0200
Subject: OMAPDSS: remove dsi videomode from dssdev

DSI videomode is no longer needed in the omap_dss_device, so remove it.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
---
 include/video/omapdss.h | 1 -
 1 file changed, 1 deletion(-)

(limited to 'include')

diff --git a/include/video/omapdss.h b/include/video/omapdss.h
index 8821d933d9fd..62ca9a77c1d6 100644
--- a/include/video/omapdss.h
+++ b/include/video/omapdss.h
@@ -632,7 +632,6 @@ struct omap_dss_device {
 
 		enum omap_dss_dsi_pixel_format dsi_pix_fmt;
 		enum omap_dss_dsi_mode dsi_mode;
-		struct omap_dss_dsi_videomode_timings dsi_vm_timings;
 	} panel;
 
 	struct {
-- 
cgit v1.2.3


From a4419f3eac875e66a01b654024f6eda17d0dd558 Mon Sep 17 00:00:00 2001
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
Date: Tue, 12 Feb 2013 16:45:23 +0530
Subject: OMAPDSS: n8x0: remove platform backlight calls

The n8x0 panel contains support to call platform backlight functions.
These are not used by any board, and can be removed.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Archit Taneja <archit@ti.com>
---
 drivers/video/omap2/displays/panel-n8x0.c | 74 -------------------------------
 include/video/omap-panel-data.h           |  2 -
 2 files changed, 76 deletions(-)

(limited to 'include')

diff --git a/drivers/video/omap2/displays/panel-n8x0.c b/drivers/video/omap2/displays/panel-n8x0.c
index 9c6b5fafeb2e..9c55c9152863 100644
--- a/drivers/video/omap2/displays/panel-n8x0.c
+++ b/drivers/video/omap2/displays/panel-n8x0.c
@@ -5,7 +5,6 @@
 #include <linux/slab.h>
 #include <linux/gpio.h>
 #include <linux/spi/spi.h>
-#include <linux/backlight.h>
 #include <linux/fb.h>
 
 #include <video/omapdss.h>
@@ -69,7 +68,6 @@ static struct panel_drv_data {
 
 	struct omap_dss_device *dssdev;
 	struct spi_device *spidev;
-	struct backlight_device *bldev;
 
 	int blizzard_ver;
 } s_drv_data;
@@ -424,55 +422,10 @@ static const struct rfbi_timings n8x0_panel_timings = {
 	.cs_pulse_width = 0,
 };
 
-static int n8x0_bl_update_status(struct backlight_device *dev)
-{
-	struct omap_dss_device *dssdev = dev_get_drvdata(&dev->dev);
-	struct panel_n8x0_data *bdata = get_board_data(dssdev);
-	struct panel_drv_data *ddata = get_drv_data(dssdev);
-	int r;
-	int level;
-
-	mutex_lock(&ddata->lock);
-
-	if (dev->props.fb_blank == FB_BLANK_UNBLANK &&
-			dev->props.power == FB_BLANK_UNBLANK)
-		level = dev->props.brightness;
-	else
-		level = 0;
-
-	dev_dbg(&dssdev->dev, "update brightness to %d\n", level);
-
-	if (!bdata->set_backlight)
-		r = -EINVAL;
-	else
-		r = bdata->set_backlight(dssdev, level);
-
-	mutex_unlock(&ddata->lock);
-
-	return r;
-}
-
-static int n8x0_bl_get_intensity(struct backlight_device *dev)
-{
-	if (dev->props.fb_blank == FB_BLANK_UNBLANK &&
-			dev->props.power == FB_BLANK_UNBLANK)
-		return dev->props.brightness;
-
-	return 0;
-}
-
-static const struct backlight_ops n8x0_bl_ops = {
-	.get_brightness = n8x0_bl_get_intensity,
-	.update_status  = n8x0_bl_update_status,
-};
-
 static int n8x0_panel_probe(struct omap_dss_device *dssdev)
 {
 	struct panel_n8x0_data *bdata = get_board_data(dssdev);
 	struct panel_drv_data *ddata;
-	struct backlight_device *bldev;
-	struct backlight_properties props;
-	int r;
 
 	dev_dbg(&dssdev->dev, "probe\n");
 
@@ -491,40 +444,13 @@ static int n8x0_panel_probe(struct omap_dss_device *dssdev)
 	dssdev->ctrl.rfbi_timings = n8x0_panel_timings;
 	dssdev->caps = OMAP_DSS_DISPLAY_CAP_MANUAL_UPDATE;
 
-	memset(&props, 0, sizeof(props));
-	props.max_brightness = 127;
-	props.type = BACKLIGHT_PLATFORM;
-	bldev = backlight_device_register(dev_name(&dssdev->dev), &dssdev->dev,
-			dssdev, &n8x0_bl_ops, &props);
-	if (IS_ERR(bldev)) {
-		r = PTR_ERR(bldev);
-		dev_err(&dssdev->dev, "register backlight failed\n");
-		return r;
-	}
-
-	ddata->bldev = bldev;
-
-	bldev->props.fb_blank = FB_BLANK_UNBLANK;
-	bldev->props.power = FB_BLANK_UNBLANK;
-	bldev->props.brightness = 127;
-
-	n8x0_bl_update_status(bldev);
-
 	return 0;
 }
 
 static void n8x0_panel_remove(struct omap_dss_device *dssdev)
 {
-	struct panel_drv_data *ddata = get_drv_data(dssdev);
-	struct backlight_device *bldev;
-
 	dev_dbg(&dssdev->dev, "remove\n");
 
-	bldev = ddata->bldev;
-	bldev->props.power = FB_BLANK_POWERDOWN;
-	n8x0_bl_update_status(bldev);
-	backlight_device_unregister(bldev);
-
 	dev_set_drvdata(&dssdev->dev, NULL);
 }
 
diff --git a/include/video/omap-panel-data.h b/include/video/omap-panel-data.h
index 6b55839b73fc..0c3b46d3daf3 100644
--- a/include/video/omap-panel-data.h
+++ b/include/video/omap-panel-data.h
@@ -56,8 +56,6 @@ struct panel_n8x0_data {
 	void (*platform_disable)(struct omap_dss_device *dssdev);
 	int panel_reset;
 	int ctrl_pwrdown;
-
-	int (*set_backlight)(struct omap_dss_device *dssdev, int level);
 };
 
 /**
-- 
cgit v1.2.3


From a0a53aa8c7b491a43e2ef66786f9511bae8cbc35 Mon Sep 17 00:00:00 2001
From: Samuel Li <samuel.li@amd.com>
Date: Mon, 8 Apr 2013 17:25:47 -0400
Subject: drm/radeon: Use direct mapping for fast fb access on RS690
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This patch allows the CPU to map the stolen vram segment
directly rather than going through the PCI BAR.  This
significantly improves performance for certain workloads with
a properly patched ddx.

Use radeon.fastfb=1 to enable it (disabled by default).
Currently only supported on RS690, but support for RS780/880
and newer APUs may be added eventually.

Signed-off-by: Samuel Li <samuel.li@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h        |  2 ++
 drivers/gpu/drm/radeon/radeon_drv.c    |  7 ++++++-
 drivers/gpu/drm/radeon/radeon_kms.c    |  3 +++
 drivers/gpu/drm/radeon/radeon_object.c |  4 +++-
 drivers/gpu/drm/radeon/rs690.c         | 23 +++++++++++++++++++++++
 drivers/gpu/drm/radeon/rs690d.h        |  3 +++
 include/uapi/drm/radeon_drm.h          |  3 +++
 7 files changed, 43 insertions(+), 2 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 8bd875304441..730d3359af60 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -95,6 +95,7 @@ extern int radeon_hw_i2c;
 extern int radeon_pcie_gen2;
 extern int radeon_msi;
 extern int radeon_lockup_timeout;
+extern int radeon_fastfb;
 
 /*
  * Copy from radeon_drv.h so we don't have to include both and have conflicting
@@ -1616,6 +1617,7 @@ struct radeon_device {
 	bool				suspend;
 	bool				need_dma32;
 	bool				accel_working;
+	bool				fastfb_working; /* IGP feature*/
 	struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES];
 	const struct firmware *me_fw;	/* all family ME firmware */
 	const struct firmware *pfp_fw;	/* r6/700 PFP firmware */
diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c
index 66a7f0fd9620..b500bbc3e411 100644
--- a/drivers/gpu/drm/radeon/radeon_drv.c
+++ b/drivers/gpu/drm/radeon/radeon_drv.c
@@ -71,9 +71,10 @@
  *   2.28.0 - r600-eg: Add MEM_WRITE packet support
  *   2.29.0 - R500 FP16 color clear registers
  *   2.30.0 - fix for FMASK texturing
+ *   2.31.0 - Add fastfb support for rs690
  */
 #define KMS_DRIVER_MAJOR	2
-#define KMS_DRIVER_MINOR	30
+#define KMS_DRIVER_MINOR	31
 #define KMS_DRIVER_PATCHLEVEL	0
 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
 int radeon_driver_unload_kms(struct drm_device *dev);
@@ -160,6 +161,7 @@ int radeon_hw_i2c = 0;
 int radeon_pcie_gen2 = -1;
 int radeon_msi = -1;
 int radeon_lockup_timeout = 10000;
+int radeon_fastfb = 0;
 
 MODULE_PARM_DESC(no_wb, "Disable AGP writeback for scratch registers");
 module_param_named(no_wb, radeon_no_wb, int, 0444);
@@ -212,6 +214,9 @@ module_param_named(msi, radeon_msi, int, 0444);
 MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (defaul 10000 = 10 seconds, 0 = disable)");
 module_param_named(lockup_timeout, radeon_lockup_timeout, int, 0444);
 
+MODULE_PARM_DESC(fastfb, "Direct FB access for IGP chips (0 = disable, 1 = enable)");
+module_param_named(fastfb, radeon_fastfb, int, 0444);
+
 static struct pci_device_id pciidlist[] = {
 	radeon_PCI_IDS
 };
diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c
index c75cb2c6ba71..f5464482dee8 100644
--- a/drivers/gpu/drm/radeon/radeon_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_kms.c
@@ -376,6 +376,9 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		else
 			return -EINVAL;
 		break;
+	case RADEON_INFO_FASTFB_WORKING:
+		value = rdev->fastfb_working;
+		break;
 	default:
 		DRM_DEBUG_KMS("Invalid request %d\n", info->request);
 		return -EINVAL;
diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c
index d3aface2d12d..58e026afec17 100644
--- a/drivers/gpu/drm/radeon/radeon_object.c
+++ b/drivers/gpu/drm/radeon/radeon_object.c
@@ -321,8 +321,10 @@ void radeon_bo_force_delete(struct radeon_device *rdev)
 int radeon_bo_init(struct radeon_device *rdev)
 {
 	/* Add an MTRR for the VRAM */
-	rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size,
+	if (!rdev->fastfb_working) {
+		rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size,
 			MTRR_TYPE_WRCOMB, 1);
+	}
 	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
 		rdev->mc.mc_vram_size >> 20,
 		(unsigned long long)rdev->mc.aper_size >> 20);
diff --git a/drivers/gpu/drm/radeon/rs690.c b/drivers/gpu/drm/radeon/rs690.c
index 5706d2ac75ab..ab4c86cfd552 100644
--- a/drivers/gpu/drm/radeon/rs690.c
+++ b/drivers/gpu/drm/radeon/rs690.c
@@ -148,6 +148,8 @@ void rs690_pm_info(struct radeon_device *rdev)
 static void rs690_mc_init(struct radeon_device *rdev)
 {
 	u64 base;
+	uint32_t h_addr, l_addr;
+	unsigned long long k8_addr;
 
 	rs400_gart_adjust_size(rdev);
 	rdev->mc.vram_is_ddr = true;
@@ -160,6 +162,27 @@ static void rs690_mc_init(struct radeon_device *rdev)
 	base = RREG32_MC(R_000100_MCCFG_FB_LOCATION);
 	base = G_000100_MC_FB_START(base) << 16;
 	rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev);
+
+	/* Use K8 direct mapping for fast fb access. */ 
+	rdev->fastfb_working = false;
+	h_addr = G_00005F_K8_ADDR_EXT(RREG32_MC(R_00005F_MC_MISC_UMA_CNTL));
+	l_addr = RREG32_MC(R_00001E_K8_FB_LOCATION);
+	k8_addr = ((unsigned long long)h_addr) << 32 | l_addr;
+#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE)
+	if (k8_addr + rdev->mc.visible_vram_size < 0x100000000ULL)	
+#endif
+	{
+		/* FastFB shall be used with UMA memory. Here it is simply disabled when sideport 
+		 * memory is present.
+		 */
+		if (rdev->mc.igp_sideport_enabled == false && radeon_fastfb == 1) {
+			DRM_INFO("Direct mapping: aper base at 0x%llx, replaced by direct mapping base 0x%llx.\n", 
+					(unsigned long long)rdev->mc.aper_base, k8_addr);
+			rdev->mc.aper_base = (resource_size_t)k8_addr;
+			rdev->fastfb_working = true;
+		}
+	}  
+
 	rs690_pm_info(rdev);
 	radeon_vram_location(rdev, &rdev->mc, base);
 	rdev->mc.gtt_base_align = rdev->mc.gtt_size - 1;
diff --git a/drivers/gpu/drm/radeon/rs690d.h b/drivers/gpu/drm/radeon/rs690d.h
index 36e6398a98ae..8af3ccf20cc0 100644
--- a/drivers/gpu/drm/radeon/rs690d.h
+++ b/drivers/gpu/drm/radeon/rs690d.h
@@ -29,6 +29,9 @@
 #define __RS690D_H__
 
 /* Registers */
+#define R_00001E_K8_FB_LOCATION                      0x00001E
+#define R_00005F_MC_MISC_UMA_CNTL                    0x00005F
+#define   G_00005F_K8_ADDR_EXT(x)                      (((x) >> 0) & 0xFF)
 #define R_000078_MC_INDEX                            0x000078
 #define   S_000078_MC_IND_ADDR(x)                      (((x) & 0x1FF) << 0)
 #define   G_000078_MC_IND_ADDR(x)                      (((x) >> 0) & 0x1FF)
diff --git a/include/uapi/drm/radeon_drm.h b/include/uapi/drm/radeon_drm.h
index eeda91774c8a..6fd25563f301 100644
--- a/include/uapi/drm/radeon_drm.h
+++ b/include/uapi/drm/radeon_drm.h
@@ -972,6 +972,9 @@ struct drm_radeon_cs {
 #define RADEON_INFO_MAX_SE		0x12
 /* max SH per SE */
 #define RADEON_INFO_MAX_SH_PER_SE	0x13
+/* fast fb access is enabled */
+#define RADEON_INFO_FASTFB_WORKING	0x14
+
 
 struct drm_radeon_info {
 	uint32_t		request;
-- 
cgit v1.2.3


From f2ba57b5eab8817d86d0f108fdf1878e51dc0a37 Mon Sep 17 00:00:00 2001
From: Christian König <deathsimple@vodafone.de>
Date: Mon, 8 Apr 2013 12:41:29 +0200
Subject: drm/radeon: UVD bringup v8
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Just everything needed to decode videos using UVD.

v6: just all the bugfixes and support for R7xx-SI merged in one patch
v7: UVD_CGC_GATE is a write only register, lockup detection fix
v8: split out VRAM fallback changes, remove support for RV770,
    add support for HEMLOCK, add buffer sizes checks

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/radeon/Makefile        |   2 +-
 drivers/gpu/drm/radeon/evergreen.c     |  40 +-
 drivers/gpu/drm/radeon/evergreend.h    |   7 +
 drivers/gpu/drm/radeon/ni.c            |  49 +++
 drivers/gpu/drm/radeon/nid.h           |   9 +
 drivers/gpu/drm/radeon/r600.c          | 291 +++++++++++++++
 drivers/gpu/drm/radeon/r600d.h         |  61 +++
 drivers/gpu/drm/radeon/radeon.h        |  41 +-
 drivers/gpu/drm/radeon/radeon_asic.c   |  63 ++++
 drivers/gpu/drm/radeon/radeon_asic.h   |  19 +
 drivers/gpu/drm/radeon/radeon_cs.c     |  30 +-
 drivers/gpu/drm/radeon/radeon_fence.c  |  23 +-
 drivers/gpu/drm/radeon/radeon_kms.c    |   1 +
 drivers/gpu/drm/radeon/radeon_object.c |   4 +-
 drivers/gpu/drm/radeon/radeon_object.h |   2 +-
 drivers/gpu/drm/radeon/radeon_ring.c   |  24 +-
 drivers/gpu/drm/radeon/radeon_test.c   |  72 ++--
 drivers/gpu/drm/radeon/radeon_uvd.c    | 664 +++++++++++++++++++++++++++++++++
 drivers/gpu/drm/radeon/rv770.c         | 132 +++++++
 drivers/gpu/drm/radeon/rv770d.h        |  14 +
 drivers/gpu/drm/radeon/si.c            |  32 ++
 drivers/gpu/drm/radeon/sid.h           |   6 +
 include/uapi/drm/radeon_drm.h          |   1 +
 23 files changed, 1534 insertions(+), 53 deletions(-)
 create mode 100644 drivers/gpu/drm/radeon/radeon_uvd.c

(limited to 'include')

diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile
index bf172522ea68..86c5e3611892 100644
--- a/drivers/gpu/drm/radeon/Makefile
+++ b/drivers/gpu/drm/radeon/Makefile
@@ -76,7 +76,7 @@ radeon-y += radeon_device.o radeon_asic.o radeon_kms.o \
 	evergreen.o evergreen_cs.o evergreen_blit_shaders.o evergreen_blit_kms.o \
 	evergreen_hdmi.o radeon_trace_points.o ni.o cayman_blit_shaders.o \
 	atombios_encoders.o radeon_semaphore.o radeon_sa.o atombios_i2c.o si.o \
-	si_blit_shaders.o radeon_prime.o
+	si_blit_shaders.o radeon_prime.o radeon_uvd.o
 
 radeon-$(CONFIG_COMPAT) += radeon_ioc32.o
 radeon-$(CONFIG_VGA_SWITCHEROO) += radeon_atpx_handler.o
diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c
index 305a657bf215..18b66ff59dcf 100644
--- a/drivers/gpu/drm/radeon/evergreen.c
+++ b/drivers/gpu/drm/radeon/evergreen.c
@@ -3360,6 +3360,9 @@ restart_ih:
 				DRM_ERROR("Unhandled interrupt: %d %d\n", src_id, src_data);
 				break;
 			}
+		case 124: /* UVD */
+			DRM_DEBUG("IH: UVD int: 0x%08x\n", src_data);
+			radeon_fence_process(rdev, R600_RING_TYPE_UVD_INDEX);
 			break;
 		case 146:
 		case 147:
@@ -3571,7 +3574,7 @@ int evergreen_copy_dma(struct radeon_device *rdev,
 
 static int evergreen_startup(struct radeon_device *rdev)
 {
-	struct radeon_ring *ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
+	struct radeon_ring *ring;
 	int r;
 
 	/* enable pcie gen2 link */
@@ -3638,6 +3641,17 @@ static int evergreen_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = r600_irq_init(rdev);
 	if (r) {
@@ -3647,6 +3661,7 @@ static int evergreen_startup(struct radeon_device *rdev)
 	}
 	evergreen_irq_set(rdev);
 
+	ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
 	r = radeon_ring_init(rdev, ring, ring->ring_size, RADEON_WB_CP_RPTR_OFFSET,
 			     R600_CP_RB_RPTR, R600_CP_RB_WPTR,
 			     0, 0xfffff, RADEON_CP_PACKET2);
@@ -3670,6 +3685,19 @@ static int evergreen_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+
+		if (r)
+			DRM_ERROR("radeon: error initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -3716,8 +3744,10 @@ int evergreen_resume(struct radeon_device *rdev)
 int evergreen_suspend(struct radeon_device *rdev)
 {
 	r600_audio_fini(rdev);
+	radeon_uvd_suspend(rdev);
 	r700_cp_stop(rdev);
 	r600_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
 	evergreen_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	evergreen_pcie_gart_disable(rdev);
@@ -3797,6 +3827,13 @@ int evergreen_init(struct radeon_device *rdev)
 	rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL;
 	r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_obj = NULL;
+		r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_UVD_INDEX],
+			       4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -3843,6 +3880,7 @@ void evergreen_fini(struct radeon_device *rdev)
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
 	evergreen_pcie_gart_fini(rdev);
+	radeon_uvd_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
 	radeon_fence_driver_fini(rdev);
diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h
index 982d25ad9af3..c5d873e525c9 100644
--- a/drivers/gpu/drm/radeon/evergreend.h
+++ b/drivers/gpu/drm/radeon/evergreend.h
@@ -992,6 +992,13 @@
 #       define TARGET_LINK_SPEED_MASK                     (0xf << 0)
 #       define SELECTABLE_DEEMPHASIS                      (1 << 6)
 
+
+/*
+ * UVD
+ */
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+
 /*
  * PM4
  */
diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c
index 02e958063682..35d7caa60c48 100644
--- a/drivers/gpu/drm/radeon/ni.c
+++ b/drivers/gpu/drm/radeon/ni.c
@@ -933,6 +933,23 @@ void cayman_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 	radeon_ring_write(ring, 10); /* poll interval */
 }
 
+void cayman_uvd_semaphore_emit(struct radeon_device *rdev,
+			       struct radeon_ring *ring,
+			       struct radeon_semaphore *semaphore,
+			       bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_LOW, 0));
+	radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_HIGH, 0));
+	radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
+	radeon_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+}
+
 static void cayman_cp_enable(struct radeon_device *rdev, bool enable)
 {
 	if (enable)
@@ -1684,6 +1701,16 @@ static int cayman_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	r = radeon_fence_driver_start_ring(rdev, CAYMAN_RING_TYPE_CP1_INDEX);
 	if (r) {
 		dev_err(rdev->dev, "failed initializing CP fences (%d).\n", r);
@@ -1750,6 +1777,18 @@ static int cayman_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -1796,6 +1835,8 @@ int cayman_suspend(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	cayman_cp_enable(rdev, false);
 	cayman_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
+	radeon_uvd_suspend(rdev);
 	evergreen_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	cayman_pcie_gart_disable(rdev);
@@ -1870,6 +1911,13 @@ int cayman_init(struct radeon_device *rdev)
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+		ring->ring_obj = NULL;
+		r600_ring_init(rdev, ring, 4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -1921,6 +1969,7 @@ void cayman_fini(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
+	radeon_uvd_fini(rdev);
 	cayman_pcie_gart_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
diff --git a/drivers/gpu/drm/radeon/nid.h b/drivers/gpu/drm/radeon/nid.h
index 445b235c4323..f2555bc44ada 100644
--- a/drivers/gpu/drm/radeon/nid.h
+++ b/drivers/gpu/drm/radeon/nid.h
@@ -489,6 +489,15 @@
 #       define CACHE_FLUSH_AND_INV_EVENT_TS                     (0x14 << 0)
 #       define CACHE_FLUSH_AND_INV_EVENT                        (0x16 << 0)
 
+/*
+ * UVD
+ */
+#define UVD_SEMA_ADDR_LOW				0xEF00
+#define UVD_SEMA_ADDR_HIGH				0xEF04
+#define UVD_SEMA_CMD					0xEF08
+#define UVD_RBC_RB_RPTR					0xF690
+#define UVD_RBC_RB_WPTR					0xF694
+
 /*
  * PM4
  */
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index 1c5308778948..7ce7b83c76f5 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -2551,6 +2551,185 @@ void r600_dma_fini(struct radeon_device *rdev)
 	radeon_ring_fini(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX]);
 }
 
+/*
+ * UVD
+ */
+int r600_uvd_rbc_start(struct radeon_device *rdev)
+{
+	struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	uint64_t rptr_addr;
+	uint32_t rb_bufsz, tmp;
+	int r;
+
+	rptr_addr = rdev->wb.gpu_addr + R600_WB_UVD_RPTR_OFFSET;
+
+	if (upper_32_bits(rptr_addr) != upper_32_bits(ring->gpu_addr)) {
+		DRM_ERROR("UVD ring and rptr not in the same 4GB segment!\n");
+		return -EINVAL;
+	}
+
+	/* force RBC into idle state */
+	WREG32(UVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Set the write pointer delay */
+	WREG32(UVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(UVD_RBC_RB_RPTR_ADDR, rptr_addr >> 2);
+
+	/* programm the 4GB memory segment for rptr and ring buffer */
+	WREG32(UVD_LMI_EXT40_ADDR, upper_32_bits(rptr_addr) |
+				   (0x7 << 16) | (0x1 << 31));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(UVD_RBC_RB_RPTR, 0x0);
+
+	ring->wptr = ring->rptr = RREG32(UVD_RBC_RB_RPTR);
+	WREG32(UVD_RBC_RB_WPTR, ring->wptr);
+
+	/* set the ring address */
+	WREG32(UVD_RBC_RB_BASE, ring->gpu_addr);
+
+	/* Set ring buffer size */
+	rb_bufsz = drm_order(ring->ring_size);
+	rb_bufsz = (0x1 << 8) | rb_bufsz;
+	WREG32(UVD_RBC_RB_CNTL, rb_bufsz);
+
+	ring->ready = true;
+	r = radeon_ring_test(rdev, R600_RING_TYPE_UVD_INDEX, ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	r = radeon_ring_lock(rdev, ring, 10);
+	if (r) {
+		DRM_ERROR("radeon: ring failed to lock UVD ring (%d).\n", r);
+		return r;
+	}
+
+	tmp = PACKET0(UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_TIMEOUT_STATUS, 0));
+	radeon_ring_write(ring, 0x8);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CNTL, 0));
+	radeon_ring_write(ring, 1);
+
+	radeon_ring_unlock_commit(rdev, ring);
+
+	return 0;
+}
+
+void r600_uvd_rbc_stop(struct radeon_device *rdev)
+{
+	struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+
+	/* force RBC into idle state */
+	WREG32(UVD_RBC_RB_CNTL, 0x11010101);
+	ring->ready = false;
+}
+
+int r600_uvd_init(struct radeon_device *rdev)
+{
+	int i, j, r;
+
+	/* disable clock gating */
+	WREG32(UVD_CGC_GATE, 0);
+
+	/* disable interupt */
+	WREG32_P(UVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(UVD_SOFT_RESET, LMI_SOFT_RESET | VCPU_SOFT_RESET |
+	       LBSI_SOFT_RESET | RBC_SOFT_RESET | CSM_SOFT_RESET |
+	       CXW_SOFT_RESET | TAP_SOFT_RESET | LMI_UMC_SOFT_RESET);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(SRBM_SOFT_RESET, 0, ~SOFT_RESET_UVD);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(UVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+	/* disable byte swapping */
+	WREG32(UVD_LMI_SWAP_CNTL, 0);
+	WREG32(UVD_MP_SWAP_CNTL, 0);
+
+	WREG32(UVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(UVD_MPC_SET_MUXA1, 0x0);
+	WREG32(UVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(UVD_MPC_SET_MUXB1, 0x0);
+	WREG32(UVD_MPC_SET_ALU, 0);
+	WREG32(UVD_MPC_SET_MUX, 0x88);
+
+	/* Stall UMC */
+	WREG32_P(UVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	WREG32_P(UVD_RB_ARB_CTRL, 1 << 3, ~(1 << 3));
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(UVD_SOFT_RESET, VCPU_SOFT_RESET);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(UVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(UVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(UVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	WREG32_P(UVD_RB_ARB_CTRL, 0, ~(1 << 3));
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(UVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(UVD_SOFT_RESET, VCPU_SOFT_RESET, ~VCPU_SOFT_RESET);
+		mdelay(10);
+		WREG32_P(UVD_SOFT_RESET, 0, ~VCPU_SOFT_RESET);
+		mdelay(10);
+		r = -1;
+	}
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable interupt */
+	WREG32_P(UVD_MASTINT_EN, 3<<1, ~(3 << 1));
+
+	r = r600_uvd_rbc_start(rdev);
+	if (r)
+		return r;
+
+	DRM_INFO("UVD initialized successfully.\n");
+	return 0;
+}
+
 /*
  * GPU scratch registers helpers function.
  */
@@ -2660,6 +2839,40 @@ int r600_dma_ring_test(struct radeon_device *rdev,
 	return r;
 }
 
+int r600_uvd_ring_test(struct radeon_device *rdev, struct radeon_ring *ring)
+{
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(UVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = radeon_ring_lock(rdev, ring, 3);
+	if (r) {
+		DRM_ERROR("radeon: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	radeon_ring_write(ring, PACKET0(UVD_CONTEXT_ID, 0));
+	radeon_ring_write(ring, 0xDEADBEEF);
+	radeon_ring_unlock_commit(rdev, ring);
+	for (i = 0; i < rdev->usec_timeout; i++) {
+		tmp = RREG32(UVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < rdev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("radeon: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
 /*
  * CP fences/semaphores
  */
@@ -2711,6 +2924,30 @@ void r600_fence_ring_emit(struct radeon_device *rdev,
 	}
 }
 
+void r600_uvd_fence_emit(struct radeon_device *rdev,
+			 struct radeon_fence *fence)
+{
+	struct radeon_ring *ring = &rdev->ring[fence->ring];
+	uint32_t addr = rdev->fence_drv[fence->ring].gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_CONTEXT_ID, 0));
+	radeon_ring_write(ring, fence->seq);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA0, 0));
+	radeon_ring_write(ring, addr & 0xffffffff);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA1, 0));
+	radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_CMD, 0));
+	radeon_ring_write(ring, 0);
+
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA0, 0));
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA1, 0));
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_CMD, 0));
+	radeon_ring_write(ring, 2);
+	return;
+}
+
 void r600_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *ring,
 			      struct radeon_semaphore *semaphore,
@@ -2780,6 +3017,23 @@ void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
 }
 
+void r600_uvd_semaphore_emit(struct radeon_device *rdev,
+			     struct radeon_ring *ring,
+			     struct radeon_semaphore *semaphore,
+			     bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_LOW, 0));
+	radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_HIGH, 0));
+	radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
+	radeon_ring_write(ring, emit_wait ? 1 : 0);
+}
+
 int r600_copy_blit(struct radeon_device *rdev,
 		   uint64_t src_offset,
 		   uint64_t dst_offset,
@@ -3183,6 +3437,16 @@ void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 	radeon_ring_write(ring, ib->length_dw);
 }
 
+void r600_uvd_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
+{
+	struct radeon_ring *ring = &rdev->ring[ib->ring];
+
+	radeon_ring_write(ring, PACKET0(UVD_RBC_IB_BASE, 0));
+	radeon_ring_write(ring, ib->gpu_addr);
+	radeon_ring_write(ring, PACKET0(UVD_RBC_IB_SIZE, 0));
+	radeon_ring_write(ring, ib->length_dw);
+}
+
 int r600_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
 {
 	struct radeon_ib ib;
@@ -3300,6 +3564,33 @@ int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
 	return r;
 }
 
+int r600_uvd_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
+{
+	struct radeon_fence *fence;
+	int r;
+
+	r = radeon_uvd_get_create_msg(rdev, ring->idx, 1, NULL);
+	if (r) {
+		DRM_ERROR("radeon: failed to get create msg (%d).\n", r);
+		return r;
+	}
+
+	r = radeon_uvd_get_destroy_msg(rdev, ring->idx, 1, &fence);
+	if (r) {
+		DRM_ERROR("radeon: failed to get destroy ib (%d).\n", r);
+		return r;
+	}
+
+	r = radeon_fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("radeon: fence wait failed (%d).\n", r);
+		return r;
+	}
+	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+	radeon_fence_unref(&fence);
+	return r;
+}
+
 /**
  * r600_dma_ring_ib_execute - Schedule an IB on the DMA engine
  *
diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h
index a42ba11a3bed..441bdb809a0b 100644
--- a/drivers/gpu/drm/radeon/r600d.h
+++ b/drivers/gpu/drm/radeon/r600d.h
@@ -691,6 +691,7 @@
 #define SRBM_SOFT_RESET                                   0xe60
 #       define SOFT_RESET_DMA                             (1 << 12)
 #       define SOFT_RESET_RLC                             (1 << 13)
+#       define SOFT_RESET_UVD                             (1 << 18)
 #       define RV770_SOFT_RESET_DMA                       (1 << 20)
 
 #define CP_INT_CNTL                                       0xc124
@@ -1142,6 +1143,66 @@
 #       define AFMT_AZ_FORMAT_WTRIG_ACK      (1 << 29)
 #       define AFMT_AZ_AUDIO_ENABLE_CHG_ACK  (1 << 30)
 
+/*
+ * UVD
+ */
+#define UVD_SEMA_ADDR_LOW				0xef00
+#define UVD_SEMA_ADDR_HIGH				0xef04
+#define UVD_SEMA_CMD					0xef08
+
+#define UVD_GPCOM_VCPU_CMD				0xef0c
+#define UVD_GPCOM_VCPU_DATA0				0xef10
+#define UVD_GPCOM_VCPU_DATA1				0xef14
+#define UVD_ENGINE_CNTL					0xef18
+
+#define UVD_SEMA_CNTL					0xf400
+#define UVD_RB_ARB_CTRL					0xf480
+
+#define UVD_LMI_EXT40_ADDR				0xf498
+#define UVD_CGC_GATE					0xf4a8
+#define UVD_LMI_CTRL2					0xf4f4
+#define UVD_MASTINT_EN					0xf500
+#define UVD_LMI_ADDR_EXT				0xf594
+#define UVD_LMI_CTRL					0xf598
+#define UVD_LMI_SWAP_CNTL				0xf5b4
+#define UVD_MP_SWAP_CNTL				0xf5bC
+#define UVD_MPC_CNTL					0xf5dC
+#define UVD_MPC_SET_MUXA0				0xf5e4
+#define UVD_MPC_SET_MUXA1				0xf5e8
+#define UVD_MPC_SET_MUXB0				0xf5eC
+#define UVD_MPC_SET_MUXB1				0xf5f0
+#define UVD_MPC_SET_MUX					0xf5f4
+#define UVD_MPC_SET_ALU					0xf5f8
+
+#define UVD_VCPU_CNTL					0xf660
+#define UVD_SOFT_RESET					0xf680
+#define		RBC_SOFT_RESET					(1<<0)
+#define		LBSI_SOFT_RESET					(1<<1)
+#define		LMI_SOFT_RESET					(1<<2)
+#define		VCPU_SOFT_RESET					(1<<3)
+#define		CSM_SOFT_RESET					(1<<5)
+#define		CXW_SOFT_RESET					(1<<6)
+#define		TAP_SOFT_RESET					(1<<7)
+#define		LMI_UMC_SOFT_RESET				(1<<13)
+#define UVD_RBC_IB_BASE					0xf684
+#define UVD_RBC_IB_SIZE					0xf688
+#define UVD_RBC_RB_BASE					0xf68c
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+#define UVD_RBC_RB_WPTR_CNTL				0xf698
+
+#define UVD_STATUS					0xf6bc
+
+#define UVD_SEMA_TIMEOUT_STATUS				0xf6c0
+#define UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL		0xf6c4
+#define UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL		0xf6c8
+#define UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL		0xf6cc
+
+#define UVD_RBC_RB_CNTL					0xf6a4
+#define UVD_RBC_RB_RPTR_ADDR				0xf6a8
+
+#define UVD_CONTEXT_ID					0xf6f4
+
 /*
  * PM4
  */
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 3db6b02c4263..66e68c1a578f 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -110,24 +110,27 @@ extern int radeon_fastfb;
 #define RADEON_BIOS_NUM_SCRATCH			8
 
 /* max number of rings */
-#define RADEON_NUM_RINGS			5
+#define RADEON_NUM_RINGS			6
 
 /* fence seq are set to this number when signaled */
 #define RADEON_FENCE_SIGNALED_SEQ		0LL
 
 /* internal ring indices */
 /* r1xx+ has gfx CP ring */
-#define RADEON_RING_TYPE_GFX_INDEX		0
+#define RADEON_RING_TYPE_GFX_INDEX	0
 
 /* cayman has 2 compute CP rings */
-#define CAYMAN_RING_TYPE_CP1_INDEX		1
-#define CAYMAN_RING_TYPE_CP2_INDEX		2
+#define CAYMAN_RING_TYPE_CP1_INDEX	1
+#define CAYMAN_RING_TYPE_CP2_INDEX	2
 
 /* R600+ has an async dma ring */
 #define R600_RING_TYPE_DMA_INDEX		3
 /* cayman add a second async dma ring */
 #define CAYMAN_RING_TYPE_DMA1_INDEX		4
 
+/* R600+ */
+#define R600_RING_TYPE_UVD_INDEX	5
+
 /* hardcode those limit for now */
 #define RADEON_VA_IB_OFFSET			(1 << 20)
 #define RADEON_VA_RESERVED_SIZE			(8 << 20)
@@ -921,6 +924,7 @@ struct radeon_wb {
 #define R600_WB_DMA_RPTR_OFFSET   1792
 #define R600_WB_IH_WPTR_OFFSET   2048
 #define CAYMAN_WB_DMA1_RPTR_OFFSET   2304
+#define R600_WB_UVD_RPTR_OFFSET  2560
 #define R600_WB_EVENT_OFFSET     3072
 
 /**
@@ -1121,6 +1125,33 @@ struct radeon_pm {
 int radeon_pm_get_type_index(struct radeon_device *rdev,
 			     enum radeon_pm_state_type ps_type,
 			     int instance);
+/*
+ * UVD
+ */
+#define RADEON_MAX_UVD_HANDLES	10
+#define RADEON_UVD_STACK_SIZE	(1024*1024)
+#define RADEON_UVD_HEAP_SIZE	(1024*1024)
+
+struct radeon_uvd {
+	struct radeon_bo	*vcpu_bo;
+	void			*cpu_addr;
+	uint64_t		gpu_addr;
+	atomic_t		handles[RADEON_MAX_UVD_HANDLES];
+	struct drm_file		*filp[RADEON_MAX_UVD_HANDLES];
+};
+
+int radeon_uvd_init(struct radeon_device *rdev);
+void radeon_uvd_fini(struct radeon_device *rdev);
+int radeon_uvd_suspend(struct radeon_device *rdev);
+int radeon_uvd_resume(struct radeon_device *rdev);
+int radeon_uvd_get_create_msg(struct radeon_device *rdev, int ring,
+			      uint32_t handle, struct radeon_fence **fence);
+int radeon_uvd_get_destroy_msg(struct radeon_device *rdev, int ring,
+			       uint32_t handle, struct radeon_fence **fence);
+void radeon_uvd_force_into_uvd_segment(struct radeon_bo *rbo);
+void radeon_uvd_free_handles(struct radeon_device *rdev,
+			     struct drm_file *filp);
+int radeon_uvd_cs_parse(struct radeon_cs_parser *parser);
 
 struct r600_audio {
 	int			channels;
@@ -1611,6 +1642,7 @@ struct radeon_device {
 	struct radeon_asic		*asic;
 	struct radeon_gem		gem;
 	struct radeon_pm		pm;
+	struct radeon_uvd		uvd;
 	uint32_t			bios_scratch[RADEON_BIOS_NUM_SCRATCH];
 	struct radeon_wb		wb;
 	struct radeon_dummy_page	dummy_page;
@@ -1625,6 +1657,7 @@ struct radeon_device {
 	const struct firmware *rlc_fw;	/* r6/700 RLC firmware */
 	const struct firmware *mc_fw;	/* NI MC firmware */
 	const struct firmware *ce_fw;	/* SI CE firmware */
+	const struct firmware *uvd_fw;	/* UVD firmware */
 	struct r600_blit r600_blit;
 	struct r600_vram_scratch vram_scratch;
 	int msi_enabled; /* msi enabled */
diff --git a/drivers/gpu/drm/radeon/radeon_asic.c b/drivers/gpu/drm/radeon/radeon_asic.c
index aba0a893ea98..a7a7b2bc4204 100644
--- a/drivers/gpu/drm/radeon/radeon_asic.c
+++ b/drivers/gpu/drm/radeon/radeon_asic.c
@@ -1130,6 +1130,15 @@ static struct radeon_asic rv770_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &r600_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1216,6 +1225,15 @@ static struct radeon_asic evergreen_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1302,6 +1320,15 @@ static struct radeon_asic sumo_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1388,6 +1415,15 @@ static struct radeon_asic btc_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1517,6 +1553,15 @@ static struct radeon_asic cayman_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &cayman_dma_is_lockup,
 			.vm_flush = &cayman_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1646,6 +1691,15 @@ static struct radeon_asic trinity_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &cayman_dma_is_lockup,
 			.vm_flush = &cayman_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1775,6 +1829,15 @@ static struct radeon_asic si_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &si_dma_is_lockup,
 			.vm_flush = &si_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h
index 3535f73ad3e2..515db96e3e2c 100644
--- a/drivers/gpu/drm/radeon/radeon_asic.h
+++ b/drivers/gpu/drm/radeon/radeon_asic.h
@@ -330,6 +330,7 @@ int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
 void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib);
 int r600_ring_test(struct radeon_device *rdev, struct radeon_ring *cp);
 int r600_dma_ring_test(struct radeon_device *rdev, struct radeon_ring *cp);
+int r600_uvd_ring_test(struct radeon_device *rdev, struct radeon_ring *ring);
 int r600_copy_blit(struct radeon_device *rdev,
 		   uint64_t src_offset, uint64_t dst_offset,
 		   unsigned num_gpu_pages, struct radeon_fence **fence);
@@ -392,6 +393,19 @@ int r600_mc_wait_for_idle(struct radeon_device *rdev);
 u32 r600_get_xclk(struct radeon_device *rdev);
 uint64_t r600_get_gpu_clock_counter(struct radeon_device *rdev);
 
+/* uvd */
+int r600_uvd_init(struct radeon_device *rdev);
+int r600_uvd_rbc_start(struct radeon_device *rdev);
+void r600_uvd_rbc_stop(struct radeon_device *rdev);
+int r600_uvd_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
+void r600_uvd_fence_emit(struct radeon_device *rdev,
+			 struct radeon_fence *fence);
+void r600_uvd_semaphore_emit(struct radeon_device *rdev,
+			     struct radeon_ring *ring,
+			     struct radeon_semaphore *semaphore,
+			     bool emit_wait);
+void r600_uvd_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib);
+
 /*
  * rv770,rv730,rv710,rv740
  */
@@ -409,6 +423,7 @@ int rv770_copy_dma(struct radeon_device *rdev,
 		  unsigned num_gpu_pages,
 		   struct radeon_fence **fence);
 u32 rv770_get_xclk(struct radeon_device *rdev);
+int rv770_uvd_resume(struct radeon_device *rdev);
 
 /*
  * evergreen
@@ -465,6 +480,10 @@ int evergreen_copy_dma(struct radeon_device *rdev,
  */
 void cayman_fence_ring_emit(struct radeon_device *rdev,
 			    struct radeon_fence *fence);
+void cayman_uvd_semaphore_emit(struct radeon_device *rdev,
+			       struct radeon_ring *ring,
+			       struct radeon_semaphore *semaphore,
+			       bool emit_wait);
 void cayman_pcie_gart_tlb_flush(struct radeon_device *rdev);
 int cayman_init(struct radeon_device *rdev);
 void cayman_fini(struct radeon_device *rdev);
diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c
index c9ee4c02522a..c7407074c09b 100644
--- a/drivers/gpu/drm/radeon/radeon_cs.c
+++ b/drivers/gpu/drm/radeon/radeon_cs.c
@@ -53,7 +53,6 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p)
 	}
 	for (i = 0; i < p->nrelocs; i++) {
 		struct drm_radeon_cs_reloc *r;
-		uint32_t domain;
 
 		duplicate = false;
 		r = (struct drm_radeon_cs_reloc *)&chunk->kdata[i*4];
@@ -81,11 +80,25 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p)
 		p->relocs[i].lobj.bo = p->relocs[i].robj;
 		p->relocs[i].lobj.written = !!r->write_domain;
 
-		domain = r->write_domain ? r->write_domain : r->read_domains;
-		p->relocs[i].lobj.domain = domain;
-		if (domain == RADEON_GEM_DOMAIN_VRAM)
-			domain |= RADEON_GEM_DOMAIN_GTT;
-		p->relocs[i].lobj.alt_domain = domain;
+		/* the first reloc of an UVD job is the
+		   msg and that must be in VRAM */
+		if (p->ring == R600_RING_TYPE_UVD_INDEX && i == 0) {
+			/* TODO: is this still needed for NI+ ? */
+			p->relocs[i].lobj.domain =
+				RADEON_GEM_DOMAIN_VRAM;
+
+			p->relocs[i].lobj.alt_domain =
+				RADEON_GEM_DOMAIN_VRAM;
+
+		} else {
+			uint32_t domain = r->write_domain ?
+				r->write_domain : r->read_domains;
+
+			p->relocs[i].lobj.domain = domain;
+			if (domain == RADEON_GEM_DOMAIN_VRAM)
+				domain |= RADEON_GEM_DOMAIN_GTT;
+			p->relocs[i].lobj.alt_domain = domain;
+		}
 
 		p->relocs[i].lobj.tv.bo = &p->relocs[i].robj->tbo;
 		p->relocs[i].handle = r->handle;
@@ -93,7 +106,7 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p)
 		radeon_bo_list_add_object(&p->relocs[i].lobj,
 					  &p->validated);
 	}
-	return radeon_bo_list_validate(&p->validated);
+	return radeon_bo_list_validate(&p->validated, p->ring);
 }
 
 static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority)
@@ -128,6 +141,9 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority
 			return -EINVAL;
 		}
 		break;
+	case RADEON_CS_RING_UVD:
+		p->ring = R600_RING_TYPE_UVD_INDEX;
+		break;
 	}
 	return 0;
 }
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 34356252567a..82fe1835ff8c 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -31,9 +31,9 @@
 #include <linux/seq_file.h>
 #include <linux/atomic.h>
 #include <linux/wait.h>
-#include <linux/list.h>
 #include <linux/kref.h>
 #include <linux/slab.h>
+#include <linux/firmware.h>
 #include <drm/drmP.h>
 #include "radeon_reg.h"
 #include "radeon.h"
@@ -767,8 +767,21 @@ int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring)
 
 	radeon_scratch_free(rdev, rdev->fence_drv[ring].scratch_reg);
 	if (rdev->wb.use_event || !radeon_ring_supports_scratch_reg(rdev, &rdev->ring[ring])) {
-		rdev->fence_drv[ring].scratch_reg = 0;
-		index = R600_WB_EVENT_OFFSET + ring * 4;
+		if (ring != R600_RING_TYPE_UVD_INDEX) {
+			rdev->fence_drv[ring].scratch_reg = 0;
+			index = R600_WB_EVENT_OFFSET + ring * 4;
+			rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
+			rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr +
+							 index;
+
+		} else {
+			/* put fence directly behind firmware */
+			rdev->fence_drv[ring].cpu_addr = rdev->uvd.cpu_addr +
+							 rdev->uvd_fw->size;
+			rdev->fence_drv[ring].gpu_addr = rdev->uvd.gpu_addr +
+							 rdev->uvd_fw->size;
+		}
+
 	} else {
 		r = radeon_scratch_get(rdev, &rdev->fence_drv[ring].scratch_reg);
 		if (r) {
@@ -778,9 +791,9 @@ int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring)
 		index = RADEON_WB_SCRATCH_OFFSET +
 			rdev->fence_drv[ring].scratch_reg -
 			rdev->scratch.reg_base;
+		rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
+		rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr + index;
 	}
-	rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
-	rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr + index;
 	radeon_fence_write(rdev, atomic64_read(&rdev->fence_drv[ring].last_seq), ring);
 	rdev->fence_drv[ring].initialized = true;
 	dev_info(rdev->dev, "fence driver on ring %d use gpu addr 0x%016llx and cpu addr 0x%p\n",
diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c
index f5464482dee8..8365c75b31b6 100644
--- a/drivers/gpu/drm/radeon/radeon_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_kms.c
@@ -516,6 +516,7 @@ void radeon_driver_preclose_kms(struct drm_device *dev,
 		rdev->hyperz_filp = NULL;
 	if (rdev->cmask_filp == file_priv)
 		rdev->cmask_filp = NULL;
+	radeon_uvd_free_handles(rdev, file_priv);
 }
 
 /*
diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c
index 4466477f1e7b..1424ccde2377 100644
--- a/drivers/gpu/drm/radeon/radeon_object.c
+++ b/drivers/gpu/drm/radeon/radeon_object.c
@@ -348,7 +348,7 @@ void radeon_bo_list_add_object(struct radeon_bo_list *lobj,
 	}
 }
 
-int radeon_bo_list_validate(struct list_head *head)
+int radeon_bo_list_validate(struct list_head *head, int ring)
 {
 	struct radeon_bo_list *lobj;
 	struct radeon_bo *bo;
@@ -366,6 +366,8 @@ int radeon_bo_list_validate(struct list_head *head)
 			
 		retry:
 			radeon_ttm_placement_from_domain(bo, domain);
+			if (ring == R600_RING_TYPE_UVD_INDEX)
+				radeon_uvd_force_into_uvd_segment(bo);
 			r = ttm_bo_validate(&bo->tbo, &bo->placement,
 						true, false);
 			if (unlikely(r)) {
diff --git a/drivers/gpu/drm/radeon/radeon_object.h b/drivers/gpu/drm/radeon/radeon_object.h
index 5fc86b03043b..e2cb80a96b51 100644
--- a/drivers/gpu/drm/radeon/radeon_object.h
+++ b/drivers/gpu/drm/radeon/radeon_object.h
@@ -128,7 +128,7 @@ extern int radeon_bo_init(struct radeon_device *rdev);
 extern void radeon_bo_fini(struct radeon_device *rdev);
 extern void radeon_bo_list_add_object(struct radeon_bo_list *lobj,
 				struct list_head *head);
-extern int radeon_bo_list_validate(struct list_head *head);
+extern int radeon_bo_list_validate(struct list_head *head, int ring);
 extern int radeon_bo_fbdev_mmap(struct radeon_bo *bo,
 				struct vm_area_struct *vma);
 extern int radeon_bo_set_tiling_flags(struct radeon_bo *bo,
diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c
index 8d58e268ff6d..31e47d898c46 100644
--- a/drivers/gpu/drm/radeon/radeon_ring.c
+++ b/drivers/gpu/drm/radeon/radeon_ring.c
@@ -368,7 +368,7 @@ void radeon_ring_free_size(struct radeon_device *rdev, struct radeon_ring *ring)
 {
 	u32 rptr;
 
-	if (rdev->wb.enabled)
+	if (rdev->wb.enabled && ring != &rdev->ring[R600_RING_TYPE_UVD_INDEX])
 		rptr = le32_to_cpu(rdev->wb.wb[ring->rptr_offs/4]);
 	else
 		rptr = RREG32(ring->rptr_reg);
@@ -821,18 +821,20 @@ static int radeon_debugfs_ring_info(struct seq_file *m, void *data)
 	return 0;
 }
 
-static int radeon_ring_type_gfx_index = RADEON_RING_TYPE_GFX_INDEX;
-static int cayman_ring_type_cp1_index = CAYMAN_RING_TYPE_CP1_INDEX;
-static int cayman_ring_type_cp2_index = CAYMAN_RING_TYPE_CP2_INDEX;
-static int radeon_ring_type_dma1_index = R600_RING_TYPE_DMA_INDEX;
-static int radeon_ring_type_dma2_index = CAYMAN_RING_TYPE_DMA1_INDEX;
+static int radeon_gfx_index = RADEON_RING_TYPE_GFX_INDEX;
+static int cayman_cp1_index = CAYMAN_RING_TYPE_CP1_INDEX;
+static int cayman_cp2_index = CAYMAN_RING_TYPE_CP2_INDEX;
+static int radeon_dma1_index = R600_RING_TYPE_DMA_INDEX;
+static int radeon_dma2_index = CAYMAN_RING_TYPE_DMA1_INDEX;
+static int r600_uvd_index = R600_RING_TYPE_UVD_INDEX;
 
 static struct drm_info_list radeon_debugfs_ring_info_list[] = {
-	{"radeon_ring_gfx", radeon_debugfs_ring_info, 0, &radeon_ring_type_gfx_index},
-	{"radeon_ring_cp1", radeon_debugfs_ring_info, 0, &cayman_ring_type_cp1_index},
-	{"radeon_ring_cp2", radeon_debugfs_ring_info, 0, &cayman_ring_type_cp2_index},
-	{"radeon_ring_dma1", radeon_debugfs_ring_info, 0, &radeon_ring_type_dma1_index},
-	{"radeon_ring_dma2", radeon_debugfs_ring_info, 0, &radeon_ring_type_dma2_index},
+	{"radeon_ring_gfx", radeon_debugfs_ring_info, 0, &radeon_gfx_index},
+	{"radeon_ring_cp1", radeon_debugfs_ring_info, 0, &cayman_cp1_index},
+	{"radeon_ring_cp2", radeon_debugfs_ring_info, 0, &cayman_cp2_index},
+	{"radeon_ring_dma1", radeon_debugfs_ring_info, 0, &radeon_dma1_index},
+	{"radeon_ring_dma2", radeon_debugfs_ring_info, 0, &radeon_dma2_index},
+	{"radeon_ring_uvd", radeon_debugfs_ring_info, 0, &r600_uvd_index},
 };
 
 static int radeon_debugfs_sa_info(struct seq_file *m, void *data)
diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c
index fda09c9ea689..bbed4af8d0bc 100644
--- a/drivers/gpu/drm/radeon/radeon_test.c
+++ b/drivers/gpu/drm/radeon/radeon_test.c
@@ -252,6 +252,36 @@ void radeon_test_moves(struct radeon_device *rdev)
 		radeon_do_test_moves(rdev, RADEON_TEST_COPY_BLIT);
 }
 
+static int radeon_test_create_and_emit_fence(struct radeon_device *rdev,
+					     struct radeon_ring *ring,
+					     struct radeon_fence **fence)
+{
+	int r;
+
+	if (ring->idx == R600_RING_TYPE_UVD_INDEX) {
+		r = radeon_uvd_get_create_msg(rdev, ring->idx, 1, NULL);
+		if (r) {
+			DRM_ERROR("Failed to get dummy create msg\n");
+			return r;
+		}
+
+		r = radeon_uvd_get_destroy_msg(rdev, ring->idx, 1, fence);
+		if (r) {
+			DRM_ERROR("Failed to get dummy destroy msg\n");
+			return r;
+		}
+	} else {
+		r = radeon_ring_lock(rdev, ring, 64);
+		if (r) {
+			DRM_ERROR("Failed to lock ring A %d\n", ring->idx);
+			return r;
+		}
+		radeon_fence_emit(rdev, fence, ring->idx);
+		radeon_ring_unlock_commit(rdev, ring);
+	}
+	return 0;
+}
+
 void radeon_test_ring_sync(struct radeon_device *rdev,
 			   struct radeon_ring *ringA,
 			   struct radeon_ring *ringB)
@@ -272,21 +302,24 @@ void radeon_test_ring_sync(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fence1, ringA->idx);
-	if (r) {
-		DRM_ERROR("Failed to emit fence 1\n");
-		radeon_ring_unlock_undo(rdev, ringA);
+	radeon_ring_unlock_commit(rdev, ringA);
+
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fence1);
+	if (r)
 		goto out_cleanup;
-	}
-	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fence2, ringA->idx);
+
+	r = radeon_ring_lock(rdev, ringA, 64);
 	if (r) {
-		DRM_ERROR("Failed to emit fence 2\n");
-		radeon_ring_unlock_undo(rdev, ringA);
+		DRM_ERROR("Failed to lock ring A %d\n", ringA->idx);
 		goto out_cleanup;
 	}
+	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
 	radeon_ring_unlock_commit(rdev, ringA);
 
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fence2);
+	if (r)
+		goto out_cleanup;
+
 	mdelay(1000);
 
 	if (radeon_fence_signaled(fence1)) {
@@ -364,27 +397,22 @@ static void radeon_test_ring_sync2(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fenceA, ringA->idx);
-	if (r) {
-		DRM_ERROR("Failed to emit sync fence 1\n");
-		radeon_ring_unlock_undo(rdev, ringA);
-		goto out_cleanup;
-	}
 	radeon_ring_unlock_commit(rdev, ringA);
 
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fenceA);
+	if (r)
+		goto out_cleanup;
+
 	r = radeon_ring_lock(rdev, ringB, 64);
 	if (r) {
 		DRM_ERROR("Failed to lock ring B %d\n", ringB->idx);
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringB->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fenceB, ringB->idx);
-	if (r) {
-		DRM_ERROR("Failed to create sync fence 2\n");
-		radeon_ring_unlock_undo(rdev, ringB);
-		goto out_cleanup;
-	}
 	radeon_ring_unlock_commit(rdev, ringB);
+	r = radeon_test_create_and_emit_fence(rdev, ringB, &fenceB);
+	if (r)
+		goto out_cleanup;
 
 	mdelay(1000);
 
@@ -393,7 +421,7 @@ static void radeon_test_ring_sync2(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	if (radeon_fence_signaled(fenceB)) {
-		DRM_ERROR("Fence A signaled without waiting for semaphore.\n");
+		DRM_ERROR("Fence B signaled without waiting for semaphore.\n");
 		goto out_cleanup;
 	}
 
diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c b/drivers/gpu/drm/radeon/radeon_uvd.c
new file mode 100644
index 000000000000..05a192e95e5d
--- /dev/null
+++ b/drivers/gpu/drm/radeon/radeon_uvd.c
@@ -0,0 +1,664 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "radeon.h"
+#include "r600d.h"
+
+/* Firmware Names */
+#define FIRMWARE_RV710		"radeon/RV710_uvd.bin"
+#define FIRMWARE_CYPRESS	"radeon/CYPRESS_uvd.bin"
+#define FIRMWARE_SUMO		"radeon/SUMO_uvd.bin"
+#define FIRMWARE_TAHITI		"radeon/TAHITI_uvd.bin"
+
+MODULE_FIRMWARE(FIRMWARE_RV710);
+MODULE_FIRMWARE(FIRMWARE_CYPRESS);
+MODULE_FIRMWARE(FIRMWARE_SUMO);
+MODULE_FIRMWARE(FIRMWARE_TAHITI);
+
+int radeon_uvd_init(struct radeon_device *rdev)
+{
+	struct platform_device *pdev;
+	unsigned long bo_size;
+	const char *fw_name;
+	int i, r;
+
+	pdev = platform_device_register_simple("radeon_uvd", 0, NULL, 0);
+	r = IS_ERR(pdev);
+	if (r) {
+		dev_err(rdev->dev, "radeon_uvd: Failed to register firmware\n");
+		return -EINVAL;
+	}
+
+	switch (rdev->family) {
+	case CHIP_RV710:
+	case CHIP_RV730:
+	case CHIP_RV740:
+		fw_name = FIRMWARE_RV710;
+		break;
+
+	case CHIP_CYPRESS:
+	case CHIP_HEMLOCK:
+	case CHIP_JUNIPER:
+	case CHIP_REDWOOD:
+	case CHIP_CEDAR:
+		fw_name = FIRMWARE_CYPRESS;
+		break;
+
+	case CHIP_SUMO:
+	case CHIP_SUMO2:
+	case CHIP_PALM:
+	case CHIP_CAYMAN:
+	case CHIP_BARTS:
+	case CHIP_TURKS:
+	case CHIP_CAICOS:
+		fw_name = FIRMWARE_SUMO;
+		break;
+
+	case CHIP_TAHITI:
+	case CHIP_VERDE:
+	case CHIP_PITCAIRN:
+	case CHIP_ARUBA:
+		fw_name = FIRMWARE_TAHITI;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&rdev->uvd_fw, fw_name, &pdev->dev);
+	if (r) {
+		dev_err(rdev->dev, "radeon_uvd: Can't load firmware \"%s\"\n",
+			fw_name);
+		platform_device_unregister(pdev);
+		return r;
+	}
+
+	platform_device_unregister(pdev);
+
+	bo_size = RADEON_GPU_PAGE_ALIGN(rdev->uvd_fw->size + 4) +
+		  RADEON_UVD_STACK_SIZE + RADEON_UVD_HEAP_SIZE;
+	r = radeon_bo_create(rdev, bo_size, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &rdev->uvd.vcpu_bo);
+	if (r) {
+		dev_err(rdev->dev, "(%d) failed to allocate UVD bo\n", r);
+		return r;
+	}
+
+	r = radeon_uvd_resume(rdev);
+	if (r)
+		return r;
+
+	memset(rdev->uvd.cpu_addr, 0, bo_size);
+	memcpy(rdev->uvd.cpu_addr, rdev->uvd_fw->data, rdev->uvd_fw->size);
+
+	r = radeon_uvd_suspend(rdev);
+	if (r)
+		return r;
+
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		atomic_set(&rdev->uvd.handles[i], 0);
+		rdev->uvd.filp[i] = NULL;
+	}
+
+	return 0;
+}
+
+void radeon_uvd_fini(struct radeon_device *rdev)
+{
+	radeon_uvd_suspend(rdev);
+	radeon_bo_unref(&rdev->uvd.vcpu_bo);
+}
+
+int radeon_uvd_suspend(struct radeon_device *rdev)
+{
+	int r;
+
+	if (rdev->uvd.vcpu_bo == NULL)
+		return 0;
+
+	r = radeon_bo_reserve(rdev->uvd.vcpu_bo, false);
+	if (!r) {
+		radeon_bo_kunmap(rdev->uvd.vcpu_bo);
+		radeon_bo_unpin(rdev->uvd.vcpu_bo);
+		radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+	}
+	return r;
+}
+
+int radeon_uvd_resume(struct radeon_device *rdev)
+{
+	int r;
+
+	if (rdev->uvd.vcpu_bo == NULL)
+		return -EINVAL;
+
+	r = radeon_bo_reserve(rdev->uvd.vcpu_bo, false);
+	if (r) {
+		radeon_bo_unref(&rdev->uvd.vcpu_bo);
+		dev_err(rdev->dev, "(%d) failed to reserve UVD bo\n", r);
+		return r;
+	}
+
+	r = radeon_bo_pin(rdev->uvd.vcpu_bo, RADEON_GEM_DOMAIN_VRAM,
+			  &rdev->uvd.gpu_addr);
+	if (r) {
+		radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+		radeon_bo_unref(&rdev->uvd.vcpu_bo);
+		dev_err(rdev->dev, "(%d) UVD bo pin failed\n", r);
+		return r;
+	}
+
+	r = radeon_bo_kmap(rdev->uvd.vcpu_bo, &rdev->uvd.cpu_addr);
+	if (r) {
+		dev_err(rdev->dev, "(%d) UVD map failed\n", r);
+		return r;
+	}
+
+	radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+
+	return 0;
+}
+
+void radeon_uvd_force_into_uvd_segment(struct radeon_bo *rbo)
+{
+	rbo->placement.fpfn = 0 >> PAGE_SHIFT;
+	rbo->placement.lpfn = (256 * 1024 * 1024) >> PAGE_SHIFT;
+}
+
+void radeon_uvd_free_handles(struct radeon_device *rdev, struct drm_file *filp)
+{
+	int i, r;
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (rdev->uvd.filp[i] == filp) {
+			uint32_t handle = atomic_read(&rdev->uvd.handles[i]);
+			struct radeon_fence *fence;
+
+			r = radeon_uvd_get_destroy_msg(rdev,
+				R600_RING_TYPE_UVD_INDEX, handle, &fence);
+			if (r) {
+				DRM_ERROR("Error destroying UVD (%d)!\n", r);
+				continue;
+			}
+
+			radeon_fence_wait(fence, false);
+			radeon_fence_unref(&fence);
+
+			rdev->uvd.filp[i] = NULL;
+			atomic_set(&rdev->uvd.handles[i], 0);
+		}
+	}
+}
+
+static int radeon_uvd_cs_msg_decode(uint32_t *msg, unsigned buf_sizes[])
+{
+	unsigned stream_type = msg[4];
+	unsigned width = msg[6];
+	unsigned height = msg[7];
+	unsigned dpb_size = msg[9];
+	unsigned pitch = msg[28];
+
+	unsigned width_in_mb = width / 16;
+	unsigned height_in_mb = ALIGN(height / 16, 2);
+
+	unsigned image_size, tmp, min_dpb_size;
+
+	image_size = width * height;
+	image_size += image_size / 2;
+	image_size = ALIGN(image_size, 1024);
+
+	switch (stream_type) {
+	case 0: /* H264 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 17;
+
+		/* macroblock context buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 17 * 192;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 32;
+		break;
+
+	case 1: /* VC1 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CONTEXT_BUFFER */
+		min_dpb_size += width_in_mb * height_in_mb * 128;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * 64;
+
+		/* DB surface buffer */
+		min_dpb_size += width_in_mb * 128;
+
+		/* BP */
+		tmp = max(width_in_mb, height_in_mb);
+		min_dpb_size += ALIGN(tmp * 7 * 16, 64);
+		break;
+
+	case 3: /* MPEG2 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+		break;
+
+	case 4: /* MPEG4 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CM */
+		min_dpb_size += width_in_mb * height_in_mb * 64;
+
+		/* IT surface buffer */
+		min_dpb_size += ALIGN(width_in_mb * height_in_mb * 32, 64);
+		break;
+
+	default:
+		DRM_ERROR("UVD codec not handled %d!\n", stream_type);
+		return -EINVAL;
+	}
+
+	if (width > pitch) {
+		DRM_ERROR("Invalid UVD decoding target pitch!\n");
+		return -EINVAL;
+	}
+
+	if (dpb_size < min_dpb_size) {
+		DRM_ERROR("Invalid dpb_size in UVD message (%d / %d)!\n",
+			  dpb_size, min_dpb_size);
+		return -EINVAL;
+	}
+
+	buf_sizes[0x1] = dpb_size;
+	buf_sizes[0x2] = image_size;
+	return 0;
+}
+
+static int radeon_uvd_cs_msg(struct radeon_cs_parser *p, struct radeon_bo *bo,
+			     unsigned offset, unsigned buf_sizes[])
+{
+	int32_t *msg, msg_type, handle;
+	void *ptr;
+
+	int i, r;
+
+	if (offset & 0x3F) {
+		DRM_ERROR("UVD messages must be 64 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	r = radeon_bo_kmap(bo, &ptr);
+	if (r)
+		return r;
+
+	msg = ptr + offset;
+
+	msg_type = msg[1];
+	handle = msg[2];
+
+	if (handle == 0) {
+		DRM_ERROR("Invalid UVD handle!\n");
+		return -EINVAL;
+	}
+
+	if (msg_type == 1) {
+		/* it's a decode msg, calc buffer sizes */
+		r = radeon_uvd_cs_msg_decode(msg, buf_sizes);
+		radeon_bo_kunmap(bo);
+		if (r)
+			return r;
+
+	} else if (msg_type == 2) {
+		/* it's a destroy msg, free the handle */
+		for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i)
+			atomic_cmpxchg(&p->rdev->uvd.handles[i], handle, 0);
+		radeon_bo_kunmap(bo);
+		return 0;
+	} else {
+		/* it's a create msg, no special handling needed */
+		radeon_bo_kunmap(bo);
+	}
+
+	/* create or decode, validate the handle */
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (atomic_read(&p->rdev->uvd.handles[i]) == handle)
+			return 0;
+	}
+
+	/* handle not found try to alloc a new one */
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (!atomic_cmpxchg(&p->rdev->uvd.handles[i], 0, handle)) {
+			p->rdev->uvd.filp[i] = p->filp;
+			return 0;
+		}
+	}
+
+	DRM_ERROR("No more free UVD handles!\n");
+	return -EINVAL;
+}
+
+static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p,
+			       int data0, int data1,
+			       unsigned buf_sizes[])
+{
+	struct radeon_cs_chunk *relocs_chunk;
+	struct radeon_cs_reloc *reloc;
+	unsigned idx, cmd, offset;
+	uint64_t start, end;
+	int r;
+
+	relocs_chunk = &p->chunks[p->chunk_relocs_idx];
+	offset = radeon_get_ib_value(p, data0);
+	idx = radeon_get_ib_value(p, data1);
+	if (idx >= relocs_chunk->length_dw) {
+		DRM_ERROR("Relocs at %d after relocations chunk end %d !\n",
+			  idx, relocs_chunk->length_dw);
+		return -EINVAL;
+	}
+
+	reloc = p->relocs_ptr[(idx / 4)];
+	start = reloc->lobj.gpu_offset;
+	end = start + radeon_bo_size(reloc->robj);
+	start += offset;
+
+	p->ib.ptr[data0] = start & 0xFFFFFFFF;
+	p->ib.ptr[data1] = start >> 32;
+
+	cmd = radeon_get_ib_value(p, p->idx) >> 1;
+
+	if (cmd < 0x4) {
+		if ((end - start) < buf_sizes[cmd]) {
+			DRM_ERROR("buffer to small (%d / %d)!\n",
+				  (unsigned)(end - start), buf_sizes[cmd]);
+			return -EINVAL;
+		}
+
+	} else if (cmd != 0x100) {
+		DRM_ERROR("invalid UVD command %X!\n", cmd);
+		return -EINVAL;
+	}
+
+	if (cmd == 0) {
+		if (end & 0xFFFFFFFFF0000000) {
+			DRM_ERROR("msg buffer %LX-%LX out of 256MB segment!\n",
+				  start, end);
+			return -EINVAL;
+		}
+
+		r = radeon_uvd_cs_msg(p, reloc->robj, offset, buf_sizes);
+		if (r)
+			return r;
+	}
+
+	if ((start & 0xFFFFFFFFF0000000) != (end & 0xFFFFFFFFF0000000)) {
+		DRM_ERROR("reloc %LX-%LX crossing 256MB boundary!\n",
+			  start, end);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int radeon_uvd_cs_reg(struct radeon_cs_parser *p,
+			     struct radeon_cs_packet *pkt,
+			     int *data0, int *data1,
+			     unsigned buf_sizes[])
+{
+	int i, r;
+
+	p->idx++;
+	for (i = 0; i <= pkt->count; ++i) {
+		switch (pkt->reg + i*4) {
+		case UVD_GPCOM_VCPU_DATA0:
+			*data0 = p->idx;
+			break;
+		case UVD_GPCOM_VCPU_DATA1:
+			*data1 = p->idx;
+			break;
+		case UVD_GPCOM_VCPU_CMD:
+			r = radeon_uvd_cs_reloc(p, *data0, *data1, buf_sizes);
+			if (r)
+				return r;
+			break;
+		case UVD_ENGINE_CNTL:
+			break;
+		default:
+			DRM_ERROR("Invalid reg 0x%X!\n",
+				  pkt->reg + i*4);
+			return -EINVAL;
+		}
+		p->idx++;
+	}
+	return 0;
+}
+
+int radeon_uvd_cs_parse(struct radeon_cs_parser *p)
+{
+	struct radeon_cs_packet pkt;
+	int r, data0 = 0, data1 = 0;
+
+	/* minimum buffer sizes */
+	unsigned buf_sizes[] = {
+		[0x00000000]	=	2048,
+		[0x00000001]	=	32 * 1024 * 1024,
+		[0x00000002]	=	2048 * 1152 * 3,
+		[0x00000003]	=	2048,
+	};
+
+	if (p->chunks[p->chunk_ib_idx].length_dw % 16) {
+		DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n",
+			  p->chunks[p->chunk_ib_idx].length_dw);
+		return -EINVAL;
+	}
+
+	if (p->chunk_relocs_idx == -1) {
+		DRM_ERROR("No relocation chunk !\n");
+		return -EINVAL;
+	}
+
+
+	do {
+		r = radeon_cs_packet_parse(p, &pkt, p->idx);
+		if (r)
+			return r;
+		switch (pkt.type) {
+		case RADEON_PACKET_TYPE0:
+			r = radeon_uvd_cs_reg(p, &pkt, &data0,
+					      &data1, buf_sizes);
+			if (r)
+				return r;
+			break;
+		case RADEON_PACKET_TYPE2:
+			p->idx += pkt.count + 2;
+			break;
+		default:
+			DRM_ERROR("Unknown packet type %d !\n", pkt.type);
+			return -EINVAL;
+		}
+	} while (p->idx < p->chunks[p->chunk_ib_idx].length_dw);
+	return 0;
+}
+
+static int radeon_uvd_send_msg(struct radeon_device *rdev,
+			       int ring, struct radeon_bo *bo,
+			       struct radeon_fence **fence)
+{
+	struct ttm_validate_buffer tv;
+	struct list_head head;
+	struct radeon_ib ib;
+	uint64_t addr;
+	int i, r;
+
+	memset(&tv, 0, sizeof(tv));
+	tv.bo = &bo->tbo;
+
+	INIT_LIST_HEAD(&head);
+	list_add(&tv.head, &head);
+
+	r = ttm_eu_reserve_buffers(&head);
+	if (r)
+		return r;
+
+	radeon_ttm_placement_from_domain(bo, RADEON_GEM_DOMAIN_VRAM);
+	radeon_uvd_force_into_uvd_segment(bo);
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+
+	r = radeon_ib_get(rdev, ring, &ib, NULL, 16);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+
+	addr = radeon_bo_gpu_offset(bo);
+	ib.ptr[0] = PACKET0(UVD_GPCOM_VCPU_DATA0, 0);
+	ib.ptr[1] = addr;
+	ib.ptr[2] = PACKET0(UVD_GPCOM_VCPU_DATA1, 0);
+	ib.ptr[3] = addr >> 32;
+	ib.ptr[4] = PACKET0(UVD_GPCOM_VCPU_CMD, 0);
+	ib.ptr[5] = 0;
+	for (i = 6; i < 16; ++i)
+		ib.ptr[i] = PACKET2(0);
+	ib.length_dw = 16;
+
+	r = radeon_ib_schedule(rdev, &ib, NULL);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+	ttm_eu_fence_buffer_objects(&head, ib.fence);
+
+	if (fence)
+		*fence = radeon_fence_ref(ib.fence);
+
+	radeon_ib_free(rdev, &ib);
+	radeon_bo_unref(&bo);
+	return 0;
+}
+
+/* multiple fence commands without any stream commands in between can
+   crash the vcpu so just try to emmit a dummy create/destroy msg to
+   avoid this */
+int radeon_uvd_get_create_msg(struct radeon_device *rdev, int ring,
+			      uint32_t handle, struct radeon_fence **fence)
+{
+	struct radeon_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = radeon_bo_create(rdev, 1024, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &bo);
+	if (r)
+		return r;
+
+	r = radeon_bo_reserve(bo, false);
+	if (r) {
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	r = radeon_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		radeon_bo_unreserve(bo);
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD create msg */
+	msg[0] = 0x00000de4;
+	msg[1] = 0x00000000;
+	msg[2] = handle;
+	msg[3] = 0x00000000;
+	msg[4] = 0x00000000;
+	msg[5] = 0x00000000;
+	msg[6] = 0x00000000;
+	msg[7] = 0x00000780;
+	msg[8] = 0x00000440;
+	msg[9] = 0x00000000;
+	msg[10] = 0x01b37000;
+	for (i = 11; i < 1024; ++i)
+		msg[i] = 0x0;
+
+	radeon_bo_kunmap(bo);
+	radeon_bo_unreserve(bo);
+
+	return radeon_uvd_send_msg(rdev, ring, bo, fence);
+}
+
+int radeon_uvd_get_destroy_msg(struct radeon_device *rdev, int ring,
+			       uint32_t handle, struct radeon_fence **fence)
+{
+	struct radeon_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = radeon_bo_create(rdev, 1024, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &bo);
+	if (r)
+		return r;
+
+	r = radeon_bo_reserve(bo, false);
+	if (r) {
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	r = radeon_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		radeon_bo_unreserve(bo);
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD destroy msg */
+	msg[0] = 0x00000de4;
+	msg[1] = 0x00000002;
+	msg[2] = handle;
+	msg[3] = 0x00000000;
+	for (i = 4; i < 1024; ++i)
+		msg[i] = 0x0;
+
+	radeon_bo_kunmap(bo);
+	radeon_bo_unreserve(bo);
+
+	return radeon_uvd_send_msg(rdev, ring, bo, fence);
+}
diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c
index d4d9be17cfb9..a47e7b903cbc 100644
--- a/drivers/gpu/drm/radeon/rv770.c
+++ b/drivers/gpu/drm/radeon/rv770.c
@@ -68,6 +68,105 @@ u32 rv770_get_xclk(struct radeon_device *rdev)
 	return reference_clock;
 }
 
+int rv770_uvd_resume(struct radeon_device *rdev)
+{
+	uint64_t addr;
+	uint32_t chip_id, size;
+	int r;
+
+	r = radeon_uvd_resume(rdev);
+	if (r)
+		return r;
+
+	/* programm the VCPU memory controller bits 0-27 */
+	addr = rdev->uvd.gpu_addr >> 3;
+	size = RADEON_GPU_PAGE_ALIGN(rdev->uvd_fw->size + 4) >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET0, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = RADEON_UVD_STACK_SIZE >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET1, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = RADEON_UVD_HEAP_SIZE >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET2, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE2, size);
+
+	/* bits 28-31 */
+	addr = (rdev->uvd.gpu_addr >> 28) & 0xF;
+	WREG32(UVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
+
+	/* bits 32-39 */
+	addr = (rdev->uvd.gpu_addr >> 32) & 0xFF;
+	WREG32(UVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
+
+	/* tell firmware which hardware it is running on */
+	switch (rdev->family) {
+	default:
+		return -EINVAL;
+	case CHIP_RV710:
+		chip_id = 0x01000005;
+		break;
+	case CHIP_RV730:
+		chip_id = 0x01000006;
+		break;
+	case CHIP_RV740:
+		chip_id = 0x01000007;
+		break;
+	case CHIP_CYPRESS:
+	case CHIP_HEMLOCK:
+		chip_id = 0x01000008;
+		break;
+	case CHIP_JUNIPER:
+		chip_id = 0x01000009;
+		break;
+	case CHIP_REDWOOD:
+		chip_id = 0x0100000a;
+		break;
+	case CHIP_CEDAR:
+		chip_id = 0x0100000b;
+		break;
+	case CHIP_SUMO:
+		chip_id = 0x0100000c;
+		break;
+	case CHIP_SUMO2:
+		chip_id = 0x0100000d;
+		break;
+	case CHIP_PALM:
+		chip_id = 0x0100000e;
+		break;
+	case CHIP_CAYMAN:
+		chip_id = 0x0100000f;
+		break;
+	case CHIP_BARTS:
+		chip_id = 0x01000010;
+		break;
+	case CHIP_TURKS:
+		chip_id = 0x01000011;
+		break;
+	case CHIP_CAICOS:
+		chip_id = 0x01000012;
+		break;
+	case CHIP_TAHITI:
+		chip_id = 0x01000014;
+		break;
+	case CHIP_VERDE:
+		chip_id = 0x01000015;
+		break;
+	case CHIP_PITCAIRN:
+		chip_id = 0x01000016;
+		break;
+	case CHIP_ARUBA:
+		chip_id = 0x01000017;
+		break;
+	}
+	WREG32(UVD_VCPU_CHIP_ID, chip_id);
+
+	return 0;
+}
+
 u32 rv770_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
@@ -1040,6 +1139,17 @@ static int rv770_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = r600_irq_init(rdev);
 	if (r) {
@@ -1074,6 +1184,19 @@ static int rv770_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -1115,6 +1238,7 @@ int rv770_resume(struct radeon_device *rdev)
 int rv770_suspend(struct radeon_device *rdev)
 {
 	r600_audio_fini(rdev);
+	radeon_uvd_suspend(rdev);
 	r700_cp_stop(rdev);
 	r600_dma_stop(rdev);
 	r600_irq_suspend(rdev);
@@ -1190,6 +1314,13 @@ int rv770_init(struct radeon_device *rdev)
 	rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL;
 	r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_obj = NULL;
+		r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_UVD_INDEX],
+			       4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -1224,6 +1355,7 @@ void rv770_fini(struct radeon_device *rdev)
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
 	rv770_pcie_gart_fini(rdev);
+	radeon_uvd_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
 	radeon_fence_driver_fini(rdev);
diff --git a/drivers/gpu/drm/radeon/rv770d.h b/drivers/gpu/drm/radeon/rv770d.h
index c55f950a4af7..da158b541720 100644
--- a/drivers/gpu/drm/radeon/rv770d.h
+++ b/drivers/gpu/drm/radeon/rv770d.h
@@ -671,4 +671,18 @@
 #       define TARGET_LINK_SPEED_MASK                     (0xf << 0)
 #       define SELECTABLE_DEEMPHASIS                      (1 << 6)
 
+/* UVD */
+#define UVD_LMI_EXT40_ADDR				0xf498
+#define UVD_VCPU_CHIP_ID				0xf4d4
+#define UVD_VCPU_CACHE_OFFSET0				0xf4d8
+#define UVD_VCPU_CACHE_SIZE0				0xf4dc
+#define UVD_VCPU_CACHE_OFFSET1				0xf4e0
+#define UVD_VCPU_CACHE_SIZE1				0xf4e4
+#define UVD_VCPU_CACHE_OFFSET2				0xf4e8
+#define UVD_VCPU_CACHE_SIZE2				0xf4ec
+#define UVD_LMI_ADDR_EXT				0xf594
+
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+
 #endif
diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c
index ace45da91434..3e9782dc35bf 100644
--- a/drivers/gpu/drm/radeon/si.c
+++ b/drivers/gpu/drm/radeon/si.c
@@ -4333,6 +4333,16 @@ static int si_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = si_irq_init(rdev);
 	if (r) {
@@ -4390,6 +4400,18 @@ static int si_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -4433,6 +4455,8 @@ int si_suspend(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	si_cp_enable(rdev, false);
 	cayman_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
+	radeon_uvd_suspend(rdev);
 	si_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	si_pcie_gart_disable(rdev);
@@ -4518,6 +4542,13 @@ int si_init(struct radeon_device *rdev)
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+		ring->ring_obj = NULL;
+		r600_ring_init(rdev, ring, 4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -4566,6 +4597,7 @@ void si_fini(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
+	radeon_uvd_fini(rdev);
 	si_pcie_gart_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
diff --git a/drivers/gpu/drm/radeon/sid.h b/drivers/gpu/drm/radeon/sid.h
index f84cff0aafcc..1fb8ee2c45dc 100644
--- a/drivers/gpu/drm/radeon/sid.h
+++ b/drivers/gpu/drm/radeon/sid.h
@@ -799,6 +799,12 @@
 #       define THREAD_TRACE_FLUSH                       (54 << 0)
 #       define THREAD_TRACE_FINISH                      (55 << 0)
 
+/*
+ * UVD
+ */
+#define UVD_RBC_RB_RPTR					0xF690
+#define UVD_RBC_RB_WPTR					0xF694
+
 /*
  * PM4
  */
diff --git a/include/uapi/drm/radeon_drm.h b/include/uapi/drm/radeon_drm.h
index 6fd25563f301..b1c1a2a4fe33 100644
--- a/include/uapi/drm/radeon_drm.h
+++ b/include/uapi/drm/radeon_drm.h
@@ -918,6 +918,7 @@ struct drm_radeon_gem_va {
 #define RADEON_CS_RING_GFX          0
 #define RADEON_CS_RING_COMPUTE      1
 #define RADEON_CS_RING_DMA          2
+#define RADEON_CS_RING_UVD          3
 /* The third dword of RADEON_CHUNK_ID_FLAGS is a sint32 that sets the priority */
 /* 0 = normal, + = higher priority, - = lower priority */
 
-- 
cgit v1.2.3


From 902aaef6c698ce0f04a6acab2d5396519d5de330 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Tue, 9 Apr 2013 10:35:42 -0400
Subject: drm/radeon: add ring working query
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Add new ioctl option and bumb minor version number.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/radeon/radeon_drv.c |  3 ++-
 drivers/gpu/drm/radeon/radeon_kms.c | 17 +++++++++++++++++
 include/uapi/drm/radeon_drm.h       |  2 ++
 3 files changed, 21 insertions(+), 1 deletion(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c
index b500bbc3e411..6e1b2e66e837 100644
--- a/drivers/gpu/drm/radeon/radeon_drv.c
+++ b/drivers/gpu/drm/radeon/radeon_drv.c
@@ -72,9 +72,10 @@
  *   2.29.0 - R500 FP16 color clear registers
  *   2.30.0 - fix for FMASK texturing
  *   2.31.0 - Add fastfb support for rs690
+ *   2.32.0 - new info request for rings working
  */
 #define KMS_DRIVER_MAJOR	2
-#define KMS_DRIVER_MINOR	31
+#define KMS_DRIVER_MINOR	32
 #define KMS_DRIVER_PATCHLEVEL	0
 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
 int radeon_driver_unload_kms(struct drm_device *dev);
diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c
index 8365c75b31b6..3037c38d9aa9 100644
--- a/drivers/gpu/drm/radeon/radeon_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_kms.c
@@ -379,6 +379,23 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	case RADEON_INFO_FASTFB_WORKING:
 		value = rdev->fastfb_working;
 		break;
+	case RADEON_INFO_RING_WORKING:
+		switch (value) {
+		case RADEON_CS_RING_GFX:
+		case RADEON_CS_RING_COMPUTE:
+			value = rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready;
+			break;
+		case RADEON_CS_RING_DMA:
+			value = rdev->ring[R600_RING_TYPE_DMA_INDEX].ready;
+			value |= rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready;
+			break;
+		case RADEON_CS_RING_UVD:
+			value = rdev->ring[R600_RING_TYPE_UVD_INDEX].ready;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
 	default:
 		DRM_DEBUG_KMS("Invalid request %d\n", info->request);
 		return -EINVAL;
diff --git a/include/uapi/drm/radeon_drm.h b/include/uapi/drm/radeon_drm.h
index b1c1a2a4fe33..05ed0107584d 100644
--- a/include/uapi/drm/radeon_drm.h
+++ b/include/uapi/drm/radeon_drm.h
@@ -975,6 +975,8 @@ struct drm_radeon_cs {
 #define RADEON_INFO_MAX_SH_PER_SE	0x13
 /* fast fb access is enabled */
 #define RADEON_INFO_FASTFB_WORKING	0x14
+/* query if a RADEON_CS_RING_* submission is supported */
+#define RADEON_INFO_RING_WORKING	0x15
 
 
 struct drm_radeon_info {
-- 
cgit v1.2.3


From 64d7b8bed851f55a17d15ec6cc60233c85f84357 Mon Sep 17 00:00:00 2001
From: Jerome Glisse <jglisse@redhat.com>
Date: Tue, 9 Apr 2013 11:17:08 -0400
Subject: drm/radeon: add si tile mode array query v3

Allow userspace to query for the tile mode array so userspace can properly
compute surface pitch and alignment requirement depending on tiling.

v2: Make strict aliasing safer by casting to char when copying
v3: merge fix from Christian

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h     |   1 +
 drivers/gpu/drm/radeon/radeon_drv.c |   3 +-
 drivers/gpu/drm/radeon/radeon_kms.c | 174 +++++++++++++++++++-----------------
 drivers/gpu/drm/radeon/si.c         |   2 +
 include/uapi/drm/radeon_drm.h       |  20 +++++
 5 files changed, 119 insertions(+), 81 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 394b20fa27dc..18904fb83d3a 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -1483,6 +1483,7 @@ struct si_asic {
 	unsigned multi_gpu_tile_size;
 
 	unsigned tile_config;
+	uint32_t tile_mode_array[32];
 };
 
 union radeon_asic_config {
diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c
index 6e1b2e66e837..d33f484ace48 100644
--- a/drivers/gpu/drm/radeon/radeon_drv.c
+++ b/drivers/gpu/drm/radeon/radeon_drv.c
@@ -73,9 +73,10 @@
  *   2.30.0 - fix for FMASK texturing
  *   2.31.0 - Add fastfb support for rs690
  *   2.32.0 - new info request for rings working
+ *   2.33.0 - Add SI tiling mode array query
  */
 #define KMS_DRIVER_MAJOR	2
-#define KMS_DRIVER_MINOR	32
+#define KMS_DRIVER_MINOR	33
 #define KMS_DRIVER_PATCHLEVEL	0
 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
 int radeon_driver_unload_kms(struct drm_device *dev);
diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c
index 3037c38d9aa9..a24b9ba4de73 100644
--- a/drivers/gpu/drm/radeon/radeon_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_kms.c
@@ -176,80 +176,65 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	struct radeon_device *rdev = dev->dev_private;
 	struct drm_radeon_info *info = data;
 	struct radeon_mode_info *minfo = &rdev->mode_info;
-	uint32_t value, *value_ptr;
-	uint64_t value64, *value_ptr64;
+	uint32_t *value, value_tmp, *value_ptr, value_size;
+	uint64_t value64;
 	struct drm_crtc *crtc;
 	int i, found;
 
-	/* TIMESTAMP is a 64-bit value, needs special handling. */
-	if (info->request == RADEON_INFO_TIMESTAMP) {
-		if (rdev->family >= CHIP_R600) {
-			value_ptr64 = (uint64_t*)((unsigned long)info->value);
-			value64 = radeon_get_gpu_clock_counter(rdev);
-
-			if (DRM_COPY_TO_USER(value_ptr64, &value64, sizeof(value64))) {
-				DRM_ERROR("copy_to_user %s:%u\n", __func__, __LINE__);
-				return -EFAULT;
-			}
-			return 0;
-		} else {
-			DRM_DEBUG_KMS("timestamp is r6xx+ only!\n");
-			return -EINVAL;
-		}
-	}
-
 	value_ptr = (uint32_t *)((unsigned long)info->value);
-	if (DRM_COPY_FROM_USER(&value, value_ptr, sizeof(value))) {
-		DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
-		return -EFAULT;
-	}
+	value = &value_tmp;
+	value_size = sizeof(uint32_t);
 
 	switch (info->request) {
 	case RADEON_INFO_DEVICE_ID:
-		value = dev->pci_device;
+		*value = dev->pci_device;
 		break;
 	case RADEON_INFO_NUM_GB_PIPES:
-		value = rdev->num_gb_pipes;
+		*value = rdev->num_gb_pipes;
 		break;
 	case RADEON_INFO_NUM_Z_PIPES:
-		value = rdev->num_z_pipes;
+		*value = rdev->num_z_pipes;
 		break;
 	case RADEON_INFO_ACCEL_WORKING:
 		/* xf86-video-ati 6.13.0 relies on this being false for evergreen */
 		if ((rdev->family >= CHIP_CEDAR) && (rdev->family <= CHIP_HEMLOCK))
-			value = false;
+			*value = false;
 		else
-			value = rdev->accel_working;
+			*value = rdev->accel_working;
 		break;
 	case RADEON_INFO_CRTC_FROM_ID:
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
 		for (i = 0, found = 0; i < rdev->num_crtc; i++) {
 			crtc = (struct drm_crtc *)minfo->crtcs[i];
-			if (crtc && crtc->base.id == value) {
+			if (crtc && crtc->base.id == *value) {
 				struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
-				value = radeon_crtc->crtc_id;
+				*value = radeon_crtc->crtc_id;
 				found = 1;
 				break;
 			}
 		}
 		if (!found) {
-			DRM_DEBUG_KMS("unknown crtc id %d\n", value);
+			DRM_DEBUG_KMS("unknown crtc id %d\n", *value);
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_ACCEL_WORKING2:
-		value = rdev->accel_working;
+		*value = rdev->accel_working;
 		break;
 	case RADEON_INFO_TILING_CONFIG:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.tile_config;
+			*value = rdev->config.si.tile_config;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.tile_config;
+			*value = rdev->config.cayman.tile_config;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.tile_config;
+			*value = rdev->config.evergreen.tile_config;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.tile_config;
+			*value = rdev->config.rv770.tile_config;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.tile_config;
+			*value = rdev->config.r600.tile_config;
 		else {
 			DRM_DEBUG_KMS("tiling config is r6xx+ only!\n");
 			return -EINVAL;
@@ -262,73 +247,81 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		 *
 		 * When returning, the value is 1 if filp owns hyper-z access,
 		 * 0 otherwise. */
-		if (value >= 2) {
-			DRM_DEBUG_KMS("WANT_HYPERZ: invalid value %d\n", value);
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		if (*value >= 2) {
+			DRM_DEBUG_KMS("WANT_HYPERZ: invalid value %d\n", *value);
 			return -EINVAL;
 		}
-		radeon_set_filp_rights(dev, &rdev->hyperz_filp, filp, &value);
+		radeon_set_filp_rights(dev, &rdev->hyperz_filp, filp, value);
 		break;
 	case RADEON_INFO_WANT_CMASK:
 		/* The same logic as Hyper-Z. */
-		if (value >= 2) {
-			DRM_DEBUG_KMS("WANT_CMASK: invalid value %d\n", value);
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		if (*value >= 2) {
+			DRM_DEBUG_KMS("WANT_CMASK: invalid value %d\n", *value);
 			return -EINVAL;
 		}
-		radeon_set_filp_rights(dev, &rdev->cmask_filp, filp, &value);
+		radeon_set_filp_rights(dev, &rdev->cmask_filp, filp, value);
 		break;
 	case RADEON_INFO_CLOCK_CRYSTAL_FREQ:
 		/* return clock value in KHz */
 		if (rdev->asic->get_xclk)
-			value = radeon_get_xclk(rdev) * 10;
+			*value = radeon_get_xclk(rdev) * 10;
 		else
-			value = rdev->clock.spll.reference_freq * 10;
+			*value = rdev->clock.spll.reference_freq * 10;
 		break;
 	case RADEON_INFO_NUM_BACKENDS:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_backends_per_se *
+			*value = rdev->config.si.max_backends_per_se *
 				rdev->config.si.max_shader_engines;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_backends_per_se *
+			*value = rdev->config.cayman.max_backends_per_se *
 				rdev->config.cayman.max_shader_engines;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_backends;
+			*value = rdev->config.evergreen.max_backends;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_backends;
+			*value = rdev->config.rv770.max_backends;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_backends;
+			*value = rdev->config.r600.max_backends;
 		else {
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_NUM_TILE_PIPES:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_tile_pipes;
+			*value = rdev->config.si.max_tile_pipes;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_tile_pipes;
+			*value = rdev->config.cayman.max_tile_pipes;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_tile_pipes;
+			*value = rdev->config.evergreen.max_tile_pipes;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_tile_pipes;
+			*value = rdev->config.rv770.max_tile_pipes;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_tile_pipes;
+			*value = rdev->config.r600.max_tile_pipes;
 		else {
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_FUSION_GART_WORKING:
-		value = 1;
+		*value = 1;
 		break;
 	case RADEON_INFO_BACKEND_MAP:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.backend_map;
+			*value = rdev->config.si.backend_map;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.backend_map;
+			*value = rdev->config.cayman.backend_map;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.backend_map;
+			*value = rdev->config.evergreen.backend_map;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.backend_map;
+			*value = rdev->config.rv770.backend_map;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.backend_map;
+			*value = rdev->config.r600.backend_map;
 		else {
 			return -EINVAL;
 		}
@@ -337,70 +330,91 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		/* this is where we report if vm is supported or not */
 		if (rdev->family < CHIP_CAYMAN)
 			return -EINVAL;
-		value = RADEON_VA_RESERVED_SIZE;
+		*value = RADEON_VA_RESERVED_SIZE;
 		break;
 	case RADEON_INFO_IB_VM_MAX_SIZE:
 		/* this is where we report if vm is supported or not */
 		if (rdev->family < CHIP_CAYMAN)
 			return -EINVAL;
-		value = RADEON_IB_VM_MAX_SIZE;
+		*value = RADEON_IB_VM_MAX_SIZE;
 		break;
 	case RADEON_INFO_MAX_PIPES:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_cu_per_sh;
+			*value = rdev->config.si.max_cu_per_sh;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_pipes_per_simd;
+			*value = rdev->config.cayman.max_pipes_per_simd;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_pipes;
+			*value = rdev->config.evergreen.max_pipes;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_pipes;
+			*value = rdev->config.rv770.max_pipes;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_pipes;
+			*value = rdev->config.r600.max_pipes;
 		else {
 			return -EINVAL;
 		}
 		break;
+	case RADEON_INFO_TIMESTAMP:
+		if (rdev->family < CHIP_R600) {
+			DRM_DEBUG_KMS("timestamp is r6xx+ only!\n");
+			return -EINVAL;
+		}
+		value = (uint32_t*)&value64;
+		value_size = sizeof(uint64_t);
+		value64 = radeon_get_gpu_clock_counter(rdev);
+		break;
 	case RADEON_INFO_MAX_SE:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_shader_engines;
+			*value = rdev->config.si.max_shader_engines;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_shader_engines;
+			*value = rdev->config.cayman.max_shader_engines;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.num_ses;
+			*value = rdev->config.evergreen.num_ses;
 		else
-			value = 1;
+			*value = 1;
 		break;
 	case RADEON_INFO_MAX_SH_PER_SE:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_sh_per_se;
+			*value = rdev->config.si.max_sh_per_se;
 		else
 			return -EINVAL;
 		break;
 	case RADEON_INFO_FASTFB_WORKING:
-		value = rdev->fastfb_working;
+		*value = rdev->fastfb_working;
 		break;
 	case RADEON_INFO_RING_WORKING:
-		switch (value) {
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		switch (*value) {
 		case RADEON_CS_RING_GFX:
 		case RADEON_CS_RING_COMPUTE:
-			value = rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready;
+			*value = rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready;
 			break;
 		case RADEON_CS_RING_DMA:
-			value = rdev->ring[R600_RING_TYPE_DMA_INDEX].ready;
-			value |= rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready;
+			*value = rdev->ring[R600_RING_TYPE_DMA_INDEX].ready;
+			*value |= rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready;
 			break;
 		case RADEON_CS_RING_UVD:
-			value = rdev->ring[R600_RING_TYPE_UVD_INDEX].ready;
+			*value = rdev->ring[R600_RING_TYPE_UVD_INDEX].ready;
 			break;
 		default:
 			return -EINVAL;
 		}
 		break;
+	case RADEON_INFO_SI_TILE_MODE_ARRAY:
+		if (rdev->family < CHIP_TAHITI) {
+			DRM_DEBUG_KMS("tile mode array is si only!\n");
+			return -EINVAL;
+		}
+		value = rdev->config.si.tile_mode_array;
+		value_size = sizeof(uint32_t)*32;
+		break;
 	default:
 		DRM_DEBUG_KMS("Invalid request %d\n", info->request);
 		return -EINVAL;
 	}
-	if (DRM_COPY_TO_USER(value_ptr, &value, sizeof(uint32_t))) {
+	if (DRM_COPY_TO_USER(value_ptr, (char*)value, value_size)) {
 		DRM_ERROR("copy_to_user %s:%u\n", __func__, __LINE__);
 		return -EFAULT;
 	}
diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c
index 7eda8303379f..aa2c555ba877 100644
--- a/drivers/gpu/drm/radeon/si.c
+++ b/drivers/gpu/drm/radeon/si.c
@@ -1211,6 +1211,7 @@ static void si_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.si.tile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_TILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else if ((rdev->family == CHIP_VERDE) ||
@@ -1451,6 +1452,7 @@ static void si_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.si.tile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_TILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else
diff --git a/include/uapi/drm/radeon_drm.h b/include/uapi/drm/radeon_drm.h
index 05ed0107584d..321d4ac5c512 100644
--- a/include/uapi/drm/radeon_drm.h
+++ b/include/uapi/drm/radeon_drm.h
@@ -977,6 +977,8 @@ struct drm_radeon_cs {
 #define RADEON_INFO_FASTFB_WORKING	0x14
 /* query if a RADEON_CS_RING_* submission is supported */
 #define RADEON_INFO_RING_WORKING	0x15
+/* SI tile mode array */
+#define RADEON_INFO_SI_TILE_MODE_ARRAY	0x16
 
 
 struct drm_radeon_info {
@@ -985,4 +987,22 @@ struct drm_radeon_info {
 	uint64_t		value;
 };
 
+/* Those correspond to the tile index to use, this is to explicitly state
+ * the API that is implicitly defined by the tile mode array.
+ */
+#define SI_TILE_MODE_COLOR_LINEAR_ALIGNED	8
+#define SI_TILE_MODE_COLOR_1D			13
+#define SI_TILE_MODE_COLOR_1D_SCANOUT		9
+#define SI_TILE_MODE_COLOR_2D_8BPP		14
+#define SI_TILE_MODE_COLOR_2D_16BPP		15
+#define SI_TILE_MODE_COLOR_2D_32BPP		16
+#define SI_TILE_MODE_COLOR_2D_64BPP		17
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_16BPP	11
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_32BPP	12
+#define SI_TILE_MODE_DEPTH_STENCIL_1D		4
+#define SI_TILE_MODE_DEPTH_STENCIL_2D		0
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_2AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_4AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_8AA	2
+
 #endif
-- 
cgit v1.2.3


From 97d103311aab2c652ef37bc3cd157e701057e535 Mon Sep 17 00:00:00 2001
From: Alon Levy <alevy@redhat.com>
Date: Sun, 9 Sep 2012 10:47:32 +0300
Subject: include/drm/drm_fb_helper.h: remove left over function declaration

Signed-off-by: Alon Levy <alevy@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 include/drm/drm_fb_helper.h | 6 ------
 1 file changed, 6 deletions(-)

(limited to 'include')

diff --git a/include/drm/drm_fb_helper.h b/include/drm/drm_fb_helper.h
index c09511625a11..6fb519f5cce0 100644
--- a/include/drm/drm_fb_helper.h
+++ b/include/drm/drm_fb_helper.h
@@ -102,12 +102,6 @@ int drm_fb_helper_pan_display(struct fb_var_screeninfo *var,
 int drm_fb_helper_set_par(struct fb_info *info);
 int drm_fb_helper_check_var(struct fb_var_screeninfo *var,
 			    struct fb_info *info);
-int drm_fb_helper_setcolreg(unsigned regno,
-			    unsigned red,
-			    unsigned green,
-			    unsigned blue,
-			    unsigned transp,
-			    struct fb_info *info);
 
 bool drm_fb_helper_restore_fbdev_mode(struct drm_fb_helper *fb_helper);
 void drm_fb_helper_fill_var(struct fb_info *info, struct drm_fb_helper *fb_helper,
-- 
cgit v1.2.3


From afe6804c045fbd69a1b75c681107b5d6df9190de Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@gmail.com>
Date: Tue, 22 Jan 2013 13:56:04 +1000
Subject: ttm: export functions to allow qxl do its own iomapping

qxl wants to use io mapping like i915 gem does, for now
just export the symbols so the driver can implement atomic
page maps using io mapping.

Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/ttm/ttm_bo_util.c | 13 +++++++++----
 include/drm/ttm/ttm_bo_driver.h   |  4 ++++
 2 files changed, 13 insertions(+), 4 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c
index 8be35c809c7b..af894584dd90 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_util.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
@@ -86,6 +86,7 @@ int ttm_mem_io_lock(struct ttm_mem_type_manager *man, bool interruptible)
 	mutex_lock(&man->io_reserve_mutex);
 	return 0;
 }
+EXPORT_SYMBOL(ttm_mem_io_lock);
 
 void ttm_mem_io_unlock(struct ttm_mem_type_manager *man)
 {
@@ -94,6 +95,7 @@ void ttm_mem_io_unlock(struct ttm_mem_type_manager *man)
 
 	mutex_unlock(&man->io_reserve_mutex);
 }
+EXPORT_SYMBOL(ttm_mem_io_unlock);
 
 static int ttm_mem_io_evict(struct ttm_mem_type_manager *man)
 {
@@ -111,8 +113,9 @@ static int ttm_mem_io_evict(struct ttm_mem_type_manager *man)
 	return 0;
 }
 
-static int ttm_mem_io_reserve(struct ttm_bo_device *bdev,
-			      struct ttm_mem_reg *mem)
+
+int ttm_mem_io_reserve(struct ttm_bo_device *bdev,
+		       struct ttm_mem_reg *mem)
 {
 	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
 	int ret = 0;
@@ -134,9 +137,10 @@ retry:
 	}
 	return ret;
 }
+EXPORT_SYMBOL(ttm_mem_io_reserve);
 
-static void ttm_mem_io_free(struct ttm_bo_device *bdev,
-			    struct ttm_mem_reg *mem)
+void ttm_mem_io_free(struct ttm_bo_device *bdev,
+		     struct ttm_mem_reg *mem)
 {
 	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
 
@@ -149,6 +153,7 @@ static void ttm_mem_io_free(struct ttm_bo_device *bdev,
 		bdev->driver->io_mem_free(bdev, mem);
 
 }
+EXPORT_SYMBOL(ttm_mem_io_free);
 
 int ttm_mem_io_reserve_vm(struct ttm_buffer_object *bo)
 {
diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h
index 0fbd046e7c93..9c8dca79808e 100644
--- a/include/drm/ttm/ttm_bo_driver.h
+++ b/include/drm/ttm/ttm_bo_driver.h
@@ -902,6 +902,10 @@ extern void ttm_bo_unreserve_locked(struct ttm_buffer_object *bo);
  * ttm_bo_util.c
  */
 
+int ttm_mem_io_reserve(struct ttm_bo_device *bdev,
+		       struct ttm_mem_reg *mem);
+void ttm_mem_io_free(struct ttm_bo_device *bdev,
+		     struct ttm_mem_reg *mem);
 /**
  * ttm_bo_move_ttm
  *
-- 
cgit v1.2.3


From f64122c1f6ade301585569863b4b3b18f6e4e332 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@gmail.com>
Date: Mon, 25 Feb 2013 14:47:55 +1000
Subject: drm: add new QXL driver. (v1.4)

QXL is a paravirtual graphics device used by the Spice virtual desktop
interface.

The drivers uses GEM and TTM to manage memory, the qxl hw fencing however
is quite different than normal TTM expects, we have to keep track of a number
of non-linear fence ids per bo that we need to have released by the hardware.

The releases are freed from a workqueue that wakes up and processes the
release ring.

releases are suballocated from a BO, there are 3 release categories, drawables,
surfaces and cursor cmds. The hw also has 3 rings for commands, cursor and release handling.

The hardware also have a surface id tracking mechnaism and the driver encapsulates it completely inside the kernel, userspace never sees the actual hw surface
ids.

This requires a newer version of the QXL userspace driver, so shouldn't be
enabled until that has been placed into your distro of choice.

Authors: Dave Airlie, Alon Levy

v1.1: fixup some issues in the ioctl interface with padding
v1.2: add module device table
v1.3: fix nomodeset, fbcon leak, dumb bo create, release ring irq,
      don't try flush release ring (broken hw), fix -modesetting.
v1.4: fbcon cpu usage reduction + suitable accel flags.

Signed-off-by: Alon Levy <alevy@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/Kconfig           |   2 +
 drivers/gpu/drm/Makefile          |   1 +
 drivers/gpu/drm/qxl/Kconfig       |  10 +
 drivers/gpu/drm/qxl/Makefile      |   9 +
 drivers/gpu/drm/qxl/qxl_cmd.c     | 707 +++++++++++++++++++++++++++
 drivers/gpu/drm/qxl/qxl_debugfs.c | 135 ++++++
 drivers/gpu/drm/qxl/qxl_dev.h     | 879 ++++++++++++++++++++++++++++++++++
 drivers/gpu/drm/qxl/qxl_display.c | 981 ++++++++++++++++++++++++++++++++++++++
 drivers/gpu/drm/qxl/qxl_draw.c    | 390 +++++++++++++++
 drivers/gpu/drm/qxl/qxl_drv.c     | 145 ++++++
 drivers/gpu/drm/qxl/qxl_drv.h     | 566 ++++++++++++++++++++++
 drivers/gpu/drm/qxl/qxl_dumb.c    |  93 ++++
 drivers/gpu/drm/qxl/qxl_fb.c      | 567 ++++++++++++++++++++++
 drivers/gpu/drm/qxl/qxl_fence.c   |  97 ++++
 drivers/gpu/drm/qxl/qxl_gem.c     | 178 +++++++
 drivers/gpu/drm/qxl/qxl_image.c   | 176 +++++++
 drivers/gpu/drm/qxl/qxl_ioctl.c   | 411 ++++++++++++++++
 drivers/gpu/drm/qxl/qxl_irq.c     |  97 ++++
 drivers/gpu/drm/qxl/qxl_kms.c     | 302 ++++++++++++
 drivers/gpu/drm/qxl/qxl_object.c  | 365 ++++++++++++++
 drivers/gpu/drm/qxl/qxl_object.h  | 112 +++++
 drivers/gpu/drm/qxl/qxl_release.c | 307 ++++++++++++
 drivers/gpu/drm/qxl/qxl_ttm.c     | 577 ++++++++++++++++++++++
 include/uapi/drm/Kbuild           |   1 +
 include/uapi/drm/qxl_drm.h        | 152 ++++++
 25 files changed, 7260 insertions(+)
 create mode 100644 drivers/gpu/drm/qxl/Kconfig
 create mode 100644 drivers/gpu/drm/qxl/Makefile
 create mode 100644 drivers/gpu/drm/qxl/qxl_cmd.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_debugfs.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_dev.h
 create mode 100644 drivers/gpu/drm/qxl/qxl_display.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_draw.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_drv.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_drv.h
 create mode 100644 drivers/gpu/drm/qxl/qxl_dumb.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_fb.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_fence.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_gem.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_image.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_ioctl.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_irq.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_kms.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_object.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_object.h
 create mode 100644 drivers/gpu/drm/qxl/qxl_release.c
 create mode 100644 drivers/gpu/drm/qxl/qxl_ttm.c
 create mode 100644 include/uapi/drm/qxl_drm.h

(limited to 'include')

diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index 1e82882da9de..19b8e0d5d910 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -220,3 +220,5 @@ source "drivers/gpu/drm/tegra/Kconfig"
 source "drivers/gpu/drm/omapdrm/Kconfig"
 
 source "drivers/gpu/drm/tilcdc/Kconfig"
+
+source "drivers/gpu/drm/qxl/Kconfig"
diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index 0d59b24f8d23..6a4211521011 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -52,4 +52,5 @@ obj-$(CONFIG_DRM_SHMOBILE) +=shmobile/
 obj-$(CONFIG_DRM_TEGRA) += tegra/
 obj-$(CONFIG_DRM_OMAP)	+= omapdrm/
 obj-$(CONFIG_DRM_TILCDC)	+= tilcdc/
+obj-$(CONFIG_DRM_QXL) += qxl/
 obj-y			+= i2c/
diff --git a/drivers/gpu/drm/qxl/Kconfig b/drivers/gpu/drm/qxl/Kconfig
new file mode 100644
index 000000000000..2f1a57e11140
--- /dev/null
+++ b/drivers/gpu/drm/qxl/Kconfig
@@ -0,0 +1,10 @@
+config DRM_QXL
+	tristate "QXL virtual GPU"
+	depends on DRM && PCI
+	select FB_SYS_FILLRECT
+	select FB_SYS_COPYAREA
+	select FB_SYS_IMAGEBLIT
+        select DRM_KMS_HELPER
+        select DRM_TTM
+	help
+		QXL virtual GPU for Spice virtualization desktop integration. Do not enable this driver unless your distro ships a corresponding X.org QXL driver that can handle kernel modesetting.
diff --git a/drivers/gpu/drm/qxl/Makefile b/drivers/gpu/drm/qxl/Makefile
new file mode 100644
index 000000000000..ea046ba691d2
--- /dev/null
+++ b/drivers/gpu/drm/qxl/Makefile
@@ -0,0 +1,9 @@
+#
+# Makefile for the drm device driver.  This driver provides support for the
+# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher.
+
+ccflags-y := -Iinclude/drm
+
+qxl-y := qxl_drv.o qxl_kms.o qxl_display.o qxl_ttm.o qxl_fb.o qxl_object.o qxl_gem.o qxl_cmd.o qxl_image.o qxl_draw.o qxl_debugfs.o qxl_irq.o qxl_dumb.o qxl_ioctl.o qxl_fence.o qxl_release.o
+
+obj-$(CONFIG_DRM_QXL)+= qxl.o
diff --git a/drivers/gpu/drm/qxl/qxl_cmd.c b/drivers/gpu/drm/qxl/qxl_cmd.c
new file mode 100644
index 000000000000..804b411a60ca
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_cmd.c
@@ -0,0 +1,707 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+/* QXL cmd/ring handling */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+static int qxl_reap_surface_id(struct qxl_device *qdev, int max_to_reap);
+
+struct ring {
+	struct qxl_ring_header      header;
+	uint8_t                     elements[0];
+};
+
+struct qxl_ring {
+	struct ring	       *ring;
+	int			element_size;
+	int			n_elements;
+	int			prod_notify;
+	wait_queue_head_t      *push_event;
+	spinlock_t             lock;
+};
+
+void qxl_ring_free(struct qxl_ring *ring)
+{
+	kfree(ring);
+}
+
+struct qxl_ring *
+qxl_ring_create(struct qxl_ring_header *header,
+		int element_size,
+		int n_elements,
+		int prod_notify,
+		bool set_prod_notify,
+		wait_queue_head_t *push_event)
+{
+	struct qxl_ring *ring;
+
+	ring = kmalloc(sizeof(*ring), GFP_KERNEL);
+	if (!ring)
+		return NULL;
+
+	ring->ring = (struct ring *)header;
+	ring->element_size = element_size;
+	ring->n_elements = n_elements;
+	ring->prod_notify = prod_notify;
+	ring->push_event = push_event;
+	if (set_prod_notify)
+		header->notify_on_prod = ring->n_elements;
+	spin_lock_init(&ring->lock);
+	return ring;
+}
+
+static int qxl_check_header(struct qxl_ring *ring)
+{
+	int ret;
+	struct qxl_ring_header *header = &(ring->ring->header);
+	unsigned long flags;
+	spin_lock_irqsave(&ring->lock, flags);
+	ret = header->prod - header->cons < header->num_items;
+	if (ret == 0)
+		header->notify_on_cons = header->cons + 1;
+	spin_unlock_irqrestore(&ring->lock, flags);
+	return ret;
+}
+
+static int qxl_check_idle(struct qxl_ring *ring)
+{
+	int ret;
+	struct qxl_ring_header *header = &(ring->ring->header);
+	unsigned long flags;
+	spin_lock_irqsave(&ring->lock, flags);
+	ret = header->prod == header->cons;
+	spin_unlock_irqrestore(&ring->lock, flags);
+	return ret;
+}
+
+int qxl_ring_push(struct qxl_ring *ring,
+		  const void *new_elt, bool interruptible)
+{
+	struct qxl_ring_header *header = &(ring->ring->header);
+	uint8_t *elt;
+	int idx, ret;
+	unsigned long flags;
+	spin_lock_irqsave(&ring->lock, flags);
+	if (header->prod - header->cons == header->num_items) {
+		header->notify_on_cons = header->cons + 1;
+		mb();
+		spin_unlock_irqrestore(&ring->lock, flags);
+		if (!drm_can_sleep()) {
+			while (!qxl_check_header(ring))
+				udelay(1);
+		} else {
+			if (interruptible) {
+				ret = wait_event_interruptible(*ring->push_event,
+							       qxl_check_header(ring));
+				if (ret)
+					return ret;
+			} else {
+				wait_event(*ring->push_event,
+					   qxl_check_header(ring));
+			}
+
+		}
+		spin_lock_irqsave(&ring->lock, flags);
+	}
+
+	idx = header->prod & (ring->n_elements - 1);
+	elt = ring->ring->elements + idx * ring->element_size;
+
+	memcpy((void *)elt, new_elt, ring->element_size);
+
+	header->prod++;
+
+	mb();
+
+	if (header->prod == header->notify_on_prod)
+		outb(0, ring->prod_notify);
+
+	spin_unlock_irqrestore(&ring->lock, flags);
+	return 0;
+}
+
+bool qxl_ring_pop(struct qxl_ring *ring,
+		  void *element)
+{
+	volatile struct qxl_ring_header *header = &(ring->ring->header);
+	volatile uint8_t *ring_elt;
+	int idx;
+	unsigned long flags;
+	spin_lock_irqsave(&ring->lock, flags);
+	if (header->cons == header->prod) {
+		header->notify_on_prod = header->cons + 1;
+		spin_unlock_irqrestore(&ring->lock, flags);
+		return false;
+	}
+
+	idx = header->cons & (ring->n_elements - 1);
+	ring_elt = ring->ring->elements + idx * ring->element_size;
+
+	memcpy(element, (void *)ring_elt, ring->element_size);
+
+	header->cons++;
+
+	spin_unlock_irqrestore(&ring->lock, flags);
+	return true;
+}
+
+void qxl_ring_wait_idle(struct qxl_ring *ring)
+{
+	struct qxl_ring_header *header = &(ring->ring->header);
+	unsigned long flags;
+
+	spin_lock_irqsave(&ring->lock, flags);
+	if (ring->ring->header.cons < ring->ring->header.prod) {
+		header->notify_on_cons = header->prod;
+		mb();
+		spin_unlock_irqrestore(&ring->lock, flags);
+		wait_event_interruptible(*ring->push_event,
+					 qxl_check_idle(ring));
+		spin_lock_irqsave(&ring->lock, flags);
+	}
+	spin_unlock_irqrestore(&ring->lock, flags);
+}
+
+int
+qxl_push_command_ring_release(struct qxl_device *qdev, struct qxl_release *release,
+			      uint32_t type, bool interruptible)
+{
+	struct qxl_command cmd;
+
+	cmd.type = type;
+	cmd.data = qxl_bo_physical_address(qdev, release->bos[0], release->release_offset);
+
+	return qxl_ring_push(qdev->command_ring, &cmd, interruptible);
+}
+
+int
+qxl_push_cursor_ring_release(struct qxl_device *qdev, struct qxl_release *release,
+			     uint32_t type, bool interruptible)
+{
+	struct qxl_command cmd;
+
+	cmd.type = type;
+	cmd.data = qxl_bo_physical_address(qdev, release->bos[0], release->release_offset);
+
+	return qxl_ring_push(qdev->cursor_ring, &cmd, interruptible);
+}
+
+bool qxl_queue_garbage_collect(struct qxl_device *qdev, bool flush)
+{
+	if (!qxl_check_idle(qdev->release_ring)) {
+		queue_work(qdev->gc_queue, &qdev->gc_work);
+		if (flush)
+			flush_work(&qdev->gc_work);
+		return true;
+	}
+	return false;
+}
+
+int qxl_garbage_collect(struct qxl_device *qdev)
+{
+	struct qxl_release *release;
+	uint64_t id, next_id;
+	int i = 0;
+	int ret;
+	union qxl_release_info *info;
+
+	while (qxl_ring_pop(qdev->release_ring, &id)) {
+		QXL_INFO(qdev, "popped %lld\n", id);
+		while (id) {
+			release = qxl_release_from_id_locked(qdev, id);
+			if (release == NULL)
+				break;
+
+			ret = qxl_release_reserve(qdev, release, false);
+			if (ret) {
+				qxl_io_log(qdev, "failed to reserve release on garbage collect %lld\n", id);
+				DRM_ERROR("failed to reserve release %lld\n", id);
+			}
+
+			info = qxl_release_map(qdev, release);
+			next_id = info->next;
+			qxl_release_unmap(qdev, release, info);
+
+			qxl_release_unreserve(qdev, release);
+			QXL_INFO(qdev, "popped %lld, next %lld\n", id,
+				next_id);
+
+			switch (release->type) {
+			case QXL_RELEASE_DRAWABLE:
+			case QXL_RELEASE_SURFACE_CMD:
+			case QXL_RELEASE_CURSOR_CMD:
+				break;
+			default:
+				DRM_ERROR("unexpected release type\n");
+				break;
+			}
+			id = next_id;
+
+			qxl_release_free(qdev, release);
+			++i;
+		}
+	}
+
+	QXL_INFO(qdev, "%s: %lld\n", __func__, i);
+
+	return i;
+}
+
+int qxl_alloc_bo_reserved(struct qxl_device *qdev, unsigned long size,
+			  struct qxl_bo **_bo)
+{
+	struct qxl_bo *bo;
+	int ret;
+
+	ret = qxl_bo_create(qdev, size, false /* not kernel - device */,
+			    QXL_GEM_DOMAIN_VRAM, NULL, &bo);
+	if (ret) {
+		DRM_ERROR("failed to allocate VRAM BO\n");
+		return ret;
+	}
+	ret = qxl_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		goto out_unref;
+
+	*_bo = bo;
+	return 0;
+out_unref:
+	qxl_bo_unref(&bo);
+	return 0;
+}
+
+static int wait_for_io_cmd_user(struct qxl_device *qdev, uint8_t val, long port)
+{
+	int irq_num;
+	long addr = qdev->io_base + port;
+	int ret;
+
+	mutex_lock(&qdev->async_io_mutex);
+	irq_num = atomic_read(&qdev->irq_received_io_cmd);
+
+
+	if (qdev->last_sent_io_cmd > irq_num) {
+		ret = wait_event_interruptible(qdev->io_cmd_event,
+					       atomic_read(&qdev->irq_received_io_cmd) > irq_num);
+		if (ret)
+			goto out;
+		irq_num = atomic_read(&qdev->irq_received_io_cmd);
+	}
+	outb(val, addr);
+	qdev->last_sent_io_cmd = irq_num + 1;
+	ret = wait_event_interruptible(qdev->io_cmd_event,
+				       atomic_read(&qdev->irq_received_io_cmd) > irq_num);
+out:
+	mutex_unlock(&qdev->async_io_mutex);
+	return ret;
+}
+
+static void wait_for_io_cmd(struct qxl_device *qdev, uint8_t val, long port)
+{
+	int ret;
+
+restart:
+	ret = wait_for_io_cmd_user(qdev, val, port);
+	if (ret == -ERESTARTSYS)
+		goto restart;
+}
+
+int qxl_io_update_area(struct qxl_device *qdev, struct qxl_bo *surf,
+			const struct qxl_rect *area)
+{
+	int surface_id;
+	uint32_t surface_width, surface_height;
+	int ret;
+
+	if (!surf->hw_surf_alloc)
+		DRM_ERROR("got io update area with no hw surface\n");
+
+	if (surf->is_primary)
+		surface_id = 0;
+	else
+		surface_id = surf->surface_id;
+	surface_width = surf->surf.width;
+	surface_height = surf->surf.height;
+
+	if (area->left < 0 || area->top < 0 ||
+	    area->right > surface_width || area->bottom > surface_height) {
+		qxl_io_log(qdev, "%s: not doing area update for "
+			   "%d, (%d,%d,%d,%d) (%d,%d)\n", __func__, surface_id, area->left,
+			   area->top, area->right, area->bottom, surface_width, surface_height);
+		return -EINVAL;
+	}
+	mutex_lock(&qdev->update_area_mutex);
+	qdev->ram_header->update_area = *area;
+	qdev->ram_header->update_surface = surface_id;
+	ret = wait_for_io_cmd_user(qdev, 0, QXL_IO_UPDATE_AREA_ASYNC);
+	mutex_unlock(&qdev->update_area_mutex);
+	return ret;
+}
+
+void qxl_io_notify_oom(struct qxl_device *qdev)
+{
+	outb(0, qdev->io_base + QXL_IO_NOTIFY_OOM);
+}
+
+void qxl_io_flush_release(struct qxl_device *qdev)
+{
+	outb(0, qdev->io_base + QXL_IO_FLUSH_RELEASE);
+}
+
+void qxl_io_flush_surfaces(struct qxl_device *qdev)
+{
+	wait_for_io_cmd(qdev, 0, QXL_IO_FLUSH_SURFACES_ASYNC);
+}
+
+
+void qxl_io_destroy_primary(struct qxl_device *qdev)
+{
+	wait_for_io_cmd(qdev, 0, QXL_IO_DESTROY_PRIMARY_ASYNC);
+}
+
+void qxl_io_create_primary(struct qxl_device *qdev, unsigned width,
+			   unsigned height, unsigned offset, struct qxl_bo *bo)
+{
+	struct qxl_surface_create *create;
+
+	QXL_INFO(qdev, "%s: qdev %p, ram_header %p\n", __func__, qdev,
+		 qdev->ram_header);
+	create = &qdev->ram_header->create_surface;
+	create->format = bo->surf.format;
+	create->width = width;
+	create->height = height;
+	create->stride = bo->surf.stride;
+	create->mem = qxl_bo_physical_address(qdev, bo, offset);
+
+	QXL_INFO(qdev, "%s: mem = %llx, from %p\n", __func__, create->mem,
+		 bo->kptr);
+
+	create->flags = QXL_SURF_FLAG_KEEP_DATA;
+	create->type = QXL_SURF_TYPE_PRIMARY;
+
+	wait_for_io_cmd(qdev, 0, QXL_IO_CREATE_PRIMARY_ASYNC);
+}
+
+void qxl_io_memslot_add(struct qxl_device *qdev, uint8_t id)
+{
+	QXL_INFO(qdev, "qxl_memslot_add %d\n", id);
+	wait_for_io_cmd(qdev, id, QXL_IO_MEMSLOT_ADD_ASYNC);
+}
+
+void qxl_io_log(struct qxl_device *qdev, const char *fmt, ...)
+{
+	va_list args;
+
+	va_start(args, fmt);
+	vsnprintf(qdev->ram_header->log_buf, QXL_LOG_BUF_SIZE, fmt, args);
+	va_end(args);
+	/*
+	 * DO not do a DRM output here - this will call printk, which will
+	 * call back into qxl for rendering (qxl_fb)
+	 */
+	outb(0, qdev->io_base + QXL_IO_LOG);
+}
+
+void qxl_io_reset(struct qxl_device *qdev)
+{
+	outb(0, qdev->io_base + QXL_IO_RESET);
+}
+
+void qxl_io_monitors_config(struct qxl_device *qdev)
+{
+	qxl_io_log(qdev, "%s: %d [%dx%d+%d+%d]\n", __func__,
+		   qdev->monitors_config ?
+		   qdev->monitors_config->count : -1,
+		   qdev->monitors_config && qdev->monitors_config->count ?
+		   qdev->monitors_config->heads[0].width : -1,
+		   qdev->monitors_config && qdev->monitors_config->count ?
+		   qdev->monitors_config->heads[0].height : -1,
+		   qdev->monitors_config && qdev->monitors_config->count ?
+		   qdev->monitors_config->heads[0].x : -1,
+		   qdev->monitors_config && qdev->monitors_config->count ?
+		   qdev->monitors_config->heads[0].y : -1
+		   );
+
+	wait_for_io_cmd(qdev, 0, QXL_IO_MONITORS_CONFIG_ASYNC);
+}
+
+int qxl_surface_id_alloc(struct qxl_device *qdev,
+		      struct qxl_bo *surf)
+{
+	uint32_t handle = -ENOMEM;
+	int idr_ret;
+	int count = 0;
+again:
+	if (idr_pre_get(&qdev->surf_id_idr, GFP_ATOMIC) == 0) {
+		DRM_ERROR("Out of memory for surf idr\n");
+		kfree(surf);
+		goto alloc_fail;
+	}
+
+	spin_lock(&qdev->surf_id_idr_lock);
+	idr_ret = idr_get_new_above(&qdev->surf_id_idr, NULL, 1, &handle);
+	spin_unlock(&qdev->surf_id_idr_lock);
+
+	if (idr_ret == -EAGAIN)
+		goto again;
+
+	if (handle >= qdev->rom->n_surfaces) {
+		count++;
+		spin_lock(&qdev->surf_id_idr_lock);
+		idr_remove(&qdev->surf_id_idr, handle);
+		spin_unlock(&qdev->surf_id_idr_lock);
+		qxl_reap_surface_id(qdev, 2);
+		goto again;
+	}
+	surf->surface_id = handle;
+
+	spin_lock(&qdev->surf_id_idr_lock);
+	qdev->last_alloced_surf_id = handle;
+	spin_unlock(&qdev->surf_id_idr_lock);
+ alloc_fail:
+	return 0;
+}
+
+void qxl_surface_id_dealloc(struct qxl_device *qdev,
+			    uint32_t surface_id)
+{
+	spin_lock(&qdev->surf_id_idr_lock);
+	idr_remove(&qdev->surf_id_idr, surface_id);
+	spin_unlock(&qdev->surf_id_idr_lock);
+}
+
+int qxl_hw_surface_alloc(struct qxl_device *qdev,
+			 struct qxl_bo *surf,
+			 struct ttm_mem_reg *new_mem)
+{
+	struct qxl_surface_cmd *cmd;
+	struct qxl_release *release;
+	int ret;
+
+	if (surf->hw_surf_alloc)
+		return 0;
+
+	ret = qxl_alloc_surface_release_reserved(qdev, QXL_SURFACE_CMD_CREATE,
+						 NULL,
+						 &release);
+	if (ret)
+		return ret;
+
+	cmd = (struct qxl_surface_cmd *)qxl_release_map(qdev, release);
+	cmd->type = QXL_SURFACE_CMD_CREATE;
+	cmd->u.surface_create.format = surf->surf.format;
+	cmd->u.surface_create.width = surf->surf.width;
+	cmd->u.surface_create.height = surf->surf.height;
+	cmd->u.surface_create.stride = surf->surf.stride;
+	if (new_mem) {
+		int slot_id = surf->type == QXL_GEM_DOMAIN_VRAM ? qdev->main_mem_slot : qdev->surfaces_mem_slot;
+		struct qxl_memslot *slot = &(qdev->mem_slots[slot_id]);
+
+		/* TODO - need to hold one of the locks to read tbo.offset */
+		cmd->u.surface_create.data = slot->high_bits;
+
+		cmd->u.surface_create.data |= (new_mem->start << PAGE_SHIFT) + surf->tbo.bdev->man[new_mem->mem_type].gpu_offset;
+	} else
+		cmd->u.surface_create.data = qxl_bo_physical_address(qdev, surf, 0);
+	cmd->surface_id = surf->surface_id;
+	qxl_release_unmap(qdev, release, &cmd->release_info);
+
+	surf->surf_create = release;
+
+	/* no need to add a release to the fence for this bo,
+	   since it is only released when we ask to destroy the surface
+	   and it would never signal otherwise */
+	qxl_fence_releaseable(qdev, release);
+
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_SURFACE, false);
+
+	qxl_release_unreserve(qdev, release);
+
+	surf->hw_surf_alloc = true;
+	spin_lock(&qdev->surf_id_idr_lock);
+	idr_replace(&qdev->surf_id_idr, surf, surf->surface_id);
+	spin_unlock(&qdev->surf_id_idr_lock);
+	return 0;
+}
+
+int qxl_hw_surface_dealloc(struct qxl_device *qdev,
+			   struct qxl_bo *surf)
+{
+	struct qxl_surface_cmd *cmd;
+	struct qxl_release *release;
+	int ret;
+	int id;
+
+	if (!surf->hw_surf_alloc)
+		return 0;
+
+	ret = qxl_alloc_surface_release_reserved(qdev, QXL_SURFACE_CMD_DESTROY,
+						 surf->surf_create,
+						 &release);
+	if (ret)
+		return ret;
+
+	surf->surf_create = NULL;
+	/* remove the surface from the idr, but not the surface id yet */
+	spin_lock(&qdev->surf_id_idr_lock);
+	idr_replace(&qdev->surf_id_idr, NULL, surf->surface_id);
+	spin_unlock(&qdev->surf_id_idr_lock);
+	surf->hw_surf_alloc = false;
+
+	id = surf->surface_id;
+	surf->surface_id = 0;
+
+	release->surface_release_id = id;
+	cmd = (struct qxl_surface_cmd *)qxl_release_map(qdev, release);
+	cmd->type = QXL_SURFACE_CMD_DESTROY;
+	cmd->surface_id = id;
+	qxl_release_unmap(qdev, release, &cmd->release_info);
+
+	qxl_fence_releaseable(qdev, release);
+
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_SURFACE, false);
+
+	qxl_release_unreserve(qdev, release);
+
+
+	return 0;
+}
+
+int qxl_update_surface(struct qxl_device *qdev, struct qxl_bo *surf)
+{
+	struct qxl_rect rect;
+	int ret;
+
+	/* if we are evicting, we need to make sure the surface is up
+	   to date */
+	rect.left = 0;
+	rect.right = surf->surf.width;
+	rect.top = 0;
+	rect.bottom = surf->surf.height;
+retry:
+	ret = qxl_io_update_area(qdev, surf, &rect);
+	if (ret == -ERESTARTSYS)
+		goto retry;
+	return ret;
+}
+
+void qxl_surface_evict_locked(struct qxl_device *qdev, struct qxl_bo *surf, bool do_update_area)
+{
+	/* no need to update area if we are just freeing the surface normally */
+	if (do_update_area)
+		qxl_update_surface(qdev, surf);
+
+	/* nuke the surface id at the hw */
+	qxl_hw_surface_dealloc(qdev, surf);
+}
+
+void qxl_surface_evict(struct qxl_device *qdev, struct qxl_bo *surf, bool do_update_area)
+{
+	mutex_lock(&qdev->surf_evict_mutex);
+	qxl_surface_evict_locked(qdev, surf, do_update_area);
+	mutex_unlock(&qdev->surf_evict_mutex);
+}
+
+static int qxl_reap_surf(struct qxl_device *qdev, struct qxl_bo *surf, bool stall)
+{
+	int ret;
+
+	ret = qxl_bo_reserve(surf, false);
+	if (ret == -EBUSY)
+		return -EBUSY;
+
+	if (surf->fence.num_active_releases > 0 && stall == false) {
+		qxl_bo_unreserve(surf);
+		return -EBUSY;
+	}
+
+	if (stall)
+		mutex_unlock(&qdev->surf_evict_mutex);
+
+	spin_lock(&surf->tbo.bdev->fence_lock);
+	ret = ttm_bo_wait(&surf->tbo, true, true, !stall);
+	spin_unlock(&surf->tbo.bdev->fence_lock);
+
+	if (stall)
+		mutex_lock(&qdev->surf_evict_mutex);
+	if (ret == -EBUSY) {
+		qxl_bo_unreserve(surf);
+		return -EBUSY;
+	}
+
+	qxl_surface_evict_locked(qdev, surf, true);
+	qxl_bo_unreserve(surf);
+	return 0;
+}
+
+static int qxl_reap_surface_id(struct qxl_device *qdev, int max_to_reap)
+{
+	int num_reaped = 0;
+	int i, ret;
+	bool stall = false;
+	int start = 0;
+
+	mutex_lock(&qdev->surf_evict_mutex);
+again:
+
+	spin_lock(&qdev->surf_id_idr_lock);
+	start = qdev->last_alloced_surf_id + 1;
+	spin_unlock(&qdev->surf_id_idr_lock);
+
+	for (i = start; i < start + qdev->rom->n_surfaces; i++) {
+		void *objptr;
+		int surfid = i % qdev->rom->n_surfaces;
+
+		/* this avoids the case where the objects is in the
+		   idr but has been evicted half way - its makes
+		   the idr lookup atomic with the eviction */
+		spin_lock(&qdev->surf_id_idr_lock);
+		objptr = idr_find(&qdev->surf_id_idr, surfid);
+		spin_unlock(&qdev->surf_id_idr_lock);
+
+		if (!objptr)
+			continue;
+
+		ret = qxl_reap_surf(qdev, objptr, stall);
+		if (ret == 0)
+			num_reaped++;
+		if (num_reaped >= max_to_reap)
+			break;
+	}
+	if (num_reaped == 0 && stall == false) {
+		stall = true;
+		goto again;
+	}
+
+	mutex_unlock(&qdev->surf_evict_mutex);
+	if (num_reaped) {
+		usleep_range(500, 1000);
+		qxl_queue_garbage_collect(qdev, true);
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/qxl/qxl_debugfs.c b/drivers/gpu/drm/qxl/qxl_debugfs.c
new file mode 100644
index 000000000000..c630152f2d2f
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_debugfs.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright (C) 2009 Red Hat <bskeggs@redhat.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/*
+ * Authors:
+ *  Alon Levy <alevy@redhat.com>
+ */
+
+#include <linux/debugfs.h>
+
+#include "drmP.h"
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+
+static int
+qxl_debugfs_irq_received(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct qxl_device *qdev = node->minor->dev->dev_private;
+
+	seq_printf(m, "%d\n", atomic_read(&qdev->irq_received));
+	seq_printf(m, "%d\n", atomic_read(&qdev->irq_received_display));
+	seq_printf(m, "%d\n", atomic_read(&qdev->irq_received_cursor));
+	seq_printf(m, "%d\n", atomic_read(&qdev->irq_received_io_cmd));
+	seq_printf(m, "%d\n", qdev->irq_received_error);
+	return 0;
+}
+
+static int
+qxl_debugfs_buffers_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct qxl_device *qdev = node->minor->dev->dev_private;
+	struct qxl_bo *bo;
+
+	list_for_each_entry(bo, &qdev->gem.objects, list) {
+		seq_printf(m, "size %ld, pc %d, sync obj %p, num releases %d\n",
+			   (unsigned long)bo->gem_base.size, bo->pin_count,
+			   bo->tbo.sync_obj, bo->fence.num_active_releases);
+	}
+	return 0;
+}
+
+static struct drm_info_list qxl_debugfs_list[] = {
+	{ "irq_received", qxl_debugfs_irq_received, 0, NULL },
+	{ "qxl_buffers", qxl_debugfs_buffers_info, 0, NULL },
+};
+#define QXL_DEBUGFS_ENTRIES ARRAY_SIZE(qxl_debugfs_list)
+
+int
+qxl_debugfs_init(struct drm_minor *minor)
+{
+	drm_debugfs_create_files(qxl_debugfs_list, QXL_DEBUGFS_ENTRIES,
+				 minor->debugfs_root, minor);
+	return 0;
+}
+
+void
+qxl_debugfs_takedown(struct drm_minor *minor)
+{
+	drm_debugfs_remove_files(qxl_debugfs_list, QXL_DEBUGFS_ENTRIES,
+				 minor);
+}
+
+int qxl_debugfs_add_files(struct qxl_device *qdev,
+			  struct drm_info_list *files,
+			  unsigned nfiles)
+{
+	unsigned i;
+
+	for (i = 0; i < qdev->debugfs_count; i++) {
+		if (qdev->debugfs[i].files == files) {
+			/* Already registered */
+			return 0;
+		}
+	}
+
+	i = qdev->debugfs_count + 1;
+	if (i > QXL_DEBUGFS_MAX_COMPONENTS) {
+		DRM_ERROR("Reached maximum number of debugfs components.\n");
+		DRM_ERROR("Report so we increase QXL_DEBUGFS_MAX_COMPONENTS.\n");
+		return -EINVAL;
+	}
+	qdev->debugfs[qdev->debugfs_count].files = files;
+	qdev->debugfs[qdev->debugfs_count].num_files = nfiles;
+	qdev->debugfs_count = i;
+#if defined(CONFIG_DEBUG_FS)
+	drm_debugfs_create_files(files, nfiles,
+				 qdev->ddev->control->debugfs_root,
+				 qdev->ddev->control);
+	drm_debugfs_create_files(files, nfiles,
+				 qdev->ddev->primary->debugfs_root,
+				 qdev->ddev->primary);
+#endif
+	return 0;
+}
+
+void qxl_debugfs_remove_files(struct qxl_device *qdev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned i;
+
+	for (i = 0; i < qdev->debugfs_count; i++) {
+		drm_debugfs_remove_files(qdev->debugfs[i].files,
+					 qdev->debugfs[i].num_files,
+					 qdev->ddev->control);
+		drm_debugfs_remove_files(qdev->debugfs[i].files,
+					 qdev->debugfs[i].num_files,
+					 qdev->ddev->primary);
+	}
+#endif
+}
diff --git a/drivers/gpu/drm/qxl/qxl_dev.h b/drivers/gpu/drm/qxl/qxl_dev.h
new file mode 100644
index 000000000000..94c5aec71920
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_dev.h
@@ -0,0 +1,879 @@
+/*
+   Copyright (C) 2009 Red Hat, Inc.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions are
+   met:
+
+       * Redistributions of source code must retain the above copyright
+	 notice, this list of conditions and the following disclaimer.
+       * Redistributions in binary form must reproduce the above copyright
+	 notice, this list of conditions and the following disclaimer in
+	 the documentation and/or other materials provided with the
+	 distribution.
+       * Neither the name of the copyright holder nor the names of its
+	 contributors may be used to endorse or promote products derived
+	 from this software without specific prior written permission.
+
+   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS "AS
+   IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
+   TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
+   PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+   HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+
+#ifndef H_QXL_DEV
+#define H_QXL_DEV
+
+#include <linux/types.h>
+
+/*
+ * from spice-protocol
+ * Release 0.10.0
+ */
+
+/* enums.h */
+
+enum SpiceImageType {
+	SPICE_IMAGE_TYPE_BITMAP,
+	SPICE_IMAGE_TYPE_QUIC,
+	SPICE_IMAGE_TYPE_RESERVED,
+	SPICE_IMAGE_TYPE_LZ_PLT = 100,
+	SPICE_IMAGE_TYPE_LZ_RGB,
+	SPICE_IMAGE_TYPE_GLZ_RGB,
+	SPICE_IMAGE_TYPE_FROM_CACHE,
+	SPICE_IMAGE_TYPE_SURFACE,
+	SPICE_IMAGE_TYPE_JPEG,
+	SPICE_IMAGE_TYPE_FROM_CACHE_LOSSLESS,
+	SPICE_IMAGE_TYPE_ZLIB_GLZ_RGB,
+	SPICE_IMAGE_TYPE_JPEG_ALPHA,
+
+	SPICE_IMAGE_TYPE_ENUM_END
+};
+
+enum SpiceBitmapFmt {
+	SPICE_BITMAP_FMT_INVALID,
+	SPICE_BITMAP_FMT_1BIT_LE,
+	SPICE_BITMAP_FMT_1BIT_BE,
+	SPICE_BITMAP_FMT_4BIT_LE,
+	SPICE_BITMAP_FMT_4BIT_BE,
+	SPICE_BITMAP_FMT_8BIT,
+	SPICE_BITMAP_FMT_16BIT,
+	SPICE_BITMAP_FMT_24BIT,
+	SPICE_BITMAP_FMT_32BIT,
+	SPICE_BITMAP_FMT_RGBA,
+
+	SPICE_BITMAP_FMT_ENUM_END
+};
+
+enum SpiceSurfaceFmt {
+	SPICE_SURFACE_FMT_INVALID,
+	SPICE_SURFACE_FMT_1_A,
+	SPICE_SURFACE_FMT_8_A = 8,
+	SPICE_SURFACE_FMT_16_555 = 16,
+	SPICE_SURFACE_FMT_32_xRGB = 32,
+	SPICE_SURFACE_FMT_16_565 = 80,
+	SPICE_SURFACE_FMT_32_ARGB = 96,
+
+	SPICE_SURFACE_FMT_ENUM_END
+};
+
+enum SpiceClipType {
+	SPICE_CLIP_TYPE_NONE,
+	SPICE_CLIP_TYPE_RECTS,
+
+	SPICE_CLIP_TYPE_ENUM_END
+};
+
+enum SpiceRopd {
+	SPICE_ROPD_INVERS_SRC = (1 << 0),
+	SPICE_ROPD_INVERS_BRUSH = (1 << 1),
+	SPICE_ROPD_INVERS_DEST = (1 << 2),
+	SPICE_ROPD_OP_PUT = (1 << 3),
+	SPICE_ROPD_OP_OR = (1 << 4),
+	SPICE_ROPD_OP_AND = (1 << 5),
+	SPICE_ROPD_OP_XOR = (1 << 6),
+	SPICE_ROPD_OP_BLACKNESS = (1 << 7),
+	SPICE_ROPD_OP_WHITENESS = (1 << 8),
+	SPICE_ROPD_OP_INVERS = (1 << 9),
+	SPICE_ROPD_INVERS_RES = (1 << 10),
+
+	SPICE_ROPD_MASK = 0x7ff
+};
+
+enum SpiceBrushType {
+	SPICE_BRUSH_TYPE_NONE,
+	SPICE_BRUSH_TYPE_SOLID,
+	SPICE_BRUSH_TYPE_PATTERN,
+
+	SPICE_BRUSH_TYPE_ENUM_END
+};
+
+enum SpiceCursorType {
+	SPICE_CURSOR_TYPE_ALPHA,
+	SPICE_CURSOR_TYPE_MONO,
+	SPICE_CURSOR_TYPE_COLOR4,
+	SPICE_CURSOR_TYPE_COLOR8,
+	SPICE_CURSOR_TYPE_COLOR16,
+	SPICE_CURSOR_TYPE_COLOR24,
+	SPICE_CURSOR_TYPE_COLOR32,
+
+	SPICE_CURSOR_TYPE_ENUM_END
+};
+
+/* qxl_dev.h */
+
+#pragma pack(push, 1)
+
+#define REDHAT_PCI_VENDOR_ID 0x1b36
+
+/* 0x100-0x11f reserved for spice, 0x1ff used for unstable work */
+#define QXL_DEVICE_ID_STABLE 0x0100
+
+enum {
+	QXL_REVISION_STABLE_V04 = 0x01,
+	QXL_REVISION_STABLE_V06 = 0x02,
+	QXL_REVISION_STABLE_V10 = 0x03,
+	QXL_REVISION_STABLE_V12 = 0x04,
+};
+
+#define QXL_DEVICE_ID_DEVEL 0x01ff
+#define QXL_REVISION_DEVEL 0x01
+
+#define QXL_ROM_MAGIC (*(uint32_t *)"QXRO")
+#define QXL_RAM_MAGIC (*(uint32_t *)"QXRA")
+
+enum {
+	QXL_RAM_RANGE_INDEX,
+	QXL_VRAM_RANGE_INDEX,
+	QXL_ROM_RANGE_INDEX,
+	QXL_IO_RANGE_INDEX,
+
+	QXL_PCI_RANGES
+};
+
+/* qxl-1 compat: append only */
+enum {
+	QXL_IO_NOTIFY_CMD,
+	QXL_IO_NOTIFY_CURSOR,
+	QXL_IO_UPDATE_AREA,
+	QXL_IO_UPDATE_IRQ,
+	QXL_IO_NOTIFY_OOM,
+	QXL_IO_RESET,
+	QXL_IO_SET_MODE,                  /* qxl-1 */
+	QXL_IO_LOG,
+	/* appended for qxl-2 */
+	QXL_IO_MEMSLOT_ADD,
+	QXL_IO_MEMSLOT_DEL,
+	QXL_IO_DETACH_PRIMARY,
+	QXL_IO_ATTACH_PRIMARY,
+	QXL_IO_CREATE_PRIMARY,
+	QXL_IO_DESTROY_PRIMARY,
+	QXL_IO_DESTROY_SURFACE_WAIT,
+	QXL_IO_DESTROY_ALL_SURFACES,
+	/* appended for qxl-3 */
+	QXL_IO_UPDATE_AREA_ASYNC,
+	QXL_IO_MEMSLOT_ADD_ASYNC,
+	QXL_IO_CREATE_PRIMARY_ASYNC,
+	QXL_IO_DESTROY_PRIMARY_ASYNC,
+	QXL_IO_DESTROY_SURFACE_ASYNC,
+	QXL_IO_DESTROY_ALL_SURFACES_ASYNC,
+	QXL_IO_FLUSH_SURFACES_ASYNC,
+	QXL_IO_FLUSH_RELEASE,
+	/* appended for qxl-4 */
+	QXL_IO_MONITORS_CONFIG_ASYNC,
+
+	QXL_IO_RANGE_SIZE
+};
+
+typedef uint64_t QXLPHYSICAL;
+typedef int32_t QXLFIXED; /* fixed 28.4 */
+
+struct qxl_point_fix {
+	QXLFIXED x;
+	QXLFIXED y;
+};
+
+struct qxl_point {
+	int32_t x;
+	int32_t y;
+};
+
+struct qxl_point_1_6 {
+	int16_t x;
+	int16_t y;
+};
+
+struct qxl_rect {
+	int32_t top;
+	int32_t left;
+	int32_t bottom;
+	int32_t right;
+};
+
+struct qxl_urect {
+	uint32_t top;
+	uint32_t left;
+	uint32_t bottom;
+	uint32_t right;
+};
+
+/* qxl-1 compat: append only */
+struct qxl_rom {
+	uint32_t magic;
+	uint32_t id;
+	uint32_t update_id;
+	uint32_t compression_level;
+	uint32_t log_level;
+	uint32_t mode;			  /* qxl-1 */
+	uint32_t modes_offset;
+	uint32_t num_io_pages;
+	uint32_t pages_offset;		  /* qxl-1 */
+	uint32_t draw_area_offset;	  /* qxl-1 */
+	uint32_t surface0_area_size;	  /* qxl-1 name: draw_area_size */
+	uint32_t ram_header_offset;
+	uint32_t mm_clock;
+	/* appended for qxl-2 */
+	uint32_t n_surfaces;
+	uint64_t flags;
+	uint8_t slots_start;
+	uint8_t slots_end;
+	uint8_t slot_gen_bits;
+	uint8_t slot_id_bits;
+	uint8_t slot_generation;
+	/* appended for qxl-4 */
+	uint8_t client_present;
+	uint8_t client_capabilities[58];
+	uint32_t client_monitors_config_crc;
+	struct {
+		uint16_t count;
+	uint16_t padding;
+		struct qxl_urect heads[64];
+	} client_monitors_config;
+};
+
+/* qxl-1 compat: fixed */
+struct qxl_mode {
+	uint32_t id;
+	uint32_t x_res;
+	uint32_t y_res;
+	uint32_t bits;
+	uint32_t stride;
+	uint32_t x_mili;
+	uint32_t y_mili;
+	uint32_t orientation;
+};
+
+/* qxl-1 compat: fixed */
+struct qxl_modes {
+	uint32_t n_modes;
+	struct qxl_mode modes[0];
+};
+
+/* qxl-1 compat: append only */
+enum qxl_cmd_type {
+	QXL_CMD_NOP,
+	QXL_CMD_DRAW,
+	QXL_CMD_UPDATE,
+	QXL_CMD_CURSOR,
+	QXL_CMD_MESSAGE,
+	QXL_CMD_SURFACE,
+};
+
+/* qxl-1 compat: fixed */
+struct qxl_command {
+	QXLPHYSICAL data;
+	uint32_t type;
+	uint32_t padding;
+};
+
+#define QXL_COMMAND_FLAG_COMPAT		(1<<0)
+#define QXL_COMMAND_FLAG_COMPAT_16BPP	(2<<0)
+
+struct qxl_command_ext {
+	struct qxl_command cmd;
+	uint32_t group_id;
+	uint32_t flags;
+};
+
+struct qxl_mem_slot {
+	uint64_t mem_start;
+	uint64_t mem_end;
+};
+
+#define QXL_SURF_TYPE_PRIMARY	   0
+
+#define QXL_SURF_FLAG_KEEP_DATA	   (1 << 0)
+
+struct qxl_surface_create {
+	uint32_t width;
+	uint32_t height;
+	int32_t stride;
+	uint32_t format;
+	uint32_t position;
+	uint32_t mouse_mode;
+	uint32_t flags;
+	uint32_t type;
+	QXLPHYSICAL mem;
+};
+
+#define QXL_COMMAND_RING_SIZE 32
+#define QXL_CURSOR_RING_SIZE 32
+#define QXL_RELEASE_RING_SIZE 8
+
+#define QXL_LOG_BUF_SIZE 4096
+
+#define QXL_INTERRUPT_DISPLAY (1 << 0)
+#define QXL_INTERRUPT_CURSOR (1 << 1)
+#define QXL_INTERRUPT_IO_CMD (1 << 2)
+#define QXL_INTERRUPT_ERROR  (1 << 3)
+#define QXL_INTERRUPT_CLIENT (1 << 4)
+#define QXL_INTERRUPT_CLIENT_MONITORS_CONFIG  (1 << 5)
+
+struct qxl_ring_header {
+	uint32_t num_items;
+	uint32_t prod;
+	uint32_t notify_on_prod;
+	uint32_t cons;
+	uint32_t notify_on_cons;
+};
+
+/* qxl-1 compat: append only */
+struct qxl_ram_header {
+	uint32_t magic;
+	uint32_t int_pending;
+	uint32_t int_mask;
+	uint8_t log_buf[QXL_LOG_BUF_SIZE];
+	struct qxl_ring_header  cmd_ring_hdr;
+	struct qxl_command	cmd_ring[QXL_COMMAND_RING_SIZE];
+	struct qxl_ring_header  cursor_ring_hdr;
+	struct qxl_command	cursor_ring[QXL_CURSOR_RING_SIZE];
+	struct qxl_ring_header  release_ring_hdr;
+	uint64_t		release_ring[QXL_RELEASE_RING_SIZE];
+	struct qxl_rect update_area;
+	/* appended for qxl-2 */
+	uint32_t update_surface;
+	struct qxl_mem_slot mem_slot;
+	struct qxl_surface_create create_surface;
+	uint64_t flags;
+
+	/* appended for qxl-4 */
+
+	/* used by QXL_IO_MONITORS_CONFIG_ASYNC */
+	QXLPHYSICAL monitors_config;
+	uint8_t guest_capabilities[64];
+};
+
+union qxl_release_info {
+	uint64_t id;	  /* in  */
+	uint64_t next;	  /* out */
+};
+
+struct qxl_release_info_ext {
+	union qxl_release_info *info;
+	uint32_t group_id;
+};
+
+struct qxl_data_chunk {
+	uint32_t data_size;
+	QXLPHYSICAL prev_chunk;
+	QXLPHYSICAL next_chunk;
+	uint8_t data[0];
+};
+
+struct qxl_message {
+	union qxl_release_info release_info;
+	uint8_t data[0];
+};
+
+struct qxl_compat_update_cmd {
+	union qxl_release_info release_info;
+	struct qxl_rect area;
+	uint32_t update_id;
+};
+
+struct qxl_update_cmd {
+	union qxl_release_info release_info;
+	struct qxl_rect area;
+	uint32_t update_id;
+	uint32_t surface_id;
+};
+
+struct qxl_cursor_header {
+	uint64_t unique;
+	uint16_t type;
+	uint16_t width;
+	uint16_t height;
+	uint16_t hot_spot_x;
+	uint16_t hot_spot_y;
+};
+
+struct qxl_cursor {
+	struct qxl_cursor_header header;
+	uint32_t data_size;
+	struct qxl_data_chunk chunk;
+};
+
+enum {
+	QXL_CURSOR_SET,
+	QXL_CURSOR_MOVE,
+	QXL_CURSOR_HIDE,
+	QXL_CURSOR_TRAIL,
+};
+
+#define QXL_CURSOR_DEVICE_DATA_SIZE 128
+
+struct qxl_cursor_cmd {
+	union qxl_release_info release_info;
+	uint8_t type;
+	union {
+		struct {
+			struct qxl_point_1_6 position;
+			uint8_t visible;
+			QXLPHYSICAL shape;
+		} set;
+		struct {
+			uint16_t length;
+			uint16_t frequency;
+		} trail;
+		struct qxl_point_1_6 position;
+	} u;
+	/* todo: dynamic size from rom */
+	uint8_t device_data[QXL_CURSOR_DEVICE_DATA_SIZE];
+};
+
+enum {
+	QXL_DRAW_NOP,
+	QXL_DRAW_FILL,
+	QXL_DRAW_OPAQUE,
+	QXL_DRAW_COPY,
+	QXL_COPY_BITS,
+	QXL_DRAW_BLEND,
+	QXL_DRAW_BLACKNESS,
+	QXL_DRAW_WHITENESS,
+	QXL_DRAW_INVERS,
+	QXL_DRAW_ROP3,
+	QXL_DRAW_STROKE,
+	QXL_DRAW_TEXT,
+	QXL_DRAW_TRANSPARENT,
+	QXL_DRAW_ALPHA_BLEND,
+	QXL_DRAW_COMPOSITE
+};
+
+struct qxl_raster_glyph {
+	struct qxl_point render_pos;
+	struct qxl_point glyph_origin;
+	uint16_t width;
+	uint16_t height;
+	uint8_t data[0];
+};
+
+struct qxl_string {
+	uint32_t data_size;
+	uint16_t length;
+	uint16_t flags;
+	struct qxl_data_chunk chunk;
+};
+
+struct qxl_copy_bits {
+	struct qxl_point src_pos;
+};
+
+enum qxl_effect_type {
+	QXL_EFFECT_BLEND = 0,
+	QXL_EFFECT_OPAQUE = 1,
+	QXL_EFFECT_REVERT_ON_DUP = 2,
+	QXL_EFFECT_BLACKNESS_ON_DUP = 3,
+	QXL_EFFECT_WHITENESS_ON_DUP = 4,
+	QXL_EFFECT_NOP_ON_DUP = 5,
+	QXL_EFFECT_NOP = 6,
+	QXL_EFFECT_OPAQUE_BRUSH = 7
+};
+
+struct qxl_pattern {
+	QXLPHYSICAL pat;
+	struct qxl_point pos;
+};
+
+struct qxl_brush {
+	uint32_t type;
+	union {
+		uint32_t color;
+		struct qxl_pattern pattern;
+	} u;
+};
+
+struct qxl_q_mask {
+	uint8_t flags;
+	struct qxl_point pos;
+	QXLPHYSICAL bitmap;
+};
+
+struct qxl_fill {
+	struct qxl_brush brush;
+	uint16_t rop_descriptor;
+	struct qxl_q_mask mask;
+};
+
+struct qxl_opaque {
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+	struct qxl_brush brush;
+	uint16_t rop_descriptor;
+	uint8_t scale_mode;
+	struct qxl_q_mask mask;
+};
+
+struct qxl_copy {
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+	uint16_t rop_descriptor;
+	uint8_t scale_mode;
+	struct qxl_q_mask mask;
+};
+
+struct qxl_transparent {
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+	uint32_t src_color;
+	uint32_t true_color;
+};
+
+struct qxl_alpha_blend {
+	uint16_t alpha_flags;
+	uint8_t alpha;
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+};
+
+struct qxl_compat_alpha_blend {
+	uint8_t alpha;
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+};
+
+struct qxl_rop_3 {
+	QXLPHYSICAL src_bitmap;
+	struct qxl_rect src_area;
+	struct qxl_brush brush;
+	uint8_t rop3;
+	uint8_t scale_mode;
+	struct qxl_q_mask mask;
+};
+
+struct qxl_line_attr {
+	uint8_t flags;
+	uint8_t join_style;
+	uint8_t end_style;
+	uint8_t style_nseg;
+	QXLFIXED width;
+	QXLFIXED miter_limit;
+	QXLPHYSICAL style;
+};
+
+struct qxl_stroke {
+	QXLPHYSICAL path;
+	struct qxl_line_attr attr;
+	struct qxl_brush brush;
+	uint16_t fore_mode;
+	uint16_t back_mode;
+};
+
+struct qxl_text {
+	QXLPHYSICAL str;
+	struct qxl_rect back_area;
+	struct qxl_brush fore_brush;
+	struct qxl_brush back_brush;
+	uint16_t fore_mode;
+	uint16_t back_mode;
+};
+
+struct qxl_mask {
+	struct qxl_q_mask mask;
+};
+
+struct qxl_clip {
+	uint32_t type;
+	QXLPHYSICAL data;
+};
+
+enum qxl_operator {
+	QXL_OP_CLEAR			 = 0x00,
+	QXL_OP_SOURCE			 = 0x01,
+	QXL_OP_DST			 = 0x02,
+	QXL_OP_OVER			 = 0x03,
+	QXL_OP_OVER_REVERSE		 = 0x04,
+	QXL_OP_IN			 = 0x05,
+	QXL_OP_IN_REVERSE		 = 0x06,
+	QXL_OP_OUT			 = 0x07,
+	QXL_OP_OUT_REVERSE		 = 0x08,
+	QXL_OP_ATOP			 = 0x09,
+	QXL_OP_ATOP_REVERSE		 = 0x0a,
+	QXL_OP_XOR			 = 0x0b,
+	QXL_OP_ADD			 = 0x0c,
+	QXL_OP_SATURATE			 = 0x0d,
+	/* Note the jump here from 0x0d to 0x30 */
+	QXL_OP_MULTIPLY			 = 0x30,
+	QXL_OP_SCREEN			 = 0x31,
+	QXL_OP_OVERLAY			 = 0x32,
+	QXL_OP_DARKEN			 = 0x33,
+	QXL_OP_LIGHTEN			 = 0x34,
+	QXL_OP_COLOR_DODGE		 = 0x35,
+	QXL_OP_COLOR_BURN		 = 0x36,
+	QXL_OP_HARD_LIGHT		 = 0x37,
+	QXL_OP_SOFT_LIGHT		 = 0x38,
+	QXL_OP_DIFFERENCE		 = 0x39,
+	QXL_OP_EXCLUSION		 = 0x3a,
+	QXL_OP_HSL_HUE			 = 0x3b,
+	QXL_OP_HSL_SATURATION		 = 0x3c,
+	QXL_OP_HSL_COLOR		 = 0x3d,
+	QXL_OP_HSL_LUMINOSITY		 = 0x3e
+};
+
+struct qxl_transform {
+	uint32_t	t00;
+	uint32_t	t01;
+	uint32_t	t02;
+	uint32_t	t10;
+	uint32_t	t11;
+	uint32_t	t12;
+};
+
+/* The flags field has the following bit fields:
+ *
+ *     operator:		[  0 -  7 ]
+ *     src_filter:		[  8 - 10 ]
+ *     mask_filter:		[ 11 - 13 ]
+ *     src_repeat:		[ 14 - 15 ]
+ *     mask_repeat:		[ 16 - 17 ]
+ *     component_alpha:		[ 18 - 18 ]
+ *     reserved:		[ 19 - 31 ]
+ *
+ * The repeat and filter values are those of pixman:
+ *		REPEAT_NONE =		0
+ *              REPEAT_NORMAL =		1
+ *		REPEAT_PAD =		2
+ *		REPEAT_REFLECT =	3
+ *
+ * The filter values are:
+ *		FILTER_NEAREST =	0
+ *		FILTER_BILINEAR	=	1
+ */
+struct qxl_composite {
+	uint32_t		flags;
+
+	QXLPHYSICAL			src;
+	QXLPHYSICAL			src_transform;	/* May be NULL */
+	QXLPHYSICAL			mask;		/* May be NULL */
+	QXLPHYSICAL			mask_transform;	/* May be NULL */
+	struct qxl_point_1_6	src_origin;
+	struct qxl_point_1_6	mask_origin;
+};
+
+struct qxl_compat_drawable {
+	union qxl_release_info release_info;
+	uint8_t effect;
+	uint8_t type;
+	uint16_t bitmap_offset;
+	struct qxl_rect bitmap_area;
+	struct qxl_rect bbox;
+	struct qxl_clip clip;
+	uint32_t mm_time;
+	union {
+		struct qxl_fill fill;
+		struct qxl_opaque opaque;
+		struct qxl_copy copy;
+		struct qxl_transparent transparent;
+		struct qxl_compat_alpha_blend alpha_blend;
+		struct qxl_copy_bits copy_bits;
+		struct qxl_copy blend;
+		struct qxl_rop_3 rop3;
+		struct qxl_stroke stroke;
+		struct qxl_text text;
+		struct qxl_mask blackness;
+		struct qxl_mask invers;
+		struct qxl_mask whiteness;
+	} u;
+};
+
+struct qxl_drawable {
+	union qxl_release_info release_info;
+	uint32_t surface_id;
+	uint8_t effect;
+	uint8_t type;
+	uint8_t self_bitmap;
+	struct qxl_rect self_bitmap_area;
+	struct qxl_rect bbox;
+	struct qxl_clip clip;
+	uint32_t mm_time;
+	int32_t surfaces_dest[3];
+	struct qxl_rect surfaces_rects[3];
+	union {
+		struct qxl_fill fill;
+		struct qxl_opaque opaque;
+		struct qxl_copy copy;
+		struct qxl_transparent transparent;
+		struct qxl_alpha_blend alpha_blend;
+		struct qxl_copy_bits copy_bits;
+		struct qxl_copy blend;
+		struct qxl_rop_3 rop3;
+		struct qxl_stroke stroke;
+		struct qxl_text text;
+		struct qxl_mask blackness;
+		struct qxl_mask invers;
+		struct qxl_mask whiteness;
+		struct qxl_composite composite;
+	} u;
+};
+
+enum qxl_surface_cmd_type {
+	QXL_SURFACE_CMD_CREATE,
+	QXL_SURFACE_CMD_DESTROY,
+};
+
+struct qxl_surface {
+	uint32_t format;
+	uint32_t width;
+	uint32_t height;
+	int32_t stride;
+	QXLPHYSICAL data;
+};
+
+struct qxl_surface_cmd {
+	union qxl_release_info release_info;
+	uint32_t surface_id;
+	uint8_t type;
+	uint32_t flags;
+	union {
+		struct qxl_surface surface_create;
+	} u;
+};
+
+struct qxl_clip_rects {
+	uint32_t num_rects;
+	struct qxl_data_chunk chunk;
+};
+
+enum {
+	QXL_PATH_BEGIN = (1 << 0),
+	QXL_PATH_END = (1 << 1),
+	QXL_PATH_CLOSE = (1 << 3),
+	QXL_PATH_BEZIER = (1 << 4),
+};
+
+struct qxl_path_seg {
+	uint32_t flags;
+	uint32_t count;
+	struct qxl_point_fix points[0];
+};
+
+struct qxl_path {
+	uint32_t data_size;
+	struct qxl_data_chunk chunk;
+};
+
+enum {
+	QXL_IMAGE_GROUP_DRIVER,
+	QXL_IMAGE_GROUP_DEVICE,
+	QXL_IMAGE_GROUP_RED,
+	QXL_IMAGE_GROUP_DRIVER_DONT_CACHE,
+};
+
+struct qxl_image_id {
+	uint32_t group;
+	uint32_t unique;
+};
+
+union qxl_image_id_union {
+	struct qxl_image_id id;
+	uint64_t value;
+};
+
+enum qxl_image_flags {
+	QXL_IMAGE_CACHE = (1 << 0),
+	QXL_IMAGE_HIGH_BITS_SET = (1 << 1),
+};
+
+enum qxl_bitmap_flags {
+	QXL_BITMAP_DIRECT = (1 << 0),
+	QXL_BITMAP_UNSTABLE = (1 << 1),
+	QXL_BITMAP_TOP_DOWN = (1 << 2), /* == SPICE_BITMAP_FLAGS_TOP_DOWN */
+};
+
+#define QXL_SET_IMAGE_ID(image, _group, _unique) {              \
+	(image)->descriptor.id = (((uint64_t)_unique) << 32) | _group;	\
+}
+
+struct qxl_image_descriptor {
+	uint64_t id;
+	uint8_t type;
+	uint8_t flags;
+	uint32_t width;
+	uint32_t height;
+};
+
+struct qxl_palette {
+	uint64_t unique;
+	uint16_t num_ents;
+	uint32_t ents[0];
+};
+
+struct qxl_bitmap {
+	uint8_t format;
+	uint8_t flags;
+	uint32_t x;
+	uint32_t y;
+	uint32_t stride;
+	QXLPHYSICAL palette;
+	QXLPHYSICAL data; /* data[0] ? */
+};
+
+struct qxl_surface_id {
+	uint32_t surface_id;
+};
+
+struct qxl_encoder_data {
+	uint32_t data_size;
+	uint8_t data[0];
+};
+
+struct qxl_image {
+	struct qxl_image_descriptor descriptor;
+	union { /* variable length */
+		struct qxl_bitmap bitmap;
+		struct qxl_encoder_data quic;
+		struct qxl_surface_id surface_image;
+	} u;
+};
+
+/* A QXLHead is a single monitor output backed by a QXLSurface.
+ * x and y offsets are unsigned since they are used in relation to
+ * the given surface, not the same as the x, y coordinates in the guest
+ * screen reference frame. */
+struct qxl_head {
+	uint32_t id;
+	uint32_t surface_id;
+	uint32_t width;
+	uint32_t height;
+	uint32_t x;
+	uint32_t y;
+	uint32_t flags;
+};
+
+struct qxl_monitors_config {
+	uint16_t count;
+	uint16_t max_allowed; /* If it is 0 no fixed limit is given by the
+				 driver */
+	struct qxl_head heads[0];
+};
+
+#pragma pack(pop)
+
+#endif /* _H_QXL_DEV */
diff --git a/drivers/gpu/drm/qxl/qxl_display.c b/drivers/gpu/drm/qxl/qxl_display.c
new file mode 100644
index 000000000000..c80ddfedbbab
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_display.c
@@ -0,0 +1,981 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+
+#include "linux/crc32.h"
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+#include "drm_crtc_helper.h"
+
+static void qxl_crtc_set_to_mode(struct qxl_device *qdev,
+				 struct drm_connector *connector,
+				 struct qxl_head *head)
+{
+	struct drm_device *dev = connector->dev;
+	struct drm_display_mode *mode, *t;
+	int width = head->width;
+	int height = head->height;
+
+	if (width < 320 || height < 240) {
+		qxl_io_log(qdev, "%s: bad head: %dx%d", width, height);
+		width = 1024;
+		height = 768;
+	}
+	if (width * height * 4 > 16*1024*1024) {
+		width = 1024;
+		height = 768;
+	}
+	/* TODO: go over regular modes and removed preferred? */
+	list_for_each_entry_safe(mode, t, &connector->probed_modes, head)
+		drm_mode_remove(connector, mode);
+	mode = drm_cvt_mode(dev, width, height, 60, false, false, false);
+	mode->type |= DRM_MODE_TYPE_PREFERRED;
+	mode->status = MODE_OK;
+	drm_mode_probed_add(connector, mode);
+	qxl_io_log(qdev, "%s: %d x %d\n", __func__, width, height);
+}
+
+void qxl_crtc_set_from_monitors_config(struct qxl_device *qdev)
+{
+	struct drm_connector *connector;
+	int i;
+	struct drm_device *dev = qdev->ddev;
+
+	i = 0;
+	qxl_io_log(qdev, "%s: %d, %d\n", __func__,
+		   dev->mode_config.num_connector,
+		   qdev->monitors_config->count);
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		if (i > qdev->monitors_config->count) {
+			/* crtc will be reported as disabled */
+			continue;
+		}
+		qxl_crtc_set_to_mode(qdev, connector,
+				     &qdev->monitors_config->heads[i]);
+		++i;
+	}
+}
+
+void qxl_alloc_client_monitors_config(struct qxl_device *qdev, unsigned count)
+{
+	if (qdev->client_monitors_config &&
+	    count > qdev->client_monitors_config->count) {
+		kfree(qdev->client_monitors_config);
+	}
+	if (!qdev->client_monitors_config) {
+		qdev->client_monitors_config = kzalloc(
+				sizeof(struct qxl_monitors_config) +
+				sizeof(struct qxl_head) * count, GFP_KERNEL);
+		if (!qdev->client_monitors_config) {
+			qxl_io_log(qdev,
+				   "%s: allocation failure for %u heads\n",
+				   __func__, count);
+			return;
+		}
+	}
+	qdev->client_monitors_config->count = count;
+}
+
+static int qxl_display_copy_rom_client_monitors_config(struct qxl_device *qdev)
+{
+	int i;
+	int num_monitors;
+	uint32_t crc;
+
+	BUG_ON(!qdev->monitors_config);
+	num_monitors = qdev->rom->client_monitors_config.count;
+	crc = crc32(0, (const uint8_t *)&qdev->rom->client_monitors_config,
+		  sizeof(qdev->rom->client_monitors_config));
+	if (crc != qdev->rom->client_monitors_config_crc) {
+		qxl_io_log(qdev, "crc mismatch: have %X (%d) != %X\n", crc,
+			   sizeof(qdev->rom->client_monitors_config),
+			   qdev->rom->client_monitors_config_crc);
+		return 1;
+	}
+	if (num_monitors > qdev->monitors_config->max_allowed) {
+		DRM_INFO("client monitors list will be truncated: %d < %d\n",
+			 qdev->monitors_config->max_allowed, num_monitors);
+		num_monitors = qdev->monitors_config->max_allowed;
+	} else {
+		num_monitors = qdev->rom->client_monitors_config.count;
+	}
+	qxl_alloc_client_monitors_config(qdev, num_monitors);
+	/* we copy max from the client but it isn't used */
+	qdev->client_monitors_config->max_allowed =
+				qdev->monitors_config->max_allowed;
+	for (i = 0 ; i < qdev->client_monitors_config->count ; ++i) {
+		struct qxl_urect *c_rect =
+			&qdev->rom->client_monitors_config.heads[i];
+		struct qxl_head *client_head =
+			&qdev->client_monitors_config->heads[i];
+		struct qxl_head *head = &qdev->monitors_config->heads[i];
+		client_head->x = head->x = c_rect->left;
+		client_head->y = head->y = c_rect->top;
+		client_head->width = head->width =
+						c_rect->right - c_rect->left;
+		client_head->height = head->height =
+						c_rect->bottom - c_rect->top;
+		client_head->surface_id = head->surface_id = 0;
+		client_head->id = head->id = i;
+		client_head->flags = head->flags = 0;
+		QXL_DEBUG(qdev, "read %dx%d+%d+%d\n", head->width, head->height,
+			  head->x, head->y);
+	}
+	return 0;
+}
+
+void qxl_display_read_client_monitors_config(struct qxl_device *qdev)
+{
+
+	while (qxl_display_copy_rom_client_monitors_config(qdev)) {
+		qxl_io_log(qdev, "failed crc check for client_monitors_config,"
+				 " retrying\n");
+	}
+	qxl_crtc_set_from_monitors_config(qdev);
+	/* fire off a uevent and let userspace tell us what to do */
+	qxl_io_log(qdev, "calling drm_sysfs_hotplug_event\n");
+	drm_sysfs_hotplug_event(qdev->ddev);
+}
+
+static int qxl_add_monitors_config_modes(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct qxl_device *qdev = dev->dev_private;
+	struct qxl_output *output = drm_connector_to_qxl_output(connector);
+	int h = output->index;
+	struct drm_display_mode *mode = NULL;
+	struct qxl_head *head;
+
+	if (!qdev->monitors_config)
+		return 0;
+	head = &qdev->monitors_config->heads[h];
+
+	mode = drm_cvt_mode(dev, head->width, head->height, 60, false, false,
+			    false);
+	mode->type |= DRM_MODE_TYPE_PREFERRED;
+	drm_mode_probed_add(connector, mode);
+	return 1;
+}
+
+static int qxl_add_common_modes(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct drm_display_mode *mode = NULL;
+	int i;
+	struct mode_size {
+		int w;
+		int h;
+	} common_modes[] = {
+		{ 640,  480},
+		{ 720,  480},
+		{ 800,  600},
+		{ 848,  480},
+		{1024,  768},
+		{1152,  768},
+		{1280,  720},
+		{1280,  800},
+		{1280,  854},
+		{1280,  960},
+		{1280, 1024},
+		{1440,  900},
+		{1400, 1050},
+		{1680, 1050},
+		{1600, 1200},
+		{1920, 1080},
+		{1920, 1200}
+	};
+
+	for (i = 0; i < ARRAY_SIZE(common_modes); i++) {
+		if (common_modes[i].w < 320 || common_modes[i].h < 200)
+			continue;
+
+		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h,
+				    60, false, false, false);
+		if (common_modes[i].w == 1024 && common_modes[i].h == 768)
+			mode->type |= DRM_MODE_TYPE_PREFERRED;
+		drm_mode_probed_add(connector, mode);
+	}
+	return i - 1;
+}
+
+static void qxl_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+			       u16 *blue, uint32_t start, uint32_t size)
+{
+	/* TODO */
+}
+
+static void qxl_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct qxl_crtc *qxl_crtc = to_qxl_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(qxl_crtc);
+}
+
+static void
+qxl_hide_cursor(struct qxl_device *qdev)
+{
+	struct qxl_release *release;
+	struct qxl_cursor_cmd *cmd;
+	int ret;
+
+	ret = qxl_alloc_release_reserved(qdev, sizeof(*cmd), QXL_RELEASE_CURSOR_CMD,
+					 &release, NULL);
+
+	cmd = (struct qxl_cursor_cmd *)qxl_release_map(qdev, release);
+	cmd->type = QXL_CURSOR_HIDE;
+	qxl_release_unmap(qdev, release, &cmd->release_info);
+
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_cursor_ring_release(qdev, release, QXL_CMD_CURSOR, false);
+	qxl_release_unreserve(qdev, release);
+}
+
+static int qxl_crtc_cursor_set(struct drm_crtc *crtc,
+			       struct drm_file *file_priv,
+			       uint32_t handle,
+			       uint32_t width,
+			       uint32_t height)
+{
+	struct drm_device *dev = crtc->dev;
+	struct qxl_device *qdev = dev->dev_private;
+	struct qxl_crtc *qcrtc = to_qxl_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct qxl_cursor *cursor;
+	struct qxl_cursor_cmd *cmd;
+	struct qxl_bo *cursor_bo, *user_bo;
+	struct qxl_release *release;
+	void *user_ptr;
+
+	int size = 64*64*4;
+	int ret = 0;
+	if (!handle) {
+		qxl_hide_cursor(qdev);
+		return 0;
+	}
+
+	obj = drm_gem_object_lookup(crtc->dev, file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("cannot find cursor object\n");
+		return -ENOENT;
+	}
+
+	user_bo = gem_to_qxl_bo(obj);
+
+	ret = qxl_bo_reserve(user_bo, false);
+	if (ret)
+		goto out_unref;
+
+	ret = qxl_bo_pin(user_bo, QXL_GEM_DOMAIN_CPU, NULL);
+	if (ret)
+		goto out_unreserve;
+
+	ret = qxl_bo_kmap(user_bo, &user_ptr);
+	if (ret)
+		goto out_unpin;
+
+	ret = qxl_alloc_release_reserved(qdev, sizeof(*cmd),
+					 QXL_RELEASE_CURSOR_CMD,
+					 &release, NULL);
+	if (ret)
+		goto out_kunmap;
+	ret = qxl_alloc_bo_reserved(qdev, sizeof(struct qxl_cursor) + size,
+				    &cursor_bo);
+	if (ret)
+		goto out_free_release;
+	ret = qxl_bo_kmap(cursor_bo, (void **)&cursor);
+	if (ret)
+		goto out_free_bo;
+
+	cursor->header.unique = 0;
+	cursor->header.type = SPICE_CURSOR_TYPE_ALPHA;
+	cursor->header.width = 64;
+	cursor->header.height = 64;
+	cursor->header.hot_spot_x = 0;
+	cursor->header.hot_spot_y = 0;
+	cursor->data_size = size;
+	cursor->chunk.next_chunk = 0;
+	cursor->chunk.prev_chunk = 0;
+	cursor->chunk.data_size = size;
+
+	memcpy(cursor->chunk.data, user_ptr, size);
+
+	qxl_bo_kunmap(cursor_bo);
+
+	/* finish with the userspace bo */
+	qxl_bo_kunmap(user_bo);
+	qxl_bo_unpin(user_bo);
+	qxl_bo_unreserve(user_bo);
+	drm_gem_object_unreference_unlocked(obj);
+
+	cmd = (struct qxl_cursor_cmd *)qxl_release_map(qdev, release);
+	cmd->type = QXL_CURSOR_SET;
+	cmd->u.set.position.x = qcrtc->cur_x;
+	cmd->u.set.position.y = qcrtc->cur_y;
+
+	cmd->u.set.shape = qxl_bo_physical_address(qdev, cursor_bo, 0);
+	qxl_release_add_res(qdev, release, cursor_bo);
+
+	cmd->u.set.visible = 1;
+	qxl_release_unmap(qdev, release, &cmd->release_info);
+
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_cursor_ring_release(qdev, release, QXL_CMD_CURSOR, false);
+	qxl_release_unreserve(qdev, release);
+
+	qxl_bo_unreserve(cursor_bo);
+	qxl_bo_unref(&cursor_bo);
+
+	return ret;
+out_free_bo:
+	qxl_bo_unref(&cursor_bo);
+out_free_release:
+	qxl_release_unreserve(qdev, release);
+	qxl_release_free(qdev, release);
+out_kunmap:
+	qxl_bo_kunmap(user_bo);
+out_unpin:
+	qxl_bo_unpin(user_bo);
+out_unreserve:
+	qxl_bo_unreserve(user_bo);
+out_unref:
+	drm_gem_object_unreference_unlocked(obj);
+	return ret;
+}
+
+static int qxl_crtc_cursor_move(struct drm_crtc *crtc,
+				int x, int y)
+{
+	struct drm_device *dev = crtc->dev;
+	struct qxl_device *qdev = dev->dev_private;
+	struct qxl_crtc *qcrtc = to_qxl_crtc(crtc);
+	struct qxl_release *release;
+	struct qxl_cursor_cmd *cmd;
+	int ret;
+
+	ret = qxl_alloc_release_reserved(qdev, sizeof(*cmd), QXL_RELEASE_CURSOR_CMD,
+				   &release, NULL);
+
+	qcrtc->cur_x = x;
+	qcrtc->cur_y = y;
+
+	cmd = (struct qxl_cursor_cmd *)qxl_release_map(qdev, release);
+	cmd->type = QXL_CURSOR_MOVE;
+	cmd->u.position.x = qcrtc->cur_x;
+	cmd->u.position.y = qcrtc->cur_y;
+	qxl_release_unmap(qdev, release, &cmd->release_info);
+
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_cursor_ring_release(qdev, release, QXL_CMD_CURSOR, false);
+	qxl_release_unreserve(qdev, release);
+	return 0;
+}
+
+
+static const struct drm_crtc_funcs qxl_crtc_funcs = {
+	.cursor_set = qxl_crtc_cursor_set,
+	.cursor_move = qxl_crtc_cursor_move,
+	.gamma_set = qxl_crtc_gamma_set,
+	.set_config = drm_crtc_helper_set_config,
+	.destroy = qxl_crtc_destroy,
+};
+
+static void qxl_user_framebuffer_destroy(struct drm_framebuffer *fb)
+{
+	struct qxl_framebuffer *qxl_fb = to_qxl_framebuffer(fb);
+
+	if (qxl_fb->obj)
+		drm_gem_object_unreference_unlocked(qxl_fb->obj);
+	drm_framebuffer_cleanup(fb);
+	kfree(qxl_fb);
+}
+
+int qxl_framebuffer_surface_dirty(struct drm_framebuffer *fb,
+				  struct drm_file *file_priv,
+				  unsigned flags, unsigned color,
+				  struct drm_clip_rect *clips,
+				  unsigned num_clips)
+{
+	/* TODO: vmwgfx where this was cribbed from had locking. Why? */
+	struct qxl_framebuffer *qxl_fb = to_qxl_framebuffer(fb);
+	struct qxl_device *qdev = qxl_fb->base.dev->dev_private;
+	struct drm_clip_rect norect;
+	struct qxl_bo *qobj;
+	int inc = 1;
+
+	qobj = gem_to_qxl_bo(qxl_fb->obj);
+	if (qxl_fb != qdev->active_user_framebuffer) {
+		DRM_INFO("%s: qxl_fb 0x%p != qdev->active_user_framebuffer 0x%p\n",
+			__func__, qxl_fb, qdev->active_user_framebuffer);
+	}
+	if (!num_clips) {
+		num_clips = 1;
+		clips = &norect;
+		norect.x1 = norect.y1 = 0;
+		norect.x2 = fb->width;
+		norect.y2 = fb->height;
+	} else if (flags & DRM_MODE_FB_DIRTY_ANNOTATE_COPY) {
+		num_clips /= 2;
+		inc = 2; /* skip source rects */
+	}
+
+	qxl_draw_dirty_fb(qdev, qxl_fb, qobj, flags, color,
+			  clips, num_clips, inc);
+	return 0;
+}
+
+static const struct drm_framebuffer_funcs qxl_fb_funcs = {
+	.destroy = qxl_user_framebuffer_destroy,
+	.dirty = qxl_framebuffer_surface_dirty,
+/*	TODO?
+ *	.create_handle = qxl_user_framebuffer_create_handle, */
+};
+
+int
+qxl_framebuffer_init(struct drm_device *dev,
+		     struct qxl_framebuffer *qfb,
+		     struct drm_mode_fb_cmd2 *mode_cmd,
+		     struct drm_gem_object *obj)
+{
+	int ret;
+
+	qfb->obj = obj;
+	ret = drm_framebuffer_init(dev, &qfb->base, &qxl_fb_funcs);
+	if (ret) {
+		qfb->obj = NULL;
+		return ret;
+	}
+	drm_helper_mode_fill_fb_struct(&qfb->base, mode_cmd);
+	return 0;
+}
+
+static void qxl_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+}
+
+static bool qxl_crtc_mode_fixup(struct drm_crtc *crtc,
+				  const struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct qxl_device *qdev = dev->dev_private;
+
+	qxl_io_log(qdev, "%s: (%d,%d) => (%d,%d)\n",
+		   __func__,
+		   mode->hdisplay, mode->vdisplay,
+		   adjusted_mode->hdisplay,
+		   adjusted_mode->vdisplay);
+	return true;
+}
+
+void
+qxl_send_monitors_config(struct qxl_device *qdev)
+{
+	int i;
+
+	BUG_ON(!qdev->ram_header->monitors_config);
+
+	if (qdev->monitors_config->count == 0) {
+		qxl_io_log(qdev, "%s: 0 monitors??\n", __func__);
+		return;
+	}
+	for (i = 0 ; i < qdev->monitors_config->count ; ++i) {
+		struct qxl_head *head = &qdev->monitors_config->heads[i];
+
+		if (head->y > 8192 || head->y < head->x ||
+		    head->width > 8192 || head->height > 8192) {
+			DRM_ERROR("head %d wrong: %dx%d+%d+%d\n",
+				  i, head->width, head->height,
+				  head->x, head->y);
+			return;
+		}
+	}
+	qxl_io_monitors_config(qdev);
+}
+
+static void qxl_monitors_config_set_single(struct qxl_device *qdev,
+					   unsigned x, unsigned y,
+					   unsigned width, unsigned height)
+{
+	DRM_DEBUG("%dx%d+%d+%d\n", width, height, x, y);
+	qdev->monitors_config->count = 1;
+	qdev->monitors_config->heads[0].x = x;
+	qdev->monitors_config->heads[0].y = y;
+	qdev->monitors_config->heads[0].width = width;
+	qdev->monitors_config->heads[0].height = height;
+}
+
+static int qxl_crtc_mode_set(struct drm_crtc *crtc,
+			       struct drm_display_mode *mode,
+			       struct drm_display_mode *adjusted_mode,
+			       int x, int y,
+			       struct drm_framebuffer *old_fb)
+{
+	struct drm_device *dev = crtc->dev;
+	struct qxl_device *qdev = dev->dev_private;
+	struct qxl_mode *m = (void *)mode->private;
+	struct qxl_framebuffer *qfb;
+	struct qxl_bo *bo, *old_bo = NULL;
+	uint32_t width, height, base_offset;
+	bool recreate_primary = false;
+	int ret;
+
+	if (!crtc->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (old_fb) {
+		qfb = to_qxl_framebuffer(old_fb);
+		old_bo = gem_to_qxl_bo(qfb->obj);
+	}
+	qfb = to_qxl_framebuffer(crtc->fb);
+	bo = gem_to_qxl_bo(qfb->obj);
+	if (!m)
+		/* and do we care? */
+		DRM_DEBUG("%dx%d: not a native mode\n", x, y);
+	else
+		DRM_DEBUG("%dx%d: qxl id %d\n",
+			  mode->hdisplay, mode->vdisplay, m->id);
+	DRM_DEBUG("+%d+%d (%d,%d) => (%d,%d)\n",
+		  x, y,
+		  mode->hdisplay, mode->vdisplay,
+		  adjusted_mode->hdisplay,
+		  adjusted_mode->vdisplay);
+
+	recreate_primary = true;
+
+	width = mode->hdisplay;
+	height = mode->vdisplay;
+	base_offset = 0;
+
+	ret = qxl_bo_reserve(bo, false);
+	if (ret != 0)
+		return ret;
+	ret = qxl_bo_pin(bo, bo->type, NULL);
+	if (ret != 0) {
+		qxl_bo_unreserve(bo);
+		return -EINVAL;
+	}
+	qxl_bo_unreserve(bo);
+	if (recreate_primary) {
+		qxl_io_destroy_primary(qdev);
+		qxl_io_log(qdev,
+			   "recreate primary: %dx%d (was %dx%d,%d,%d)\n",
+			   width, height, bo->surf.width,
+			   bo->surf.height, bo->surf.stride, bo->surf.format);
+		qxl_io_create_primary(qdev, width, height, base_offset, bo);
+		bo->is_primary = true;
+	}
+
+	if (old_bo && old_bo != bo) {
+		old_bo->is_primary = false;
+		ret = qxl_bo_reserve(old_bo, false);
+		qxl_bo_unpin(old_bo);
+		qxl_bo_unreserve(old_bo);
+	}
+
+	if (qdev->monitors_config->count == 0) {
+		qxl_monitors_config_set_single(qdev, x, y,
+					       mode->hdisplay,
+					       mode->vdisplay);
+	}
+	qdev->mode_set = true;
+	return 0;
+}
+
+static void qxl_crtc_prepare(struct drm_crtc *crtc)
+{
+	DRM_DEBUG("current: %dx%d+%d+%d (%d).\n",
+		  crtc->mode.hdisplay, crtc->mode.vdisplay,
+		  crtc->x, crtc->y, crtc->enabled);
+}
+
+static void qxl_crtc_commit(struct drm_crtc *crtc)
+{
+	DRM_DEBUG("\n");
+}
+
+void qxl_crtc_load_lut(struct drm_crtc *crtc)
+{
+	DRM_DEBUG("\n");
+}
+
+static const struct drm_crtc_helper_funcs qxl_crtc_helper_funcs = {
+	.dpms = qxl_crtc_dpms,
+	.mode_fixup = qxl_crtc_mode_fixup,
+	.mode_set = qxl_crtc_mode_set,
+	.prepare = qxl_crtc_prepare,
+	.commit = qxl_crtc_commit,
+	.load_lut = qxl_crtc_load_lut,
+};
+
+int qdev_crtc_init(struct drm_device *dev, int num_crtc)
+{
+	struct qxl_crtc *qxl_crtc;
+
+	qxl_crtc = kzalloc(sizeof(struct qxl_crtc), GFP_KERNEL);
+	if (!qxl_crtc)
+		return -ENOMEM;
+
+	drm_crtc_init(dev, &qxl_crtc->base, &qxl_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&qxl_crtc->base, 256);
+	drm_crtc_helper_add(&qxl_crtc->base, &qxl_crtc_helper_funcs);
+	return 0;
+}
+
+static void qxl_enc_dpms(struct drm_encoder *encoder, int mode)
+{
+	DRM_DEBUG("\n");
+}
+
+static bool qxl_enc_mode_fixup(struct drm_encoder *encoder,
+			       const struct drm_display_mode *mode,
+			       struct drm_display_mode *adjusted_mode)
+{
+	DRM_DEBUG("\n");
+	return true;
+}
+
+static void qxl_enc_prepare(struct drm_encoder *encoder)
+{
+	DRM_DEBUG("\n");
+}
+
+static void qxl_write_monitors_config_for_encoder(struct qxl_device *qdev,
+		struct drm_encoder *encoder)
+{
+	int i;
+	struct qxl_head *head;
+	struct drm_display_mode *mode;
+
+	BUG_ON(!encoder);
+	/* TODO: ugly, do better */
+	for (i = 0 ; (encoder->possible_crtcs != (1 << i)) && i < 32; ++i)
+		;
+	if (encoder->possible_crtcs != (1 << i)) {
+		DRM_ERROR("encoder has wrong possible_crtcs: %x\n",
+			  encoder->possible_crtcs);
+		return;
+	}
+	if (!qdev->monitors_config ||
+	    qdev->monitors_config->max_allowed <= i) {
+		DRM_ERROR(
+		"head number too large or missing monitors config: %p, %d",
+		qdev->monitors_config,
+		qdev->monitors_config ?
+			qdev->monitors_config->max_allowed : -1);
+		return;
+	}
+	if (!encoder->crtc) {
+		DRM_ERROR("missing crtc on encoder %p\n", encoder);
+		return;
+	}
+	if (i != 0)
+		DRM_DEBUG("missing for multiple monitors: no head holes\n");
+	head = &qdev->monitors_config->heads[i];
+	head->id = i;
+	head->surface_id = 0;
+	if (encoder->crtc->enabled) {
+		mode = &encoder->crtc->mode;
+		head->width = mode->hdisplay;
+		head->height = mode->vdisplay;
+		head->x = encoder->crtc->x;
+		head->y = encoder->crtc->y;
+		if (qdev->monitors_config->count < i + 1)
+			qdev->monitors_config->count = i + 1;
+	} else {
+		head->width = 0;
+		head->height = 0;
+		head->x = 0;
+		head->y = 0;
+	}
+	DRM_DEBUG("setting head %d to +%d+%d %dx%d\n",
+		  i, head->x, head->y, head->width, head->height);
+	head->flags = 0;
+	/* TODO - somewhere else to call this for multiple monitors
+	 * (config_commit?) */
+	qxl_send_monitors_config(qdev);
+}
+
+static void qxl_enc_commit(struct drm_encoder *encoder)
+{
+	struct qxl_device *qdev = encoder->dev->dev_private;
+
+	qxl_write_monitors_config_for_encoder(qdev, encoder);
+	DRM_DEBUG("\n");
+}
+
+static void qxl_enc_mode_set(struct drm_encoder *encoder,
+				struct drm_display_mode *mode,
+				struct drm_display_mode *adjusted_mode)
+{
+	DRM_DEBUG("\n");
+}
+
+static int qxl_conn_get_modes(struct drm_connector *connector)
+{
+	int ret = 0;
+	struct qxl_device *qdev = connector->dev->dev_private;
+
+	DRM_DEBUG_KMS("monitors_config=%p\n", qdev->monitors_config);
+	/* TODO: what should we do here? only show the configured modes for the
+	 * device, or allow the full list, or both? */
+	if (qdev->monitors_config && qdev->monitors_config->count) {
+		ret = qxl_add_monitors_config_modes(connector);
+		if (ret < 0)
+			return ret;
+	}
+	ret += qxl_add_common_modes(connector);
+	return ret;
+}
+
+static int qxl_conn_mode_valid(struct drm_connector *connector,
+			       struct drm_display_mode *mode)
+{
+	/* TODO: is this called for user defined modes? (xrandr --add-mode)
+	 * TODO: check that the mode fits in the framebuffer */
+	DRM_DEBUG("%s: %dx%d status=%d\n", mode->name, mode->hdisplay,
+		  mode->vdisplay, mode->status);
+	return MODE_OK;
+}
+
+struct drm_encoder *qxl_best_encoder(struct drm_connector *connector)
+{
+	struct qxl_output *qxl_output =
+		drm_connector_to_qxl_output(connector);
+
+	DRM_DEBUG("\n");
+	return &qxl_output->enc;
+}
+
+
+static const struct drm_encoder_helper_funcs qxl_enc_helper_funcs = {
+	.dpms = qxl_enc_dpms,
+	.mode_fixup = qxl_enc_mode_fixup,
+	.prepare = qxl_enc_prepare,
+	.mode_set = qxl_enc_mode_set,
+	.commit = qxl_enc_commit,
+};
+
+static const struct drm_connector_helper_funcs qxl_connector_helper_funcs = {
+	.get_modes = qxl_conn_get_modes,
+	.mode_valid = qxl_conn_mode_valid,
+	.best_encoder = qxl_best_encoder,
+};
+
+static void qxl_conn_save(struct drm_connector *connector)
+{
+	DRM_DEBUG("\n");
+}
+
+static void qxl_conn_restore(struct drm_connector *connector)
+{
+	DRM_DEBUG("\n");
+}
+
+static enum drm_connector_status qxl_conn_detect(
+			struct drm_connector *connector,
+			bool force)
+{
+	struct qxl_output *output =
+		drm_connector_to_qxl_output(connector);
+	struct drm_device *ddev = connector->dev;
+	struct qxl_device *qdev = ddev->dev_private;
+	int connected;
+
+	/* The first monitor is always connected */
+	connected = (output->index == 0) ||
+		    (qdev->monitors_config &&
+		     qdev->monitors_config->count > output->index);
+
+	DRM_DEBUG("\n");
+	return connected ? connector_status_connected
+			 : connector_status_disconnected;
+}
+
+static int qxl_conn_set_property(struct drm_connector *connector,
+				   struct drm_property *property,
+				   uint64_t value)
+{
+	DRM_DEBUG("\n");
+	return 0;
+}
+
+static void qxl_conn_destroy(struct drm_connector *connector)
+{
+	struct qxl_output *qxl_output =
+		drm_connector_to_qxl_output(connector);
+
+	drm_sysfs_connector_remove(connector);
+	drm_connector_cleanup(connector);
+	kfree(qxl_output);
+}
+
+static const struct drm_connector_funcs qxl_connector_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.save = qxl_conn_save,
+	.restore = qxl_conn_restore,
+	.detect = qxl_conn_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = qxl_conn_set_property,
+	.destroy = qxl_conn_destroy,
+};
+
+static void qxl_enc_destroy(struct drm_encoder *encoder)
+{
+	drm_encoder_cleanup(encoder);
+}
+
+static const struct drm_encoder_funcs qxl_enc_funcs = {
+	.destroy = qxl_enc_destroy,
+};
+
+int qdev_output_init(struct drm_device *dev, int num_output)
+{
+	struct qxl_output *qxl_output;
+	struct drm_connector *connector;
+	struct drm_encoder *encoder;
+
+	qxl_output = kzalloc(sizeof(struct qxl_output), GFP_KERNEL);
+	if (!qxl_output)
+		return -ENOMEM;
+
+	qxl_output->index = num_output;
+
+	connector = &qxl_output->base;
+	encoder = &qxl_output->enc;
+	drm_connector_init(dev, &qxl_output->base,
+			   &qxl_connector_funcs, DRM_MODE_CONNECTOR_VIRTUAL);
+
+	drm_encoder_init(dev, &qxl_output->enc, &qxl_enc_funcs,
+			 DRM_MODE_ENCODER_VIRTUAL);
+
+	encoder->possible_crtcs = 1 << num_output;
+	drm_mode_connector_attach_encoder(&qxl_output->base,
+					  &qxl_output->enc);
+	drm_encoder_helper_add(encoder, &qxl_enc_helper_funcs);
+	drm_connector_helper_add(connector, &qxl_connector_helper_funcs);
+
+	drm_sysfs_connector_add(connector);
+	return 0;
+}
+
+static struct drm_framebuffer *
+qxl_user_framebuffer_create(struct drm_device *dev,
+			    struct drm_file *file_priv,
+			    struct drm_mode_fb_cmd2 *mode_cmd)
+{
+	struct drm_gem_object *obj;
+	struct qxl_framebuffer *qxl_fb;
+	struct qxl_device *qdev = dev->dev_private;
+	int ret;
+
+	obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handles[0]);
+
+	qxl_fb = kzalloc(sizeof(*qxl_fb), GFP_KERNEL);
+	if (qxl_fb == NULL)
+		return NULL;
+
+	ret = qxl_framebuffer_init(dev, qxl_fb, mode_cmd, obj);
+	if (ret) {
+		kfree(qxl_fb);
+		drm_gem_object_unreference_unlocked(obj);
+		return NULL;
+	}
+
+	if (qdev->active_user_framebuffer) {
+		DRM_INFO("%s: active_user_framebuffer %p -> %p\n",
+			 __func__,
+			 qdev->active_user_framebuffer, qxl_fb);
+	}
+	qdev->active_user_framebuffer = qxl_fb;
+
+	return &qxl_fb->base;
+}
+
+static const struct drm_mode_config_funcs qxl_mode_funcs = {
+	.fb_create = qxl_user_framebuffer_create,
+};
+
+int qxl_modeset_init(struct qxl_device *qdev)
+{
+	int i;
+	int ret;
+	struct drm_gem_object *gobj;
+	int max_allowed = QXL_NUM_OUTPUTS;
+	int monitors_config_size = sizeof(struct qxl_monitors_config) +
+				   max_allowed * sizeof(struct qxl_head);
+
+	drm_mode_config_init(qdev->ddev);
+	ret = qxl_gem_object_create(qdev, monitors_config_size, 0,
+				    QXL_GEM_DOMAIN_VRAM,
+				    false, false, NULL, &gobj);
+	if (ret) {
+		DRM_ERROR("%s: failed to create gem ret=%d\n", __func__, ret);
+		return -ENOMEM;
+	}
+	qdev->monitors_config_bo = gem_to_qxl_bo(gobj);
+	qxl_bo_kmap(qdev->monitors_config_bo, NULL);
+	qdev->monitors_config = qdev->monitors_config_bo->kptr;
+	qdev->ram_header->monitors_config =
+		qxl_bo_physical_address(qdev, qdev->monitors_config_bo, 0);
+
+	memset(qdev->monitors_config, 0, monitors_config_size);
+	qdev->monitors_config->max_allowed = max_allowed;
+
+	qdev->ddev->mode_config.funcs = (void *)&qxl_mode_funcs;
+
+	/* modes will be validated against the framebuffer size */
+	qdev->ddev->mode_config.min_width = 320;
+	qdev->ddev->mode_config.min_height = 200;
+	qdev->ddev->mode_config.max_width = 8192;
+	qdev->ddev->mode_config.max_height = 8192;
+
+	qdev->ddev->mode_config.fb_base = qdev->vram_base;
+	for (i = 0 ; i < QXL_NUM_OUTPUTS; ++i) {
+		qdev_crtc_init(qdev->ddev, i);
+		qdev_output_init(qdev->ddev, i);
+	}
+
+	qdev->mode_info.mode_config_initialized = true;
+
+	/* primary surface must be created by this point, to allow
+	 * issuing command queue commands and having them read by
+	 * spice server. */
+	qxl_fbdev_init(qdev);
+	return 0;
+}
+
+void qxl_modeset_fini(struct qxl_device *qdev)
+{
+	qxl_fbdev_fini(qdev);
+	if (qdev->mode_info.mode_config_initialized) {
+		drm_mode_config_cleanup(qdev->ddev);
+		qdev->mode_info.mode_config_initialized = false;
+	}
+}
diff --git a/drivers/gpu/drm/qxl/qxl_draw.c b/drivers/gpu/drm/qxl/qxl_draw.c
new file mode 100644
index 000000000000..3c8c3dbf9378
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_draw.c
@@ -0,0 +1,390 @@
+/*
+ * Copyright 2011 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+ * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+/* returns a pointer to the already allocated qxl_rect array inside
+ * the qxl_clip_rects. This is *not* the same as the memory allocated
+ * on the device, it is offset to qxl_clip_rects.chunk.data */
+static struct qxl_rect *drawable_set_clipping(struct qxl_device *qdev,
+					      struct qxl_drawable *drawable,
+					      unsigned num_clips,
+					      struct qxl_bo **clips_bo,
+					      struct qxl_release *release)
+{
+	struct qxl_clip_rects *dev_clips;
+	int ret;
+	int size = sizeof(*dev_clips) + sizeof(struct qxl_rect) * num_clips;
+	ret = qxl_alloc_bo_reserved(qdev, size, clips_bo);
+	if (ret)
+		return NULL;
+
+	ret = qxl_bo_kmap(*clips_bo, (void **)&dev_clips);
+	if (ret) {
+		qxl_bo_unref(clips_bo);
+		return NULL;
+	}
+	dev_clips->num_rects = num_clips;
+	dev_clips->chunk.next_chunk = 0;
+	dev_clips->chunk.prev_chunk = 0;
+	dev_clips->chunk.data_size = sizeof(struct qxl_rect) * num_clips;
+	return (struct qxl_rect *)dev_clips->chunk.data;
+}
+
+static int
+make_drawable(struct qxl_device *qdev, int surface, uint8_t type,
+	      const struct qxl_rect *rect,
+	      struct qxl_release **release)
+{
+	struct qxl_drawable *drawable;
+	int i, ret;
+
+	ret = qxl_alloc_release_reserved(qdev, sizeof(*drawable),
+					 QXL_RELEASE_DRAWABLE, release,
+					 NULL);
+	if (ret)
+		return ret;
+
+	drawable = (struct qxl_drawable *)qxl_release_map(qdev, *release);
+	drawable->type = type;
+
+	drawable->surface_id = surface;		/* Only primary for now */
+	drawable->effect = QXL_EFFECT_OPAQUE;
+	drawable->self_bitmap = 0;
+	drawable->self_bitmap_area.top = 0;
+	drawable->self_bitmap_area.left = 0;
+	drawable->self_bitmap_area.bottom = 0;
+	drawable->self_bitmap_area.right = 0;
+	/* FIXME: add clipping */
+	drawable->clip.type = SPICE_CLIP_TYPE_NONE;
+
+	/*
+	 * surfaces_dest[i] should apparently be filled out with the
+	 * surfaces that we depend on, and surface_rects should be
+	 * filled with the rectangles of those surfaces that we
+	 * are going to use.
+	 */
+	for (i = 0; i < 3; ++i)
+		drawable->surfaces_dest[i] = -1;
+
+	if (rect)
+		drawable->bbox = *rect;
+
+	drawable->mm_time = qdev->rom->mm_clock;
+	qxl_release_unmap(qdev, *release, &drawable->release_info);
+	return 0;
+}
+
+static int qxl_palette_create_1bit(struct qxl_bo **palette_bo,
+				   const struct qxl_fb_image *qxl_fb_image)
+{
+	struct qxl_device *qdev = qxl_fb_image->qdev;
+	const struct fb_image *fb_image = &qxl_fb_image->fb_image;
+	uint32_t visual = qxl_fb_image->visual;
+	const uint32_t *pseudo_palette = qxl_fb_image->pseudo_palette;
+	struct qxl_palette *pal;
+	int ret;
+	uint32_t fgcolor, bgcolor;
+	static uint64_t unique; /* we make no attempt to actually set this
+				 * correctly globaly, since that would require
+				 * tracking all of our palettes. */
+
+	ret = qxl_alloc_bo_reserved(qdev,
+				    sizeof(struct qxl_palette) + sizeof(uint32_t) * 2,
+				    palette_bo);
+
+	ret = qxl_bo_kmap(*palette_bo, (void **)&pal);
+	pal->num_ents = 2;
+	pal->unique = unique++;
+	if (visual == FB_VISUAL_TRUECOLOR || visual == FB_VISUAL_DIRECTCOLOR) {
+		/* NB: this is the only used branch currently. */
+		fgcolor = pseudo_palette[fb_image->fg_color];
+		bgcolor = pseudo_palette[fb_image->bg_color];
+	} else {
+		fgcolor = fb_image->fg_color;
+		bgcolor = fb_image->bg_color;
+	}
+	pal->ents[0] = bgcolor;
+	pal->ents[1] = fgcolor;
+	qxl_bo_kunmap(*palette_bo);
+	return 0;
+}
+
+void qxl_draw_opaque_fb(const struct qxl_fb_image *qxl_fb_image,
+			int stride /* filled in if 0 */)
+{
+	struct qxl_device *qdev = qxl_fb_image->qdev;
+	struct qxl_drawable *drawable;
+	struct qxl_rect rect;
+	const struct fb_image *fb_image = &qxl_fb_image->fb_image;
+	int x = fb_image->dx;
+	int y = fb_image->dy;
+	int width = fb_image->width;
+	int height = fb_image->height;
+	const char *src = fb_image->data;
+	int depth = fb_image->depth;
+	struct qxl_release *release;
+	struct qxl_bo *image_bo;
+	struct qxl_image *image;
+	int ret;
+
+	if (stride == 0)
+		stride = depth * width / 8;
+
+	rect.left = x;
+	rect.right = x + width;
+	rect.top = y;
+	rect.bottom = y + height;
+
+	ret = make_drawable(qdev, 0, QXL_DRAW_COPY, &rect, &release);
+	if (ret)
+		return;
+
+	ret = qxl_image_create(qdev, release, &image_bo,
+			       (const uint8_t *)src, 0, 0,
+			       width, height, depth, stride);
+	if (ret) {
+		qxl_release_unreserve(qdev, release);
+		qxl_release_free(qdev, release);
+		return;
+	}
+
+	if (depth == 1) {
+		struct qxl_bo *palette_bo;
+		void *ptr;
+		ret = qxl_palette_create_1bit(&palette_bo, qxl_fb_image);
+		qxl_release_add_res(qdev, release, palette_bo);
+
+		ptr = qxl_bo_kmap_atomic_page(qdev, image_bo, 0);
+		image = ptr;
+		image->u.bitmap.palette =
+			qxl_bo_physical_address(qdev, palette_bo, 0);
+		qxl_bo_kunmap_atomic_page(qdev, image_bo, ptr);
+		qxl_bo_unreserve(palette_bo);
+		qxl_bo_unref(&palette_bo);
+	}
+
+	drawable = (struct qxl_drawable *)qxl_release_map(qdev, release);
+
+	drawable->u.copy.src_area.top = 0;
+	drawable->u.copy.src_area.bottom = height;
+	drawable->u.copy.src_area.left = 0;
+	drawable->u.copy.src_area.right = width;
+
+	drawable->u.copy.rop_descriptor = SPICE_ROPD_OP_PUT;
+	drawable->u.copy.scale_mode = 0;
+	drawable->u.copy.mask.flags = 0;
+	drawable->u.copy.mask.pos.x = 0;
+	drawable->u.copy.mask.pos.y = 0;
+	drawable->u.copy.mask.bitmap = 0;
+
+	drawable->u.copy.src_bitmap =
+		qxl_bo_physical_address(qdev, image_bo, 0);
+	qxl_release_unmap(qdev, release, &drawable->release_info);
+
+	qxl_release_add_res(qdev, release, image_bo);
+	qxl_bo_unreserve(image_bo);
+	qxl_bo_unref(&image_bo);
+
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_DRAW, false);
+	qxl_release_unreserve(qdev, release);
+}
+
+/* push a draw command using the given clipping rectangles as
+ * the sources from the shadow framebuffer.
+ *
+ * Right now implementing with a single draw and a clip list. Clip
+ * lists are known to be a problem performance wise, this can be solved
+ * by treating them differently in the server.
+ */
+void qxl_draw_dirty_fb(struct qxl_device *qdev,
+		       struct qxl_framebuffer *qxl_fb,
+		       struct qxl_bo *bo,
+		       unsigned flags, unsigned color,
+		       struct drm_clip_rect *clips,
+		       unsigned num_clips, int inc)
+{
+	/*
+	 * TODO: if flags & DRM_MODE_FB_DIRTY_ANNOTATE_FILL then we should
+	 * send a fill command instead, much cheaper.
+	 *
+	 * See include/drm/drm_mode.h
+	 */
+	struct drm_clip_rect *clips_ptr;
+	int i;
+	int left, right, top, bottom;
+	int width, height;
+	struct qxl_drawable *drawable;
+	struct qxl_rect drawable_rect;
+	struct qxl_rect *rects;
+	int stride = qxl_fb->base.pitches[0];
+	/* depth is not actually interesting, we don't mask with it */
+	int depth = qxl_fb->base.bits_per_pixel;
+	uint8_t *surface_base;
+	struct qxl_release *release;
+	struct qxl_bo *image_bo;
+	struct qxl_bo *clips_bo;
+	int ret;
+
+	left = clips->x1;
+	right = clips->x2;
+	top = clips->y1;
+	bottom = clips->y2;
+
+	/* skip the first clip rect */
+	for (i = 1, clips_ptr = clips + inc;
+	     i < num_clips; i++, clips_ptr += inc) {
+		left = min_t(int, left, (int)clips_ptr->x1);
+		right = max_t(int, right, (int)clips_ptr->x2);
+		top = min_t(int, top, (int)clips_ptr->y1);
+		bottom = max_t(int, bottom, (int)clips_ptr->y2);
+	}
+
+	width = right - left;
+	height = bottom - top;
+	drawable_rect.left = left;
+	drawable_rect.right = right;
+	drawable_rect.top = top;
+	drawable_rect.bottom = bottom;
+	ret = make_drawable(qdev, 0, QXL_DRAW_COPY, &drawable_rect,
+			    &release);
+	if (ret)
+		return;
+
+	ret = qxl_bo_kmap(bo, (void **)&surface_base);
+	if (ret)
+		goto out_unref;
+
+	ret = qxl_image_create(qdev, release, &image_bo, surface_base,
+			       left, top, width, height, depth, stride);
+	qxl_bo_kunmap(bo);
+	if (ret)
+		goto out_unref;
+
+	rects = drawable_set_clipping(qdev, drawable, num_clips, &clips_bo, release);
+	if (!rects) {
+		qxl_bo_unref(&image_bo);
+		goto out_unref;
+	}
+	drawable = (struct qxl_drawable *)qxl_release_map(qdev, release);
+
+	drawable->clip.type = SPICE_CLIP_TYPE_RECTS;
+	drawable->clip.data = qxl_bo_physical_address(qdev,
+						      clips_bo, 0);
+	qxl_release_add_res(qdev, release, clips_bo);
+
+	drawable->u.copy.src_area.top = 0;
+	drawable->u.copy.src_area.bottom = height;
+	drawable->u.copy.src_area.left = 0;
+	drawable->u.copy.src_area.right = width;
+
+	drawable->u.copy.rop_descriptor = SPICE_ROPD_OP_PUT;
+	drawable->u.copy.scale_mode = 0;
+	drawable->u.copy.mask.flags = 0;
+	drawable->u.copy.mask.pos.x = 0;
+	drawable->u.copy.mask.pos.y = 0;
+	drawable->u.copy.mask.bitmap = 0;
+
+	drawable->u.copy.src_bitmap = qxl_bo_physical_address(qdev, image_bo, 0);
+	qxl_release_unmap(qdev, release, &drawable->release_info);
+	qxl_release_add_res(qdev, release, image_bo);
+	qxl_bo_unreserve(image_bo);
+	qxl_bo_unref(&image_bo);
+	clips_ptr = clips;
+	for (i = 0; i < num_clips; i++, clips_ptr += inc) {
+		rects[i].left   = clips_ptr->x1;
+		rects[i].right  = clips_ptr->x2;
+		rects[i].top    = clips_ptr->y1;
+		rects[i].bottom = clips_ptr->y2;
+	}
+	qxl_bo_kunmap(clips_bo);
+	qxl_bo_unreserve(clips_bo);
+	qxl_bo_unref(&clips_bo);
+
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_DRAW, false);
+	qxl_release_unreserve(qdev, release);
+	return;
+
+out_unref:
+	qxl_release_unreserve(qdev, release);
+	qxl_release_free(qdev, release);
+}
+
+void qxl_draw_copyarea(struct qxl_device *qdev,
+		       u32 width, u32 height,
+		       u32 sx, u32 sy,
+		       u32 dx, u32 dy)
+{
+	struct qxl_drawable *drawable;
+	struct qxl_rect rect;
+	struct qxl_release *release;
+	int ret;
+
+	rect.left = dx;
+	rect.top = dy;
+	rect.right = dx + width;
+	rect.bottom = dy + height;
+	ret = make_drawable(qdev, 0, QXL_COPY_BITS, &rect, &release);
+	if (ret)
+		return;
+
+	drawable = (struct qxl_drawable *)qxl_release_map(qdev, release);
+	drawable->u.copy_bits.src_pos.x = sx;
+	drawable->u.copy_bits.src_pos.y = sy;
+
+	qxl_release_unmap(qdev, release, &drawable->release_info);
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_DRAW, false);
+	qxl_release_unreserve(qdev, release);
+}
+
+void qxl_draw_fill(struct qxl_draw_fill *qxl_draw_fill_rec)
+{
+	struct qxl_device *qdev = qxl_draw_fill_rec->qdev;
+	struct qxl_rect rect = qxl_draw_fill_rec->rect;
+	uint32_t color = qxl_draw_fill_rec->color;
+	uint16_t rop = qxl_draw_fill_rec->rop;
+	struct qxl_drawable *drawable;
+	struct qxl_release *release;
+	int ret;
+
+	ret = make_drawable(qdev, 0, QXL_DRAW_FILL, &rect, &release);
+	if (ret)
+		return;
+
+	drawable = (struct qxl_drawable *)qxl_release_map(qdev, release);
+	drawable->u.fill.brush.type = SPICE_BRUSH_TYPE_SOLID;
+	drawable->u.fill.brush.u.color = color;
+	drawable->u.fill.rop_descriptor = rop;
+	drawable->u.fill.mask.flags = 0;
+	drawable->u.fill.mask.pos.x = 0;
+	drawable->u.fill.mask.pos.y = 0;
+	drawable->u.fill.mask.bitmap = 0;
+
+	qxl_release_unmap(qdev, release, &drawable->release_info);
+	qxl_fence_releaseable(qdev, release);
+	qxl_push_command_ring_release(qdev, release, QXL_CMD_DRAW, false);
+	qxl_release_unreserve(qdev, release);
+}
diff --git a/drivers/gpu/drm/qxl/qxl_drv.c b/drivers/gpu/drm/qxl/qxl_drv.c
new file mode 100644
index 000000000000..d337da0a9759
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_drv.c
@@ -0,0 +1,145 @@
+/* vim: set ts=8 sw=8 tw=78 ai noexpandtab */
+/* qxl_drv.c -- QXL driver -*- linux-c -*-
+ *
+ * Copyright 2011 Red Hat, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Dave Airlie <airlie@redhat.com>
+ *    Alon Levy <alevy@redhat.com>
+ */
+
+#include <linux/module.h>
+#include <linux/console.h>
+
+#include "drmP.h"
+#include "drm/drm.h"
+
+#include "qxl_drv.h"
+
+extern int qxl_max_ioctls;
+static DEFINE_PCI_DEVICE_TABLE(pciidlist) = {
+	{ 0x1b36, 0x100, PCI_ANY_ID, PCI_ANY_ID, PCI_CLASS_DISPLAY_VGA << 8,
+	  0xffff00, 0 },
+	{ 0x1b36, 0x100, PCI_ANY_ID, PCI_ANY_ID, PCI_CLASS_DISPLAY_OTHER << 8,
+	  0xffff00, 0 },
+	{ 0, 0, 0 },
+};
+MODULE_DEVICE_TABLE(pci, pciidlist);
+
+int qxl_modeset = -1;
+
+MODULE_PARM_DESC(modeset, "Disable/Enable modesetting");
+module_param_named(modeset, qxl_modeset, int, 0400);
+
+static struct drm_driver qxl_driver;
+static struct pci_driver qxl_pci_driver;
+
+static int
+qxl_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
+{
+	if (pdev->revision < 4) {
+		DRM_ERROR("qxl too old, doesn't support client_monitors_config,"
+			  " use xf86-video-qxl in user mode");
+		return -EINVAL; /* TODO: ENODEV ? */
+	}
+	return drm_get_pci_dev(pdev, ent, &qxl_driver);
+}
+
+static void
+qxl_pci_remove(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	drm_put_dev(dev);
+}
+
+static struct pci_driver qxl_pci_driver = {
+	 .name = DRIVER_NAME,
+	 .id_table = pciidlist,
+	 .probe = qxl_pci_probe,
+	 .remove = qxl_pci_remove,
+};
+
+static const struct file_operations qxl_fops = {
+	.owner = THIS_MODULE,
+	.open = drm_open,
+	.release = drm_release,
+	.unlocked_ioctl = drm_ioctl,
+	.poll = drm_poll,
+	.fasync = drm_fasync,
+	.mmap = qxl_mmap,
+};
+
+static struct drm_driver qxl_driver = {
+	.driver_features = DRIVER_GEM | DRIVER_MODESET |
+			   DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED,
+	.dev_priv_size = 0,
+	.load = qxl_driver_load,
+	.unload = qxl_driver_unload,
+
+	.dumb_create = qxl_mode_dumb_create,
+	.dumb_map_offset = qxl_mode_dumb_mmap,
+	.dumb_destroy = qxl_mode_dumb_destroy,
+#if defined(CONFIG_DEBUG_FS)
+	.debugfs_init = qxl_debugfs_init,
+	.debugfs_cleanup = qxl_debugfs_takedown,
+#endif
+	.gem_init_object = qxl_gem_object_init,
+	.gem_free_object = qxl_gem_object_free,
+	.gem_open_object = qxl_gem_object_open,
+	.gem_close_object = qxl_gem_object_close,
+	.fops = &qxl_fops,
+	.ioctls = qxl_ioctls,
+	.irq_handler = qxl_irq_handler,
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.date = DRIVER_DATE,
+	.major = 0,
+	.minor = 1,
+	.patchlevel = 0,
+};
+
+static int __init qxl_init(void)
+{
+#ifdef CONFIG_VGA_CONSOLE
+	if (vgacon_text_force() && qxl_modeset == -1)
+		return -EINVAL;
+#endif
+
+	if (qxl_modeset == 0)
+		return -EINVAL;
+	qxl_driver.num_ioctls = qxl_max_ioctls;
+	return drm_pci_init(&qxl_driver, &qxl_pci_driver);
+}
+
+static void __exit qxl_exit(void)
+{
+	drm_pci_exit(&qxl_driver, &qxl_pci_driver);
+}
+
+module_init(qxl_init);
+module_exit(qxl_exit);
+
+MODULE_AUTHOR(DRIVER_AUTHOR);
+MODULE_DESCRIPTION(DRIVER_DESC);
+MODULE_LICENSE("GPL and additional rights");
diff --git a/drivers/gpu/drm/qxl/qxl_drv.h b/drivers/gpu/drm/qxl/qxl_drv.h
new file mode 100644
index 000000000000..52b582c211da
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_drv.h
@@ -0,0 +1,566 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+
+#ifndef QXL_DRV_H
+#define QXL_DRV_H
+
+/*
+ * Definitions taken from spice-protocol, plus kernel driver specific bits.
+ */
+
+#include <linux/workqueue.h>
+#include <linux/firmware.h>
+#include <linux/platform_device.h>
+
+#include "drmP.h"
+#include "drm_crtc.h"
+#include <ttm/ttm_bo_api.h>
+#include <ttm/ttm_bo_driver.h>
+#include <ttm/ttm_placement.h>
+#include <ttm/ttm_module.h>
+
+#include <drm/qxl_drm.h>
+#include "qxl_dev.h"
+
+#define DRIVER_AUTHOR		"Dave Airlie"
+
+#define DRIVER_NAME		"qxl"
+#define DRIVER_DESC		"RH QXL"
+#define DRIVER_DATE		"20120117"
+
+#define DRIVER_MAJOR 0
+#define DRIVER_MINOR 1
+#define DRIVER_PATCHLEVEL 0
+
+#define QXL_NUM_OUTPUTS 1
+
+#define QXL_DEBUGFS_MAX_COMPONENTS		32
+
+extern int qxl_log_level;
+
+enum {
+	QXL_INFO_LEVEL = 1,
+	QXL_DEBUG_LEVEL = 2,
+};
+
+#define QXL_INFO(qdev, fmt, ...) do { \
+		if (qxl_log_level >= QXL_INFO_LEVEL) {	\
+			qxl_io_log(qdev, fmt, __VA_ARGS__); \
+		}	\
+	} while (0)
+#define QXL_DEBUG(qdev, fmt, ...) do { \
+		if (qxl_log_level >= QXL_DEBUG_LEVEL) {	\
+			qxl_io_log(qdev, fmt, __VA_ARGS__); \
+		}	\
+	} while (0)
+#define QXL_INFO_ONCE(qdev, fmt, ...) do { \
+		static int done;		\
+		if (!done) {			\
+			done = 1;			\
+			QXL_INFO(qdev, fmt, __VA_ARGS__);	\
+		}						\
+	} while (0)
+
+#define DRM_FILE_OFFSET 0x100000000ULL
+#define DRM_FILE_PAGE_OFFSET (DRM_FILE_OFFSET >> PAGE_SHIFT)
+
+#define QXL_INTERRUPT_MASK (\
+	QXL_INTERRUPT_DISPLAY |\
+	QXL_INTERRUPT_CURSOR |\
+	QXL_INTERRUPT_IO_CMD |\
+	QXL_INTERRUPT_CLIENT_MONITORS_CONFIG)
+
+struct qxl_fence {
+	struct qxl_device *qdev;
+	uint32_t num_active_releases;
+	uint32_t *release_ids;
+	struct radix_tree_root tree;
+};
+
+struct qxl_bo {
+	/* Protected by gem.mutex */
+	struct list_head		list;
+	/* Protected by tbo.reserved */
+	u32				placements[3];
+	struct ttm_placement		placement;
+	struct ttm_buffer_object	tbo;
+	struct ttm_bo_kmap_obj		kmap;
+	unsigned			pin_count;
+	void				*kptr;
+	int                             type;
+	/* Constant after initialization */
+	struct drm_gem_object		gem_base;
+	bool is_primary; /* is this now a primary surface */
+	bool hw_surf_alloc;
+	struct qxl_surface surf;
+	uint32_t surface_id;
+	struct qxl_fence fence; /* per bo fence  - list of releases */
+	struct qxl_release *surf_create;
+	atomic_t reserve_count;
+};
+#define gem_to_qxl_bo(gobj) container_of((gobj), struct qxl_bo, gem_base)
+
+struct qxl_gem {
+	struct mutex		mutex;
+	struct list_head	objects;
+};
+
+struct qxl_bo_list {
+	struct list_head lhead;
+	struct qxl_bo *bo;
+};
+
+struct qxl_reloc_list {
+	struct list_head bos;
+};
+
+struct qxl_crtc {
+	struct drm_crtc base;
+	int cur_x;
+	int cur_y;
+};
+
+struct qxl_output {
+	int index;
+	struct drm_connector base;
+	struct drm_encoder enc;
+};
+
+struct qxl_framebuffer {
+	struct drm_framebuffer base;
+	struct drm_gem_object *obj;
+};
+
+#define to_qxl_crtc(x) container_of(x, struct qxl_crtc, base)
+#define drm_connector_to_qxl_output(x) container_of(x, struct qxl_output, base)
+#define drm_encoder_to_qxl_output(x) container_of(x, struct qxl_output, base)
+#define to_qxl_framebuffer(x) container_of(x, struct qxl_framebuffer, base)
+
+struct qxl_mman {
+	struct ttm_bo_global_ref        bo_global_ref;
+	struct drm_global_reference	mem_global_ref;
+	bool				mem_global_referenced;
+	struct ttm_bo_device		bdev;
+};
+
+struct qxl_mode_info {
+	int num_modes;
+	struct qxl_mode *modes;
+	bool mode_config_initialized;
+
+	/* pointer to fbdev info structure */
+	struct qxl_fbdev *qfbdev;
+};
+
+
+struct qxl_memslot {
+	uint8_t		generation;
+	uint64_t	start_phys_addr;
+	uint64_t	end_phys_addr;
+	uint64_t	high_bits;
+};
+
+enum {
+	QXL_RELEASE_DRAWABLE,
+	QXL_RELEASE_SURFACE_CMD,
+	QXL_RELEASE_CURSOR_CMD,
+};
+
+/* drm_ prefix to differentiate from qxl_release_info in
+ * spice-protocol/qxl_dev.h */
+#define QXL_MAX_RES 96
+struct qxl_release {
+	int id;
+	int type;
+	int bo_count;
+	uint32_t release_offset;
+	uint32_t surface_release_id;
+	struct qxl_bo *bos[QXL_MAX_RES];
+};
+
+struct qxl_fb_image {
+	struct qxl_device *qdev;
+	uint32_t pseudo_palette[16];
+	struct fb_image fb_image;
+	uint32_t visual;
+};
+
+struct qxl_draw_fill {
+	struct qxl_device *qdev;
+	struct qxl_rect rect;
+	uint32_t color;
+	uint16_t rop;
+};
+
+/*
+ * Debugfs
+ */
+struct qxl_debugfs {
+	struct drm_info_list	*files;
+	unsigned		num_files;
+};
+
+int qxl_debugfs_add_files(struct qxl_device *rdev,
+			     struct drm_info_list *files,
+			     unsigned nfiles);
+int qxl_debugfs_fence_init(struct qxl_device *rdev);
+void qxl_debugfs_remove_files(struct qxl_device *qdev);
+
+struct qxl_device;
+
+struct qxl_device {
+	struct device			*dev;
+	struct drm_device		*ddev;
+	struct pci_dev			*pdev;
+	unsigned long flags;
+
+	resource_size_t vram_base, vram_size;
+	resource_size_t surfaceram_base, surfaceram_size;
+	resource_size_t rom_base, rom_size;
+	struct qxl_rom *rom;
+
+	struct qxl_mode *modes;
+	struct qxl_bo *monitors_config_bo;
+	struct qxl_monitors_config *monitors_config;
+
+	/* last received client_monitors_config */
+	struct qxl_monitors_config *client_monitors_config;
+
+	int io_base;
+	void *ram;
+	struct qxl_mman		mman;
+	struct qxl_gem		gem;
+	struct qxl_mode_info mode_info;
+
+	/*
+	 * last created framebuffer with fb_create
+	 * only used by debugfs dumbppm
+	 */
+	struct qxl_framebuffer *active_user_framebuffer;
+
+	struct fb_info			*fbdev_info;
+	struct qxl_framebuffer	*fbdev_qfb;
+	void *ram_physical;
+
+	struct qxl_ring *release_ring;
+	struct qxl_ring *command_ring;
+	struct qxl_ring *cursor_ring;
+
+	struct qxl_ram_header *ram_header;
+	bool mode_set;
+
+	bool primary_created;
+
+	struct qxl_memslot	*mem_slots;
+	uint8_t		n_mem_slots;
+
+	uint8_t		main_mem_slot;
+	uint8_t		surfaces_mem_slot;
+	uint8_t		slot_id_bits;
+	uint8_t		slot_gen_bits;
+	uint64_t	va_slot_mask;
+
+	struct idr	release_idr;
+	spinlock_t release_idr_lock;
+	struct mutex	async_io_mutex;
+	unsigned int last_sent_io_cmd;
+
+	/* interrupt handling */
+	atomic_t irq_received;
+	atomic_t irq_received_display;
+	atomic_t irq_received_cursor;
+	atomic_t irq_received_io_cmd;
+	unsigned irq_received_error;
+	wait_queue_head_t display_event;
+	wait_queue_head_t cursor_event;
+	wait_queue_head_t io_cmd_event;
+	struct work_struct client_monitors_config_work;
+
+	/* debugfs */
+	struct qxl_debugfs	debugfs[QXL_DEBUGFS_MAX_COMPONENTS];
+	unsigned		debugfs_count;
+
+	struct mutex		update_area_mutex;
+
+	struct idr	surf_id_idr;
+	spinlock_t surf_id_idr_lock;
+	int last_alloced_surf_id;
+
+	struct mutex surf_evict_mutex;
+	struct io_mapping *vram_mapping;
+	struct io_mapping *surface_mapping;
+
+	/* */
+	struct mutex release_mutex;
+	struct qxl_bo *current_release_bo[3];
+	int current_release_bo_offset[3];
+
+	struct workqueue_struct *gc_queue;
+	struct work_struct gc_work;
+
+};
+
+/* forward declaration for QXL_INFO_IO */
+void qxl_io_log(struct qxl_device *qdev, const char *fmt, ...);
+
+extern struct drm_ioctl_desc qxl_ioctls[];
+extern int qxl_max_ioctl;
+
+int qxl_driver_load(struct drm_device *dev, unsigned long flags);
+int qxl_driver_unload(struct drm_device *dev);
+
+int qxl_modeset_init(struct qxl_device *qdev);
+void qxl_modeset_fini(struct qxl_device *qdev);
+
+int qxl_bo_init(struct qxl_device *qdev);
+void qxl_bo_fini(struct qxl_device *qdev);
+
+struct qxl_ring *qxl_ring_create(struct qxl_ring_header *header,
+				 int element_size,
+				 int n_elements,
+				 int prod_notify,
+				 bool set_prod_notify,
+				 wait_queue_head_t *push_event);
+void qxl_ring_free(struct qxl_ring *ring);
+
+static inline void *
+qxl_fb_virtual_address(struct qxl_device *qdev, unsigned long physical)
+{
+	QXL_INFO(qdev, "not implemented (%lu)\n", physical);
+	return 0;
+}
+
+static inline uint64_t
+qxl_bo_physical_address(struct qxl_device *qdev, struct qxl_bo *bo,
+			unsigned long offset)
+{
+	int slot_id = bo->type == QXL_GEM_DOMAIN_VRAM ? qdev->main_mem_slot : qdev->surfaces_mem_slot;
+	struct qxl_memslot *slot = &(qdev->mem_slots[slot_id]);
+
+	/* TODO - need to hold one of the locks to read tbo.offset */
+	return slot->high_bits | (bo->tbo.offset + offset);
+}
+
+/* qxl_fb.c */
+#define QXLFB_CONN_LIMIT 1
+
+int qxl_fbdev_init(struct qxl_device *qdev);
+void qxl_fbdev_fini(struct qxl_device *qdev);
+int qxl_get_handle_for_primary_fb(struct qxl_device *qdev,
+				  struct drm_file *file_priv,
+				  uint32_t *handle);
+
+/* qxl_display.c */
+int
+qxl_framebuffer_init(struct drm_device *dev,
+		     struct qxl_framebuffer *rfb,
+		     struct drm_mode_fb_cmd2 *mode_cmd,
+		     struct drm_gem_object *obj);
+void qxl_display_read_client_monitors_config(struct qxl_device *qdev);
+void qxl_send_monitors_config(struct qxl_device *qdev);
+
+/* used by qxl_debugfs only */
+void qxl_crtc_set_from_monitors_config(struct qxl_device *qdev);
+void qxl_alloc_client_monitors_config(struct qxl_device *qdev, unsigned count);
+
+/* qxl_gem.c */
+int qxl_gem_init(struct qxl_device *qdev);
+void qxl_gem_fini(struct qxl_device *qdev);
+int qxl_gem_object_create(struct qxl_device *qdev, int size,
+			  int alignment, int initial_domain,
+			  bool discardable, bool kernel,
+			  struct qxl_surface *surf,
+			  struct drm_gem_object **obj);
+int qxl_gem_object_pin(struct drm_gem_object *obj, uint32_t pin_domain,
+			  uint64_t *gpu_addr);
+void qxl_gem_object_unpin(struct drm_gem_object *obj);
+int qxl_gem_object_create_with_handle(struct qxl_device *qdev,
+				      struct drm_file *file_priv,
+				      u32 domain,
+				      size_t size,
+				      struct qxl_surface *surf,
+				      struct qxl_bo **qobj,
+				      uint32_t *handle);
+int qxl_gem_object_init(struct drm_gem_object *obj);
+void qxl_gem_object_free(struct drm_gem_object *gobj);
+int qxl_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_priv);
+void qxl_gem_object_close(struct drm_gem_object *obj,
+			  struct drm_file *file_priv);
+void qxl_bo_force_delete(struct qxl_device *qdev);
+int qxl_bo_kmap(struct qxl_bo *bo, void **ptr);
+
+/* qxl_dumb.c */
+int qxl_mode_dumb_create(struct drm_file *file_priv,
+			 struct drm_device *dev,
+			 struct drm_mode_create_dumb *args);
+int qxl_mode_dumb_destroy(struct drm_file *file_priv,
+			  struct drm_device *dev,
+			  uint32_t handle);
+int qxl_mode_dumb_mmap(struct drm_file *filp,
+		       struct drm_device *dev,
+		       uint32_t handle, uint64_t *offset_p);
+
+
+/* qxl ttm */
+int qxl_ttm_init(struct qxl_device *qdev);
+void qxl_ttm_fini(struct qxl_device *qdev);
+int qxl_mmap(struct file *filp, struct vm_area_struct *vma);
+
+/* qxl image */
+
+int qxl_image_create(struct qxl_device *qdev,
+		     struct qxl_release *release,
+		     struct qxl_bo **image_bo,
+		     const uint8_t *data,
+		     int x, int y, int width, int height,
+		     int depth, int stride);
+void qxl_update_screen(struct qxl_device *qxl);
+
+/* qxl io operations (qxl_cmd.c) */
+
+void qxl_io_create_primary(struct qxl_device *qdev,
+			   unsigned width, unsigned height, unsigned offset,
+			   struct qxl_bo *bo);
+void qxl_io_destroy_primary(struct qxl_device *qdev);
+void qxl_io_memslot_add(struct qxl_device *qdev, uint8_t id);
+void qxl_io_notify_oom(struct qxl_device *qdev);
+
+int qxl_io_update_area(struct qxl_device *qdev, struct qxl_bo *surf,
+		       const struct qxl_rect *area);
+
+void qxl_io_reset(struct qxl_device *qdev);
+void qxl_io_monitors_config(struct qxl_device *qdev);
+int qxl_ring_push(struct qxl_ring *ring, const void *new_elt, bool interruptible);
+void qxl_io_flush_release(struct qxl_device *qdev);
+void qxl_io_flush_surfaces(struct qxl_device *qdev);
+
+int qxl_release_reserve(struct qxl_device *qdev,
+			struct qxl_release *release, bool no_wait);
+void qxl_release_unreserve(struct qxl_device *qdev,
+			   struct qxl_release *release);
+union qxl_release_info *qxl_release_map(struct qxl_device *qdev,
+					struct qxl_release *release);
+void qxl_release_unmap(struct qxl_device *qdev,
+		       struct qxl_release *release,
+		       union qxl_release_info *info);
+/*
+ * qxl_bo_add_resource.
+ *
+ */
+void qxl_bo_add_resource(struct qxl_bo *main_bo, struct qxl_bo *resource);
+
+int qxl_alloc_surface_release_reserved(struct qxl_device *qdev,
+				       enum qxl_surface_cmd_type surface_cmd_type,
+				       struct qxl_release *create_rel,
+				       struct qxl_release **release);
+int qxl_alloc_release_reserved(struct qxl_device *qdev, unsigned long size,
+			       int type, struct qxl_release **release,
+			       struct qxl_bo **rbo);
+int qxl_fence_releaseable(struct qxl_device *qdev,
+			  struct qxl_release *release);
+int
+qxl_push_command_ring_release(struct qxl_device *qdev, struct qxl_release *release,
+			      uint32_t type, bool interruptible);
+int
+qxl_push_cursor_ring_release(struct qxl_device *qdev, struct qxl_release *release,
+			     uint32_t type, bool interruptible);
+int qxl_alloc_bo_reserved(struct qxl_device *qdev, unsigned long size,
+			  struct qxl_bo **_bo);
+/* qxl drawing commands */
+
+void qxl_draw_opaque_fb(const struct qxl_fb_image *qxl_fb_image,
+			int stride /* filled in if 0 */);
+
+void qxl_draw_dirty_fb(struct qxl_device *qdev,
+		       struct qxl_framebuffer *qxl_fb,
+		       struct qxl_bo *bo,
+		       unsigned flags, unsigned color,
+		       struct drm_clip_rect *clips,
+		       unsigned num_clips, int inc);
+
+void qxl_draw_fill(struct qxl_draw_fill *qxl_draw_fill_rec);
+
+void qxl_draw_copyarea(struct qxl_device *qdev,
+		       u32 width, u32 height,
+		       u32 sx, u32 sy,
+		       u32 dx, u32 dy);
+
+uint64_t
+qxl_release_alloc(struct qxl_device *qdev, int type,
+		  struct qxl_release **ret);
+
+void qxl_release_free(struct qxl_device *qdev,
+		      struct qxl_release *release);
+void qxl_release_add_res(struct qxl_device *qdev,
+			 struct qxl_release *release,
+			 struct qxl_bo *bo);
+/* used by qxl_debugfs_release */
+struct qxl_release *qxl_release_from_id_locked(struct qxl_device *qdev,
+						   uint64_t id);
+
+bool qxl_queue_garbage_collect(struct qxl_device *qdev, bool flush);
+int qxl_garbage_collect(struct qxl_device *qdev);
+
+/* debugfs */
+
+int qxl_debugfs_init(struct drm_minor *minor);
+void qxl_debugfs_takedown(struct drm_minor *minor);
+
+/* qxl_irq.c */
+int qxl_irq_init(struct qxl_device *qdev);
+irqreturn_t qxl_irq_handler(DRM_IRQ_ARGS);
+
+/* qxl_fb.c */
+int qxl_fb_init(struct qxl_device *qdev);
+
+int qxl_debugfs_add_files(struct qxl_device *qdev,
+			  struct drm_info_list *files,
+			  unsigned nfiles);
+
+int qxl_surface_id_alloc(struct qxl_device *qdev,
+			 struct qxl_bo *surf);
+void qxl_surface_id_dealloc(struct qxl_device *qdev,
+			    uint32_t surface_id);
+int qxl_hw_surface_alloc(struct qxl_device *qdev,
+			 struct qxl_bo *surf,
+			 struct ttm_mem_reg *mem);
+int qxl_hw_surface_dealloc(struct qxl_device *qdev,
+			   struct qxl_bo *surf);
+
+int qxl_bo_check_id(struct qxl_device *qdev, struct qxl_bo *bo);
+
+struct qxl_drv_surface *
+qxl_surface_lookup(struct drm_device *dev, int surface_id);
+void qxl_surface_evict(struct qxl_device *qdev, struct qxl_bo *surf, bool freeing);
+int qxl_update_surface(struct qxl_device *qdev, struct qxl_bo *surf);
+
+/* qxl_fence.c */
+int qxl_fence_add_release(struct qxl_fence *qfence, uint32_t rel_id);
+int qxl_fence_remove_release(struct qxl_fence *qfence, uint32_t rel_id);
+int qxl_fence_init(struct qxl_device *qdev, struct qxl_fence *qfence);
+void qxl_fence_fini(struct qxl_fence *qfence);
+
+#endif
diff --git a/drivers/gpu/drm/qxl/qxl_dumb.c b/drivers/gpu/drm/qxl/qxl_dumb.c
new file mode 100644
index 000000000000..847c4ee798f7
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_dumb.c
@@ -0,0 +1,93 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+/* dumb ioctls implementation */
+
+int qxl_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct qxl_bo *qobj;
+	uint32_t handle;
+	int r;
+	struct qxl_surface surf;
+	uint32_t pitch, format;
+	pitch = args->width * ((args->bpp + 1) / 8);
+	args->size = pitch * args->height;
+	args->size = ALIGN(args->size, PAGE_SIZE);
+
+	switch (args->bpp) {
+	case 16:
+		format = SPICE_SURFACE_FMT_16_565;
+		break;
+	case 32:
+		format = SPICE_SURFACE_FMT_32_xRGB;
+		break;
+	default:
+		return -EINVAL;
+	}
+	  
+	surf.width = args->width;
+	surf.height = args->height;
+	surf.stride = pitch;
+	surf.format = format;
+	r = qxl_gem_object_create_with_handle(qdev, file_priv,
+					      QXL_GEM_DOMAIN_VRAM,
+					      args->size, &surf, &qobj,
+					      &handle);
+	if (r)
+		return r;
+	args->pitch = pitch;
+	args->handle = handle;
+	return 0;
+}
+
+int qxl_mode_dumb_destroy(struct drm_file *file_priv,
+			     struct drm_device *dev,
+			     uint32_t handle)
+{
+	return drm_gem_handle_delete(file_priv, handle);
+}
+
+int qxl_mode_dumb_mmap(struct drm_file *file_priv,
+		       struct drm_device *dev,
+		       uint32_t handle, uint64_t *offset_p)
+{
+	struct drm_gem_object *gobj;
+	struct qxl_bo *qobj;
+
+	BUG_ON(!offset_p);
+	gobj = drm_gem_object_lookup(dev, file_priv, handle);
+	if (gobj == NULL)
+		return -ENOENT;
+	qobj = gem_to_qxl_bo(gobj);
+	*offset_p = qxl_bo_mmap_offset(qobj);
+	drm_gem_object_unreference_unlocked(gobj);
+	return 0;
+}
diff --git a/drivers/gpu/drm/qxl/qxl_fb.c b/drivers/gpu/drm/qxl/qxl_fb.c
new file mode 100644
index 000000000000..232b52b50194
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_fb.c
@@ -0,0 +1,567 @@
+/*
+ * Copyright © 2013 Red Hat
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *     David Airlie
+ */
+#include <linux/module.h>
+#include <linux/fb.h>
+
+#include "drmP.h"
+#include "drm/drm.h"
+#include "drm/drm_crtc.h"
+#include "drm/drm_crtc_helper.h"
+#include "qxl_drv.h"
+
+#include "qxl_object.h"
+#include "drm_fb_helper.h"
+
+#define QXL_DIRTY_DELAY (HZ / 30)
+
+struct qxl_fbdev {
+	struct drm_fb_helper helper;
+	struct qxl_framebuffer	qfb;
+	struct list_head	fbdev_list;
+	struct qxl_device	*qdev;
+
+	void *shadow;
+	int size;
+
+	/* dirty memory logging */
+	struct {
+		spinlock_t lock;
+		bool active;
+		unsigned x1;
+		unsigned y1;
+		unsigned x2;
+		unsigned y2;
+	} dirty;
+};
+
+static void qxl_fb_image_init(struct qxl_fb_image *qxl_fb_image,
+			      struct qxl_device *qdev, struct fb_info *info,
+			      const struct fb_image *image)
+{
+	qxl_fb_image->qdev = qdev;
+	if (info) {
+		qxl_fb_image->visual = info->fix.visual;
+		if (qxl_fb_image->visual == FB_VISUAL_TRUECOLOR ||
+		    qxl_fb_image->visual == FB_VISUAL_DIRECTCOLOR)
+			memcpy(&qxl_fb_image->pseudo_palette,
+			       info->pseudo_palette,
+			       sizeof(qxl_fb_image->pseudo_palette));
+	} else {
+		 /* fallback */
+		if (image->depth == 1)
+			qxl_fb_image->visual = FB_VISUAL_MONO10;
+		else
+			qxl_fb_image->visual = FB_VISUAL_DIRECTCOLOR;
+	}
+	if (image) {
+		memcpy(&qxl_fb_image->fb_image, image,
+		       sizeof(qxl_fb_image->fb_image));
+	}
+}
+
+static void qxl_fb_dirty_flush(struct fb_info *info)
+{
+	struct qxl_fbdev *qfbdev = info->par;
+	struct qxl_device *qdev = qfbdev->qdev;
+	struct qxl_fb_image qxl_fb_image;
+	struct fb_image *image = &qxl_fb_image.fb_image;
+	u32 x1, x2, y1, y2;
+
+	/* TODO: hard coding 32 bpp */
+	int stride = qfbdev->qfb.base.pitches[0] * 4;
+
+	x1 = qfbdev->dirty.x1;
+	x2 = qfbdev->dirty.x2;
+	y1 = qfbdev->dirty.y1;
+	y2 = qfbdev->dirty.y2;
+	/*
+	 * we are using a shadow draw buffer, at qdev->surface0_shadow
+	 */
+	qxl_io_log(qdev, "dirty x[%d, %d], y[%d, %d]", x1, x2, y1, y2);
+	image->dx = x1;
+	image->dy = y1;
+	image->width = x2 - x1;
+	image->height = y2 - y1;
+	image->fg_color = 0xffffffff; /* unused, just to avoid uninitialized
+					 warnings */
+	image->bg_color = 0;
+	image->depth = 32;	     /* TODO: take from somewhere? */
+	image->cmap.start = 0;
+	image->cmap.len = 0;
+	image->cmap.red = NULL;
+	image->cmap.green = NULL;
+	image->cmap.blue = NULL;
+	image->cmap.transp = NULL;
+	image->data = qfbdev->shadow + (x1 * 4) + (stride * y1);
+
+	qxl_fb_image_init(&qxl_fb_image, qdev, info, NULL);
+	qxl_draw_opaque_fb(&qxl_fb_image, stride);
+	qfbdev->dirty.x1 = 0;
+	qfbdev->dirty.x2 = 0;
+	qfbdev->dirty.y1 = 0;
+	qfbdev->dirty.y2 = 0;
+}
+
+static void qxl_deferred_io(struct fb_info *info,
+			    struct list_head *pagelist)
+{
+	struct qxl_fbdev *qfbdev = info->par;
+	unsigned long start, end, min, max;
+	struct page *page;
+	int y1, y2;
+
+	min = ULONG_MAX;
+	max = 0;
+	list_for_each_entry(page, pagelist, lru) {
+		start = page->index << PAGE_SHIFT;
+		end = start + PAGE_SIZE - 1;
+		min = min(min, start);
+		max = max(max, end);
+	}
+
+	if (min < max) {
+		y1 = min / info->fix.line_length;
+		y2 = (max / info->fix.line_length) + 1;
+
+		/* TODO: add spin lock? */
+		/* spin_lock_irqsave(&qfbdev->dirty.lock, flags); */
+		qfbdev->dirty.x1 = 0;
+		qfbdev->dirty.y1 = y1;
+		qfbdev->dirty.x2 = info->var.xres;
+		qfbdev->dirty.y2 = y2;
+		/* spin_unlock_irqrestore(&qfbdev->dirty.lock, flags); */
+	}
+
+	qxl_fb_dirty_flush(info);
+};
+
+
+struct fb_deferred_io qxl_defio = {
+	.delay		= QXL_DIRTY_DELAY,
+	.deferred_io	= qxl_deferred_io,
+};
+
+static void qxl_fb_fillrect(struct fb_info *info,
+			    const struct fb_fillrect *fb_rect)
+{
+	struct qxl_fbdev *qfbdev = info->par;
+	struct qxl_device *qdev = qfbdev->qdev;
+	struct qxl_rect rect;
+	uint32_t color;
+	int x = fb_rect->dx;
+	int y = fb_rect->dy;
+	int width = fb_rect->width;
+	int height = fb_rect->height;
+	uint16_t rop;
+	struct qxl_draw_fill qxl_draw_fill_rec;
+
+	if (info->fix.visual == FB_VISUAL_TRUECOLOR ||
+	    info->fix.visual == FB_VISUAL_DIRECTCOLOR)
+		color = ((u32 *) (info->pseudo_palette))[fb_rect->color];
+	else
+		color = fb_rect->color;
+	rect.left = x;
+	rect.right = x + width;
+	rect.top = y;
+	rect.bottom = y + height;
+	switch (fb_rect->rop) {
+	case ROP_XOR:
+		rop = SPICE_ROPD_OP_XOR;
+		break;
+	case ROP_COPY:
+		rop = SPICE_ROPD_OP_PUT;
+		break;
+	default:
+		pr_err("qxl_fb_fillrect(): unknown rop, "
+		       "defaulting to SPICE_ROPD_OP_PUT\n");
+		rop = SPICE_ROPD_OP_PUT;
+	}
+	qxl_draw_fill_rec.qdev = qdev;
+	qxl_draw_fill_rec.rect = rect;
+	qxl_draw_fill_rec.color = color;
+	qxl_draw_fill_rec.rop = rop;
+	if (!drm_can_sleep()) {
+		qxl_io_log(qdev,
+			"%s: TODO use RCU, mysterious locks with spin_lock\n",
+			__func__);
+		return;
+	}
+	qxl_draw_fill(&qxl_draw_fill_rec);
+}
+
+static void qxl_fb_copyarea(struct fb_info *info,
+			    const struct fb_copyarea *region)
+{
+	struct qxl_fbdev *qfbdev = info->par;
+
+	qxl_draw_copyarea(qfbdev->qdev,
+			  region->width, region->height,
+			  region->sx, region->sy,
+			  region->dx, region->dy);
+}
+
+static void qxl_fb_imageblit_safe(struct qxl_fb_image *qxl_fb_image)
+{
+	qxl_draw_opaque_fb(qxl_fb_image, 0);
+}
+
+static void qxl_fb_imageblit(struct fb_info *info,
+			     const struct fb_image *image)
+{
+	struct qxl_fbdev *qfbdev = info->par;
+	struct qxl_device *qdev = qfbdev->qdev;
+	struct qxl_fb_image qxl_fb_image;
+
+	if (!drm_can_sleep()) {
+		/* we cannot do any ttm_bo allocation since that will fail on
+		 * ioremap_wc..__get_vm_area_node, so queue the work item
+		 * instead This can happen from printk inside an interrupt
+		 * context, i.e.: smp_apic_timer_interrupt..check_cpu_stall */
+		qxl_io_log(qdev,
+			"%s: TODO use RCU, mysterious locks with spin_lock\n",
+			   __func__);
+		return;
+	}
+
+	/* ensure proper order of rendering operations - TODO: must do this
+	 * for everything. */
+	qxl_fb_image_init(&qxl_fb_image, qfbdev->qdev, info, image);
+	qxl_fb_imageblit_safe(&qxl_fb_image);
+}
+
+int qxl_fb_init(struct qxl_device *qdev)
+{
+	return 0;
+}
+
+static struct fb_ops qxlfb_ops = {
+	.owner = THIS_MODULE,
+	.fb_check_var = drm_fb_helper_check_var,
+	.fb_set_par = drm_fb_helper_set_par, /* TODO: copy vmwgfx */
+	.fb_fillrect = qxl_fb_fillrect,
+	.fb_copyarea = qxl_fb_copyarea,
+	.fb_imageblit = qxl_fb_imageblit,
+	.fb_pan_display = drm_fb_helper_pan_display,
+	.fb_blank = drm_fb_helper_blank,
+	.fb_setcmap = drm_fb_helper_setcmap,
+	.fb_debug_enter = drm_fb_helper_debug_enter,
+	.fb_debug_leave = drm_fb_helper_debug_leave,
+};
+
+static void qxlfb_destroy_pinned_object(struct drm_gem_object *gobj)
+{
+	struct qxl_bo *qbo = gem_to_qxl_bo(gobj);
+	int ret;
+
+	ret = qxl_bo_reserve(qbo, false);
+	if (likely(ret == 0)) {
+		qxl_bo_kunmap(qbo);
+		qxl_bo_unpin(qbo);
+		qxl_bo_unreserve(qbo);
+	}
+	drm_gem_object_unreference_unlocked(gobj);
+}
+
+int qxl_get_handle_for_primary_fb(struct qxl_device *qdev,
+				  struct drm_file *file_priv,
+				  uint32_t *handle)
+{
+	int r;
+	struct drm_gem_object *gobj = qdev->fbdev_qfb->obj;
+
+	BUG_ON(!gobj);
+	/* drm_get_handle_create adds a reference - good */
+	r = drm_gem_handle_create(file_priv, gobj, handle);
+	if (r)
+		return r;
+	return 0;
+}
+
+static int qxlfb_create_pinned_object(struct qxl_fbdev *qfbdev,
+				      struct drm_mode_fb_cmd2 *mode_cmd,
+				      struct drm_gem_object **gobj_p)
+{
+	struct qxl_device *qdev = qfbdev->qdev;
+	struct drm_gem_object *gobj = NULL;
+	struct qxl_bo *qbo = NULL;
+	int ret;
+	int aligned_size, size;
+	int height = mode_cmd->height;
+	int bpp;
+	int depth;
+
+	drm_fb_get_bpp_depth(mode_cmd->pixel_format, &bpp, &depth);
+
+	size = mode_cmd->pitches[0] * height;
+	aligned_size = ALIGN(size, PAGE_SIZE);
+	/* TODO: unallocate and reallocate surface0 for real. Hack to just
+	 * have a large enough surface0 for 1024x768 Xorg 32bpp mode */
+	ret = qxl_gem_object_create(qdev, aligned_size, 0,
+				    QXL_GEM_DOMAIN_SURFACE,
+				    false, /* is discardable */
+				    false, /* is kernel (false means device) */
+				    NULL,
+				    &gobj);
+	if (ret) {
+		pr_err("failed to allocate framebuffer (%d)\n",
+		       aligned_size);
+		return -ENOMEM;
+	}
+	qbo = gem_to_qxl_bo(gobj);
+
+	qbo->surf.width = mode_cmd->width;
+	qbo->surf.height = mode_cmd->height;
+	qbo->surf.stride = mode_cmd->pitches[0];
+	qbo->surf.format = SPICE_SURFACE_FMT_32_xRGB;
+	ret = qxl_bo_reserve(qbo, false);
+	if (unlikely(ret != 0))
+		goto out_unref;
+	ret = qxl_bo_pin(qbo, QXL_GEM_DOMAIN_SURFACE, NULL);
+	if (ret) {
+		qxl_bo_unreserve(qbo);
+		goto out_unref;
+	}
+	ret = qxl_bo_kmap(qbo, NULL);
+	qxl_bo_unreserve(qbo); /* unreserve, will be mmaped */
+	if (ret)
+		goto out_unref;
+
+	*gobj_p = gobj;
+	return 0;
+out_unref:
+	qxlfb_destroy_pinned_object(gobj);
+	*gobj_p = NULL;
+	return ret;
+}
+
+static int qxlfb_create(struct qxl_fbdev *qfbdev,
+			struct drm_fb_helper_surface_size *sizes)
+{
+	struct qxl_device *qdev = qfbdev->qdev;
+	struct fb_info *info;
+	struct drm_framebuffer *fb = NULL;
+	struct drm_mode_fb_cmd2 mode_cmd;
+	struct drm_gem_object *gobj = NULL;
+	struct qxl_bo *qbo = NULL;
+	struct device *device = &qdev->pdev->dev;
+	int ret;
+	int size;
+	int bpp = sizes->surface_bpp;
+	int depth = sizes->surface_depth;
+	void *shadow;
+
+	mode_cmd.width = sizes->surface_width;
+	mode_cmd.height = sizes->surface_height;
+
+	mode_cmd.pitches[0] = ALIGN(mode_cmd.width * ((bpp + 1) / 8), 64);
+	mode_cmd.pixel_format = drm_mode_legacy_fb_format(bpp, depth);
+
+	ret = qxlfb_create_pinned_object(qfbdev, &mode_cmd, &gobj);
+	qbo = gem_to_qxl_bo(gobj);
+	QXL_INFO(qdev, "%s: %dx%d %d\n", __func__, mode_cmd.width,
+		 mode_cmd.height, mode_cmd.pitches[0]);
+
+	shadow = vmalloc(mode_cmd.pitches[0] * mode_cmd.height);
+	/* TODO: what's the usual response to memory allocation errors? */
+	BUG_ON(!shadow);
+	QXL_INFO(qdev,
+	"surface0 at gpu offset %lld, mmap_offset %lld (virt %p, shadow %p)\n",
+		 qxl_bo_gpu_offset(qbo),
+		 qxl_bo_mmap_offset(qbo),
+		 qbo->kptr,
+		 shadow);
+	size = mode_cmd.pitches[0] * mode_cmd.height;
+
+	info = framebuffer_alloc(0, device);
+	if (info == NULL) {
+		ret = -ENOMEM;
+		goto out_unref;
+	}
+
+	info->par = qfbdev;
+
+	qxl_framebuffer_init(qdev->ddev, &qfbdev->qfb, &mode_cmd, gobj);
+
+	fb = &qfbdev->qfb.base;
+
+	/* setup helper with fb data */
+	qfbdev->helper.fb = fb;
+	qfbdev->helper.fbdev = info;
+	qfbdev->shadow = shadow;
+	strcpy(info->fix.id, "qxldrmfb");
+
+	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->depth);
+
+	info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA | FBINFO_HWACCEL_FILLRECT;
+	info->fbops = &qxlfb_ops;
+
+	/*
+	 * TODO: using gobj->size in various places in this function. Not sure
+	 * what the difference between the different sizes is.
+	 */
+	info->fix.smem_start = qdev->vram_base; /* TODO - correct? */
+	info->fix.smem_len = gobj->size;
+	info->screen_base = qfbdev->shadow;
+	info->screen_size = gobj->size;
+
+	drm_fb_helper_fill_var(info, &qfbdev->helper, sizes->fb_width,
+			       sizes->fb_height);
+
+	/* setup aperture base/size for vesafb takeover */
+	info->apertures = alloc_apertures(1);
+	if (!info->apertures) {
+		ret = -ENOMEM;
+		goto out_unref;
+	}
+	info->apertures->ranges[0].base = qdev->ddev->mode_config.fb_base;
+	info->apertures->ranges[0].size = qdev->vram_size;
+
+	info->fix.mmio_start = 0;
+	info->fix.mmio_len = 0;
+
+	if (info->screen_base == NULL) {
+		ret = -ENOSPC;
+		goto out_unref;
+	}
+
+	ret = fb_alloc_cmap(&info->cmap, 256, 0);
+	if (ret) {
+		ret = -ENOMEM;
+		goto out_unref;
+	}
+
+	info->fbdefio = &qxl_defio;
+	fb_deferred_io_init(info);
+
+	qdev->fbdev_info = info;
+	qdev->fbdev_qfb = &qfbdev->qfb;
+	DRM_INFO("fb mappable at 0x%lX, size %lu\n",  info->fix.smem_start, (unsigned long)info->screen_size);
+	DRM_INFO("fb: depth %d, pitch %d, width %d, height %d\n", fb->depth, fb->pitches[0], fb->width, fb->height);
+	return 0;
+
+out_unref:
+	if (qbo) {
+		ret = qxl_bo_reserve(qbo, false);
+		if (likely(ret == 0)) {
+			qxl_bo_kunmap(qbo);
+			qxl_bo_unpin(qbo);
+			qxl_bo_unreserve(qbo);
+		}
+	}
+	if (fb && ret) {
+		drm_gem_object_unreference(gobj);
+		drm_framebuffer_cleanup(fb);
+		kfree(fb);
+	}
+	drm_gem_object_unreference(gobj);
+	return ret;
+}
+
+static int qxl_fb_find_or_create_single(
+		struct drm_fb_helper *helper,
+		struct drm_fb_helper_surface_size *sizes)
+{
+	struct qxl_fbdev *qfbdev = (struct qxl_fbdev *)helper;
+	int new_fb = 0;
+	int ret;
+
+	if (!helper->fb) {
+		ret = qxlfb_create(qfbdev, sizes);
+		if (ret)
+			return ret;
+		new_fb = 1;
+	}
+	return new_fb;
+}
+
+static int qxl_fbdev_destroy(struct drm_device *dev, struct qxl_fbdev *qfbdev)
+{
+	struct fb_info *info;
+	struct qxl_framebuffer *qfb = &qfbdev->qfb;
+
+	if (qfbdev->helper.fbdev) {
+		info = qfbdev->helper.fbdev;
+
+		unregister_framebuffer(info);
+		framebuffer_release(info);
+	}
+	if (qfb->obj) {
+		qxlfb_destroy_pinned_object(qfb->obj);
+		qfb->obj = NULL;
+	}
+	drm_fb_helper_fini(&qfbdev->helper);
+	vfree(qfbdev->shadow);
+	drm_framebuffer_cleanup(&qfb->base);
+
+	return 0;
+}
+
+static struct drm_fb_helper_funcs qxl_fb_helper_funcs = {
+	/* TODO
+	.gamma_set = qxl_crtc_fb_gamma_set,
+	.gamma_get = qxl_crtc_fb_gamma_get,
+	*/
+	.fb_probe = qxl_fb_find_or_create_single,
+};
+
+int qxl_fbdev_init(struct qxl_device *qdev)
+{
+	struct qxl_fbdev *qfbdev;
+	int bpp_sel = 32; /* TODO: parameter from somewhere? */
+	int ret;
+
+	qfbdev = kzalloc(sizeof(struct qxl_fbdev), GFP_KERNEL);
+	if (!qfbdev)
+		return -ENOMEM;
+
+	qfbdev->qdev = qdev;
+	qdev->mode_info.qfbdev = qfbdev;
+	qfbdev->helper.funcs = &qxl_fb_helper_funcs;
+
+	ret = drm_fb_helper_init(qdev->ddev, &qfbdev->helper,
+				 1 /* num_crtc - QXL supports just 1 */,
+				 QXLFB_CONN_LIMIT);
+	if (ret) {
+		kfree(qfbdev);
+		return ret;
+	}
+
+	drm_fb_helper_single_add_all_connectors(&qfbdev->helper);
+	drm_fb_helper_initial_config(&qfbdev->helper, bpp_sel);
+	return 0;
+}
+
+void qxl_fbdev_fini(struct qxl_device *qdev)
+{
+	if (!qdev->mode_info.qfbdev)
+		return;
+
+	qxl_fbdev_destroy(qdev->ddev, qdev->mode_info.qfbdev);
+	kfree(qdev->mode_info.qfbdev);
+	qdev->mode_info.qfbdev = NULL;
+}
+
+
diff --git a/drivers/gpu/drm/qxl/qxl_fence.c b/drivers/gpu/drm/qxl/qxl_fence.c
new file mode 100644
index 000000000000..63c6715ad385
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_fence.c
@@ -0,0 +1,97 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+
+#include "qxl_drv.h"
+
+/* QXL fencing-
+
+   When we submit operations to the GPU we pass a release reference to the GPU
+   with them, the release reference is then added to the release ring when
+   the GPU is finished with that particular operation and has removed it from
+   its tree.
+
+   So we have can have multiple outstanding non linear fences per object.
+
+   From a TTM POV we only care if the object has any outstanding releases on
+   it.
+
+   we wait until all outstanding releases are processeed.
+
+   sync object is just a list of release ids that represent that fence on
+   that buffer.
+
+   we just add new releases onto the sync object attached to the object.
+
+   This currently uses a radix tree to store the list of release ids.
+
+   For some reason every so often qxl hw fails to release, things go wrong.
+*/
+
+
+int qxl_fence_add_release(struct qxl_fence *qfence, uint32_t rel_id)
+{
+	struct qxl_bo *bo = container_of(qfence, struct qxl_bo, fence);
+
+	spin_lock(&bo->tbo.bdev->fence_lock);
+	radix_tree_insert(&qfence->tree, rel_id, qfence);
+	qfence->num_active_releases++;
+	spin_unlock(&bo->tbo.bdev->fence_lock);
+	return 0;
+}
+
+int qxl_fence_remove_release(struct qxl_fence *qfence, uint32_t rel_id)
+{
+	void *ret;
+	int retval = 0;
+	struct qxl_bo *bo = container_of(qfence, struct qxl_bo, fence);
+
+	spin_lock(&bo->tbo.bdev->fence_lock);
+
+	ret = radix_tree_delete(&qfence->tree, rel_id);
+	if (ret == qfence)
+		qfence->num_active_releases--;
+	else {
+		DRM_DEBUG("didn't find fence in radix tree for %d\n", rel_id);
+		retval = -ENOENT;
+	}
+	spin_unlock(&bo->tbo.bdev->fence_lock);
+	return retval;
+}
+
+
+int qxl_fence_init(struct qxl_device *qdev, struct qxl_fence *qfence)
+{
+	qfence->qdev = qdev;
+	qfence->num_active_releases = 0;
+	INIT_RADIX_TREE(&qfence->tree, GFP_ATOMIC);
+	return 0;
+}
+
+void qxl_fence_fini(struct qxl_fence *qfence)
+{
+	kfree(qfence->release_ids);
+	qfence->num_active_releases = 0;
+}
diff --git a/drivers/gpu/drm/qxl/qxl_gem.c b/drivers/gpu/drm/qxl/qxl_gem.c
new file mode 100644
index 000000000000..adc1ee2cf7fb
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_gem.c
@@ -0,0 +1,178 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "drmP.h"
+#include "drm/drm.h"
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+int qxl_gem_object_init(struct drm_gem_object *obj)
+{
+	/* we do nothings here */
+	return 0;
+}
+
+void qxl_gem_object_free(struct drm_gem_object *gobj)
+{
+	struct qxl_bo *qobj = gem_to_qxl_bo(gobj);
+
+	if (qobj)
+		qxl_bo_unref(&qobj);
+}
+
+int qxl_gem_object_create(struct qxl_device *qdev, int size,
+			  int alignment, int initial_domain,
+			  bool discardable, bool kernel,
+			  struct qxl_surface *surf,
+			  struct drm_gem_object **obj)
+{
+	struct qxl_bo *qbo;
+	int r;
+
+	*obj = NULL;
+	/* At least align on page size */
+	if (alignment < PAGE_SIZE)
+		alignment = PAGE_SIZE;
+	r = qxl_bo_create(qdev, size, kernel, initial_domain, surf, &qbo);
+	if (r) {
+		if (r != -ERESTARTSYS)
+			DRM_ERROR(
+			"Failed to allocate GEM object (%d, %d, %u, %d)\n",
+				  size, initial_domain, alignment, r);
+		return r;
+	}
+	*obj = &qbo->gem_base;
+
+	mutex_lock(&qdev->gem.mutex);
+	list_add_tail(&qbo->list, &qdev->gem.objects);
+	mutex_unlock(&qdev->gem.mutex);
+
+	return 0;
+}
+
+int qxl_gem_object_create_with_handle(struct qxl_device *qdev,
+				      struct drm_file *file_priv,
+				      u32 domain,
+				      size_t size,
+				      struct qxl_surface *surf,
+				      struct qxl_bo **qobj,
+				      uint32_t *handle)
+{
+	struct drm_gem_object *gobj;
+	int r;
+
+	BUG_ON(!qobj);
+	BUG_ON(!handle);
+
+	r = qxl_gem_object_create(qdev, size, 0,
+				  domain,
+				  false, false, surf,
+				  &gobj);
+	if (r)
+		return -ENOMEM;
+	r = drm_gem_handle_create(file_priv, gobj, handle);
+	if (r)
+		return r;
+	/* drop reference from allocate - handle holds it now */
+	*qobj = gem_to_qxl_bo(gobj);
+	drm_gem_object_unreference_unlocked(gobj);
+	return 0;
+}
+
+int qxl_gem_object_pin(struct drm_gem_object *obj, uint32_t pin_domain,
+			  uint64_t *gpu_addr)
+{
+	struct qxl_bo *qobj = obj->driver_private;
+	int r;
+
+	r = qxl_bo_reserve(qobj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = qxl_bo_pin(qobj, pin_domain, gpu_addr);
+	qxl_bo_unreserve(qobj);
+	return r;
+}
+
+void qxl_gem_object_unpin(struct drm_gem_object *obj)
+{
+	struct qxl_bo *qobj = obj->driver_private;
+	int r;
+
+	r = qxl_bo_reserve(qobj, false);
+	if (likely(r == 0)) {
+		qxl_bo_unpin(qobj);
+		qxl_bo_unreserve(qobj);
+	}
+}
+
+int qxl_gem_set_domain(struct drm_gem_object *gobj,
+			  uint32_t rdomain, uint32_t wdomain)
+{
+	struct qxl_bo *qobj;
+	uint32_t domain;
+	int r;
+
+	/* FIXME: reeimplement */
+	qobj = gobj->driver_private;
+	/* work out where to validate the buffer to */
+	domain = wdomain;
+	if (!domain)
+		domain = rdomain;
+	if (!domain) {
+		/* Do nothings */
+		pr_warn("Set domain withou domain !\n");
+		return 0;
+	}
+	if (domain == QXL_GEM_DOMAIN_CPU) {
+		/* Asking for cpu access wait for object idle */
+		r = qxl_bo_wait(qobj, NULL, false);
+		if (r) {
+			pr_err("Failed to wait for object !\n");
+			return r;
+		}
+	}
+	return 0;
+}
+
+int qxl_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_priv)
+{
+	return 0;
+}
+
+void qxl_gem_object_close(struct drm_gem_object *obj,
+			  struct drm_file *file_priv)
+{
+}
+
+int qxl_gem_init(struct qxl_device *qdev)
+{
+	INIT_LIST_HEAD(&qdev->gem.objects);
+	return 0;
+}
+
+void qxl_gem_fini(struct qxl_device *qdev)
+{
+	qxl_bo_force_delete(qdev);
+}
diff --git a/drivers/gpu/drm/qxl/qxl_image.c b/drivers/gpu/drm/qxl/qxl_image.c
new file mode 100644
index 000000000000..cf856206996b
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_image.c
@@ -0,0 +1,176 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include <linux/gfp.h>
+#include <linux/slab.h>
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+static int
+qxl_image_create_helper(struct qxl_device *qdev,
+			struct qxl_release *release,
+			struct qxl_bo **image_bo,
+			const uint8_t *data,
+			int width, int height,
+			int depth, unsigned int hash,
+			int stride)
+{
+	struct qxl_image *image;
+	struct qxl_data_chunk *chunk;
+	int i;
+	int chunk_stride;
+	int linesize = width * depth / 8;
+	struct qxl_bo *chunk_bo;
+	int ret;
+	void *ptr;
+	/* Chunk */
+	/* FIXME: Check integer overflow */
+	/* TODO: variable number of chunks */
+	chunk_stride = stride; /* TODO: should use linesize, but it renders
+				  wrong (check the bitmaps are sent correctly
+				  first) */
+	ret = qxl_alloc_bo_reserved(qdev, sizeof(*chunk) + height * chunk_stride,
+				    &chunk_bo);
+	
+	ptr = qxl_bo_kmap_atomic_page(qdev, chunk_bo, 0);
+	chunk = ptr;
+	chunk->data_size = height * chunk_stride;
+	chunk->prev_chunk = 0;
+	chunk->next_chunk = 0;
+	qxl_bo_kunmap_atomic_page(qdev, chunk_bo, ptr);
+
+	{
+		void *k_data, *i_data;
+		int remain;
+		int page;
+		int size;
+		if (stride == linesize && chunk_stride == stride) {
+			remain = linesize * height;
+			page = 0;
+			i_data = (void *)data;
+
+			while (remain > 0) {
+				ptr = qxl_bo_kmap_atomic_page(qdev, chunk_bo, page << PAGE_SHIFT);
+
+				if (page == 0) {
+					chunk = ptr;
+					k_data = chunk->data;
+					size = PAGE_SIZE - offsetof(struct qxl_data_chunk, data);
+				} else {
+					k_data = ptr;
+					size = PAGE_SIZE;
+				}
+				size = min(size, remain);
+
+				memcpy(k_data, i_data, size);
+
+				qxl_bo_kunmap_atomic_page(qdev, chunk_bo, ptr);
+				i_data += size;
+				remain -= size;
+				page++;
+			}
+		} else {
+			unsigned page_base, page_offset, out_offset;
+			for (i = 0 ; i < height ; ++i) {
+				i_data = (void *)data + i * stride;
+				remain = linesize;
+				out_offset = offsetof(struct qxl_data_chunk, data) + i * chunk_stride;
+
+				while (remain > 0) {
+					page_base = out_offset & PAGE_MASK;
+					page_offset = offset_in_page(out_offset);
+					
+					size = min((int)(PAGE_SIZE - page_offset), remain);
+
+					ptr = qxl_bo_kmap_atomic_page(qdev, chunk_bo, page_base);
+					k_data = ptr + page_offset;
+					memcpy(k_data, i_data, size);
+					qxl_bo_kunmap_atomic_page(qdev, chunk_bo, ptr);
+					remain -= size;
+					i_data += size;
+					out_offset += size;
+				}
+			}
+		}
+	}
+
+
+	qxl_bo_kunmap(chunk_bo);
+
+	/* Image */
+	ret = qxl_alloc_bo_reserved(qdev, sizeof(*image), image_bo);
+
+	ptr = qxl_bo_kmap_atomic_page(qdev, *image_bo, 0);
+	image = ptr;
+
+	image->descriptor.id = 0;
+	image->descriptor.type = SPICE_IMAGE_TYPE_BITMAP;
+
+	image->descriptor.flags = 0;
+	image->descriptor.width = width;
+	image->descriptor.height = height;
+
+	switch (depth) {
+	case 1:
+		/* TODO: BE? check by arch? */
+		image->u.bitmap.format = SPICE_BITMAP_FMT_1BIT_BE;
+		break;
+	case 24:
+		image->u.bitmap.format = SPICE_BITMAP_FMT_24BIT;
+		break;
+	case 32:
+		image->u.bitmap.format = SPICE_BITMAP_FMT_32BIT;
+		break;
+	default:
+		DRM_ERROR("unsupported image bit depth\n");
+		return -EINVAL; /* TODO: cleanup */
+	}
+	image->u.bitmap.flags = QXL_BITMAP_TOP_DOWN;
+	image->u.bitmap.x = width;
+	image->u.bitmap.y = height;
+	image->u.bitmap.stride = chunk_stride;
+	image->u.bitmap.palette = 0;
+	image->u.bitmap.data = qxl_bo_physical_address(qdev, chunk_bo, 0);
+	qxl_release_add_res(qdev, release, chunk_bo);
+	qxl_bo_unreserve(chunk_bo);
+	qxl_bo_unref(&chunk_bo);
+
+	qxl_bo_kunmap_atomic_page(qdev, *image_bo, ptr);
+
+	return 0;
+}
+
+int qxl_image_create(struct qxl_device *qdev,
+		     struct qxl_release *release,
+		     struct qxl_bo **image_bo,
+		     const uint8_t *data,
+		     int x, int y, int width, int height,
+		     int depth, int stride)
+{
+	data += y * stride + x * (depth / 8);
+	return qxl_image_create_helper(qdev, release, image_bo, data,
+				       width, height, depth, 0, stride);
+}
diff --git a/drivers/gpu/drm/qxl/qxl_ioctl.c b/drivers/gpu/drm/qxl/qxl_ioctl.c
new file mode 100644
index 000000000000..83ca4f713f88
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_ioctl.c
@@ -0,0 +1,411 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+/*
+ * TODO: allocating a new gem(in qxl_bo) for each request.
+ * This is wasteful since bo's are page aligned.
+ */
+int qxl_alloc_ioctl(struct drm_device *dev, void *data,
+		    struct drm_file *file_priv)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_alloc *qxl_alloc = data;
+	int ret;
+	struct qxl_bo *qobj;
+	uint32_t handle;
+	u32 domain = QXL_GEM_DOMAIN_VRAM;
+
+	if (qxl_alloc->size == 0) {
+		DRM_ERROR("invalid size %d\n", qxl_alloc->size);
+		return -EINVAL;
+	}
+	ret = qxl_gem_object_create_with_handle(qdev, file_priv,
+						domain,
+						qxl_alloc->size,
+						NULL,
+						&qobj, &handle);
+	if (ret) {
+		DRM_ERROR("%s: failed to create gem ret=%d\n",
+			  __func__, ret);
+		return -ENOMEM;
+	}
+	qxl_alloc->handle = handle;
+	return 0;
+}
+
+int qxl_map_ioctl(struct drm_device *dev, void *data,
+		  struct drm_file *file_priv)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_map *qxl_map = data;
+
+	return qxl_mode_dumb_mmap(file_priv, qdev->ddev, qxl_map->handle,
+				  &qxl_map->offset);
+}
+
+/*
+ * dst must be validated, i.e. whole bo on vram/surfacesram (right now all bo's
+ * are on vram).
+ * *(dst + dst_off) = qxl_bo_physical_address(src, src_off)
+ */
+static void
+apply_reloc(struct qxl_device *qdev, struct qxl_bo *dst, uint64_t dst_off,
+	    struct qxl_bo *src, uint64_t src_off)
+{
+	void *reloc_page;
+
+	reloc_page = qxl_bo_kmap_atomic_page(qdev, dst, dst_off & PAGE_MASK);
+	*(uint64_t *)(reloc_page + (dst_off & ~PAGE_MASK)) = qxl_bo_physical_address(qdev,
+								     src, src_off);
+	qxl_bo_kunmap_atomic_page(qdev, dst, reloc_page);
+}
+
+static void
+apply_surf_reloc(struct qxl_device *qdev, struct qxl_bo *dst, uint64_t dst_off,
+		 struct qxl_bo *src)
+{
+	uint32_t id = 0;
+	void *reloc_page;
+
+	if (src && !src->is_primary)
+		id = src->surface_id;
+
+	reloc_page = qxl_bo_kmap_atomic_page(qdev, dst, dst_off & PAGE_MASK);
+	*(uint32_t *)(reloc_page + (dst_off & ~PAGE_MASK)) = id;
+	qxl_bo_kunmap_atomic_page(qdev, dst, reloc_page);
+}
+
+/* return holding the reference to this object */
+struct qxl_bo *qxlhw_handle_to_bo(struct qxl_device *qdev,
+				  struct drm_file *file_priv, uint64_t handle,
+				  struct qxl_reloc_list *reloc_list)
+{
+	struct drm_gem_object *gobj;
+	struct qxl_bo *qobj;
+	int ret;
+
+	gobj = drm_gem_object_lookup(qdev->ddev, file_priv, handle);
+	if (!gobj) {
+		DRM_ERROR("bad bo handle %lld\n", handle);
+		return NULL;
+	}
+	qobj = gem_to_qxl_bo(gobj);
+
+	ret = qxl_bo_list_add(reloc_list, qobj);
+	if (ret)
+		return NULL;
+
+	return qobj;
+}
+
+/*
+ * Usage of execbuffer:
+ * Relocations need to take into account the full QXLDrawable size.
+ * However, the command as passed from user space must *not* contain the initial
+ * QXLReleaseInfo struct (first XXX bytes)
+ */
+int qxl_execbuffer_ioctl(struct drm_device *dev, void *data,
+			 struct drm_file *file_priv)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_execbuffer *execbuffer = data;
+	struct drm_qxl_command user_cmd;
+	int cmd_num;
+	struct qxl_bo *reloc_src_bo;
+	struct qxl_bo *reloc_dst_bo;
+	struct drm_qxl_reloc reloc;
+	void *fb_cmd;
+	int i, ret;
+	struct qxl_reloc_list reloc_list;
+	int unwritten;
+	uint32_t reloc_dst_offset;
+	INIT_LIST_HEAD(&reloc_list.bos);
+
+	for (cmd_num = 0; cmd_num < execbuffer->commands_num; ++cmd_num) {
+		struct qxl_release *release;
+		struct qxl_bo *cmd_bo;
+		int release_type;
+		struct drm_qxl_command *commands =
+			(struct drm_qxl_command *)execbuffer->commands;
+
+		if (DRM_COPY_FROM_USER(&user_cmd, &commands[cmd_num],
+				       sizeof(user_cmd)))
+			return -EFAULT;
+		switch (user_cmd.type) {
+		case QXL_CMD_DRAW:
+			release_type = QXL_RELEASE_DRAWABLE;
+			break;
+		case QXL_CMD_SURFACE:
+		case QXL_CMD_CURSOR:
+		default:
+			DRM_DEBUG("Only draw commands in execbuffers\n");
+			return -EINVAL;
+			break;
+		}
+
+		if (user_cmd.command_size > PAGE_SIZE - sizeof(union qxl_release_info))
+			return -EINVAL;
+
+		ret = qxl_alloc_release_reserved(qdev,
+						 sizeof(union qxl_release_info) +
+						 user_cmd.command_size,
+						 release_type,
+						 &release,
+						 &cmd_bo);
+		if (ret)
+			return ret;
+
+		/* TODO copy slow path code from i915 */
+		fb_cmd = qxl_bo_kmap_atomic_page(qdev, cmd_bo, (release->release_offset & PAGE_SIZE));
+		unwritten = __copy_from_user_inatomic_nocache(fb_cmd + sizeof(union qxl_release_info) + (release->release_offset & ~PAGE_SIZE), (void *)(unsigned long)user_cmd.command, user_cmd.command_size);
+		qxl_bo_kunmap_atomic_page(qdev, cmd_bo, fb_cmd);
+		if (unwritten) {
+			DRM_ERROR("got unwritten %d\n", unwritten);
+			qxl_release_unreserve(qdev, release);
+			qxl_release_free(qdev, release);
+			return -EFAULT;
+		}
+
+		for (i = 0 ; i < user_cmd.relocs_num; ++i) {
+			if (DRM_COPY_FROM_USER(&reloc,
+					       &((struct drm_qxl_reloc *)user_cmd.relocs)[i],
+					       sizeof(reloc))) {
+				qxl_bo_list_unreserve(&reloc_list, true);
+				qxl_release_unreserve(qdev, release);
+				qxl_release_free(qdev, release);
+				return -EFAULT;
+			}
+
+			/* add the bos to the list of bos to validate -
+			   need to validate first then process relocs? */
+			if (reloc.dst_handle) {
+				reloc_dst_bo = qxlhw_handle_to_bo(qdev, file_priv,
+								  reloc.dst_handle, &reloc_list);
+				if (!reloc_dst_bo) {
+					qxl_bo_list_unreserve(&reloc_list, true);
+					qxl_release_unreserve(qdev, release);
+					qxl_release_free(qdev, release);
+					return -EINVAL;
+				}
+				reloc_dst_offset = 0;
+			} else {
+				reloc_dst_bo = cmd_bo;
+				reloc_dst_offset = release->release_offset;
+			}
+
+			/* reserve and validate the reloc dst bo */
+			if (reloc.reloc_type == QXL_RELOC_TYPE_BO || reloc.src_handle > 0) {
+				reloc_src_bo =
+					qxlhw_handle_to_bo(qdev, file_priv,
+							   reloc.src_handle, &reloc_list);
+				if (!reloc_src_bo) {
+					if (reloc_dst_bo != cmd_bo)
+						drm_gem_object_unreference_unlocked(&reloc_dst_bo->gem_base);
+					qxl_bo_list_unreserve(&reloc_list, true);
+					qxl_release_unreserve(qdev, release);
+					qxl_release_free(qdev, release);
+					return -EINVAL;
+				}
+			} else
+				reloc_src_bo = NULL;
+			if (reloc.reloc_type == QXL_RELOC_TYPE_BO) {
+				apply_reloc(qdev, reloc_dst_bo, reloc_dst_offset + reloc.dst_offset,
+					    reloc_src_bo, reloc.src_offset);
+			} else if (reloc.reloc_type == QXL_RELOC_TYPE_SURF) {
+				apply_surf_reloc(qdev, reloc_dst_bo, reloc_dst_offset + reloc.dst_offset, reloc_src_bo);
+			} else {
+				DRM_ERROR("unknown reloc type %d\n", reloc.reloc_type);
+				return -EINVAL;
+			}
+
+			if (reloc_src_bo && reloc_src_bo != cmd_bo) {
+				qxl_release_add_res(qdev, release, reloc_src_bo);
+				drm_gem_object_unreference_unlocked(&reloc_src_bo->gem_base);
+			}
+
+			if (reloc_dst_bo != cmd_bo)
+				drm_gem_object_unreference_unlocked(&reloc_dst_bo->gem_base);
+		}
+		qxl_fence_releaseable(qdev, release);
+
+		ret = qxl_push_command_ring_release(qdev, release, user_cmd.type, true);
+		if (ret == -ERESTARTSYS) {
+			qxl_release_unreserve(qdev, release);
+			qxl_release_free(qdev, release);
+			qxl_bo_list_unreserve(&reloc_list, true);
+			return ret;
+		}
+		qxl_release_unreserve(qdev, release);
+	}
+	qxl_bo_list_unreserve(&reloc_list, 0);
+	return 0;
+}
+
+int qxl_update_area_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *file)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_update_area *update_area = data;
+	struct qxl_rect area = {.left = update_area->left,
+				.top = update_area->top,
+				.right = update_area->right,
+				.bottom = update_area->bottom};
+	int ret;
+	struct drm_gem_object *gobj = NULL;
+	struct qxl_bo *qobj = NULL;
+
+	if (update_area->left >= update_area->right ||
+	    update_area->top >= update_area->bottom)
+		return -EINVAL;
+
+	gobj = drm_gem_object_lookup(dev, file, update_area->handle);
+	if (gobj == NULL)
+		return -ENOENT;
+
+	qobj = gem_to_qxl_bo(gobj);
+
+	ret = qxl_bo_reserve(qobj, false);
+	if (ret)
+		goto out;
+
+	if (!qobj->pin_count) {
+		ret = ttm_bo_validate(&qobj->tbo, &qobj->placement,
+				      true, false);
+		if (unlikely(ret))
+			goto out;
+	}
+
+	ret = qxl_bo_check_id(qdev, qobj);
+	if (ret)
+		goto out2;
+	if (!qobj->surface_id)
+		DRM_ERROR("got update area for surface with no id %d\n", update_area->handle);
+	ret = qxl_io_update_area(qdev, qobj, &area);
+
+out2:
+	qxl_bo_unreserve(qobj);
+
+out:
+	drm_gem_object_unreference_unlocked(gobj);
+	return ret;
+}
+
+static int qxl_getparam_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *file_priv)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_getparam *param = data;
+
+	switch (param->param) {
+	case QXL_PARAM_NUM_SURFACES:
+		param->value = qdev->rom->n_surfaces;
+		break;
+	case QXL_PARAM_MAX_RELOCS:
+		param->value = QXL_MAX_RES;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int qxl_clientcap_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *file_priv)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_clientcap *param = data;
+	int byte, idx;
+
+	byte = param->index / 8;
+	idx = param->index % 8;
+
+	if (qdev->pdev->revision < 4)
+		return -ENOSYS;
+
+	if (byte > 58)
+		return -ENOSYS;
+
+	if (qdev->rom->client_capabilities[byte] & (1 << idx))
+		return 0;
+	return -ENOSYS;
+}
+
+static int qxl_alloc_surf_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *file)
+{
+	struct qxl_device *qdev = dev->dev_private;
+	struct drm_qxl_alloc_surf *param = data;
+	struct qxl_bo *qobj;
+	int handle;
+	int ret;
+	int size, actual_stride;
+	struct qxl_surface surf;
+
+	/* work out size allocate bo with handle */
+	actual_stride = param->stride < 0 ? -param->stride : param->stride;
+	size = actual_stride * param->height + actual_stride;
+
+	surf.format = param->format;
+	surf.width = param->width;
+	surf.height = param->height;
+	surf.stride = param->stride;
+	surf.data = 0;
+
+	ret = qxl_gem_object_create_with_handle(qdev, file,
+						QXL_GEM_DOMAIN_SURFACE,
+						size,
+						&surf,
+						&qobj, &handle);
+	if (ret) {
+		DRM_ERROR("%s: failed to create gem ret=%d\n",
+			  __func__, ret);
+		return -ENOMEM;
+	} else
+		param->handle = handle;
+	return ret;
+}
+
+struct drm_ioctl_desc qxl_ioctls[] = {
+	DRM_IOCTL_DEF_DRV(QXL_ALLOC, qxl_alloc_ioctl, DRM_AUTH|DRM_UNLOCKED),
+
+	DRM_IOCTL_DEF_DRV(QXL_MAP, qxl_map_ioctl, DRM_AUTH|DRM_UNLOCKED),
+
+	DRM_IOCTL_DEF_DRV(QXL_EXECBUFFER, qxl_execbuffer_ioctl,
+							DRM_AUTH|DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(QXL_UPDATE_AREA, qxl_update_area_ioctl,
+							DRM_AUTH|DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(QXL_GETPARAM, qxl_getparam_ioctl,
+							DRM_AUTH|DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(QXL_CLIENTCAP, qxl_clientcap_ioctl,
+							DRM_AUTH|DRM_UNLOCKED),
+
+	DRM_IOCTL_DEF_DRV(QXL_ALLOC_SURF, qxl_alloc_surf_ioctl,
+			  DRM_AUTH|DRM_UNLOCKED),
+};
+
+int qxl_max_ioctls = DRM_ARRAY_SIZE(qxl_ioctls);
diff --git a/drivers/gpu/drm/qxl/qxl_irq.c b/drivers/gpu/drm/qxl/qxl_irq.c
new file mode 100644
index 000000000000..21393dc4700a
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_irq.c
@@ -0,0 +1,97 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "qxl_drv.h"
+
+irqreturn_t qxl_irq_handler(DRM_IRQ_ARGS)
+{
+	struct drm_device *dev = (struct drm_device *) arg;
+	struct qxl_device *qdev = (struct qxl_device *)dev->dev_private;
+	uint32_t pending;
+
+	pending = xchg(&qdev->ram_header->int_pending, 0);
+
+	atomic_inc(&qdev->irq_received);
+
+	if (pending & QXL_INTERRUPT_DISPLAY) {
+		atomic_inc(&qdev->irq_received_display);
+		wake_up_all(&qdev->display_event);
+		qxl_queue_garbage_collect(qdev, false);
+	}
+	if (pending & QXL_INTERRUPT_CURSOR) {
+		atomic_inc(&qdev->irq_received_cursor);
+		wake_up_all(&qdev->cursor_event);
+	}
+	if (pending & QXL_INTERRUPT_IO_CMD) {
+		atomic_inc(&qdev->irq_received_io_cmd);
+		wake_up_all(&qdev->io_cmd_event);
+	}
+	if (pending & QXL_INTERRUPT_ERROR) {
+		/* TODO: log it, reset device (only way to exit this condition)
+		 * (do it a certain number of times, afterwards admit defeat,
+		 * to avoid endless loops).
+		 */
+		qdev->irq_received_error++;
+		qxl_io_log(qdev, "%s: driver is in bug mode.\n", __func__);
+	}
+	if (pending & QXL_INTERRUPT_CLIENT_MONITORS_CONFIG) {
+		qxl_io_log(qdev, "QXL_INTERRUPT_CLIENT_MONITORS_CONFIG\n");
+		schedule_work(&qdev->client_monitors_config_work);
+	}
+	qdev->ram_header->int_mask = QXL_INTERRUPT_MASK;
+	outb(0, qdev->io_base + QXL_IO_UPDATE_IRQ);
+	return IRQ_HANDLED;
+}
+
+static void qxl_client_monitors_config_work_func(struct work_struct *work)
+{
+	struct qxl_device *qdev = container_of(work, struct qxl_device,
+					       client_monitors_config_work);
+
+	qxl_display_read_client_monitors_config(qdev);
+}
+
+int qxl_irq_init(struct qxl_device *qdev)
+{
+	int ret;
+
+	init_waitqueue_head(&qdev->display_event);
+	init_waitqueue_head(&qdev->cursor_event);
+	init_waitqueue_head(&qdev->io_cmd_event);
+	INIT_WORK(&qdev->client_monitors_config_work,
+		  qxl_client_monitors_config_work_func);
+	atomic_set(&qdev->irq_received, 0);
+	atomic_set(&qdev->irq_received_display, 0);
+	atomic_set(&qdev->irq_received_cursor, 0);
+	atomic_set(&qdev->irq_received_io_cmd, 0);
+	qdev->irq_received_error = 0;
+	ret = drm_irq_install(qdev->ddev);
+	qdev->ram_header->int_mask = QXL_INTERRUPT_MASK;
+	if (unlikely(ret != 0)) {
+		DRM_ERROR("Failed installing irq: %d\n", ret);
+		return 1;
+	}
+	return 0;
+}
diff --git a/drivers/gpu/drm/qxl/qxl_kms.c b/drivers/gpu/drm/qxl/qxl_kms.c
new file mode 100644
index 000000000000..036e0de13412
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_kms.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+#include <linux/io-mapping.h>
+
+int qxl_log_level;
+
+static void qxl_dump_mode(struct qxl_device *qdev, void *p)
+{
+	struct qxl_mode *m = p;
+	DRM_DEBUG_KMS("%d: %dx%d %d bits, stride %d, %dmm x %dmm, orientation %d\n",
+		      m->id, m->x_res, m->y_res, m->bits, m->stride, m->x_mili,
+		      m->y_mili, m->orientation);
+}
+
+static bool qxl_check_device(struct qxl_device *qdev)
+{
+	struct qxl_rom *rom = qdev->rom;
+	int mode_offset;
+	int i;
+
+	if (rom->magic != 0x4f525851) {
+		DRM_ERROR("bad rom signature %x\n", rom->magic);
+		return false;
+	}
+
+	DRM_INFO("Device Version %d.%d\n", rom->id, rom->update_id);
+	DRM_INFO("Compression level %d log level %d\n", rom->compression_level,
+		 rom->log_level);
+	DRM_INFO("Currently using mode #%d, list at 0x%x\n",
+		 rom->mode, rom->modes_offset);
+	DRM_INFO("%d io pages at offset 0x%x\n",
+		 rom->num_io_pages, rom->pages_offset);
+	DRM_INFO("%d byte draw area at offset 0x%x\n",
+		 rom->surface0_area_size, rom->draw_area_offset);
+
+	qdev->vram_size = rom->surface0_area_size;
+	DRM_INFO("RAM header offset: 0x%x\n", rom->ram_header_offset);
+
+	mode_offset = rom->modes_offset / 4;
+	qdev->mode_info.num_modes = ((u32 *)rom)[mode_offset];
+	DRM_INFO("rom modes offset 0x%x for %d modes\n", rom->modes_offset,
+		 qdev->mode_info.num_modes);
+	qdev->mode_info.modes = (void *)((uint32_t *)rom + mode_offset + 1);
+	for (i = 0; i < qdev->mode_info.num_modes; i++)
+		qxl_dump_mode(qdev, qdev->mode_info.modes + i);
+	return true;
+}
+
+static uint8_t setup_slot(struct qxl_device *qdev, uint8_t slot_index_offset,
+	unsigned long start_phys_addr, unsigned long end_phys_addr)
+{
+	uint64_t high_bits;
+	struct qxl_memslot *slot;
+	uint8_t slot_index;
+	struct qxl_ram_header *ram_header = qdev->ram_header;
+
+	slot_index = qdev->rom->slots_start + slot_index_offset;
+	slot = &qdev->mem_slots[slot_index];
+	slot->start_phys_addr = start_phys_addr;
+	slot->end_phys_addr = end_phys_addr;
+	ram_header->mem_slot.mem_start = slot->start_phys_addr;
+	ram_header->mem_slot.mem_end = slot->end_phys_addr;
+	qxl_io_memslot_add(qdev, slot_index);
+	slot->generation = qdev->rom->slot_generation;
+	high_bits = slot_index << qdev->slot_gen_bits;
+	high_bits |= slot->generation;
+	high_bits <<= (64 - (qdev->slot_gen_bits + qdev->slot_id_bits));
+	slot->high_bits = high_bits;
+	return slot_index;
+}
+
+static void qxl_gc_work(struct work_struct *work)
+{
+	struct qxl_device *qdev = container_of(work, struct qxl_device, gc_work);
+	qxl_garbage_collect(qdev);
+}
+
+int qxl_device_init(struct qxl_device *qdev,
+		    struct drm_device *ddev,
+		    struct pci_dev *pdev,
+		    unsigned long flags)
+{
+	int r;
+
+	qdev->dev = &pdev->dev;
+	qdev->ddev = ddev;
+	qdev->pdev = pdev;
+	qdev->flags = flags;
+
+	mutex_init(&qdev->gem.mutex);
+	mutex_init(&qdev->update_area_mutex);
+	mutex_init(&qdev->release_mutex);
+	mutex_init(&qdev->surf_evict_mutex);
+	INIT_LIST_HEAD(&qdev->gem.objects);
+
+	qdev->rom_base = pci_resource_start(pdev, 2);
+	qdev->rom_size = pci_resource_len(pdev, 2);
+	qdev->vram_base = pci_resource_start(pdev, 0);
+	qdev->surfaceram_base = pci_resource_start(pdev, 1);
+	qdev->surfaceram_size = pci_resource_len(pdev, 1);
+	qdev->io_base = pci_resource_start(pdev, 3);
+
+	qdev->vram_mapping = io_mapping_create_wc(qdev->vram_base, pci_resource_len(pdev, 0));
+	qdev->surface_mapping = io_mapping_create_wc(qdev->surfaceram_base, qdev->surfaceram_size);
+	DRM_DEBUG_KMS("qxl: vram %p-%p(%dM %dk), surface %p-%p(%dM %dk)\n",
+		 (void *)qdev->vram_base, (void *)pci_resource_end(pdev, 0),
+		 (int)pci_resource_len(pdev, 0) / 1024 / 1024,
+		 (int)pci_resource_len(pdev, 0) / 1024,
+		 (void *)qdev->surfaceram_base,
+		 (void *)pci_resource_end(pdev, 1),
+		 (int)qdev->surfaceram_size / 1024 / 1024,
+		 (int)qdev->surfaceram_size / 1024);
+
+	qdev->rom = ioremap(qdev->rom_base, qdev->rom_size);
+	if (!qdev->rom) {
+		pr_err("Unable to ioremap ROM\n");
+		return -ENOMEM;
+	}
+
+	qxl_check_device(qdev);
+
+	r = qxl_bo_init(qdev);
+	if (r) {
+		DRM_ERROR("bo init failed %d\n", r);
+		return r;
+	}
+
+	qdev->ram_header = ioremap(qdev->vram_base +
+				   qdev->rom->ram_header_offset,
+				   sizeof(*qdev->ram_header));
+
+	qdev->command_ring = qxl_ring_create(&(qdev->ram_header->cmd_ring_hdr),
+					     sizeof(struct qxl_command),
+					     QXL_COMMAND_RING_SIZE,
+					     qdev->io_base + QXL_IO_NOTIFY_CMD,
+					     false,
+					     &qdev->display_event);
+
+	qdev->cursor_ring = qxl_ring_create(
+				&(qdev->ram_header->cursor_ring_hdr),
+				sizeof(struct qxl_command),
+				QXL_CURSOR_RING_SIZE,
+				qdev->io_base + QXL_IO_NOTIFY_CMD,
+				false,
+				&qdev->cursor_event);
+
+	qdev->release_ring = qxl_ring_create(
+				&(qdev->ram_header->release_ring_hdr),
+				sizeof(uint64_t),
+				QXL_RELEASE_RING_SIZE, 0, true,
+				NULL);
+
+	/* TODO - slot initialization should happen on reset. where is our
+	 * reset handler? */
+	qdev->n_mem_slots = qdev->rom->slots_end;
+	qdev->slot_gen_bits = qdev->rom->slot_gen_bits;
+	qdev->slot_id_bits = qdev->rom->slot_id_bits;
+	qdev->va_slot_mask =
+		(~(uint64_t)0) >> (qdev->slot_id_bits + qdev->slot_gen_bits);
+
+	qdev->mem_slots =
+		kmalloc(qdev->n_mem_slots * sizeof(struct qxl_memslot),
+			GFP_KERNEL);
+
+	idr_init(&qdev->release_idr);
+	spin_lock_init(&qdev->release_idr_lock);
+
+	idr_init(&qdev->surf_id_idr);
+	spin_lock_init(&qdev->surf_id_idr_lock);
+
+	mutex_init(&qdev->async_io_mutex);
+
+	/* reset the device into a known state - no memslots, no primary
+	 * created, no surfaces. */
+	qxl_io_reset(qdev);
+
+	/* must initialize irq before first async io - slot creation */
+	r = qxl_irq_init(qdev);
+	if (r)
+		return r;
+
+	/*
+	 * Note that virtual is surface0. We rely on the single ioremap done
+	 * before.
+	 */
+	qdev->main_mem_slot = setup_slot(qdev, 0,
+		(unsigned long)qdev->vram_base,
+		(unsigned long)qdev->vram_base + qdev->rom->ram_header_offset);
+	qdev->surfaces_mem_slot = setup_slot(qdev, 1,
+		(unsigned long)qdev->surfaceram_base,
+		(unsigned long)qdev->surfaceram_base + qdev->surfaceram_size);
+	DRM_INFO("main mem slot %d [%lx,%x)\n",
+		qdev->main_mem_slot,
+		(unsigned long)qdev->vram_base, qdev->rom->ram_header_offset);
+
+
+	qdev->gc_queue = create_singlethread_workqueue("qxl_gc");
+	INIT_WORK(&qdev->gc_work, qxl_gc_work);
+
+	r = qxl_fb_init(qdev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+void qxl_device_fini(struct qxl_device *qdev)
+{
+	if (qdev->current_release_bo[0])
+		qxl_bo_unref(&qdev->current_release_bo[0]);
+	if (qdev->current_release_bo[1])
+		qxl_bo_unref(&qdev->current_release_bo[1]);
+	flush_workqueue(qdev->gc_queue);
+	destroy_workqueue(qdev->gc_queue);
+	qdev->gc_queue = NULL;
+
+	qxl_ring_free(qdev->command_ring);
+	qxl_ring_free(qdev->cursor_ring);
+	qxl_ring_free(qdev->release_ring);
+	qxl_bo_fini(qdev);
+	io_mapping_free(qdev->surface_mapping);
+	io_mapping_free(qdev->vram_mapping);
+	iounmap(qdev->ram_header);
+	iounmap(qdev->rom);
+	qdev->rom = NULL;
+	qdev->mode_info.modes = NULL;
+	qdev->mode_info.num_modes = 0;
+	qxl_debugfs_remove_files(qdev);
+}
+
+int qxl_driver_unload(struct drm_device *dev)
+{
+	struct qxl_device *qdev = dev->dev_private;
+
+	if (qdev == NULL)
+		return 0;
+	qxl_modeset_fini(qdev);
+	qxl_device_fini(qdev);
+
+	kfree(qdev);
+	dev->dev_private = NULL;
+	return 0;
+}
+
+int qxl_driver_load(struct drm_device *dev, unsigned long flags)
+{
+	struct qxl_device *qdev;
+	int r;
+
+	/* require kms */
+	if (!drm_core_check_feature(dev, DRIVER_MODESET))
+		return -ENODEV;
+
+	qdev = kzalloc(sizeof(struct qxl_device), GFP_KERNEL);
+	if (qdev == NULL)
+		return -ENOMEM;
+
+	dev->dev_private = qdev;
+
+	r = qxl_device_init(qdev, dev, dev->pdev, flags);
+	if (r)
+		goto out;
+
+	r = qxl_modeset_init(qdev);
+	if (r) {
+		qxl_driver_unload(dev);
+		goto out;
+	}
+
+	return 0;
+out:
+	kfree(qdev);
+	return r;
+}
+
+
diff --git a/drivers/gpu/drm/qxl/qxl_object.c b/drivers/gpu/drm/qxl/qxl_object.c
new file mode 100644
index 000000000000..51efb94a5dee
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_object.c
@@ -0,0 +1,365 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+#include <linux/io-mapping.h>
+static void qxl_ttm_bo_destroy(struct ttm_buffer_object *tbo)
+{
+	struct qxl_bo *bo;
+	struct qxl_device *qdev;
+
+	bo = container_of(tbo, struct qxl_bo, tbo);
+	qdev = (struct qxl_device *)bo->gem_base.dev->dev_private;
+
+	qxl_surface_evict(qdev, bo, false);
+	qxl_fence_fini(&bo->fence);
+	mutex_lock(&qdev->gem.mutex);
+	list_del_init(&bo->list);
+	mutex_unlock(&qdev->gem.mutex);
+	drm_gem_object_release(&bo->gem_base);
+	kfree(bo);
+}
+
+bool qxl_ttm_bo_is_qxl_bo(struct ttm_buffer_object *bo)
+{
+	if (bo->destroy == &qxl_ttm_bo_destroy)
+		return true;
+	return false;
+}
+
+void qxl_ttm_placement_from_domain(struct qxl_bo *qbo, u32 domain)
+{
+	u32 c = 0;
+
+	qbo->placement.fpfn = 0;
+	qbo->placement.lpfn = 0;
+	qbo->placement.placement = qbo->placements;
+	qbo->placement.busy_placement = qbo->placements;
+	if (domain & QXL_GEM_DOMAIN_VRAM)
+		qbo->placements[c++] = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_VRAM;
+	if (domain & QXL_GEM_DOMAIN_SURFACE)
+		qbo->placements[c++] = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_PRIV0;
+	if (domain & QXL_GEM_DOMAIN_CPU)
+		qbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;
+	if (!c)
+		qbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;
+	qbo->placement.num_placement = c;
+	qbo->placement.num_busy_placement = c;
+}
+
+
+int qxl_bo_create(struct qxl_device *qdev,
+		  unsigned long size, bool kernel, u32 domain,
+		  struct qxl_surface *surf,
+		  struct qxl_bo **bo_ptr)
+{
+	struct qxl_bo *bo;
+	enum ttm_bo_type type;
+	int r;
+
+	if (unlikely(qdev->mman.bdev.dev_mapping == NULL))
+		qdev->mman.bdev.dev_mapping = qdev->ddev->dev_mapping;
+	if (kernel)
+		type = ttm_bo_type_kernel;
+	else
+		type = ttm_bo_type_device;
+	*bo_ptr = NULL;
+	bo = kzalloc(sizeof(struct qxl_bo), GFP_KERNEL);
+	if (bo == NULL)
+		return -ENOMEM;
+	size = roundup(size, PAGE_SIZE);
+	r = drm_gem_object_init(qdev->ddev, &bo->gem_base, size);
+	if (unlikely(r)) {
+		kfree(bo);
+		return r;
+	}
+	bo->gem_base.driver_private = NULL;
+	bo->type = domain;
+	bo->pin_count = 0;
+	bo->surface_id = 0;
+	qxl_fence_init(qdev, &bo->fence);
+	INIT_LIST_HEAD(&bo->list);
+	atomic_set(&bo->reserve_count, 0);
+	if (surf)
+		bo->surf = *surf;
+
+	qxl_ttm_placement_from_domain(bo, domain);
+
+	r = ttm_bo_init(&qdev->mman.bdev, &bo->tbo, size, type,
+			&bo->placement, 0, !kernel, NULL, size,
+			NULL, &qxl_ttm_bo_destroy);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			dev_err(qdev->dev,
+				"object_init failed for (%lu, 0x%08X)\n",
+				size, domain);
+		return r;
+	}
+	*bo_ptr = bo;
+	return 0;
+}
+
+int qxl_bo_kmap(struct qxl_bo *bo, void **ptr)
+{
+	bool is_iomem;
+	int r;
+
+	if (bo->kptr) {
+		if (ptr)
+			*ptr = bo->kptr;
+		return 0;
+	}
+	r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap);
+	if (r)
+		return r;
+	bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
+	if (ptr)
+		*ptr = bo->kptr;
+	return 0;
+}
+
+void *qxl_bo_kmap_atomic_page(struct qxl_device *qdev,
+			      struct qxl_bo *bo, int page_offset)
+{
+	struct ttm_mem_type_manager *man = &bo->tbo.bdev->man[bo->tbo.mem.mem_type];
+	void *rptr;
+	int ret;
+	struct io_mapping *map;
+
+	if (bo->tbo.mem.mem_type == TTM_PL_VRAM)
+		map = qdev->vram_mapping;
+	else if (bo->tbo.mem.mem_type == TTM_PL_PRIV0)
+		map = qdev->surface_mapping;
+	else
+		goto fallback;
+
+	(void) ttm_mem_io_lock(man, false);
+	ret = ttm_mem_io_reserve(bo->tbo.bdev, &bo->tbo.mem);
+	ttm_mem_io_unlock(man);
+
+	return io_mapping_map_atomic_wc(map, bo->tbo.mem.bus.offset + page_offset);
+fallback:
+	if (bo->kptr) {
+		rptr = bo->kptr + (page_offset * PAGE_SIZE);
+		return rptr;
+	}
+
+	ret = qxl_bo_kmap(bo, &rptr);
+	if (ret)
+		return NULL;
+
+	rptr += page_offset * PAGE_SIZE;
+	return rptr;
+}
+
+void qxl_bo_kunmap(struct qxl_bo *bo)
+{
+	if (bo->kptr == NULL)
+		return;
+	bo->kptr = NULL;
+	ttm_bo_kunmap(&bo->kmap);
+}
+
+void qxl_bo_kunmap_atomic_page(struct qxl_device *qdev,
+			       struct qxl_bo *bo, void *pmap)
+{
+	struct ttm_mem_type_manager *man = &bo->tbo.bdev->man[bo->tbo.mem.mem_type];
+	struct io_mapping *map;
+
+	if (bo->tbo.mem.mem_type == TTM_PL_VRAM)
+		map = qdev->vram_mapping;
+	else if (bo->tbo.mem.mem_type == TTM_PL_PRIV0)
+		map = qdev->surface_mapping;
+	else
+		goto fallback;
+
+	io_mapping_unmap_atomic(pmap);
+
+	(void) ttm_mem_io_lock(man, false);
+	ttm_mem_io_free(bo->tbo.bdev, &bo->tbo.mem);
+	ttm_mem_io_unlock(man);
+	return ;
+ fallback:
+	qxl_bo_kunmap(bo);
+}
+
+void qxl_bo_unref(struct qxl_bo **bo)
+{
+	struct ttm_buffer_object *tbo;
+
+	if ((*bo) == NULL)
+		return;
+	tbo = &((*bo)->tbo);
+	ttm_bo_unref(&tbo);
+	if (tbo == NULL)
+		*bo = NULL;
+}
+
+struct qxl_bo *qxl_bo_ref(struct qxl_bo *bo)
+{
+	ttm_bo_reference(&bo->tbo);
+	return bo;
+}
+
+int qxl_bo_pin(struct qxl_bo *bo, u32 domain, u64 *gpu_addr)
+{
+	struct qxl_device *qdev = (struct qxl_device *)bo->gem_base.dev->dev_private;
+	int r, i;
+
+	if (bo->pin_count) {
+		bo->pin_count++;
+		if (gpu_addr)
+			*gpu_addr = qxl_bo_gpu_offset(bo);
+		return 0;
+	}
+	qxl_ttm_placement_from_domain(bo, domain);
+	for (i = 0; i < bo->placement.num_placement; i++)
+		bo->placements[i] |= TTM_PL_FLAG_NO_EVICT;
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	if (likely(r == 0)) {
+		bo->pin_count = 1;
+		if (gpu_addr != NULL)
+			*gpu_addr = qxl_bo_gpu_offset(bo);
+	}
+	if (unlikely(r != 0))
+		dev_err(qdev->dev, "%p pin failed\n", bo);
+	return r;
+}
+
+int qxl_bo_unpin(struct qxl_bo *bo)
+{
+	struct qxl_device *qdev = (struct qxl_device *)bo->gem_base.dev->dev_private;
+	int r, i;
+
+	if (!bo->pin_count) {
+		dev_warn(qdev->dev, "%p unpin not necessary\n", bo);
+		return 0;
+	}
+	bo->pin_count--;
+	if (bo->pin_count)
+		return 0;
+	for (i = 0; i < bo->placement.num_placement; i++)
+		bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT;
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	if (unlikely(r != 0))
+		dev_err(qdev->dev, "%p validate failed for unpin\n", bo);
+	return r;
+}
+
+void qxl_bo_force_delete(struct qxl_device *qdev)
+{
+	struct qxl_bo *bo, *n;
+
+	if (list_empty(&qdev->gem.objects))
+		return;
+	dev_err(qdev->dev, "Userspace still has active objects !\n");
+	list_for_each_entry_safe(bo, n, &qdev->gem.objects, list) {
+		mutex_lock(&qdev->ddev->struct_mutex);
+		dev_err(qdev->dev, "%p %p %lu %lu force free\n",
+			&bo->gem_base, bo, (unsigned long)bo->gem_base.size,
+			*((unsigned long *)&bo->gem_base.refcount));
+		mutex_lock(&qdev->gem.mutex);
+		list_del_init(&bo->list);
+		mutex_unlock(&qdev->gem.mutex);
+		/* this should unref the ttm bo */
+		drm_gem_object_unreference(&bo->gem_base);
+		mutex_unlock(&qdev->ddev->struct_mutex);
+	}
+}
+
+int qxl_bo_init(struct qxl_device *qdev)
+{
+	return qxl_ttm_init(qdev);
+}
+
+void qxl_bo_fini(struct qxl_device *qdev)
+{
+	qxl_ttm_fini(qdev);
+}
+
+int qxl_bo_check_id(struct qxl_device *qdev, struct qxl_bo *bo)
+{
+	int ret;
+	if (bo->type == QXL_GEM_DOMAIN_SURFACE && bo->surface_id == 0) {
+		/* allocate a surface id for this surface now */
+		ret = qxl_surface_id_alloc(qdev, bo);
+		if (ret)
+			return ret;
+
+		ret = qxl_hw_surface_alloc(qdev, bo, NULL);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
+void qxl_bo_list_unreserve(struct qxl_reloc_list *reloc_list, bool failed)
+{
+	struct qxl_bo_list *entry, *sf;
+
+	list_for_each_entry_safe(entry, sf, &reloc_list->bos, lhead) {
+		qxl_bo_unreserve(entry->bo);
+		list_del(&entry->lhead);
+		kfree(entry);
+	}
+}
+
+int qxl_bo_list_add(struct qxl_reloc_list *reloc_list, struct qxl_bo *bo)
+{
+	struct qxl_bo_list *entry;
+	int ret;
+
+	list_for_each_entry(entry, &reloc_list->bos, lhead) {
+		if (entry->bo == bo)
+			return 0;
+	}
+
+	entry = kmalloc(sizeof(struct qxl_bo_list), GFP_KERNEL);
+	if (!entry)
+		return -ENOMEM;
+
+	entry->bo = bo;
+	list_add(&entry->lhead, &reloc_list->bos);
+
+	ret = qxl_bo_reserve(bo, false);
+	if (ret)
+		return ret;
+
+	if (!bo->pin_count) {
+		qxl_ttm_placement_from_domain(bo, bo->type);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement,
+				      true, false);
+		if (ret)
+			return ret;
+	}
+
+	/* allocate a surface for reserved + validated buffers */
+	ret = qxl_bo_check_id(bo->gem_base.dev->dev_private, bo);
+	if (ret)
+		return ret;
+	return 0;
+}
diff --git a/drivers/gpu/drm/qxl/qxl_object.h b/drivers/gpu/drm/qxl/qxl_object.h
new file mode 100644
index 000000000000..b4fd89fbd8b7
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_object.h
@@ -0,0 +1,112 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+#ifndef QXL_OBJECT_H
+#define QXL_OBJECT_H
+
+#include "qxl_drv.h"
+
+static inline int qxl_bo_reserve(struct qxl_bo *bo, bool no_wait)
+{
+	int r;
+
+	r = ttm_bo_reserve(&bo->tbo, true, no_wait, false, 0);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS) {
+			struct qxl_device *qdev = (struct qxl_device *)bo->gem_base.dev->dev_private;
+			dev_err(qdev->dev, "%p reserve failed\n", bo);
+		}
+		return r;
+	}
+	return 0;
+}
+
+static inline void qxl_bo_unreserve(struct qxl_bo *bo)
+{
+	ttm_bo_unreserve(&bo->tbo);
+}
+
+static inline u64 qxl_bo_gpu_offset(struct qxl_bo *bo)
+{
+	return bo->tbo.offset;
+}
+
+static inline unsigned long qxl_bo_size(struct qxl_bo *bo)
+{
+	return bo->tbo.num_pages << PAGE_SHIFT;
+}
+
+static inline bool qxl_bo_is_reserved(struct qxl_bo *bo)
+{
+	return !!atomic_read(&bo->tbo.reserved);
+}
+
+static inline u64 qxl_bo_mmap_offset(struct qxl_bo *bo)
+{
+	return bo->tbo.addr_space_offset;
+}
+
+static inline int qxl_bo_wait(struct qxl_bo *bo, u32 *mem_type,
+			      bool no_wait)
+{
+	int r;
+
+	r = ttm_bo_reserve(&bo->tbo, true, no_wait, false, 0);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS) {
+			struct qxl_device *qdev = (struct qxl_device *)bo->gem_base.dev->dev_private;
+			dev_err(qdev->dev, "%p reserve failed for wait\n",
+				bo);
+		}
+		return r;
+	}
+	spin_lock(&bo->tbo.bdev->fence_lock);
+	if (mem_type)
+		*mem_type = bo->tbo.mem.mem_type;
+	if (bo->tbo.sync_obj)
+		r = ttm_bo_wait(&bo->tbo, true, true, no_wait);
+	spin_unlock(&bo->tbo.bdev->fence_lock);
+	ttm_bo_unreserve(&bo->tbo);
+	return r;
+}
+
+extern int qxl_bo_create(struct qxl_device *qdev,
+			 unsigned long size,
+			 bool kernel, u32 domain,
+			 struct qxl_surface *surf,
+			 struct qxl_bo **bo_ptr);
+extern int qxl_bo_kmap(struct qxl_bo *bo, void **ptr);
+extern void qxl_bo_kunmap(struct qxl_bo *bo);
+void *qxl_bo_kmap_atomic_page(struct qxl_device *qdev, struct qxl_bo *bo, int page_offset);
+void qxl_bo_kunmap_atomic_page(struct qxl_device *qdev, struct qxl_bo *bo, void *map);
+extern struct qxl_bo *qxl_bo_ref(struct qxl_bo *bo);
+extern void qxl_bo_unref(struct qxl_bo **bo);
+extern int qxl_bo_pin(struct qxl_bo *bo, u32 domain, u64 *gpu_addr);
+extern int qxl_bo_unpin(struct qxl_bo *bo);
+extern void qxl_ttm_placement_from_domain(struct qxl_bo *qbo, u32 domain);
+extern bool qxl_ttm_bo_is_qxl_bo(struct ttm_buffer_object *bo);
+
+extern int qxl_bo_list_add(struct qxl_reloc_list *reloc_list, struct qxl_bo *bo);
+extern void qxl_bo_list_unreserve(struct qxl_reloc_list *reloc_list, bool failed);
+#endif
diff --git a/drivers/gpu/drm/qxl/qxl_release.c b/drivers/gpu/drm/qxl/qxl_release.c
new file mode 100644
index 000000000000..1600781d8cbc
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_release.c
@@ -0,0 +1,307 @@
+/*
+ * Copyright 2011 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+ * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+/*
+ * drawable cmd cache - allocate a bunch of VRAM pages, suballocate
+ * into 256 byte chunks for now - gives 16 cmds per page.
+ *
+ * use an ida to index into the chunks?
+ */
+/* manage releaseables */
+/* stack them 16 high for now -drawable object is 191 */
+#define RELEASE_SIZE 256
+#define RELEASES_PER_BO (4096 / RELEASE_SIZE)
+/* put an alloc/dealloc surface cmd into one bo and round up to 128 */
+#define SURFACE_RELEASE_SIZE 128
+#define SURFACE_RELEASES_PER_BO (4096 / SURFACE_RELEASE_SIZE)
+
+static const int release_size_per_bo[] = { RELEASE_SIZE, SURFACE_RELEASE_SIZE, RELEASE_SIZE };
+static const int releases_per_bo[] = { RELEASES_PER_BO, SURFACE_RELEASES_PER_BO, RELEASES_PER_BO };
+uint64_t
+qxl_release_alloc(struct qxl_device *qdev, int type,
+		  struct qxl_release **ret)
+{
+	struct qxl_release *release;
+	int handle = 0;
+	size_t size = sizeof(*release);
+	int idr_ret;
+
+	release = kmalloc(size, GFP_KERNEL);
+	if (!release) {
+		DRM_ERROR("Out of memory\n");
+		return 0;
+	}
+	release->type = type;
+	release->bo_count = 0;
+	release->release_offset = 0;
+	release->surface_release_id = 0;
+again:
+	if (idr_pre_get(&qdev->release_idr, GFP_KERNEL) == 0) {
+		DRM_ERROR("Out of memory for release idr\n");
+		kfree(release);
+		goto release_fail;
+	}
+	spin_lock(&qdev->release_idr_lock);
+	idr_ret = idr_get_new_above(&qdev->release_idr, release, 1, &handle);
+	spin_unlock(&qdev->release_idr_lock);
+	if (idr_ret == -EAGAIN)
+		goto again;
+	if (ret)
+		*ret = release;
+	QXL_INFO(qdev, "allocated release %lld\n", handle);
+	release->id = handle;
+release_fail:
+
+	return handle;
+}
+
+void
+qxl_release_free(struct qxl_device *qdev,
+		 struct qxl_release *release)
+{
+	int i;
+
+	QXL_INFO(qdev, "release %d, type %d, %d bos\n", release->id,
+		 release->type, release->bo_count);
+
+	if (release->surface_release_id)
+		qxl_surface_id_dealloc(qdev, release->surface_release_id);
+
+	for (i = 0 ; i < release->bo_count; ++i) {
+		QXL_INFO(qdev, "release %llx\n",
+			release->bos[i]->tbo.addr_space_offset
+						- DRM_FILE_OFFSET);
+		qxl_fence_remove_release(&release->bos[i]->fence, release->id);
+		qxl_bo_unref(&release->bos[i]);
+	}
+	spin_lock(&qdev->release_idr_lock);
+	idr_remove(&qdev->release_idr, release->id);
+	spin_unlock(&qdev->release_idr_lock);
+	kfree(release);
+}
+
+void
+qxl_release_add_res(struct qxl_device *qdev, struct qxl_release *release,
+		    struct qxl_bo *bo)
+{
+	int i;
+	for (i = 0; i < release->bo_count; i++)
+		if (release->bos[i] == bo)
+			return;
+
+	if (release->bo_count >= QXL_MAX_RES) {
+		DRM_ERROR("exceeded max resource on a qxl_release item\n");
+		return;
+	}
+	release->bos[release->bo_count++] = qxl_bo_ref(bo);
+}
+
+int qxl_release_bo_alloc(struct qxl_device *qdev,
+			 struct qxl_bo **bo)
+{
+	int ret;
+	ret = qxl_bo_create(qdev, PAGE_SIZE, false, QXL_GEM_DOMAIN_VRAM, NULL,
+			    bo);
+	return ret;
+}
+
+int qxl_release_reserve(struct qxl_device *qdev,
+			struct qxl_release *release, bool no_wait)
+{
+	int ret;
+	if (atomic_inc_return(&release->bos[0]->reserve_count) == 1) {
+		ret = qxl_bo_reserve(release->bos[0], no_wait);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
+void qxl_release_unreserve(struct qxl_device *qdev,
+			  struct qxl_release *release)
+{
+	if (atomic_dec_and_test(&release->bos[0]->reserve_count))
+		qxl_bo_unreserve(release->bos[0]);
+}
+
+int qxl_alloc_surface_release_reserved(struct qxl_device *qdev,
+				       enum qxl_surface_cmd_type surface_cmd_type,
+				       struct qxl_release *create_rel,
+				       struct qxl_release **release)
+{
+	int ret;
+
+	if (surface_cmd_type == QXL_SURFACE_CMD_DESTROY && create_rel) {
+		int idr_ret;
+		struct qxl_bo *bo;
+		union qxl_release_info *info;
+
+		/* stash the release after the create command */
+		idr_ret = qxl_release_alloc(qdev, QXL_RELEASE_SURFACE_CMD, release);
+		bo = qxl_bo_ref(create_rel->bos[0]);
+
+		(*release)->release_offset = create_rel->release_offset + 64;
+
+		qxl_release_add_res(qdev, *release, bo);
+
+		ret = qxl_release_reserve(qdev, *release, false);
+		if (ret) {
+			DRM_ERROR("release reserve failed\n");
+			goto out_unref;
+		}
+		info = qxl_release_map(qdev, *release);
+		info->id = idr_ret;
+		qxl_release_unmap(qdev, *release, info);
+
+
+out_unref:
+		qxl_bo_unref(&bo);
+		return ret;
+	}
+
+	return qxl_alloc_release_reserved(qdev, sizeof(struct qxl_surface_cmd),
+					 QXL_RELEASE_SURFACE_CMD, release, NULL);
+}
+
+int qxl_alloc_release_reserved(struct qxl_device *qdev, unsigned long size,
+				       int type, struct qxl_release **release,
+				       struct qxl_bo **rbo)
+{
+	struct qxl_bo *bo;
+	int idr_ret;
+	int ret;
+	union qxl_release_info *info;
+	int cur_idx;
+
+	if (type == QXL_RELEASE_DRAWABLE)
+		cur_idx = 0;
+	else if (type == QXL_RELEASE_SURFACE_CMD)
+		cur_idx = 1;
+	else if (type == QXL_RELEASE_CURSOR_CMD)
+		cur_idx = 2;
+	else {
+		DRM_ERROR("got illegal type: %d\n", type);
+		return -EINVAL;
+	}
+
+	idr_ret = qxl_release_alloc(qdev, type, release);
+
+	mutex_lock(&qdev->release_mutex);
+	if (qdev->current_release_bo_offset[cur_idx] + 1 >= releases_per_bo[cur_idx]) {
+		qxl_bo_unref(&qdev->current_release_bo[cur_idx]);
+		qdev->current_release_bo_offset[cur_idx] = 0;
+		qdev->current_release_bo[cur_idx] = NULL;
+	}
+	if (!qdev->current_release_bo[cur_idx]) {
+		ret = qxl_release_bo_alloc(qdev, &qdev->current_release_bo[cur_idx]);
+		if (ret) {
+			mutex_unlock(&qdev->release_mutex);
+			return ret;
+		}
+
+		/* pin releases bo's they are too messy to evict */
+		ret = qxl_bo_reserve(qdev->current_release_bo[cur_idx], false);
+		qxl_bo_pin(qdev->current_release_bo[cur_idx], QXL_GEM_DOMAIN_VRAM, NULL);
+		qxl_bo_unreserve(qdev->current_release_bo[cur_idx]);
+	}
+
+	bo = qxl_bo_ref(qdev->current_release_bo[cur_idx]);
+
+	(*release)->release_offset = qdev->current_release_bo_offset[cur_idx] * release_size_per_bo[cur_idx];
+	qdev->current_release_bo_offset[cur_idx]++;
+
+	if (rbo)
+		*rbo = bo;
+
+	qxl_release_add_res(qdev, *release, bo);
+
+	ret = qxl_release_reserve(qdev, *release, false);
+	mutex_unlock(&qdev->release_mutex);
+	if (ret)
+		goto out_unref;
+
+	info = qxl_release_map(qdev, *release);
+	info->id = idr_ret;
+	qxl_release_unmap(qdev, *release, info);
+
+out_unref:
+	qxl_bo_unref(&bo);
+	return ret;
+}
+
+int qxl_fence_releaseable(struct qxl_device *qdev,
+			  struct qxl_release *release)
+{
+	int i, ret;
+	for (i = 0; i < release->bo_count; i++) {
+		if (!release->bos[i]->tbo.sync_obj)
+			release->bos[i]->tbo.sync_obj = &release->bos[i]->fence;
+		ret = qxl_fence_add_release(&release->bos[i]->fence, release->id);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
+struct qxl_release *qxl_release_from_id_locked(struct qxl_device *qdev,
+						   uint64_t id)
+{
+	struct qxl_release *release;
+
+	spin_lock(&qdev->release_idr_lock);
+	release = idr_find(&qdev->release_idr, id);
+	spin_unlock(&qdev->release_idr_lock);
+	if (!release) {
+		DRM_ERROR("failed to find id in release_idr\n");
+		return NULL;
+	}
+	if (release->bo_count < 1) {
+		DRM_ERROR("read a released resource with 0 bos\n");
+		return NULL;
+	}
+	return release;
+}
+
+union qxl_release_info *qxl_release_map(struct qxl_device *qdev,
+					struct qxl_release *release)
+{
+	void *ptr;
+	union qxl_release_info *info;
+	struct qxl_bo *bo = release->bos[0];
+
+	ptr = qxl_bo_kmap_atomic_page(qdev, bo, release->release_offset & PAGE_SIZE);
+	info = ptr + (release->release_offset & ~PAGE_SIZE);
+	return info;
+}
+
+void qxl_release_unmap(struct qxl_device *qdev,
+		       struct qxl_release *release,
+		       union qxl_release_info *info)
+{
+	struct qxl_bo *bo = release->bos[0];
+	void *ptr;
+
+	ptr = ((void *)info) - (release->release_offset & ~PAGE_SIZE);
+	qxl_bo_kunmap_atomic_page(qdev, bo, ptr);
+}
diff --git a/drivers/gpu/drm/qxl/qxl_ttm.c b/drivers/gpu/drm/qxl/qxl_ttm.c
new file mode 100644
index 000000000000..aa9fb9afca0b
--- /dev/null
+++ b/drivers/gpu/drm/qxl/qxl_ttm.c
@@ -0,0 +1,577 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alon Levy
+ */
+
+#include <ttm/ttm_bo_api.h>
+#include <ttm/ttm_bo_driver.h>
+#include <ttm/ttm_placement.h>
+#include <ttm/ttm_page_alloc.h>
+#include <ttm/ttm_module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+#include <drm/qxl_drm.h>
+#include "qxl_drv.h"
+#include "qxl_object.h"
+
+#include <linux/delay.h>
+static int qxl_ttm_debugfs_init(struct qxl_device *qdev);
+
+static struct qxl_device *qxl_get_qdev(struct ttm_bo_device *bdev)
+{
+	struct qxl_mman *mman;
+	struct qxl_device *qdev;
+
+	mman = container_of(bdev, struct qxl_mman, bdev);
+	qdev = container_of(mman, struct qxl_device, mman);
+	return qdev;
+}
+
+static int qxl_ttm_mem_global_init(struct drm_global_reference *ref)
+{
+	return ttm_mem_global_init(ref->object);
+}
+
+static void qxl_ttm_mem_global_release(struct drm_global_reference *ref)
+{
+	ttm_mem_global_release(ref->object);
+}
+
+static int qxl_ttm_global_init(struct qxl_device *qdev)
+{
+	struct drm_global_reference *global_ref;
+	int r;
+
+	qdev->mman.mem_global_referenced = false;
+	global_ref = &qdev->mman.mem_global_ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_MEM;
+	global_ref->size = sizeof(struct ttm_mem_global);
+	global_ref->init = &qxl_ttm_mem_global_init;
+	global_ref->release = &qxl_ttm_mem_global_release;
+
+	r = drm_global_item_ref(global_ref);
+	if (r != 0) {
+		DRM_ERROR("Failed setting up TTM memory accounting "
+			  "subsystem.\n");
+		return r;
+	}
+
+	qdev->mman.bo_global_ref.mem_glob =
+		qdev->mman.mem_global_ref.object;
+	global_ref = &qdev->mman.bo_global_ref.ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_BO;
+	global_ref->size = sizeof(struct ttm_bo_global);
+	global_ref->init = &ttm_bo_global_init;
+	global_ref->release = &ttm_bo_global_release;
+	r = drm_global_item_ref(global_ref);
+	if (r != 0) {
+		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
+		drm_global_item_unref(&qdev->mman.mem_global_ref);
+		return r;
+	}
+
+	qdev->mman.mem_global_referenced = true;
+	return 0;
+}
+
+static void qxl_ttm_global_fini(struct qxl_device *qdev)
+{
+	if (qdev->mman.mem_global_referenced) {
+		drm_global_item_unref(&qdev->mman.bo_global_ref.ref);
+		drm_global_item_unref(&qdev->mman.mem_global_ref);
+		qdev->mman.mem_global_referenced = false;
+	}
+}
+
+static struct vm_operations_struct qxl_ttm_vm_ops;
+static const struct vm_operations_struct *ttm_vm_ops;
+
+static int qxl_ttm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
+{
+	struct ttm_buffer_object *bo;
+	struct qxl_device *qdev;
+	int r;
+
+	bo = (struct ttm_buffer_object *)vma->vm_private_data;
+	if (bo == NULL)
+		return VM_FAULT_NOPAGE;
+	qdev = qxl_get_qdev(bo->bdev);
+	r = ttm_vm_ops->fault(vma, vmf);
+	return r;
+}
+
+int qxl_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct drm_file *file_priv;
+	struct qxl_device *qdev;
+	int r;
+
+	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET)) {
+		pr_info("%s: vma->vm_pgoff (%ld) < DRM_FILE_PAGE_OFFSET\n",
+			__func__, vma->vm_pgoff);
+		return drm_mmap(filp, vma);
+	}
+
+	file_priv = filp->private_data;
+	qdev = file_priv->minor->dev->dev_private;
+	if (qdev == NULL) {
+		DRM_ERROR(
+		 "filp->private_data->minor->dev->dev_private == NULL\n");
+		return -EINVAL;
+	}
+	QXL_INFO(qdev, "%s: filp->private_data = 0x%p, vma->vm_pgoff = %lx\n",
+		 __func__, filp->private_data, vma->vm_pgoff);
+
+	r = ttm_bo_mmap(filp, vma, &qdev->mman.bdev);
+	if (unlikely(r != 0))
+		return r;
+	if (unlikely(ttm_vm_ops == NULL)) {
+		ttm_vm_ops = vma->vm_ops;
+		qxl_ttm_vm_ops = *ttm_vm_ops;
+		qxl_ttm_vm_ops.fault = &qxl_ttm_fault;
+	}
+	vma->vm_ops = &qxl_ttm_vm_ops;
+	return 0;
+}
+
+static int qxl_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags)
+{
+	return 0;
+}
+
+static int qxl_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
+			     struct ttm_mem_type_manager *man)
+{
+	struct qxl_device *qdev;
+
+	qdev = qxl_get_qdev(bdev);
+
+	switch (type) {
+	case TTM_PL_SYSTEM:
+		/* System memory */
+		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		break;
+	case TTM_PL_VRAM:
+	case TTM_PL_PRIV0:
+		/* "On-card" video ram */
+		man->func = &ttm_bo_manager_func;
+		man->gpu_offset = 0;
+		man->flags = TTM_MEMTYPE_FLAG_FIXED |
+			     TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		break;
+	default:
+		DRM_ERROR("Unsupported memory type %u\n", (unsigned)type);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void qxl_evict_flags(struct ttm_buffer_object *bo,
+				struct ttm_placement *placement)
+{
+	struct qxl_bo *qbo;
+	static u32 placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;
+
+	if (!qxl_ttm_bo_is_qxl_bo(bo)) {
+		placement->fpfn = 0;
+		placement->lpfn = 0;
+		placement->placement = &placements;
+		placement->busy_placement = &placements;
+		placement->num_placement = 1;
+		placement->num_busy_placement = 1;
+		return;
+	}
+	qbo = container_of(bo, struct qxl_bo, tbo);
+	qxl_ttm_placement_from_domain(qbo, QXL_GEM_DOMAIN_CPU);
+	*placement = qbo->placement;
+}
+
+static int qxl_verify_access(struct ttm_buffer_object *bo, struct file *filp)
+{
+	return 0;
+}
+
+static int qxl_ttm_io_mem_reserve(struct ttm_bo_device *bdev,
+				  struct ttm_mem_reg *mem)
+{
+	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
+	struct qxl_device *qdev = qxl_get_qdev(bdev);
+
+	mem->bus.addr = NULL;
+	mem->bus.offset = 0;
+	mem->bus.size = mem->num_pages << PAGE_SHIFT;
+	mem->bus.base = 0;
+	mem->bus.is_iomem = false;
+	if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE))
+		return -EINVAL;
+	switch (mem->mem_type) {
+	case TTM_PL_SYSTEM:
+		/* system memory */
+		return 0;
+	case TTM_PL_VRAM:
+		mem->bus.is_iomem = true;
+		mem->bus.base = qdev->vram_base;
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		break;
+	case TTM_PL_PRIV0:
+		mem->bus.is_iomem = true;
+		mem->bus.base = qdev->surfaceram_base;
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void qxl_ttm_io_mem_free(struct ttm_bo_device *bdev,
+				struct ttm_mem_reg *mem)
+{
+}
+
+/*
+ * TTM backend functions.
+ */
+struct qxl_ttm_tt {
+	struct ttm_dma_tt		ttm;
+	struct qxl_device		*qdev;
+	u64				offset;
+};
+
+static int qxl_ttm_backend_bind(struct ttm_tt *ttm,
+				struct ttm_mem_reg *bo_mem)
+{
+	struct qxl_ttm_tt *gtt = (void *)ttm;
+
+	gtt->offset = (unsigned long)(bo_mem->start << PAGE_SHIFT);
+	if (!ttm->num_pages) {
+		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
+		     ttm->num_pages, bo_mem, ttm);
+	}
+	/* Not implemented */
+	return -1;
+}
+
+static int qxl_ttm_backend_unbind(struct ttm_tt *ttm)
+{
+	/* Not implemented */
+	return -1;
+}
+
+static void qxl_ttm_backend_destroy(struct ttm_tt *ttm)
+{
+	struct qxl_ttm_tt *gtt = (void *)ttm;
+
+	ttm_dma_tt_fini(&gtt->ttm);
+	kfree(gtt);
+}
+
+static struct ttm_backend_func qxl_backend_func = {
+	.bind = &qxl_ttm_backend_bind,
+	.unbind = &qxl_ttm_backend_unbind,
+	.destroy = &qxl_ttm_backend_destroy,
+};
+
+static int qxl_ttm_tt_populate(struct ttm_tt *ttm)
+{
+	int r;
+
+	if (ttm->state != tt_unpopulated)
+		return 0;
+
+	r = ttm_pool_populate(ttm);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void qxl_ttm_tt_unpopulate(struct ttm_tt *ttm)
+{
+	ttm_pool_unpopulate(ttm);
+}
+
+struct ttm_tt *qxl_ttm_tt_create(struct ttm_bo_device *bdev,
+				 unsigned long size, uint32_t page_flags,
+				 struct page *dummy_read_page)
+{
+	struct qxl_device *qdev;
+	struct qxl_ttm_tt *gtt;
+
+	qdev = qxl_get_qdev(bdev);
+	gtt = kzalloc(sizeof(struct qxl_ttm_tt), GFP_KERNEL);
+	if (gtt == NULL)
+		return NULL;
+	gtt->ttm.ttm.func = &qxl_backend_func;
+	gtt->qdev = qdev;
+	if (ttm_dma_tt_init(&gtt->ttm, bdev, size, page_flags,
+			    dummy_read_page)) {
+		kfree(gtt);
+		return NULL;
+	}
+	return &gtt->ttm.ttm;
+}
+
+static void qxl_move_null(struct ttm_buffer_object *bo,
+			     struct ttm_mem_reg *new_mem)
+{
+	struct ttm_mem_reg *old_mem = &bo->mem;
+
+	BUG_ON(old_mem->mm_node != NULL);
+	*old_mem = *new_mem;
+	new_mem->mm_node = NULL;
+}
+
+static int qxl_bo_move(struct ttm_buffer_object *bo,
+		       bool evict, bool interruptible,
+		       bool no_wait_gpu,
+		       struct ttm_mem_reg *new_mem)
+{
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
+		qxl_move_null(bo, new_mem);
+		return 0;
+	}
+	return ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem);
+}
+
+
+static int qxl_sync_obj_wait(void *sync_obj,
+			     bool lazy, bool interruptible)
+{
+	struct qxl_fence *qfence = (struct qxl_fence *)sync_obj;
+	int count = 0, sc = 0;
+	struct qxl_bo *bo = container_of(qfence, struct qxl_bo, fence);
+
+	if (qfence->num_active_releases == 0)
+		return 0;
+
+retry:
+	if (sc == 0) {
+		if (bo->type == QXL_GEM_DOMAIN_SURFACE)
+			qxl_update_surface(qfence->qdev, bo);
+	} else if (sc >= 1) {
+		qxl_io_notify_oom(qfence->qdev);
+	}
+
+	sc++;
+
+	for (count = 0; count < 10; count++) {
+		bool ret;
+		ret = qxl_queue_garbage_collect(qfence->qdev, true);
+		if (ret == false)
+			break;
+
+		if (qfence->num_active_releases == 0)
+			return 0;
+	}
+
+	if (qfence->num_active_releases) {
+		bool have_drawable_releases = false;
+		void **slot;
+		struct radix_tree_iter iter;
+		int release_id;
+
+		radix_tree_for_each_slot(slot, &qfence->tree, &iter, 0) {
+			struct qxl_release *release;
+
+			release_id = iter.index;
+			release = qxl_release_from_id_locked(qfence->qdev, release_id);
+			if (release == NULL)
+				continue;
+
+			if (release->type == QXL_RELEASE_DRAWABLE)
+				have_drawable_releases = true;
+		}
+
+		qxl_queue_garbage_collect(qfence->qdev, true);
+
+		if (have_drawable_releases || sc < 4) {
+			if (sc > 2)
+				/* back off */
+				usleep_range(500, 1000);
+			if (have_drawable_releases && sc > 300) {
+				WARN(1, "sync obj %d still has outstanding releases %d %d %d %ld %d\n", sc, bo->surface_id, bo->is_primary, bo->pin_count, (unsigned long)bo->gem_base.size, qfence->num_active_releases);
+				return -EBUSY;
+			}
+			goto retry;
+		}
+	}
+	return 0;
+}
+
+static int qxl_sync_obj_flush(void *sync_obj)
+{
+	return 0;
+}
+
+static void qxl_sync_obj_unref(void **sync_obj)
+{
+}
+
+static void *qxl_sync_obj_ref(void *sync_obj)
+{
+	return sync_obj;
+}
+
+static bool qxl_sync_obj_signaled(void *sync_obj)
+{
+	struct qxl_fence *qfence = (struct qxl_fence *)sync_obj;
+	return (qfence->num_active_releases == 0);
+}
+
+static void qxl_bo_move_notify(struct ttm_buffer_object *bo,
+			       struct ttm_mem_reg *new_mem)
+{
+	struct qxl_bo *qbo;
+	struct qxl_device *qdev;
+
+	if (!qxl_ttm_bo_is_qxl_bo(bo))
+		return;
+	qbo = container_of(bo, struct qxl_bo, tbo);
+	qdev = qbo->gem_base.dev->dev_private;
+
+	if (bo->mem.mem_type == TTM_PL_PRIV0 && qbo->surface_id)
+		qxl_surface_evict(qdev, qbo, new_mem ? true : false);
+}
+
+static struct ttm_bo_driver qxl_bo_driver = {
+	.ttm_tt_create = &qxl_ttm_tt_create,
+	.ttm_tt_populate = &qxl_ttm_tt_populate,
+	.ttm_tt_unpopulate = &qxl_ttm_tt_unpopulate,
+	.invalidate_caches = &qxl_invalidate_caches,
+	.init_mem_type = &qxl_init_mem_type,
+	.evict_flags = &qxl_evict_flags,
+	.move = &qxl_bo_move,
+	.verify_access = &qxl_verify_access,
+	.io_mem_reserve = &qxl_ttm_io_mem_reserve,
+	.io_mem_free = &qxl_ttm_io_mem_free,
+	.sync_obj_signaled = &qxl_sync_obj_signaled,
+	.sync_obj_wait = &qxl_sync_obj_wait,
+	.sync_obj_flush = &qxl_sync_obj_flush,
+	.sync_obj_unref = &qxl_sync_obj_unref,
+	.sync_obj_ref = &qxl_sync_obj_ref,
+	.move_notify = &qxl_bo_move_notify,
+};
+
+
+
+int qxl_ttm_init(struct qxl_device *qdev)
+{
+	int r;
+	int num_io_pages; /* != rom->num_io_pages, we include surface0 */
+
+	r = qxl_ttm_global_init(qdev);
+	if (r)
+		return r;
+	/* No others user of address space so set it to 0 */
+	r = ttm_bo_device_init(&qdev->mman.bdev,
+			       qdev->mman.bo_global_ref.ref.object,
+			       &qxl_bo_driver, DRM_FILE_PAGE_OFFSET, 0);
+	if (r) {
+		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		return r;
+	}
+	/* NOTE: this includes the framebuffer (aka surface 0) */
+	num_io_pages = qdev->rom->ram_header_offset / PAGE_SIZE;
+	r = ttm_bo_init_mm(&qdev->mman.bdev, TTM_PL_VRAM,
+			   num_io_pages);
+	if (r) {
+		DRM_ERROR("Failed initializing VRAM heap.\n");
+		return r;
+	}
+	r = ttm_bo_init_mm(&qdev->mman.bdev, TTM_PL_PRIV0,
+			   qdev->surfaceram_size / PAGE_SIZE);
+	if (r) {
+		DRM_ERROR("Failed initializing Surfaces heap.\n");
+		return r;
+	}
+	DRM_INFO("qxl: %uM of VRAM memory size\n",
+		 (unsigned)qdev->vram_size / (1024 * 1024));
+	DRM_INFO("qxl: %luM of IO pages memory ready (VRAM domain)\n",
+		 ((unsigned)num_io_pages * PAGE_SIZE) / (1024 * 1024));
+	if (unlikely(qdev->mman.bdev.dev_mapping == NULL))
+		qdev->mman.bdev.dev_mapping = qdev->ddev->dev_mapping;
+	r = qxl_ttm_debugfs_init(qdev);
+	if (r) {
+		DRM_ERROR("Failed to init debugfs\n");
+		return r;
+	}
+	return 0;
+}
+
+void qxl_ttm_fini(struct qxl_device *qdev)
+{
+	ttm_bo_clean_mm(&qdev->mman.bdev, TTM_PL_VRAM);
+	ttm_bo_clean_mm(&qdev->mman.bdev, TTM_PL_PRIV0);
+	ttm_bo_device_release(&qdev->mman.bdev);
+	qxl_ttm_global_fini(qdev);
+	DRM_INFO("qxl: ttm finalized\n");
+}
+
+
+#define QXL_DEBUGFS_MEM_TYPES 2
+
+#if defined(CONFIG_DEBUG_FS)
+static int qxl_mm_dump_table(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_mm *mm = (struct drm_mm *)node->info_ent->data;
+	struct drm_device *dev = node->minor->dev;
+	struct qxl_device *rdev = dev->dev_private;
+	int ret;
+	struct ttm_bo_global *glob = rdev->mman.bdev.glob;
+
+	spin_lock(&glob->lru_lock);
+	ret = drm_mm_dump_table(m, mm);
+	spin_unlock(&glob->lru_lock);
+	return ret;
+}
+#endif
+
+static int qxl_ttm_debugfs_init(struct qxl_device *qdev)
+{
+	static struct drm_info_list qxl_mem_types_list[QXL_DEBUGFS_MEM_TYPES];
+	static char qxl_mem_types_names[QXL_DEBUGFS_MEM_TYPES][32];
+	unsigned i;
+
+	for (i = 0; i < QXL_DEBUGFS_MEM_TYPES; i++) {
+		if (i == 0)
+			sprintf(qxl_mem_types_names[i], "qxl_mem_mm");
+		else
+			sprintf(qxl_mem_types_names[i], "qxl_surf_mm");
+		qxl_mem_types_list[i].name = qxl_mem_types_names[i];
+		qxl_mem_types_list[i].show = &qxl_mm_dump_table;
+		qxl_mem_types_list[i].driver_features = 0;
+		if (i == 0)
+			qxl_mem_types_list[i].data = qdev->mman.bdev.man[TTM_PL_VRAM].priv;
+		else
+			qxl_mem_types_list[i].data = qdev->mman.bdev.man[TTM_PL_PRIV0].priv;
+
+	}
+	return qxl_debugfs_add_files(qdev, qxl_mem_types_list, i);
+}
diff --git a/include/uapi/drm/Kbuild b/include/uapi/drm/Kbuild
index ba99ce3f7372..a042a957296d 100644
--- a/include/uapi/drm/Kbuild
+++ b/include/uapi/drm/Kbuild
@@ -8,6 +8,7 @@ header-y += i810_drm.h
 header-y += i915_drm.h
 header-y += mga_drm.h
 header-y += nouveau_drm.h
+header-y += qxl_drm.h
 header-y += r128_drm.h
 header-y += radeon_drm.h
 header-y += savage_drm.h
diff --git a/include/uapi/drm/qxl_drm.h b/include/uapi/drm/qxl_drm.h
new file mode 100644
index 000000000000..ebebd36c4117
--- /dev/null
+++ b/include/uapi/drm/qxl_drm.h
@@ -0,0 +1,152 @@
+/*
+ * Copyright 2013 Red Hat
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef QXL_DRM_H
+#define QXL_DRM_H
+
+#include <stddef.h>
+#include "drm/drm.h"
+
+/* Please note that modifications to all structs defined here are
+ * subject to backwards-compatibility constraints.
+ *
+ * Do not use pointers, use uint64_t instead for 32 bit / 64 bit user/kernel
+ * compatibility Keep fields aligned to their size
+ */
+
+#define QXL_GEM_DOMAIN_CPU 0
+#define QXL_GEM_DOMAIN_VRAM 1
+#define QXL_GEM_DOMAIN_SURFACE 2
+
+#define DRM_QXL_ALLOC       0x00
+#define DRM_QXL_MAP         0x01
+#define DRM_QXL_EXECBUFFER  0x02
+#define DRM_QXL_UPDATE_AREA 0x03
+#define DRM_QXL_GETPARAM    0x04
+#define DRM_QXL_CLIENTCAP   0x05
+
+#define DRM_QXL_ALLOC_SURF  0x06
+
+struct drm_qxl_alloc {
+	uint32_t size;
+	uint32_t handle; /* 0 is an invalid handle */
+};
+
+struct drm_qxl_map {
+	uint64_t offset; /* use for mmap system call */
+	uint32_t handle;
+	uint32_t pad;
+};
+
+/*
+ * dest is the bo we are writing the relocation into
+ * src is bo we are relocating.
+ * *(dest_handle.base_addr + dest_offset) = physical_address(src_handle.addr +
+ * src_offset)
+ */
+#define QXL_RELOC_TYPE_BO 1
+#define QXL_RELOC_TYPE_SURF 2
+
+struct drm_qxl_reloc {
+	uint64_t src_offset; /* offset into src_handle or src buffer */
+	uint64_t dst_offset; /* offset in dest handle */
+	uint32_t src_handle; /* dest handle to compute address from */
+	uint32_t dst_handle; /* 0 if to command buffer */
+	uint32_t reloc_type;
+	uint32_t pad;
+};
+
+struct drm_qxl_command {
+	uint64_t	 __user command; /* void* */
+	uint64_t	 __user relocs; /* struct drm_qxl_reloc* */
+	uint32_t		type;
+	uint32_t		command_size;
+	uint32_t		relocs_num;
+	uint32_t                pad;
+};
+
+/* XXX: call it drm_qxl_commands? */
+struct drm_qxl_execbuffer {
+	uint32_t		flags;		/* for future use */
+	uint32_t		commands_num;
+	uint64_t	 __user commands;	/* struct drm_qxl_command* */
+};
+
+struct drm_qxl_update_area {
+	uint32_t handle;
+	uint32_t top;
+	uint32_t left;
+	uint32_t bottom;
+	uint32_t right;
+	uint32_t pad;
+};
+
+#define QXL_PARAM_NUM_SURFACES 1 /* rom->n_surfaces */
+#define QXL_PARAM_MAX_RELOCS 2
+struct drm_qxl_getparam {
+	uint64_t param;
+	uint64_t value;
+};
+
+/* these are one bit values */
+struct drm_qxl_clientcap {
+	uint32_t index;
+	uint32_t pad;
+};
+
+struct drm_qxl_alloc_surf {
+	uint32_t format;
+	uint32_t width;
+	uint32_t height;
+	int32_t stride;
+	uint32_t handle;
+	uint32_t pad;
+};
+
+#define DRM_IOCTL_QXL_ALLOC \
+	DRM_IOWR(DRM_COMMAND_BASE + DRM_QXL_ALLOC, struct drm_qxl_alloc)
+
+#define DRM_IOCTL_QXL_MAP \
+	DRM_IOWR(DRM_COMMAND_BASE + DRM_QXL_MAP, struct drm_qxl_map)
+
+#define DRM_IOCTL_QXL_EXECBUFFER \
+	DRM_IOW(DRM_COMMAND_BASE + DRM_QXL_EXECBUFFER,\
+		struct drm_qxl_execbuffer)
+
+#define DRM_IOCTL_QXL_UPDATE_AREA \
+	DRM_IOW(DRM_COMMAND_BASE + DRM_QXL_UPDATE_AREA,\
+		struct drm_qxl_update_area)
+
+#define DRM_IOCTL_QXL_GETPARAM \
+	DRM_IOWR(DRM_COMMAND_BASE + DRM_QXL_GETPARAM,\
+		struct drm_qxl_getparam)
+
+#define DRM_IOCTL_QXL_CLIENTCAP \
+	DRM_IOW(DRM_COMMAND_BASE + DRM_QXL_CLIENTCAP,\
+		struct drm_qxl_clientcap)
+
+#define DRM_IOCTL_QXL_ALLOC_SURF \
+	DRM_IOWR(DRM_COMMAND_BASE + DRM_QXL_ALLOC_SURF,\
+		struct drm_qxl_alloc_surf)
+
+#endif
-- 
cgit v1.2.3


From 715f59cc2391cb4e28a4d464a0564a60a00ecca4 Mon Sep 17 00:00:00 2001
From: Christopher Harvey <charvey@matrox.com>
Date: Fri, 5 Apr 2013 15:28:32 +0000
Subject: drm: Misc comment cleanup

Signed-off-by: Christopher Harvey <charvey@matrox.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/mgag200/mgag200_drv.h | 6 +++---
 include/drm/drm_crtc.h                | 2 +-
 include/uapi/drm/drm_mode.h           | 6 +++---
 3 files changed, 7 insertions(+), 7 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/mgag200/mgag200_drv.h b/drivers/gpu/drm/mgag200/mgag200_drv.h
index 4d932c46725d..dcfc973e29f7 100644
--- a/drivers/gpu/drm/mgag200/mgag200_drv.h
+++ b/drivers/gpu/drm/mgag200/mgag200_drv.h
@@ -215,7 +215,7 @@ mgag200_bo(struct ttm_buffer_object *bo)
 {
 	return container_of(bo, struct mgag200_bo, bo);
 }
-				/* mga_crtc.c */
+				/* mgag200_crtc.c */
 void mga_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green,
 			     u16 blue, int regno);
 void mga_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green,
@@ -225,7 +225,7 @@ void mga_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green,
 int mgag200_modeset_init(struct mga_device *mdev);
 void mgag200_modeset_fini(struct mga_device *mdev);
 
-				/* mga_fbdev.c */
+				/* mgag200_fb.c */
 int mgag200_fbdev_init(struct mga_device *mdev);
 void mgag200_fbdev_fini(struct mga_device *mdev);
 
@@ -254,7 +254,7 @@ mgag200_dumb_mmap_offset(struct drm_file *file,
 			 struct drm_device *dev,
 			 uint32_t handle,
 			 uint64_t *offset);
-				/* mga_i2c.c */
+				/* mgag200_i2c.c */
 struct mga_i2c_chan *mgag200_i2c_create(struct drm_device *dev);
 void mgag200_i2c_destroy(struct mga_i2c_chan *i2c);
 
diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index e3e0d651c6ca..8c7846bd74f4 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -310,7 +310,7 @@ struct drm_plane;
  * drm_crtc_funcs - control CRTCs for a given device
  * @save: save CRTC state
  * @restore: restore CRTC state
- * @reset: reset CRTC after state has been invalidate (e.g. resume)
+ * @reset: reset CRTC after state has been invalidated (e.g. resume)
  * @cursor_set: setup the cursor
  * @cursor_move: move the cursor
  * @gamma_set: specify color ramp for CRTC
diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
index 3d6301b6ec16..090e5331ab7e 100644
--- a/include/uapi/drm/drm_mode.h
+++ b/include/uapi/drm/drm_mode.h
@@ -367,13 +367,13 @@ struct drm_mode_mode_cmd {
  * depending on the value in flags different members are used.
  *
  * CURSOR_BO uses
- *    crtc
+ *    crtc_id
  *    width
  *    height
- *    handle - if 0 turns the cursor of
+ *    handle - if 0 turns the cursor off
  *
  * CURSOR_MOVE uses
- *    crtc
+ *    crtc_id
  *    x
  *    y
  */
-- 
cgit v1.2.3


From b6330548383cf18d608faf1c2dc0c81c60476233 Mon Sep 17 00:00:00 2001
From: Paul Sokolovsky <paul.sokolovsky@linaro.org>
Date: Fri, 12 Apr 2013 18:28:26 -0500
Subject: drm.h: Fix DRM compilation with bare-metal toolchain.

An ifdef in drm.h expects to be compiled with full-fledged Linux
toolchain, but it's common to compile kernel with just bare-metal
toolchain which doesn't define __linux__. So, also add __KERNEL__
check.

[nm@ti.com: port forward to 3.9-rc6 and post to dri devel for feedback as RFC]
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 include/uapi/drm/drm.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index 8d1e2bbee83a..73a99e4664be 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -36,7 +36,7 @@
 #ifndef _DRM_H_
 #define _DRM_H_
 
-#if defined(__linux__)
+#if defined(__KERNEL__) || defined(__linux__)
 
 #include <linux/types.h>
 #include <asm/ioctl.h>
-- 
cgit v1.2.3


From 754716874389ccbea5ee03174df8ad9e72e41880 Mon Sep 17 00:00:00 2001
From: Terje Bergstrom <tbergstrom@nvidia.com>
Date: Fri, 22 Mar 2013 16:34:01 +0200
Subject: gpu: host1x: Add host1x driver

Add host1x, the driver for host1x and its client unit 2D. The Tegra
host1x module is the DMA engine for register access to Tegra's
graphics- and multimedia-related modules. The modules served by
host1x are referred to as clients. host1x includes some other
functionality, such as synchronization.

Signed-off-by: Arto Merilainen <amerilainen@nvidia.com>
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Erik Faye-Lund <kusmabite@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
---
 drivers/gpu/Makefile                      |   1 +
 drivers/gpu/host1x/Kconfig                |  10 ++
 drivers/gpu/host1x/Makefile               |   8 ++
 drivers/gpu/host1x/dev.c                  | 153 +++++++++++++++++++++
 drivers/gpu/host1x/dev.h                  |  96 ++++++++++++++
 drivers/gpu/host1x/hw/Makefile            |   6 +
 drivers/gpu/host1x/hw/host1x01.c          |  33 +++++
 drivers/gpu/host1x/hw/host1x01.h          |  25 ++++
 drivers/gpu/host1x/hw/host1x01_hardware.h |  27 ++++
 drivers/gpu/host1x/hw/hw_host1x01_sync.h  |  74 +++++++++++
 drivers/gpu/host1x/hw/syncpt_hw.c         | 102 ++++++++++++++
 drivers/gpu/host1x/syncpt.c               | 212 ++++++++++++++++++++++++++++++
 drivers/gpu/host1x/syncpt.h               | 147 +++++++++++++++++++++
 drivers/video/Kconfig                     |   2 +
 include/trace/events/host1x.h             |  61 +++++++++
 15 files changed, 957 insertions(+)
 create mode 100644 drivers/gpu/host1x/Kconfig
 create mode 100644 drivers/gpu/host1x/Makefile
 create mode 100644 drivers/gpu/host1x/dev.c
 create mode 100644 drivers/gpu/host1x/dev.h
 create mode 100644 drivers/gpu/host1x/hw/Makefile
 create mode 100644 drivers/gpu/host1x/hw/host1x01.c
 create mode 100644 drivers/gpu/host1x/hw/host1x01.h
 create mode 100644 drivers/gpu/host1x/hw/host1x01_hardware.h
 create mode 100644 drivers/gpu/host1x/hw/hw_host1x01_sync.h
 create mode 100644 drivers/gpu/host1x/hw/syncpt_hw.c
 create mode 100644 drivers/gpu/host1x/syncpt.c
 create mode 100644 drivers/gpu/host1x/syncpt.h
 create mode 100644 include/trace/events/host1x.h

(limited to 'include')

diff --git a/drivers/gpu/Makefile b/drivers/gpu/Makefile
index 30879df3daea..d8a22c2a579d 100644
--- a/drivers/gpu/Makefile
+++ b/drivers/gpu/Makefile
@@ -1 +1,2 @@
 obj-y			+= drm/ vga/
+obj-$(CONFIG_TEGRA_HOST1X)	+= host1x/
diff --git a/drivers/gpu/host1x/Kconfig b/drivers/gpu/host1x/Kconfig
new file mode 100644
index 000000000000..567ae8b7c7c9
--- /dev/null
+++ b/drivers/gpu/host1x/Kconfig
@@ -0,0 +1,10 @@
+config TEGRA_HOST1X
+	tristate "NVIDIA Tegra host1x driver"
+	depends on ARCH_TEGRA || ARCH_MULTIPLATFORM
+	help
+	  Driver for the NVIDIA Tegra host1x hardware.
+
+	  The Tegra host1x module is the DMA engine for register access to
+	  Tegra's graphics- and multimedia-related modules. The modules served
+	  by host1x are referred to as clients. host1x includes some other
+	  functionality, such as synchronization.
diff --git a/drivers/gpu/host1x/Makefile b/drivers/gpu/host1x/Makefile
new file mode 100644
index 000000000000..363e6ab31517
--- /dev/null
+++ b/drivers/gpu/host1x/Makefile
@@ -0,0 +1,8 @@
+ccflags-y = -Idrivers/gpu/host1x
+
+host1x-y = \
+	syncpt.o \
+	dev.o \
+	hw/host1x01.o
+
+obj-$(CONFIG_TEGRA_HOST1X) += host1x.o
diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c
new file mode 100644
index 000000000000..0d6002cb67c1
--- /dev/null
+++ b/drivers/gpu/host1x/dev.c
@@ -0,0 +1,153 @@
+/*
+ * Tegra host1x driver
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/module.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/clk.h>
+#include <linux/io.h>
+
+#define CREATE_TRACE_POINTS
+#include <trace/events/host1x.h>
+
+#include "dev.h"
+#include "hw/host1x01.h"
+
+void host1x_sync_writel(struct host1x *host1x, u32 v, u32 r)
+{
+	void __iomem *sync_regs = host1x->regs + host1x->info->sync_offset;
+
+	writel(v, sync_regs + r);
+}
+
+u32 host1x_sync_readl(struct host1x *host1x, u32 r)
+{
+	void __iomem *sync_regs = host1x->regs + host1x->info->sync_offset;
+
+	return readl(sync_regs + r);
+}
+
+static const struct host1x_info host1x01_info = {
+	.nb_channels	= 8,
+	.nb_pts		= 32,
+	.nb_mlocks	= 16,
+	.nb_bases	= 8,
+	.init		= host1x01_init,
+	.sync_offset	= 0x3000,
+};
+
+static struct of_device_id host1x_of_match[] = {
+	{ .compatible = "nvidia,tegra30-host1x", .data = &host1x01_info, },
+	{ .compatible = "nvidia,tegra20-host1x", .data = &host1x01_info, },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, host1x_of_match);
+
+static int host1x_probe(struct platform_device *pdev)
+{
+	const struct of_device_id *id;
+	struct host1x *host;
+	struct resource *regs;
+	int syncpt_irq;
+	int err;
+
+	id = of_match_device(host1x_of_match, &pdev->dev);
+	if (!id)
+		return -EINVAL;
+
+	regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!regs) {
+		dev_err(&pdev->dev, "failed to get registers\n");
+		return -ENXIO;
+	}
+
+	syncpt_irq = platform_get_irq(pdev, 0);
+	if (syncpt_irq < 0) {
+		dev_err(&pdev->dev, "failed to get IRQ\n");
+		return -ENXIO;
+	}
+
+	host = devm_kzalloc(&pdev->dev, sizeof(*host), GFP_KERNEL);
+	if (!host)
+		return -ENOMEM;
+
+	host->dev = &pdev->dev;
+	host->info = id->data;
+
+	/* set common host1x device data */
+	platform_set_drvdata(pdev, host);
+
+	host->regs = devm_ioremap_resource(&pdev->dev, regs);
+	if (IS_ERR(host->regs))
+		return PTR_ERR(host->regs);
+
+	if (host->info->init) {
+		err = host->info->init(host);
+		if (err)
+			return err;
+	}
+
+	host->clk = devm_clk_get(&pdev->dev, NULL);
+	if (IS_ERR(host->clk)) {
+		dev_err(&pdev->dev, "failed to get clock\n");
+		err = PTR_ERR(host->clk);
+		return err;
+	}
+
+	err = clk_prepare_enable(host->clk);
+	if (err < 0) {
+		dev_err(&pdev->dev, "failed to enable clock\n");
+		return err;
+	}
+
+	err = host1x_syncpt_init(host);
+	if (err) {
+		dev_err(&pdev->dev, "failed to initialize syncpts\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static int __exit host1x_remove(struct platform_device *pdev)
+{
+	struct host1x *host = platform_get_drvdata(pdev);
+
+	host1x_syncpt_deinit(host);
+	clk_disable_unprepare(host->clk);
+
+	return 0;
+}
+
+static struct platform_driver platform_driver = {
+	.probe = host1x_probe,
+	.remove = __exit_p(host1x_remove),
+	.driver = {
+		.owner = THIS_MODULE,
+		.name = "tegra-host1x",
+		.of_match_table = host1x_of_match,
+	},
+};
+
+module_platform_driver(platform_driver);
+
+MODULE_AUTHOR("Terje Bergstrom <tbergstrom@nvidia.com>");
+MODULE_DESCRIPTION("Host1x driver for Tegra products");
+MODULE_LICENSE("GPL");
diff --git a/drivers/gpu/host1x/dev.h b/drivers/gpu/host1x/dev.h
new file mode 100644
index 000000000000..eaf602657f76
--- /dev/null
+++ b/drivers/gpu/host1x/dev.h
@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef HOST1X_DEV_H
+#define HOST1X_DEV_H
+
+#include <linux/platform_device.h>
+#include <linux/device.h>
+
+#include "syncpt.h"
+
+struct host1x_syncpt;
+
+struct host1x_syncpt_ops {
+	void (*restore)(struct host1x_syncpt *syncpt);
+	void (*restore_wait_base)(struct host1x_syncpt *syncpt);
+	void (*load_wait_base)(struct host1x_syncpt *syncpt);
+	u32 (*load)(struct host1x_syncpt *syncpt);
+	void (*cpu_incr)(struct host1x_syncpt *syncpt);
+	int (*patch_wait)(struct host1x_syncpt *syncpt, void *patch_addr);
+};
+
+struct host1x_info {
+	int	nb_channels;		/* host1x: num channels supported */
+	int	nb_pts;			/* host1x: num syncpoints supported */
+	int	nb_bases;		/* host1x: num syncpoints supported */
+	int	nb_mlocks;		/* host1x: number of mlocks */
+	int	(*init)(struct host1x *); /* initialize per SoC ops */
+	int	sync_offset;
+};
+
+struct host1x {
+	const struct host1x_info *info;
+
+	void __iomem *regs;
+	struct host1x_syncpt *syncpt;
+	struct device *dev;
+	struct clk *clk;
+
+	const struct host1x_syncpt_ops *syncpt_op;
+};
+
+void host1x_sync_writel(struct host1x *host1x, u32 r, u32 v);
+u32 host1x_sync_readl(struct host1x *host1x, u32 r);
+
+static inline void host1x_hw_syncpt_restore(struct host1x *host,
+					    struct host1x_syncpt *sp)
+{
+	host->syncpt_op->restore(sp);
+}
+
+static inline void host1x_hw_syncpt_restore_wait_base(struct host1x *host,
+						      struct host1x_syncpt *sp)
+{
+	host->syncpt_op->restore_wait_base(sp);
+}
+
+static inline void host1x_hw_syncpt_load_wait_base(struct host1x *host,
+						   struct host1x_syncpt *sp)
+{
+	host->syncpt_op->load_wait_base(sp);
+}
+
+static inline u32 host1x_hw_syncpt_load(struct host1x *host,
+					struct host1x_syncpt *sp)
+{
+	return host->syncpt_op->load(sp);
+}
+
+static inline void host1x_hw_syncpt_cpu_incr(struct host1x *host,
+					     struct host1x_syncpt *sp)
+{
+	host->syncpt_op->cpu_incr(sp);
+}
+
+static inline int host1x_hw_syncpt_patch_wait(struct host1x *host,
+					      struct host1x_syncpt *sp,
+					      void *patch_addr)
+{
+	return host->syncpt_op->patch_wait(sp, patch_addr);
+}
+
+#endif
diff --git a/drivers/gpu/host1x/hw/Makefile b/drivers/gpu/host1x/hw/Makefile
new file mode 100644
index 000000000000..9b50863a2236
--- /dev/null
+++ b/drivers/gpu/host1x/hw/Makefile
@@ -0,0 +1,6 @@
+ccflags-y = -Idrivers/gpu/host1x
+
+host1x-hw-objs  = \
+	host1x01.o
+
+obj-$(CONFIG_TEGRA_HOST1X) += host1x-hw.o
diff --git a/drivers/gpu/host1x/hw/host1x01.c b/drivers/gpu/host1x/hw/host1x01.c
new file mode 100644
index 000000000000..612b4574c4b6
--- /dev/null
+++ b/drivers/gpu/host1x/hw/host1x01.c
@@ -0,0 +1,33 @@
+/*
+ * Host1x init for T20 and T30 Architecture Chips
+ *
+ * Copyright (c) 2011-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+/* include hw specification */
+#include "hw/host1x01.h"
+#include "hw/host1x01_hardware.h"
+
+/* include code */
+#include "hw/syncpt_hw.c"
+
+#include "dev.h"
+
+int host1x01_init(struct host1x *host)
+{
+	host->syncpt_op = &host1x_syncpt_ops;
+
+	return 0;
+}
diff --git a/drivers/gpu/host1x/hw/host1x01.h b/drivers/gpu/host1x/hw/host1x01.h
new file mode 100644
index 000000000000..2706b6743250
--- /dev/null
+++ b/drivers/gpu/host1x/hw/host1x01.h
@@ -0,0 +1,25 @@
+/*
+ * Host1x init for T20 and T30 Architecture Chips
+ *
+ * Copyright (c) 2011-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef HOST1X_HOST1X01_H
+#define HOST1X_HOST1X01_H
+
+struct host1x;
+
+int host1x01_init(struct host1x *host);
+
+#endif /* HOST1X_HOST1X01_H_ */
diff --git a/drivers/gpu/host1x/hw/host1x01_hardware.h b/drivers/gpu/host1x/hw/host1x01_hardware.h
new file mode 100644
index 000000000000..8cecbee7a270
--- /dev/null
+++ b/drivers/gpu/host1x/hw/host1x01_hardware.h
@@ -0,0 +1,27 @@
+/*
+ * Tegra host1x Register Offsets for Tegra20 and Tegra30
+ *
+ * Copyright (c) 2010-2013 NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __HOST1X_HOST1X01_HARDWARE_H
+#define __HOST1X_HOST1X01_HARDWARE_H
+
+#include <linux/types.h>
+#include <linux/bitops.h>
+
+#include "hw_host1x01_sync.h"
+
+#endif
diff --git a/drivers/gpu/host1x/hw/hw_host1x01_sync.h b/drivers/gpu/host1x/hw/hw_host1x01_sync.h
new file mode 100644
index 000000000000..3af258b46e62
--- /dev/null
+++ b/drivers/gpu/host1x/hw/hw_host1x01_sync.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+ /*
+  * Function naming determines intended use:
+  *
+  *     <x>_r(void) : Returns the offset for register <x>.
+  *
+  *     <x>_w(void) : Returns the word offset for word (4 byte) element <x>.
+  *
+  *     <x>_<y>_s(void) : Returns size of field <y> of register <x> in bits.
+  *
+  *     <x>_<y>_f(u32 v) : Returns a value based on 'v' which has been shifted
+  *         and masked to place it at field <y> of register <x>.  This value
+  *         can be |'d with others to produce a full register value for
+  *         register <x>.
+  *
+  *     <x>_<y>_m(void) : Returns a mask for field <y> of register <x>.  This
+  *         value can be ~'d and then &'d to clear the value of field <y> for
+  *         register <x>.
+  *
+  *     <x>_<y>_<z>_f(void) : Returns the constant value <z> after being shifted
+  *         to place it at field <y> of register <x>.  This value can be |'d
+  *         with others to produce a full register value for <x>.
+  *
+  *     <x>_<y>_v(u32 r) : Returns the value of field <y> from a full register
+  *         <x> value 'r' after being shifted to place its LSB at bit 0.
+  *         This value is suitable for direct comparison with other unshifted
+  *         values appropriate for use in field <y> of register <x>.
+  *
+  *     <x>_<y>_<z>_v(void) : Returns the constant value for <z> defined for
+  *         field <y> of register <x>.  This value is suitable for direct
+  *         comparison with unshifted values appropriate for use in field <y>
+  *         of register <x>.
+  */
+
+#ifndef __hw_host1x01_sync_h__
+#define __hw_host1x01_sync_h__
+
+#define REGISTER_STRIDE	4
+
+static inline u32 host1x_sync_syncpt_r(unsigned int id)
+{
+	return 0x400 + id * REGISTER_STRIDE;
+}
+#define HOST1X_SYNC_SYNCPT(id) \
+	host1x_sync_syncpt_r(id)
+static inline u32 host1x_sync_syncpt_base_r(unsigned int id)
+{
+	return 0x600 + id * REGISTER_STRIDE;
+}
+#define HOST1X_SYNC_SYNCPT_BASE(id) \
+	host1x_sync_syncpt_base_r(id)
+static inline u32 host1x_sync_syncpt_cpu_incr_r(unsigned int id)
+{
+	return 0x700 + id * REGISTER_STRIDE;
+}
+#define HOST1X_SYNC_SYNCPT_CPU_INCR(id) \
+	host1x_sync_syncpt_cpu_incr_r(id)
+#endif /* __hw_host1x01_sync_h__ */
diff --git a/drivers/gpu/host1x/hw/syncpt_hw.c b/drivers/gpu/host1x/hw/syncpt_hw.c
new file mode 100644
index 000000000000..885b2578dc86
--- /dev/null
+++ b/drivers/gpu/host1x/hw/syncpt_hw.c
@@ -0,0 +1,102 @@
+/*
+ * Tegra host1x Syncpoints
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/io.h>
+
+#include "dev.h"
+#include "syncpt.h"
+
+/*
+ * Write the current syncpoint value back to hw.
+ */
+static void syncpt_restore(struct host1x_syncpt *sp)
+{
+	struct host1x *host = sp->host;
+	int min = host1x_syncpt_read_min(sp);
+	host1x_sync_writel(host, min, HOST1X_SYNC_SYNCPT(sp->id));
+}
+
+/*
+ * Write the current waitbase value back to hw.
+ */
+static void syncpt_restore_wait_base(struct host1x_syncpt *sp)
+{
+	struct host1x *host = sp->host;
+	host1x_sync_writel(host, sp->base_val,
+			   HOST1X_SYNC_SYNCPT_BASE(sp->id));
+}
+
+/*
+ * Read waitbase value from hw.
+ */
+static void syncpt_read_wait_base(struct host1x_syncpt *sp)
+{
+	struct host1x *host = sp->host;
+	sp->base_val =
+		host1x_sync_readl(host, HOST1X_SYNC_SYNCPT_BASE(sp->id));
+}
+
+/*
+ * Updates the last value read from hardware.
+ */
+static u32 syncpt_load(struct host1x_syncpt *sp)
+{
+	struct host1x *host = sp->host;
+	u32 old, live;
+
+	/* Loop in case there's a race writing to min_val */
+	do {
+		old = host1x_syncpt_read_min(sp);
+		live = host1x_sync_readl(host, HOST1X_SYNC_SYNCPT(sp->id));
+	} while ((u32)atomic_cmpxchg(&sp->min_val, old, live) != old);
+
+	if (!host1x_syncpt_check_max(sp, live))
+		dev_err(host->dev, "%s failed: id=%u, min=%d, max=%d\n",
+			__func__, sp->id, host1x_syncpt_read_min(sp),
+			host1x_syncpt_read_max(sp));
+
+	return live;
+}
+
+/*
+ * Write a cpu syncpoint increment to the hardware, without touching
+ * the cache.
+ */
+static void syncpt_cpu_incr(struct host1x_syncpt *sp)
+{
+	struct host1x *host = sp->host;
+	u32 reg_offset = sp->id / 32;
+
+	if (!host1x_syncpt_client_managed(sp) &&
+	    host1x_syncpt_idle(sp)) {
+		dev_err(host->dev, "Trying to increment syncpoint id %d beyond max\n",
+			sp->id);
+		return;
+	}
+	host1x_sync_writel(host, BIT_MASK(sp->id),
+			   HOST1X_SYNC_SYNCPT_CPU_INCR(reg_offset));
+	wmb();
+}
+
+static const struct host1x_syncpt_ops host1x_syncpt_ops = {
+	.restore = syncpt_restore,
+	.restore_wait_base = syncpt_restore_wait_base,
+	.load_wait_base = syncpt_read_wait_base,
+	.load = syncpt_load,
+	.cpu_incr = syncpt_cpu_incr,
+};
diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c
new file mode 100644
index 000000000000..30385f62daee
--- /dev/null
+++ b/drivers/gpu/host1x/syncpt.c
@@ -0,0 +1,212 @@
+/*
+ * Tegra host1x Syncpoints
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/module.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+
+#include <trace/events/host1x.h>
+
+#include "syncpt.h"
+#include "dev.h"
+
+static struct host1x_syncpt *_host1x_syncpt_alloc(struct host1x *host,
+						  struct device *dev,
+						  int client_managed)
+{
+	int i;
+	struct host1x_syncpt *sp = host->syncpt;
+	char *name;
+
+	for (i = 0; i < host->info->nb_pts && sp->name; i++, sp++)
+		;
+	if (sp->dev)
+		return NULL;
+
+	name = kasprintf(GFP_KERNEL, "%02d-%s", sp->id,
+			dev ? dev_name(dev) : NULL);
+	if (!name)
+		return NULL;
+
+	sp->dev = dev;
+	sp->name = name;
+	sp->client_managed = client_managed;
+
+	return sp;
+}
+
+u32 host1x_syncpt_id(struct host1x_syncpt *sp)
+{
+	return sp->id;
+}
+
+/*
+ * Updates the value sent to hardware.
+ */
+u32 host1x_syncpt_incr_max(struct host1x_syncpt *sp, u32 incrs)
+{
+	return (u32)atomic_add_return(incrs, &sp->max_val);
+}
+
+ /*
+ * Write cached syncpoint and waitbase values to hardware.
+ */
+void host1x_syncpt_restore(struct host1x *host)
+{
+	struct host1x_syncpt *sp_base = host->syncpt;
+	u32 i;
+
+	for (i = 0; i < host1x_syncpt_nb_pts(host); i++)
+		host1x_hw_syncpt_restore(host, sp_base + i);
+	for (i = 0; i < host1x_syncpt_nb_bases(host); i++)
+		host1x_hw_syncpt_restore_wait_base(host, sp_base + i);
+	wmb();
+}
+
+/*
+ * Update the cached syncpoint and waitbase values by reading them
+ * from the registers.
+  */
+void host1x_syncpt_save(struct host1x *host)
+{
+	struct host1x_syncpt *sp_base = host->syncpt;
+	u32 i;
+
+	for (i = 0; i < host1x_syncpt_nb_pts(host); i++) {
+		if (host1x_syncpt_client_managed(sp_base + i))
+			host1x_hw_syncpt_load(host, sp_base + i);
+		else
+			WARN_ON(!host1x_syncpt_idle(sp_base + i));
+	}
+
+	for (i = 0; i < host1x_syncpt_nb_bases(host); i++)
+		host1x_hw_syncpt_load_wait_base(host, sp_base + i);
+}
+
+/*
+ * Updates the cached syncpoint value by reading a new value from the hardware
+ * register
+ */
+u32 host1x_syncpt_load(struct host1x_syncpt *sp)
+{
+	u32 val;
+	val = host1x_hw_syncpt_load(sp->host, sp);
+	trace_host1x_syncpt_load_min(sp->id, val);
+
+	return val;
+}
+
+/*
+ * Get the current syncpoint base
+ */
+u32 host1x_syncpt_load_wait_base(struct host1x_syncpt *sp)
+{
+	u32 val;
+	host1x_hw_syncpt_load_wait_base(sp->host, sp);
+	val = sp->base_val;
+	return val;
+}
+
+/*
+ * Write a cpu syncpoint increment to the hardware, without touching
+ * the cache. Caller is responsible for host being powered.
+ */
+void host1x_syncpt_cpu_incr(struct host1x_syncpt *sp)
+{
+	host1x_hw_syncpt_cpu_incr(sp->host, sp);
+}
+
+/*
+ * Increment syncpoint value from cpu, updating cache
+ */
+void host1x_syncpt_incr(struct host1x_syncpt *sp)
+{
+	if (host1x_syncpt_client_managed(sp))
+		host1x_syncpt_incr_max(sp, 1);
+	host1x_syncpt_cpu_incr(sp);
+}
+
+int host1x_syncpt_init(struct host1x *host)
+{
+	struct host1x_syncpt *syncpt;
+	int i;
+
+	syncpt = devm_kzalloc(host->dev, sizeof(*syncpt) * host->info->nb_pts,
+		GFP_KERNEL);
+	if (!syncpt)
+		return -ENOMEM;
+
+	for (i = 0; i < host->info->nb_pts; ++i) {
+		syncpt[i].id = i;
+		syncpt[i].host = host;
+	}
+
+	host->syncpt = syncpt;
+
+	host1x_syncpt_restore(host);
+
+	return 0;
+}
+
+struct host1x_syncpt *host1x_syncpt_request(struct device *dev,
+					    int client_managed)
+{
+	struct host1x *host = dev_get_drvdata(dev->parent);
+	return _host1x_syncpt_alloc(host, dev, client_managed);
+}
+
+void host1x_syncpt_free(struct host1x_syncpt *sp)
+{
+	if (!sp)
+		return;
+
+	kfree(sp->name);
+	sp->dev = NULL;
+	sp->name = NULL;
+	sp->client_managed = 0;
+}
+
+void host1x_syncpt_deinit(struct host1x *host)
+{
+	int i;
+	struct host1x_syncpt *sp = host->syncpt;
+	for (i = 0; i < host->info->nb_pts; i++, sp++)
+		kfree(sp->name);
+}
+
+int host1x_syncpt_nb_pts(struct host1x *host)
+{
+	return host->info->nb_pts;
+}
+
+int host1x_syncpt_nb_bases(struct host1x *host)
+{
+	return host->info->nb_bases;
+}
+
+int host1x_syncpt_nb_mlocks(struct host1x *host)
+{
+	return host->info->nb_mlocks;
+}
+
+struct host1x_syncpt *host1x_syncpt_get(struct host1x *host, u32 id)
+{
+	if (host->info->nb_pts < id)
+		return NULL;
+	return host->syncpt + id;
+}
diff --git a/drivers/gpu/host1x/syncpt.h b/drivers/gpu/host1x/syncpt.h
new file mode 100644
index 000000000000..efa2b6e659d6
--- /dev/null
+++ b/drivers/gpu/host1x/syncpt.h
@@ -0,0 +1,147 @@
+/*
+ * Tegra host1x Syncpoints
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __HOST1X_SYNCPT_H
+#define __HOST1X_SYNCPT_H
+
+#include <linux/atomic.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+
+struct host1x;
+
+struct host1x_syncpt {
+	int id;
+	atomic_t min_val;
+	atomic_t max_val;
+	u32 base_val;
+	const char *name;
+	int client_managed;
+	struct host1x *host;
+	struct device *dev;
+};
+
+/* Initialize sync point array  */
+int host1x_syncpt_init(struct host1x *host);
+
+/*  Free sync point array */
+void host1x_syncpt_deinit(struct host1x *host);
+
+/*
+ * Read max. It indicates how many operations there are in queue, either in
+ * channel or in a software thread.
+ * */
+static inline u32 host1x_syncpt_read_max(struct host1x_syncpt *sp)
+{
+	smp_rmb();
+	return (u32)atomic_read(&sp->max_val);
+}
+
+/*
+ * Read min, which is a shadow of the current sync point value in hardware.
+ */
+static inline u32 host1x_syncpt_read_min(struct host1x_syncpt *sp)
+{
+	smp_rmb();
+	return (u32)atomic_read(&sp->min_val);
+}
+
+/* Return number of sync point supported. */
+int host1x_syncpt_nb_pts(struct host1x *host);
+
+/* Return number of wait bases supported. */
+int host1x_syncpt_nb_bases(struct host1x *host);
+
+/* Return number of mlocks supported. */
+int host1x_syncpt_nb_mlocks(struct host1x *host);
+
+/*
+ * Check sync point sanity. If max is larger than min, there have too many
+ * sync point increments.
+ *
+ * Client managed sync point are not tracked.
+ * */
+static inline bool host1x_syncpt_check_max(struct host1x_syncpt *sp, u32 real)
+{
+	u32 max;
+	if (sp->client_managed)
+		return true;
+	max = host1x_syncpt_read_max(sp);
+	return (s32)(max - real) >= 0;
+}
+
+/* Return true if sync point is client managed. */
+static inline int host1x_syncpt_client_managed(struct host1x_syncpt *sp)
+{
+	return sp->client_managed;
+}
+
+/*
+ * Returns true if syncpoint min == max, which means that there are no
+ * outstanding operations.
+ */
+static inline bool host1x_syncpt_idle(struct host1x_syncpt *sp)
+{
+	int min, max;
+	smp_rmb();
+	min = atomic_read(&sp->min_val);
+	max = atomic_read(&sp->max_val);
+	return (min == max);
+}
+
+/* Return pointer to struct denoting sync point id. */
+struct host1x_syncpt *host1x_syncpt_get(struct host1x *host, u32 id);
+
+/* Request incrementing a sync point. */
+void host1x_syncpt_cpu_incr(struct host1x_syncpt *sp);
+
+/* Load current value from hardware to the shadow register. */
+u32 host1x_syncpt_load(struct host1x_syncpt *sp);
+
+/* Save host1x sync point state into shadow registers. */
+void host1x_syncpt_save(struct host1x *host);
+
+/* Reset host1x sync point state from shadow registers. */
+void host1x_syncpt_restore(struct host1x *host);
+
+/* Read current wait base value into shadow register and return it. */
+u32 host1x_syncpt_load_wait_base(struct host1x_syncpt *sp);
+
+/* Increment sync point and its max. */
+void host1x_syncpt_incr(struct host1x_syncpt *sp);
+
+/* Indicate future operations by incrementing the sync point max. */
+u32 host1x_syncpt_incr_max(struct host1x_syncpt *sp, u32 incrs);
+
+/* Check if sync point id is valid. */
+static inline int host1x_syncpt_is_valid(struct host1x_syncpt *sp)
+{
+	return sp->id < host1x_syncpt_nb_pts(sp->host);
+}
+
+/* Return id of the sync point */
+u32 host1x_syncpt_id(struct host1x_syncpt *sp);
+
+/* Allocate a sync point for a device. */
+struct host1x_syncpt *host1x_syncpt_request(struct device *dev,
+		int client_managed);
+
+/* Free a sync point. */
+void host1x_syncpt_free(struct host1x_syncpt *sp);
+
+#endif
diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
index 4c1546f71d56..3e0e3f088686 100644
--- a/drivers/video/Kconfig
+++ b/drivers/video/Kconfig
@@ -21,6 +21,8 @@ source "drivers/gpu/vga/Kconfig"
 
 source "drivers/gpu/drm/Kconfig"
 
+source "drivers/gpu/host1x/Kconfig"
+
 config VGASTATE
        tristate
        default n
diff --git a/include/trace/events/host1x.h b/include/trace/events/host1x.h
new file mode 100644
index 000000000000..4bd2b0253d27
--- /dev/null
+++ b/include/trace/events/host1x.h
@@ -0,0 +1,61 @@
+/*
+ * include/trace/events/host1x.h
+ *
+ * host1x event logging to ftrace.
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM host1x
+
+#if !defined(_TRACE_HOST1X_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_HOST1X_H
+
+#include <linux/ktime.h>
+#include <linux/tracepoint.h>
+
+DECLARE_EVENT_CLASS(host1x,
+	TP_PROTO(const char *name),
+	TP_ARGS(name),
+	TP_STRUCT__entry(__field(const char *, name)),
+	TP_fast_assign(__entry->name = name;),
+	TP_printk("name=%s", __entry->name)
+);
+
+TRACE_EVENT(host1x_syncpt_load_min,
+	TP_PROTO(u32 id, u32 val),
+
+	TP_ARGS(id, val),
+
+	TP_STRUCT__entry(
+		__field(u32, id)
+		__field(u32, val)
+	),
+
+	TP_fast_assign(
+		__entry->id = id;
+		__entry->val = val;
+	),
+
+	TP_printk("id=%d, val=%d", __entry->id, __entry->val)
+);
+
+#endif /*  _TRACE_HOST1X_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>
-- 
cgit v1.2.3


From 6579324a41cc414009a601738b70a53d6376325c Mon Sep 17 00:00:00 2001
From: Terje Bergstrom <tbergstrom@nvidia.com>
Date: Fri, 22 Mar 2013 16:34:03 +0200
Subject: gpu: host1x: Add channel support

Add support for host1x client modules, and host1x channels to submit
work to the clients.

Signed-off-by: Arto Merilainen <amerilainen@nvidia.com>
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Erik Faye-Lund <kusmabite@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
---
 drivers/gpu/host1x/Kconfig                  |  12 +
 drivers/gpu/host1x/Makefile                 |   3 +
 drivers/gpu/host1x/cdma.c                   | 487 ++++++++++++++++++++++
 drivers/gpu/host1x/cdma.h                   | 100 +++++
 drivers/gpu/host1x/channel.c                | 126 ++++++
 drivers/gpu/host1x/channel.h                |  52 +++
 drivers/gpu/host1x/dev.c                    |  17 +
 drivers/gpu/host1x/dev.h                    | 113 ++++++
 drivers/gpu/host1x/host1x.h                 |  28 ++
 drivers/gpu/host1x/host1x_bo.h              |  87 ++++
 drivers/gpu/host1x/hw/cdma_hw.c             | 324 +++++++++++++++
 drivers/gpu/host1x/hw/channel_hw.c          | 143 +++++++
 drivers/gpu/host1x/hw/host1x01.c            |   5 +
 drivers/gpu/host1x/hw/host1x01_hardware.h   | 116 ++++++
 drivers/gpu/host1x/hw/hw_host1x01_channel.h | 102 +++++
 drivers/gpu/host1x/hw/hw_host1x01_sync.h    |  12 +
 drivers/gpu/host1x/hw/hw_host1x01_uclass.h  | 168 ++++++++
 drivers/gpu/host1x/hw/syncpt_hw.c           |  11 +
 drivers/gpu/host1x/intr.c                   |  28 +-
 drivers/gpu/host1x/intr.h                   |   6 +
 drivers/gpu/host1x/job.c                    | 603 ++++++++++++++++++++++++++++
 drivers/gpu/host1x/job.h                    | 162 ++++++++
 drivers/gpu/host1x/syncpt.c                 |  11 +
 drivers/gpu/host1x/syncpt.h                 |   6 +
 include/trace/events/host1x.h               | 192 +++++++++
 25 files changed, 2913 insertions(+), 1 deletion(-)
 create mode 100644 drivers/gpu/host1x/cdma.c
 create mode 100644 drivers/gpu/host1x/cdma.h
 create mode 100644 drivers/gpu/host1x/channel.c
 create mode 100644 drivers/gpu/host1x/channel.h
 create mode 100644 drivers/gpu/host1x/host1x.h
 create mode 100644 drivers/gpu/host1x/host1x_bo.h
 create mode 100644 drivers/gpu/host1x/hw/cdma_hw.c
 create mode 100644 drivers/gpu/host1x/hw/channel_hw.c
 create mode 100644 drivers/gpu/host1x/hw/hw_host1x01_channel.h
 create mode 100644 drivers/gpu/host1x/hw/hw_host1x01_uclass.h
 create mode 100644 drivers/gpu/host1x/job.c
 create mode 100644 drivers/gpu/host1x/job.h

(limited to 'include')

diff --git a/drivers/gpu/host1x/Kconfig b/drivers/gpu/host1x/Kconfig
index 567ae8b7c7c9..7d6bed222542 100644
--- a/drivers/gpu/host1x/Kconfig
+++ b/drivers/gpu/host1x/Kconfig
@@ -8,3 +8,15 @@ config TEGRA_HOST1X
 	  Tegra's graphics- and multimedia-related modules. The modules served
 	  by host1x are referred to as clients. host1x includes some other
 	  functionality, such as synchronization.
+
+if TEGRA_HOST1X
+
+config TEGRA_HOST1X_FIREWALL
+	bool "Enable HOST1X security firewall"
+	default y
+	help
+	  Say yes if kernel should protect command streams from tampering.
+
+	  If unsure, choose Y.
+
+endif
diff --git a/drivers/gpu/host1x/Makefile b/drivers/gpu/host1x/Makefile
index 5ef47ffce674..06a995b60902 100644
--- a/drivers/gpu/host1x/Makefile
+++ b/drivers/gpu/host1x/Makefile
@@ -4,6 +4,9 @@ host1x-y = \
 	syncpt.o \
 	dev.o \
 	intr.o \
+	cdma.o \
+	channel.o \
+	job.o \
 	hw/host1x01.o
 
 obj-$(CONFIG_TEGRA_HOST1X) += host1x.o
diff --git a/drivers/gpu/host1x/cdma.c b/drivers/gpu/host1x/cdma.c
new file mode 100644
index 000000000000..33935de91bb1
--- /dev/null
+++ b/drivers/gpu/host1x/cdma.c
@@ -0,0 +1,487 @@
+/*
+ * Tegra host1x Command DMA
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+
+#include <asm/cacheflush.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/kfifo.h>
+#include <linux/slab.h>
+#include <trace/events/host1x.h>
+
+#include "cdma.h"
+#include "channel.h"
+#include "dev.h"
+#include "debug.h"
+#include "host1x_bo.h"
+#include "job.h"
+
+/*
+ * push_buffer
+ *
+ * The push buffer is a circular array of words to be fetched by command DMA.
+ * Note that it works slightly differently to the sync queue; fence == pos
+ * means that the push buffer is full, not empty.
+ */
+
+#define HOST1X_PUSHBUFFER_SLOTS	512
+
+/*
+ * Clean up push buffer resources
+ */
+static void host1x_pushbuffer_destroy(struct push_buffer *pb)
+{
+	struct host1x_cdma *cdma = pb_to_cdma(pb);
+	struct host1x *host1x = cdma_to_host1x(cdma);
+
+	if (pb->phys != 0)
+		dma_free_writecombine(host1x->dev, pb->size_bytes + 4,
+				      pb->mapped, pb->phys);
+
+	pb->mapped = NULL;
+	pb->phys = 0;
+}
+
+/*
+ * Init push buffer resources
+ */
+static int host1x_pushbuffer_init(struct push_buffer *pb)
+{
+	struct host1x_cdma *cdma = pb_to_cdma(pb);
+	struct host1x *host1x = cdma_to_host1x(cdma);
+
+	pb->mapped = NULL;
+	pb->phys = 0;
+	pb->size_bytes = HOST1X_PUSHBUFFER_SLOTS * 8;
+
+	/* initialize buffer pointers */
+	pb->fence = pb->size_bytes - 8;
+	pb->pos = 0;
+
+	/* allocate and map pushbuffer memory */
+	pb->mapped = dma_alloc_writecombine(host1x->dev, pb->size_bytes + 4,
+					    &pb->phys, GFP_KERNEL);
+	if (!pb->mapped)
+		goto fail;
+
+	host1x_hw_pushbuffer_init(host1x, pb);
+
+	return 0;
+
+fail:
+	host1x_pushbuffer_destroy(pb);
+	return -ENOMEM;
+}
+
+/*
+ * Push two words to the push buffer
+ * Caller must ensure push buffer is not full
+ */
+static void host1x_pushbuffer_push(struct push_buffer *pb, u32 op1, u32 op2)
+{
+	u32 pos = pb->pos;
+	u32 *p = (u32 *)((u32)pb->mapped + pos);
+	WARN_ON(pos == pb->fence);
+	*(p++) = op1;
+	*(p++) = op2;
+	pb->pos = (pos + 8) & (pb->size_bytes - 1);
+}
+
+/*
+ * Pop a number of two word slots from the push buffer
+ * Caller must ensure push buffer is not empty
+ */
+static void host1x_pushbuffer_pop(struct push_buffer *pb, unsigned int slots)
+{
+	/* Advance the next write position */
+	pb->fence = (pb->fence + slots * 8) & (pb->size_bytes - 1);
+}
+
+/*
+ * Return the number of two word slots free in the push buffer
+ */
+static u32 host1x_pushbuffer_space(struct push_buffer *pb)
+{
+	return ((pb->fence - pb->pos) & (pb->size_bytes - 1)) / 8;
+}
+
+/*
+ * Sleep (if necessary) until the requested event happens
+ *   - CDMA_EVENT_SYNC_QUEUE_EMPTY : sync queue is completely empty.
+ *     - Returns 1
+ *   - CDMA_EVENT_PUSH_BUFFER_SPACE : there is space in the push buffer
+ *     - Return the amount of space (> 0)
+ * Must be called with the cdma lock held.
+ */
+unsigned int host1x_cdma_wait_locked(struct host1x_cdma *cdma,
+				     enum cdma_event event)
+{
+	for (;;) {
+		unsigned int space;
+
+		if (event == CDMA_EVENT_SYNC_QUEUE_EMPTY)
+			space = list_empty(&cdma->sync_queue) ? 1 : 0;
+		else if (event == CDMA_EVENT_PUSH_BUFFER_SPACE) {
+			struct push_buffer *pb = &cdma->push_buffer;
+			space = host1x_pushbuffer_space(pb);
+		} else {
+			WARN_ON(1);
+			return -EINVAL;
+		}
+
+		if (space)
+			return space;
+
+		trace_host1x_wait_cdma(dev_name(cdma_to_channel(cdma)->dev),
+				       event);
+
+		/* If somebody has managed to already start waiting, yield */
+		if (cdma->event != CDMA_EVENT_NONE) {
+			mutex_unlock(&cdma->lock);
+			schedule();
+			mutex_lock(&cdma->lock);
+			continue;
+		}
+		cdma->event = event;
+
+		mutex_unlock(&cdma->lock);
+		down(&cdma->sem);
+		mutex_lock(&cdma->lock);
+	}
+	return 0;
+}
+
+/*
+ * Start timer that tracks the time spent by the job.
+ * Must be called with the cdma lock held.
+ */
+static void cdma_start_timer_locked(struct host1x_cdma *cdma,
+				    struct host1x_job *job)
+{
+	struct host1x *host = cdma_to_host1x(cdma);
+
+	if (cdma->timeout.client) {
+		/* timer already started */
+		return;
+	}
+
+	cdma->timeout.client = job->client;
+	cdma->timeout.syncpt = host1x_syncpt_get(host, job->syncpt_id);
+	cdma->timeout.syncpt_val = job->syncpt_end;
+	cdma->timeout.start_ktime = ktime_get();
+
+	schedule_delayed_work(&cdma->timeout.wq,
+			      msecs_to_jiffies(job->timeout));
+}
+
+/*
+ * Stop timer when a buffer submission completes.
+ * Must be called with the cdma lock held.
+ */
+static void stop_cdma_timer_locked(struct host1x_cdma *cdma)
+{
+	cancel_delayed_work(&cdma->timeout.wq);
+	cdma->timeout.client = 0;
+}
+
+/*
+ * For all sync queue entries that have already finished according to the
+ * current sync point registers:
+ *  - unpin & unref their mems
+ *  - pop their push buffer slots
+ *  - remove them from the sync queue
+ * This is normally called from the host code's worker thread, but can be
+ * called manually if necessary.
+ * Must be called with the cdma lock held.
+ */
+static void update_cdma_locked(struct host1x_cdma *cdma)
+{
+	bool signal = false;
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	struct host1x_job *job, *n;
+
+	/* If CDMA is stopped, queue is cleared and we can return */
+	if (!cdma->running)
+		return;
+
+	/*
+	 * Walk the sync queue, reading the sync point registers as necessary,
+	 * to consume as many sync queue entries as possible without blocking
+	 */
+	list_for_each_entry_safe(job, n, &cdma->sync_queue, list) {
+		struct host1x_syncpt *sp =
+			host1x_syncpt_get(host1x, job->syncpt_id);
+
+		/* Check whether this syncpt has completed, and bail if not */
+		if (!host1x_syncpt_is_expired(sp, job->syncpt_end)) {
+			/* Start timer on next pending syncpt */
+			if (job->timeout)
+				cdma_start_timer_locked(cdma, job);
+			break;
+		}
+
+		/* Cancel timeout, when a buffer completes */
+		if (cdma->timeout.client)
+			stop_cdma_timer_locked(cdma);
+
+		/* Unpin the memory */
+		host1x_job_unpin(job);
+
+		/* Pop push buffer slots */
+		if (job->num_slots) {
+			struct push_buffer *pb = &cdma->push_buffer;
+			host1x_pushbuffer_pop(pb, job->num_slots);
+			if (cdma->event == CDMA_EVENT_PUSH_BUFFER_SPACE)
+				signal = true;
+		}
+
+		list_del(&job->list);
+		host1x_job_put(job);
+	}
+
+	if (cdma->event == CDMA_EVENT_SYNC_QUEUE_EMPTY &&
+	    list_empty(&cdma->sync_queue))
+		signal = true;
+
+	if (signal) {
+		cdma->event = CDMA_EVENT_NONE;
+		up(&cdma->sem);
+	}
+}
+
+void host1x_cdma_update_sync_queue(struct host1x_cdma *cdma,
+				   struct device *dev)
+{
+	u32 restart_addr;
+	u32 syncpt_incrs;
+	struct host1x_job *job = NULL;
+	u32 syncpt_val;
+	struct host1x *host1x = cdma_to_host1x(cdma);
+
+	syncpt_val = host1x_syncpt_load(cdma->timeout.syncpt);
+
+	dev_dbg(dev, "%s: starting cleanup (thresh %d)\n",
+		__func__, syncpt_val);
+
+	/*
+	 * Move the sync_queue read pointer to the first entry that hasn't
+	 * completed based on the current HW syncpt value. It's likely there
+	 * won't be any (i.e. we're still at the head), but covers the case
+	 * where a syncpt incr happens just prior/during the teardown.
+	 */
+
+	dev_dbg(dev, "%s: skip completed buffers still in sync_queue\n",
+		__func__);
+
+	list_for_each_entry(job, &cdma->sync_queue, list) {
+		if (syncpt_val < job->syncpt_end)
+			break;
+
+		host1x_job_dump(dev, job);
+	}
+
+	/*
+	 * Walk the sync_queue, first incrementing with the CPU syncpts that
+	 * are partially executed (the first buffer) or fully skipped while
+	 * still in the current context (slots are also NOP-ed).
+	 *
+	 * At the point contexts are interleaved, syncpt increments must be
+	 * done inline with the pushbuffer from a GATHER buffer to maintain
+	 * the order (slots are modified to be a GATHER of syncpt incrs).
+	 *
+	 * Note: save in restart_addr the location where the timed out buffer
+	 * started in the PB, so we can start the refetch from there (with the
+	 * modified NOP-ed PB slots). This lets things appear to have completed
+	 * properly for this buffer and resources are freed.
+	 */
+
+	dev_dbg(dev, "%s: perform CPU incr on pending same ctx buffers\n",
+		__func__);
+
+	if (!list_empty(&cdma->sync_queue))
+		restart_addr = job->first_get;
+	else
+		restart_addr = cdma->last_pos;
+
+	/* do CPU increments as long as this context continues */
+	list_for_each_entry_from(job, &cdma->sync_queue, list) {
+		/* different context, gets us out of this loop */
+		if (job->client != cdma->timeout.client)
+			break;
+
+		/* won't need a timeout when replayed */
+		job->timeout = 0;
+
+		syncpt_incrs = job->syncpt_end - syncpt_val;
+		dev_dbg(dev, "%s: CPU incr (%d)\n", __func__, syncpt_incrs);
+
+		host1x_job_dump(dev, job);
+
+		/* safe to use CPU to incr syncpts */
+		host1x_hw_cdma_timeout_cpu_incr(host1x, cdma, job->first_get,
+						syncpt_incrs, job->syncpt_end,
+						job->num_slots);
+
+		syncpt_val += syncpt_incrs;
+	}
+
+	/* The following sumbits from the same client may be dependent on the
+	 * failed submit and therefore they may fail. Force a small timeout
+	 * to make the queue cleanup faster */
+
+	list_for_each_entry_from(job, &cdma->sync_queue, list)
+		if (job->client == cdma->timeout.client)
+			job->timeout = min_t(unsigned int, job->timeout, 500);
+
+	dev_dbg(dev, "%s: finished sync_queue modification\n", __func__);
+
+	/* roll back DMAGET and start up channel again */
+	host1x_hw_cdma_resume(host1x, cdma, restart_addr);
+}
+
+/*
+ * Create a cdma
+ */
+int host1x_cdma_init(struct host1x_cdma *cdma)
+{
+	int err;
+
+	mutex_init(&cdma->lock);
+	sema_init(&cdma->sem, 0);
+
+	INIT_LIST_HEAD(&cdma->sync_queue);
+
+	cdma->event = CDMA_EVENT_NONE;
+	cdma->running = false;
+	cdma->torndown = false;
+
+	err = host1x_pushbuffer_init(&cdma->push_buffer);
+	if (err)
+		return err;
+	return 0;
+}
+
+/*
+ * Destroy a cdma
+ */
+int host1x_cdma_deinit(struct host1x_cdma *cdma)
+{
+	struct push_buffer *pb = &cdma->push_buffer;
+	struct host1x *host1x = cdma_to_host1x(cdma);
+
+	if (cdma->running) {
+		pr_warn("%s: CDMA still running\n", __func__);
+		return -EBUSY;
+	}
+
+	host1x_pushbuffer_destroy(pb);
+	host1x_hw_cdma_timeout_destroy(host1x, cdma);
+
+	return 0;
+}
+
+/*
+ * Begin a cdma submit
+ */
+int host1x_cdma_begin(struct host1x_cdma *cdma, struct host1x_job *job)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+
+	mutex_lock(&cdma->lock);
+
+	if (job->timeout) {
+		/* init state on first submit with timeout value */
+		if (!cdma->timeout.initialized) {
+			int err;
+			err = host1x_hw_cdma_timeout_init(host1x, cdma,
+							  job->syncpt_id);
+			if (err) {
+				mutex_unlock(&cdma->lock);
+				return err;
+			}
+		}
+	}
+	if (!cdma->running)
+		host1x_hw_cdma_start(host1x, cdma);
+
+	cdma->slots_free = 0;
+	cdma->slots_used = 0;
+	cdma->first_get = cdma->push_buffer.pos;
+
+	trace_host1x_cdma_begin(dev_name(job->channel->dev));
+	return 0;
+}
+
+/*
+ * Push two words into a push buffer slot
+ * Blocks as necessary if the push buffer is full.
+ */
+void host1x_cdma_push(struct host1x_cdma *cdma, u32 op1, u32 op2)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	struct push_buffer *pb = &cdma->push_buffer;
+	u32 slots_free = cdma->slots_free;
+
+	if (slots_free == 0) {
+		host1x_hw_cdma_flush(host1x, cdma);
+		slots_free = host1x_cdma_wait_locked(cdma,
+						CDMA_EVENT_PUSH_BUFFER_SPACE);
+	}
+	cdma->slots_free = slots_free - 1;
+	cdma->slots_used++;
+	host1x_pushbuffer_push(pb, op1, op2);
+}
+
+/*
+ * End a cdma submit
+ * Kick off DMA, add job to the sync queue, and a number of slots to be freed
+ * from the pushbuffer. The handles for a submit must all be pinned at the same
+ * time, but they can be unpinned in smaller chunks.
+ */
+void host1x_cdma_end(struct host1x_cdma *cdma,
+		     struct host1x_job *job)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	bool idle = list_empty(&cdma->sync_queue);
+
+	host1x_hw_cdma_flush(host1x, cdma);
+
+	job->first_get = cdma->first_get;
+	job->num_slots = cdma->slots_used;
+	host1x_job_get(job);
+	list_add_tail(&job->list, &cdma->sync_queue);
+
+	/* start timer on idle -> active transitions */
+	if (job->timeout && idle)
+		cdma_start_timer_locked(cdma, job);
+
+	trace_host1x_cdma_end(dev_name(job->channel->dev));
+	mutex_unlock(&cdma->lock);
+}
+
+/*
+ * Update cdma state according to current sync point values
+ */
+void host1x_cdma_update(struct host1x_cdma *cdma)
+{
+	mutex_lock(&cdma->lock);
+	update_cdma_locked(cdma);
+	mutex_unlock(&cdma->lock);
+}
diff --git a/drivers/gpu/host1x/cdma.h b/drivers/gpu/host1x/cdma.h
new file mode 100644
index 000000000000..313c4b784348
--- /dev/null
+++ b/drivers/gpu/host1x/cdma.h
@@ -0,0 +1,100 @@
+/*
+ * Tegra host1x Command DMA
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __HOST1X_CDMA_H
+#define __HOST1X_CDMA_H
+
+#include <linux/sched.h>
+#include <linux/semaphore.h>
+#include <linux/list.h>
+
+struct host1x_syncpt;
+struct host1x_userctx_timeout;
+struct host1x_job;
+
+/*
+ * cdma
+ *
+ * This is in charge of a host command DMA channel.
+ * Sends ops to a push buffer, and takes responsibility for unpinning
+ * (& possibly freeing) of memory after those ops have completed.
+ * Producer:
+ *	begin
+ *		push - send ops to the push buffer
+ *	end - start command DMA and enqueue handles to be unpinned
+ * Consumer:
+ *	update - call to update sync queue and push buffer, unpin memory
+ */
+
+struct push_buffer {
+	u32 *mapped;			/* mapped pushbuffer memory */
+	dma_addr_t phys;		/* physical address of pushbuffer */
+	u32 fence;			/* index we've written */
+	u32 pos;			/* index to write to */
+	u32 size_bytes;
+};
+
+struct buffer_timeout {
+	struct delayed_work wq;		/* work queue */
+	bool initialized;		/* timer one-time setup flag */
+	struct host1x_syncpt *syncpt;	/* buffer completion syncpt */
+	u32 syncpt_val;			/* syncpt value when completed */
+	ktime_t start_ktime;		/* starting time */
+	/* context timeout information */
+	int client;
+};
+
+enum cdma_event {
+	CDMA_EVENT_NONE,		/* not waiting for any event */
+	CDMA_EVENT_SYNC_QUEUE_EMPTY,	/* wait for empty sync queue */
+	CDMA_EVENT_PUSH_BUFFER_SPACE	/* wait for space in push buffer */
+};
+
+struct host1x_cdma {
+	struct mutex lock;		/* controls access to shared state */
+	struct semaphore sem;		/* signalled when event occurs */
+	enum cdma_event event;		/* event that sem is waiting for */
+	unsigned int slots_used;	/* pb slots used in current submit */
+	unsigned int slots_free;	/* pb slots free in current submit */
+	unsigned int first_get;		/* DMAGET value, where submit begins */
+	unsigned int last_pos;		/* last value written to DMAPUT */
+	struct push_buffer push_buffer;	/* channel's push buffer */
+	struct list_head sync_queue;	/* job queue */
+	struct buffer_timeout timeout;	/* channel's timeout state/wq */
+	bool running;
+	bool torndown;
+};
+
+#define cdma_to_channel(cdma) container_of(cdma, struct host1x_channel, cdma)
+#define cdma_to_host1x(cdma) dev_get_drvdata(cdma_to_channel(cdma)->dev->parent)
+#define pb_to_cdma(pb) container_of(pb, struct host1x_cdma, push_buffer)
+
+int host1x_cdma_init(struct host1x_cdma *cdma);
+int host1x_cdma_deinit(struct host1x_cdma *cdma);
+void host1x_cdma_stop(struct host1x_cdma *cdma);
+int host1x_cdma_begin(struct host1x_cdma *cdma, struct host1x_job *job);
+void host1x_cdma_push(struct host1x_cdma *cdma, u32 op1, u32 op2);
+void host1x_cdma_end(struct host1x_cdma *cdma, struct host1x_job *job);
+void host1x_cdma_update(struct host1x_cdma *cdma);
+void host1x_cdma_peek(struct host1x_cdma *cdma, u32 dmaget, int slot,
+		      u32 *out);
+unsigned int host1x_cdma_wait_locked(struct host1x_cdma *cdma,
+				     enum cdma_event event);
+void host1x_cdma_update_sync_queue(struct host1x_cdma *cdma,
+				   struct device *dev);
+#endif
diff --git a/drivers/gpu/host1x/channel.c b/drivers/gpu/host1x/channel.c
new file mode 100644
index 000000000000..83ea51b9f0fc
--- /dev/null
+++ b/drivers/gpu/host1x/channel.c
@@ -0,0 +1,126 @@
+/*
+ * Tegra host1x Channel
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/slab.h>
+#include <linux/module.h>
+
+#include "channel.h"
+#include "dev.h"
+#include "job.h"
+
+/* Constructor for the host1x device list */
+int host1x_channel_list_init(struct host1x *host)
+{
+	INIT_LIST_HEAD(&host->chlist.list);
+	mutex_init(&host->chlist_mutex);
+
+	if (host->info->nb_channels > BITS_PER_LONG) {
+		WARN(1, "host1x hardware has more channels than supported by the driver\n");
+		return -ENOSYS;
+	}
+
+	return 0;
+}
+
+int host1x_job_submit(struct host1x_job *job)
+{
+	struct host1x *host = dev_get_drvdata(job->channel->dev->parent);
+
+	return host1x_hw_channel_submit(host, job);
+}
+
+struct host1x_channel *host1x_channel_get(struct host1x_channel *channel)
+{
+	int err = 0;
+
+	mutex_lock(&channel->reflock);
+
+	if (channel->refcount == 0)
+		err = host1x_cdma_init(&channel->cdma);
+
+	if (!err)
+		channel->refcount++;
+
+	mutex_unlock(&channel->reflock);
+
+	return err ? NULL : channel;
+}
+
+void host1x_channel_put(struct host1x_channel *channel)
+{
+	mutex_lock(&channel->reflock);
+
+	if (channel->refcount == 1) {
+		struct host1x *host = dev_get_drvdata(channel->dev->parent);
+
+		host1x_hw_cdma_stop(host, &channel->cdma);
+		host1x_cdma_deinit(&channel->cdma);
+	}
+
+	channel->refcount--;
+
+	mutex_unlock(&channel->reflock);
+}
+
+struct host1x_channel *host1x_channel_request(struct device *dev)
+{
+	struct host1x *host = dev_get_drvdata(dev->parent);
+	int max_channels = host->info->nb_channels;
+	struct host1x_channel *channel = NULL;
+	int index, err;
+
+	mutex_lock(&host->chlist_mutex);
+
+	index = find_first_zero_bit(&host->allocated_channels, max_channels);
+	if (index >= max_channels)
+		goto fail;
+
+	channel = kzalloc(sizeof(*channel), GFP_KERNEL);
+	if (!channel)
+		goto fail;
+
+	err = host1x_hw_channel_init(host, channel, index);
+	if (err < 0)
+		goto fail;
+
+	/* Link device to host1x_channel */
+	channel->dev = dev;
+
+	/* Add to channel list */
+	list_add_tail(&channel->list, &host->chlist.list);
+
+	host->allocated_channels |= BIT(index);
+
+	mutex_unlock(&host->chlist_mutex);
+	return channel;
+
+fail:
+	dev_err(dev, "failed to init channel\n");
+	kfree(channel);
+	mutex_unlock(&host->chlist_mutex);
+	return NULL;
+}
+
+void host1x_channel_free(struct host1x_channel *channel)
+{
+	struct host1x *host = dev_get_drvdata(channel->dev->parent);
+
+	host->allocated_channels &= ~BIT(channel->id);
+	list_del(&channel->list);
+	kfree(channel);
+}
diff --git a/drivers/gpu/host1x/channel.h b/drivers/gpu/host1x/channel.h
new file mode 100644
index 000000000000..48723b8eea42
--- /dev/null
+++ b/drivers/gpu/host1x/channel.h
@@ -0,0 +1,52 @@
+/*
+ * Tegra host1x Channel
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __HOST1X_CHANNEL_H
+#define __HOST1X_CHANNEL_H
+
+#include <linux/io.h>
+
+#include "cdma.h"
+
+struct host1x;
+
+struct host1x_channel {
+	struct list_head list;
+
+	unsigned int refcount;
+	unsigned int id;
+	struct mutex reflock;
+	struct mutex submitlock;
+	void __iomem *regs;
+	struct device *dev;
+	struct host1x_cdma cdma;
+};
+
+/* channel list operations */
+int host1x_channel_list_init(struct host1x *host);
+
+struct host1x_channel *host1x_channel_request(struct device *dev);
+void host1x_channel_free(struct host1x_channel *channel);
+struct host1x_channel *host1x_channel_get(struct host1x_channel *channel);
+void host1x_channel_put(struct host1x_channel *channel);
+int host1x_job_submit(struct host1x_job *job);
+
+#define host1x_for_each_channel(host, channel)				\
+	list_for_each_entry(channel, &host->chlist.list, list)
+
+#endif
diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c
index b967f6e8df55..4e522c532bc8 100644
--- a/drivers/gpu/host1x/dev.c
+++ b/drivers/gpu/host1x/dev.c
@@ -29,6 +29,7 @@
 
 #include "dev.h"
 #include "intr.h"
+#include "channel.h"
 #include "hw/host1x01.h"
 
 void host1x_sync_writel(struct host1x *host1x, u32 v, u32 r)
@@ -45,6 +46,16 @@ u32 host1x_sync_readl(struct host1x *host1x, u32 r)
 	return readl(sync_regs + r);
 }
 
+void host1x_ch_writel(struct host1x_channel *ch, u32 v, u32 r)
+{
+	writel(v, ch->regs + r);
+}
+
+u32 host1x_ch_readl(struct host1x_channel *ch, u32 r)
+{
+	return readl(ch->regs + r);
+}
+
 static const struct host1x_info host1x01_info = {
 	.nb_channels	= 8,
 	.nb_pts		= 32,
@@ -112,6 +123,12 @@ static int host1x_probe(struct platform_device *pdev)
 		return err;
 	}
 
+	err = host1x_channel_list_init(host);
+	if (err) {
+		dev_err(&pdev->dev, "failed to initialize channel list\n");
+		return err;
+	}
+
 	err = clk_prepare_enable(host->clk);
 	if (err < 0) {
 		dev_err(&pdev->dev, "failed to enable clock\n");
diff --git a/drivers/gpu/host1x/dev.h b/drivers/gpu/host1x/dev.h
index caf9cc62eb17..1a9b4383dc3b 100644
--- a/drivers/gpu/host1x/dev.h
+++ b/drivers/gpu/host1x/dev.h
@@ -20,10 +20,39 @@
 #include <linux/platform_device.h>
 #include <linux/device.h>
 
+#include "channel.h"
 #include "syncpt.h"
 #include "intr.h"
+#include "cdma.h"
+#include "job.h"
 
 struct host1x_syncpt;
+struct host1x_channel;
+struct host1x_cdma;
+struct host1x_job;
+struct push_buffer;
+
+struct host1x_channel_ops {
+	int (*init)(struct host1x_channel *channel, struct host1x *host,
+		    unsigned int id);
+	int (*submit)(struct host1x_job *job);
+};
+
+struct host1x_cdma_ops {
+	void (*start)(struct host1x_cdma *cdma);
+	void (*stop)(struct host1x_cdma *cdma);
+	void (*flush)(struct  host1x_cdma *cdma);
+	int (*timeout_init)(struct host1x_cdma *cdma, u32 syncpt_id);
+	void (*timeout_destroy)(struct host1x_cdma *cdma);
+	void (*freeze)(struct host1x_cdma *cdma);
+	void (*resume)(struct host1x_cdma *cdma, u32 getptr);
+	void (*timeout_cpu_incr)(struct host1x_cdma *cdma, u32 getptr,
+				 u32 syncpt_incrs, u32 syncval, u32 nr_slots);
+};
+
+struct host1x_pushbuffer_ops {
+	void (*init)(struct push_buffer *pb);
+};
 
 struct host1x_syncpt_ops {
 	void (*restore)(struct host1x_syncpt *syncpt);
@@ -68,11 +97,22 @@ struct host1x {
 
 	const struct host1x_syncpt_ops *syncpt_op;
 	const struct host1x_intr_ops *intr_op;
+	const struct host1x_channel_ops *channel_op;
+	const struct host1x_cdma_ops *cdma_op;
+	const struct host1x_pushbuffer_ops *cdma_pb_op;
 
+	struct host1x_syncpt *nop_sp;
+
+	struct mutex chlist_mutex;
+	struct host1x_channel chlist;
+	unsigned long allocated_channels;
+	unsigned int num_allocated_channels;
 };
 
 void host1x_sync_writel(struct host1x *host1x, u32 r, u32 v);
 u32 host1x_sync_readl(struct host1x *host1x, u32 r);
+void host1x_ch_writel(struct host1x_channel *ch, u32 r, u32 v);
+u32 host1x_ch_readl(struct host1x_channel *ch, u32 r);
 
 static inline void host1x_hw_syncpt_restore(struct host1x *host,
 					    struct host1x_syncpt *sp)
@@ -144,4 +184,77 @@ static inline int host1x_hw_intr_free_syncpt_irq(struct host1x *host)
 {
 	return host->intr_op->free_syncpt_irq(host);
 }
+
+static inline int host1x_hw_channel_init(struct host1x *host,
+					 struct host1x_channel *channel,
+					 int chid)
+{
+	return host->channel_op->init(channel, host, chid);
+}
+
+static inline int host1x_hw_channel_submit(struct host1x *host,
+					   struct host1x_job *job)
+{
+	return host->channel_op->submit(job);
+}
+
+static inline void host1x_hw_cdma_start(struct host1x *host,
+					struct host1x_cdma *cdma)
+{
+	host->cdma_op->start(cdma);
+}
+
+static inline void host1x_hw_cdma_stop(struct host1x *host,
+				       struct host1x_cdma *cdma)
+{
+	host->cdma_op->stop(cdma);
+}
+
+static inline void host1x_hw_cdma_flush(struct host1x *host,
+					struct host1x_cdma *cdma)
+{
+	host->cdma_op->flush(cdma);
+}
+
+static inline int host1x_hw_cdma_timeout_init(struct host1x *host,
+					      struct host1x_cdma *cdma,
+					      u32 syncpt_id)
+{
+	return host->cdma_op->timeout_init(cdma, syncpt_id);
+}
+
+static inline void host1x_hw_cdma_timeout_destroy(struct host1x *host,
+						  struct host1x_cdma *cdma)
+{
+	host->cdma_op->timeout_destroy(cdma);
+}
+
+static inline void host1x_hw_cdma_freeze(struct host1x *host,
+					 struct host1x_cdma *cdma)
+{
+	host->cdma_op->freeze(cdma);
+}
+
+static inline void host1x_hw_cdma_resume(struct host1x *host,
+					 struct host1x_cdma *cdma, u32 getptr)
+{
+	host->cdma_op->resume(cdma, getptr);
+}
+
+static inline void host1x_hw_cdma_timeout_cpu_incr(struct host1x *host,
+						   struct host1x_cdma *cdma,
+						   u32 getptr,
+						   u32 syncpt_incrs,
+						   u32 syncval, u32 nr_slots)
+{
+	host->cdma_op->timeout_cpu_incr(cdma, getptr, syncpt_incrs, syncval,
+					nr_slots);
+}
+
+static inline void host1x_hw_pushbuffer_init(struct host1x *host,
+					     struct push_buffer *pb)
+{
+	host->cdma_pb_op->init(pb);
+}
+
 #endif
diff --git a/drivers/gpu/host1x/host1x.h b/drivers/gpu/host1x/host1x.h
new file mode 100644
index 000000000000..bca6563f6766
--- /dev/null
+++ b/drivers/gpu/host1x/host1x.h
@@ -0,0 +1,28 @@
+/*
+ * Tegra host1x driver
+ *
+ * Copyright (c) 2009-2013, NVIDIA Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ */
+
+#ifndef __LINUX_HOST1X_H
+#define __LINUX_HOST1X_H
+
+enum host1x_class {
+	HOST1X_CLASS_HOST1X	= 0x1
+};
+
+#endif
diff --git a/drivers/gpu/host1x/host1x_bo.h b/drivers/gpu/host1x/host1x_bo.h
new file mode 100644
index 000000000000..4c1f10bd773d
--- /dev/null
+++ b/drivers/gpu/host1x/host1x_bo.h
@@ -0,0 +1,87 @@
+/*
+ * Tegra host1x Memory Management Abstraction header
+ *
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _HOST1X_BO_H
+#define _HOST1X_BO_H
+
+struct host1x_bo;
+
+struct host1x_bo_ops {
+	struct host1x_bo *(*get)(struct host1x_bo *bo);
+	void (*put)(struct host1x_bo *bo);
+	dma_addr_t (*pin)(struct host1x_bo *bo, struct sg_table **sgt);
+	void (*unpin)(struct host1x_bo *bo, struct sg_table *sgt);
+	void *(*mmap)(struct host1x_bo *bo);
+	void (*munmap)(struct host1x_bo *bo, void *addr);
+	void *(*kmap)(struct host1x_bo *bo, unsigned int pagenum);
+	void (*kunmap)(struct host1x_bo *bo, unsigned int pagenum, void *addr);
+};
+
+struct host1x_bo {
+	const struct host1x_bo_ops *ops;
+};
+
+static inline void host1x_bo_init(struct host1x_bo *bo,
+				  const struct host1x_bo_ops *ops)
+{
+	bo->ops = ops;
+}
+
+static inline struct host1x_bo *host1x_bo_get(struct host1x_bo *bo)
+{
+	return bo->ops->get(bo);
+}
+
+static inline void host1x_bo_put(struct host1x_bo *bo)
+{
+	bo->ops->put(bo);
+}
+
+static inline dma_addr_t host1x_bo_pin(struct host1x_bo *bo,
+				       struct sg_table **sgt)
+{
+	return bo->ops->pin(bo, sgt);
+}
+
+static inline void host1x_bo_unpin(struct host1x_bo *bo, struct sg_table *sgt)
+{
+	bo->ops->unpin(bo, sgt);
+}
+
+static inline void *host1x_bo_mmap(struct host1x_bo *bo)
+{
+	return bo->ops->mmap(bo);
+}
+
+static inline void host1x_bo_munmap(struct host1x_bo *bo, void *addr)
+{
+	bo->ops->munmap(bo, addr);
+}
+
+static inline void *host1x_bo_kmap(struct host1x_bo *bo, unsigned int pagenum)
+{
+	return bo->ops->kmap(bo, pagenum);
+}
+
+static inline void host1x_bo_kunmap(struct host1x_bo *bo,
+				    unsigned int pagenum, void *addr)
+{
+	bo->ops->kunmap(bo, pagenum, addr);
+}
+
+#endif
diff --git a/drivers/gpu/host1x/hw/cdma_hw.c b/drivers/gpu/host1x/hw/cdma_hw.c
new file mode 100644
index 000000000000..4eb22ef29776
--- /dev/null
+++ b/drivers/gpu/host1x/hw/cdma_hw.c
@@ -0,0 +1,324 @@
+/*
+ * Tegra host1x Command DMA
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/slab.h>
+#include <linux/scatterlist.h>
+#include <linux/dma-mapping.h>
+
+#include "cdma.h"
+#include "channel.h"
+#include "dev.h"
+#include "debug.h"
+
+/*
+ * Put the restart at the end of pushbuffer memor
+ */
+static void push_buffer_init(struct push_buffer *pb)
+{
+	*(pb->mapped + (pb->size_bytes >> 2)) = host1x_opcode_restart(0);
+}
+
+/*
+ * Increment timedout buffer's syncpt via CPU.
+ */
+static void cdma_timeout_cpu_incr(struct host1x_cdma *cdma, u32 getptr,
+				u32 syncpt_incrs, u32 syncval, u32 nr_slots)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	struct push_buffer *pb = &cdma->push_buffer;
+	u32 i;
+
+	for (i = 0; i < syncpt_incrs; i++)
+		host1x_syncpt_cpu_incr(cdma->timeout.syncpt);
+
+	/* after CPU incr, ensure shadow is up to date */
+	host1x_syncpt_load(cdma->timeout.syncpt);
+
+	/* NOP all the PB slots */
+	while (nr_slots--) {
+		u32 *p = (u32 *)((u32)pb->mapped + getptr);
+		*(p++) = HOST1X_OPCODE_NOP;
+		*(p++) = HOST1X_OPCODE_NOP;
+		dev_dbg(host1x->dev, "%s: NOP at 0x%x\n", __func__,
+			pb->phys + getptr);
+		getptr = (getptr + 8) & (pb->size_bytes - 1);
+	}
+	wmb();
+}
+
+/*
+ * Start channel DMA
+ */
+static void cdma_start(struct host1x_cdma *cdma)
+{
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+
+	if (cdma->running)
+		return;
+
+	cdma->last_pos = cdma->push_buffer.pos;
+
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP,
+			 HOST1X_CHANNEL_DMACTRL);
+
+	/* set base, put and end pointer */
+	host1x_ch_writel(ch, cdma->push_buffer.phys, HOST1X_CHANNEL_DMASTART);
+	host1x_ch_writel(ch, cdma->push_buffer.pos, HOST1X_CHANNEL_DMAPUT);
+	host1x_ch_writel(ch, cdma->push_buffer.phys +
+			 cdma->push_buffer.size_bytes + 4,
+			 HOST1X_CHANNEL_DMAEND);
+
+	/* reset GET */
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP |
+			 HOST1X_CHANNEL_DMACTRL_DMAGETRST |
+			 HOST1X_CHANNEL_DMACTRL_DMAINITGET,
+			 HOST1X_CHANNEL_DMACTRL);
+
+	/* start the command DMA */
+	host1x_ch_writel(ch, 0, HOST1X_CHANNEL_DMACTRL);
+
+	cdma->running = true;
+}
+
+/*
+ * Similar to cdma_start(), but rather than starting from an idle
+ * state (where DMA GET is set to DMA PUT), on a timeout we restore
+ * DMA GET from an explicit value (so DMA may again be pending).
+ */
+static void cdma_timeout_restart(struct host1x_cdma *cdma, u32 getptr)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+
+	if (cdma->running)
+		return;
+
+	cdma->last_pos = cdma->push_buffer.pos;
+
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP,
+			 HOST1X_CHANNEL_DMACTRL);
+
+	/* set base, end pointer (all of memory) */
+	host1x_ch_writel(ch, cdma->push_buffer.phys, HOST1X_CHANNEL_DMASTART);
+	host1x_ch_writel(ch, cdma->push_buffer.phys +
+			 cdma->push_buffer.size_bytes,
+			 HOST1X_CHANNEL_DMAEND);
+
+	/* set GET, by loading the value in PUT (then reset GET) */
+	host1x_ch_writel(ch, getptr, HOST1X_CHANNEL_DMAPUT);
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP |
+			 HOST1X_CHANNEL_DMACTRL_DMAGETRST |
+			 HOST1X_CHANNEL_DMACTRL_DMAINITGET,
+			 HOST1X_CHANNEL_DMACTRL);
+
+	dev_dbg(host1x->dev,
+		"%s: DMA GET 0x%x, PUT HW 0x%x / shadow 0x%x\n", __func__,
+		host1x_ch_readl(ch, HOST1X_CHANNEL_DMAGET),
+		host1x_ch_readl(ch, HOST1X_CHANNEL_DMAPUT),
+		cdma->last_pos);
+
+	/* deassert GET reset and set PUT */
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP,
+			 HOST1X_CHANNEL_DMACTRL);
+	host1x_ch_writel(ch, cdma->push_buffer.pos, HOST1X_CHANNEL_DMAPUT);
+
+	/* start the command DMA */
+	host1x_ch_writel(ch, 0, HOST1X_CHANNEL_DMACTRL);
+
+	cdma->running = true;
+}
+
+/*
+ * Kick channel DMA into action by writing its PUT offset (if it has changed)
+ */
+static void cdma_flush(struct host1x_cdma *cdma)
+{
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+
+	if (cdma->push_buffer.pos != cdma->last_pos) {
+		host1x_ch_writel(ch, cdma->push_buffer.pos,
+				 HOST1X_CHANNEL_DMAPUT);
+		cdma->last_pos = cdma->push_buffer.pos;
+	}
+}
+
+static void cdma_stop(struct host1x_cdma *cdma)
+{
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+
+	mutex_lock(&cdma->lock);
+	if (cdma->running) {
+		host1x_cdma_wait_locked(cdma, CDMA_EVENT_SYNC_QUEUE_EMPTY);
+		host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP,
+				 HOST1X_CHANNEL_DMACTRL);
+		cdma->running = false;
+	}
+	mutex_unlock(&cdma->lock);
+}
+
+/*
+ * Stops both channel's command processor and CDMA immediately.
+ * Also, tears down the channel and resets corresponding module.
+ */
+static void cdma_freeze(struct host1x_cdma *cdma)
+{
+	struct host1x *host = cdma_to_host1x(cdma);
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+	u32 cmdproc_stop;
+
+	if (cdma->torndown && !cdma->running) {
+		dev_warn(host->dev, "Already torn down\n");
+		return;
+	}
+
+	dev_dbg(host->dev, "freezing channel (id %d)\n", ch->id);
+
+	cmdproc_stop = host1x_sync_readl(host, HOST1X_SYNC_CMDPROC_STOP);
+	cmdproc_stop |= BIT(ch->id);
+	host1x_sync_writel(host, cmdproc_stop, HOST1X_SYNC_CMDPROC_STOP);
+
+	dev_dbg(host->dev, "%s: DMA GET 0x%x, PUT HW 0x%x / shadow 0x%x\n",
+		__func__, host1x_ch_readl(ch, HOST1X_CHANNEL_DMAGET),
+		host1x_ch_readl(ch, HOST1X_CHANNEL_DMAPUT),
+		cdma->last_pos);
+
+	host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP,
+			 HOST1X_CHANNEL_DMACTRL);
+
+	host1x_sync_writel(host, BIT(ch->id), HOST1X_SYNC_CH_TEARDOWN);
+
+	cdma->running = false;
+	cdma->torndown = true;
+}
+
+static void cdma_resume(struct host1x_cdma *cdma, u32 getptr)
+{
+	struct host1x *host1x = cdma_to_host1x(cdma);
+	struct host1x_channel *ch = cdma_to_channel(cdma);
+	u32 cmdproc_stop;
+
+	dev_dbg(host1x->dev,
+		"resuming channel (id %d, DMAGET restart = 0x%x)\n",
+		ch->id, getptr);
+
+	cmdproc_stop = host1x_sync_readl(host1x, HOST1X_SYNC_CMDPROC_STOP);
+	cmdproc_stop &= ~(BIT(ch->id));
+	host1x_sync_writel(host1x, cmdproc_stop, HOST1X_SYNC_CMDPROC_STOP);
+
+	cdma->torndown = false;
+	cdma_timeout_restart(cdma, getptr);
+}
+
+/*
+ * If this timeout fires, it indicates the current sync_queue entry has
+ * exceeded its TTL and the userctx should be timed out and remaining
+ * submits already issued cleaned up (future submits return an error).
+ */
+static void cdma_timeout_handler(struct work_struct *work)
+{
+	struct host1x_cdma *cdma;
+	struct host1x *host1x;
+	struct host1x_channel *ch;
+
+	u32 syncpt_val;
+
+	u32 prev_cmdproc, cmdproc_stop;
+
+	cdma = container_of(to_delayed_work(work), struct host1x_cdma,
+			    timeout.wq);
+	host1x = cdma_to_host1x(cdma);
+	ch = cdma_to_channel(cdma);
+
+	mutex_lock(&cdma->lock);
+
+	if (!cdma->timeout.client) {
+		dev_dbg(host1x->dev,
+			"cdma_timeout: expired, but has no clientid\n");
+		mutex_unlock(&cdma->lock);
+		return;
+	}
+
+	/* stop processing to get a clean snapshot */
+	prev_cmdproc = host1x_sync_readl(host1x, HOST1X_SYNC_CMDPROC_STOP);
+	cmdproc_stop = prev_cmdproc | BIT(ch->id);
+	host1x_sync_writel(host1x, cmdproc_stop, HOST1X_SYNC_CMDPROC_STOP);
+
+	dev_dbg(host1x->dev, "cdma_timeout: cmdproc was 0x%x is 0x%x\n",
+		prev_cmdproc, cmdproc_stop);
+
+	syncpt_val = host1x_syncpt_load(cdma->timeout.syncpt);
+
+	/* has buffer actually completed? */
+	if ((s32)(syncpt_val - cdma->timeout.syncpt_val) >= 0) {
+		dev_dbg(host1x->dev,
+			"cdma_timeout: expired, but buffer had completed\n");
+		/* restore */
+		cmdproc_stop = prev_cmdproc & ~(BIT(ch->id));
+		host1x_sync_writel(host1x, cmdproc_stop,
+				   HOST1X_SYNC_CMDPROC_STOP);
+		mutex_unlock(&cdma->lock);
+		return;
+	}
+
+	dev_warn(host1x->dev, "%s: timeout: %d (%s), HW thresh %d, done %d\n",
+		__func__, cdma->timeout.syncpt->id, cdma->timeout.syncpt->name,
+		syncpt_val, cdma->timeout.syncpt_val);
+
+	/* stop HW, resetting channel/module */
+	host1x_hw_cdma_freeze(host1x, cdma);
+
+	host1x_cdma_update_sync_queue(cdma, ch->dev);
+	mutex_unlock(&cdma->lock);
+}
+
+/*
+ * Init timeout resources
+ */
+static int cdma_timeout_init(struct host1x_cdma *cdma, u32 syncpt_id)
+{
+	INIT_DELAYED_WORK(&cdma->timeout.wq, cdma_timeout_handler);
+	cdma->timeout.initialized = true;
+
+	return 0;
+}
+
+/*
+ * Clean up timeout resources
+ */
+static void cdma_timeout_destroy(struct host1x_cdma *cdma)
+{
+	if (cdma->timeout.initialized)
+		cancel_delayed_work(&cdma->timeout.wq);
+	cdma->timeout.initialized = false;
+}
+
+static const struct host1x_cdma_ops host1x_cdma_ops = {
+	.start = cdma_start,
+	.stop = cdma_stop,
+	.flush = cdma_flush,
+
+	.timeout_init = cdma_timeout_init,
+	.timeout_destroy = cdma_timeout_destroy,
+	.freeze = cdma_freeze,
+	.resume = cdma_resume,
+	.timeout_cpu_incr = cdma_timeout_cpu_incr,
+};
+
+static const struct host1x_pushbuffer_ops host1x_pushbuffer_ops = {
+	.init = push_buffer_init,
+};
diff --git a/drivers/gpu/host1x/hw/channel_hw.c b/drivers/gpu/host1x/hw/channel_hw.c
new file mode 100644
index 000000000000..5137a5604215
--- /dev/null
+++ b/drivers/gpu/host1x/hw/channel_hw.c
@@ -0,0 +1,143 @@
+/*
+ * Tegra host1x Channel
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/slab.h>
+#include <trace/events/host1x.h>
+
+#include "host1x.h"
+#include "host1x_bo.h"
+#include "channel.h"
+#include "dev.h"
+#include "intr.h"
+#include "job.h"
+
+#define HOST1X_CHANNEL_SIZE 16384
+#define TRACE_MAX_LENGTH 128U
+
+static void submit_gathers(struct host1x_job *job)
+{
+	struct host1x_cdma *cdma = &job->channel->cdma;
+	unsigned int i;
+
+	for (i = 0; i < job->num_gathers; i++) {
+		struct host1x_job_gather *g = &job->gathers[i];
+		u32 op1 = host1x_opcode_gather(g->words);
+		u32 op2 = g->base + g->offset;
+		host1x_cdma_push(cdma, op1, op2);
+	}
+}
+
+static int channel_submit(struct host1x_job *job)
+{
+	struct host1x_channel *ch = job->channel;
+	struct host1x_syncpt *sp;
+	u32 user_syncpt_incrs = job->syncpt_incrs;
+	u32 prev_max = 0;
+	u32 syncval;
+	int err;
+	struct host1x_waitlist *completed_waiter = NULL;
+	struct host1x *host = dev_get_drvdata(ch->dev->parent);
+
+	sp = host->syncpt + job->syncpt_id;
+	trace_host1x_channel_submit(dev_name(ch->dev),
+				    job->num_gathers, job->num_relocs,
+				    job->num_waitchk, job->syncpt_id,
+				    job->syncpt_incrs);
+
+	/* before error checks, return current max */
+	prev_max = job->syncpt_end = host1x_syncpt_read_max(sp);
+
+	/* get submit lock */
+	err = mutex_lock_interruptible(&ch->submitlock);
+	if (err)
+		goto error;
+
+	completed_waiter = kzalloc(sizeof(*completed_waiter), GFP_KERNEL);
+	if (!completed_waiter) {
+		mutex_unlock(&ch->submitlock);
+		err = -ENOMEM;
+		goto error;
+	}
+
+	/* begin a CDMA submit */
+	err = host1x_cdma_begin(&ch->cdma, job);
+	if (err) {
+		mutex_unlock(&ch->submitlock);
+		goto error;
+	}
+
+	if (job->serialize) {
+		/*
+		 * Force serialization by inserting a host wait for the
+		 * previous job to finish before this one can commence.
+		 */
+		host1x_cdma_push(&ch->cdma,
+				 host1x_opcode_setclass(HOST1X_CLASS_HOST1X,
+					host1x_uclass_wait_syncpt_r(), 1),
+				 host1x_class_host_wait_syncpt(job->syncpt_id,
+					host1x_syncpt_read_max(sp)));
+	}
+
+	syncval = host1x_syncpt_incr_max(sp, user_syncpt_incrs);
+
+	job->syncpt_end = syncval;
+
+	/* add a setclass for modules that require it */
+	if (job->class)
+		host1x_cdma_push(&ch->cdma,
+				 host1x_opcode_setclass(job->class, 0, 0),
+				 HOST1X_OPCODE_NOP);
+
+	submit_gathers(job);
+
+	/* end CDMA submit & stash pinned hMems into sync queue */
+	host1x_cdma_end(&ch->cdma, job);
+
+	trace_host1x_channel_submitted(dev_name(ch->dev), prev_max, syncval);
+
+	/* schedule a submit complete interrupt */
+	err = host1x_intr_add_action(host, job->syncpt_id, syncval,
+				     HOST1X_INTR_ACTION_SUBMIT_COMPLETE, ch,
+				     completed_waiter, NULL);
+	completed_waiter = NULL;
+	WARN(err, "Failed to set submit complete interrupt");
+
+	mutex_unlock(&ch->submitlock);
+
+	return 0;
+
+error:
+	kfree(completed_waiter);
+	return err;
+}
+
+static int host1x_channel_init(struct host1x_channel *ch, struct host1x *dev,
+			       unsigned int index)
+{
+	ch->id = index;
+	mutex_init(&ch->reflock);
+	mutex_init(&ch->submitlock);
+
+	ch->regs = dev->regs + index * HOST1X_CHANNEL_SIZE;
+	return 0;
+}
+
+static const struct host1x_channel_ops host1x_channel_ops = {
+	.init = host1x_channel_init,
+	.submit = channel_submit,
+};
diff --git a/drivers/gpu/host1x/hw/host1x01.c b/drivers/gpu/host1x/hw/host1x01.c
index f5c35fa66d05..013ff381d825 100644
--- a/drivers/gpu/host1x/hw/host1x01.c
+++ b/drivers/gpu/host1x/hw/host1x01.c
@@ -21,6 +21,8 @@
 #include "hw/host1x01_hardware.h"
 
 /* include code */
+#include "hw/cdma_hw.c"
+#include "hw/channel_hw.c"
 #include "hw/intr_hw.c"
 #include "hw/syncpt_hw.c"
 
@@ -28,6 +30,9 @@
 
 int host1x01_init(struct host1x *host)
 {
+	host->channel_op = &host1x_channel_ops;
+	host->cdma_op = &host1x_cdma_ops;
+	host->cdma_pb_op = &host1x_pushbuffer_ops;
 	host->syncpt_op = &host1x_syncpt_ops;
 	host->intr_op = &host1x_intr_ops;
 
diff --git a/drivers/gpu/host1x/hw/host1x01_hardware.h b/drivers/gpu/host1x/hw/host1x01_hardware.h
index 8cecbee7a270..5f0fb866efa8 100644
--- a/drivers/gpu/host1x/hw/host1x01_hardware.h
+++ b/drivers/gpu/host1x/hw/host1x01_hardware.h
@@ -22,6 +22,122 @@
 #include <linux/types.h>
 #include <linux/bitops.h>
 
+#include "hw_host1x01_channel.h"
 #include "hw_host1x01_sync.h"
+#include "hw_host1x01_uclass.h"
+
+static inline u32 host1x_class_host_wait_syncpt(
+	unsigned indx, unsigned threshold)
+{
+	return host1x_uclass_wait_syncpt_indx_f(indx)
+		| host1x_uclass_wait_syncpt_thresh_f(threshold);
+}
+
+static inline u32 host1x_class_host_load_syncpt_base(
+	unsigned indx, unsigned threshold)
+{
+	return host1x_uclass_load_syncpt_base_base_indx_f(indx)
+		| host1x_uclass_load_syncpt_base_value_f(threshold);
+}
+
+static inline u32 host1x_class_host_wait_syncpt_base(
+	unsigned indx, unsigned base_indx, unsigned offset)
+{
+	return host1x_uclass_wait_syncpt_base_indx_f(indx)
+		| host1x_uclass_wait_syncpt_base_base_indx_f(base_indx)
+		| host1x_uclass_wait_syncpt_base_offset_f(offset);
+}
+
+static inline u32 host1x_class_host_incr_syncpt_base(
+	unsigned base_indx, unsigned offset)
+{
+	return host1x_uclass_incr_syncpt_base_base_indx_f(base_indx)
+		| host1x_uclass_incr_syncpt_base_offset_f(offset);
+}
+
+static inline u32 host1x_class_host_incr_syncpt(
+	unsigned cond, unsigned indx)
+{
+	return host1x_uclass_incr_syncpt_cond_f(cond)
+		| host1x_uclass_incr_syncpt_indx_f(indx);
+}
+
+static inline u32 host1x_class_host_indoff_reg_write(
+	unsigned mod_id, unsigned offset, bool auto_inc)
+{
+	u32 v = host1x_uclass_indoff_indbe_f(0xf)
+		| host1x_uclass_indoff_indmodid_f(mod_id)
+		| host1x_uclass_indoff_indroffset_f(offset);
+	if (auto_inc)
+		v |= host1x_uclass_indoff_autoinc_f(1);
+	return v;
+}
+
+static inline u32 host1x_class_host_indoff_reg_read(
+	unsigned mod_id, unsigned offset, bool auto_inc)
+{
+	u32 v = host1x_uclass_indoff_indmodid_f(mod_id)
+		| host1x_uclass_indoff_indroffset_f(offset)
+		| host1x_uclass_indoff_rwn_read_v();
+	if (auto_inc)
+		v |= host1x_uclass_indoff_autoinc_f(1);
+	return v;
+}
+
+
+/* cdma opcodes */
+static inline u32 host1x_opcode_setclass(
+	unsigned class_id, unsigned offset, unsigned mask)
+{
+	return (0 << 28) | (offset << 16) | (class_id << 6) | mask;
+}
+
+static inline u32 host1x_opcode_incr(unsigned offset, unsigned count)
+{
+	return (1 << 28) | (offset << 16) | count;
+}
+
+static inline u32 host1x_opcode_nonincr(unsigned offset, unsigned count)
+{
+	return (2 << 28) | (offset << 16) | count;
+}
+
+static inline u32 host1x_opcode_mask(unsigned offset, unsigned mask)
+{
+	return (3 << 28) | (offset << 16) | mask;
+}
+
+static inline u32 host1x_opcode_imm(unsigned offset, unsigned value)
+{
+	return (4 << 28) | (offset << 16) | value;
+}
+
+static inline u32 host1x_opcode_imm_incr_syncpt(unsigned cond, unsigned indx)
+{
+	return host1x_opcode_imm(host1x_uclass_incr_syncpt_r(),
+		host1x_class_host_incr_syncpt(cond, indx));
+}
+
+static inline u32 host1x_opcode_restart(unsigned address)
+{
+	return (5 << 28) | (address >> 4);
+}
+
+static inline u32 host1x_opcode_gather(unsigned count)
+{
+	return (6 << 28) | count;
+}
+
+static inline u32 host1x_opcode_gather_nonincr(unsigned offset,	unsigned count)
+{
+	return (6 << 28) | (offset << 16) | BIT(15) | count;
+}
+
+static inline u32 host1x_opcode_gather_incr(unsigned offset, unsigned count)
+{
+	return (6 << 28) | (offset << 16) | BIT(15) | BIT(14) | count;
+}
+
+#define HOST1X_OPCODE_NOP host1x_opcode_nonincr(0, 0)
 
 #endif
diff --git a/drivers/gpu/host1x/hw/hw_host1x01_channel.h b/drivers/gpu/host1x/hw/hw_host1x01_channel.h
new file mode 100644
index 000000000000..9ba133205668
--- /dev/null
+++ b/drivers/gpu/host1x/hw/hw_host1x01_channel.h
@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+ /*
+  * Function naming determines intended use:
+  *
+  *     <x>_r(void) : Returns the offset for register <x>.
+  *
+  *     <x>_w(void) : Returns the word offset for word (4 byte) element <x>.
+  *
+  *     <x>_<y>_s(void) : Returns size of field <y> of register <x> in bits.
+  *
+  *     <x>_<y>_f(u32 v) : Returns a value based on 'v' which has been shifted
+  *         and masked to place it at field <y> of register <x>.  This value
+  *         can be |'d with others to produce a full register value for
+  *         register <x>.
+  *
+  *     <x>_<y>_m(void) : Returns a mask for field <y> of register <x>.  This
+  *         value can be ~'d and then &'d to clear the value of field <y> for
+  *         register <x>.
+  *
+  *     <x>_<y>_<z>_f(void) : Returns the constant value <z> after being shifted
+  *         to place it at field <y> of register <x>.  This value can be |'d
+  *         with others to produce a full register value for <x>.
+  *
+  *     <x>_<y>_v(u32 r) : Returns the value of field <y> from a full register
+  *         <x> value 'r' after being shifted to place its LSB at bit 0.
+  *         This value is suitable for direct comparison with other unshifted
+  *         values appropriate for use in field <y> of register <x>.
+  *
+  *     <x>_<y>_<z>_v(void) : Returns the constant value for <z> defined for
+  *         field <y> of register <x>.  This value is suitable for direct
+  *         comparison with unshifted values appropriate for use in field <y>
+  *         of register <x>.
+  */
+
+#ifndef __hw_host1x_channel_host1x_h__
+#define __hw_host1x_channel_host1x_h__
+
+static inline u32 host1x_channel_dmastart_r(void)
+{
+	return 0x14;
+}
+#define HOST1X_CHANNEL_DMASTART \
+	host1x_channel_dmastart_r()
+static inline u32 host1x_channel_dmaput_r(void)
+{
+	return 0x18;
+}
+#define HOST1X_CHANNEL_DMAPUT \
+	host1x_channel_dmaput_r()
+static inline u32 host1x_channel_dmaget_r(void)
+{
+	return 0x1c;
+}
+#define HOST1X_CHANNEL_DMAGET \
+	host1x_channel_dmaget_r()
+static inline u32 host1x_channel_dmaend_r(void)
+{
+	return 0x20;
+}
+#define HOST1X_CHANNEL_DMAEND \
+	host1x_channel_dmaend_r()
+static inline u32 host1x_channel_dmactrl_r(void)
+{
+	return 0x24;
+}
+#define HOST1X_CHANNEL_DMACTRL \
+	host1x_channel_dmactrl_r()
+static inline u32 host1x_channel_dmactrl_dmastop(void)
+{
+	return 1 << 0;
+}
+#define HOST1X_CHANNEL_DMACTRL_DMASTOP \
+	host1x_channel_dmactrl_dmastop()
+static inline u32 host1x_channel_dmactrl_dmagetrst(void)
+{
+	return 1 << 1;
+}
+#define HOST1X_CHANNEL_DMACTRL_DMAGETRST \
+	host1x_channel_dmactrl_dmagetrst()
+static inline u32 host1x_channel_dmactrl_dmainitget(void)
+{
+	return 1 << 2;
+}
+#define HOST1X_CHANNEL_DMACTRL_DMAINITGET \
+	host1x_channel_dmactrl_dmainitget()
+#endif
diff --git a/drivers/gpu/host1x/hw/hw_host1x01_sync.h b/drivers/gpu/host1x/hw/hw_host1x01_sync.h
index eea0bb06052a..8f2a246c5426 100644
--- a/drivers/gpu/host1x/hw/hw_host1x01_sync.h
+++ b/drivers/gpu/host1x/hw/hw_host1x01_sync.h
@@ -77,6 +77,18 @@ static inline u32 host1x_sync_syncpt_thresh_int_enable_cpu0_r(unsigned int id)
 }
 #define HOST1X_SYNC_SYNCPT_THRESH_INT_ENABLE_CPU0(id) \
 	host1x_sync_syncpt_thresh_int_enable_cpu0_r(id)
+static inline u32 host1x_sync_cmdproc_stop_r(void)
+{
+	return 0xac;
+}
+#define HOST1X_SYNC_CMDPROC_STOP \
+	host1x_sync_cmdproc_stop_r()
+static inline u32 host1x_sync_ch_teardown_r(void)
+{
+	return 0xb0;
+}
+#define HOST1X_SYNC_CH_TEARDOWN \
+	host1x_sync_ch_teardown_r()
 static inline u32 host1x_sync_usec_clk_r(void)
 {
 	return 0x1a4;
diff --git a/drivers/gpu/host1x/hw/hw_host1x01_uclass.h b/drivers/gpu/host1x/hw/hw_host1x01_uclass.h
new file mode 100644
index 000000000000..7af660966ad6
--- /dev/null
+++ b/drivers/gpu/host1x/hw/hw_host1x01_uclass.h
@@ -0,0 +1,168 @@
+/*
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+ /*
+  * Function naming determines intended use:
+  *
+  *     <x>_r(void) : Returns the offset for register <x>.
+  *
+  *     <x>_w(void) : Returns the word offset for word (4 byte) element <x>.
+  *
+  *     <x>_<y>_s(void) : Returns size of field <y> of register <x> in bits.
+  *
+  *     <x>_<y>_f(u32 v) : Returns a value based on 'v' which has been shifted
+  *         and masked to place it at field <y> of register <x>.  This value
+  *         can be |'d with others to produce a full register value for
+  *         register <x>.
+  *
+  *     <x>_<y>_m(void) : Returns a mask for field <y> of register <x>.  This
+  *         value can be ~'d and then &'d to clear the value of field <y> for
+  *         register <x>.
+  *
+  *     <x>_<y>_<z>_f(void) : Returns the constant value <z> after being shifted
+  *         to place it at field <y> of register <x>.  This value can be |'d
+  *         with others to produce a full register value for <x>.
+  *
+  *     <x>_<y>_v(u32 r) : Returns the value of field <y> from a full register
+  *         <x> value 'r' after being shifted to place its LSB at bit 0.
+  *         This value is suitable for direct comparison with other unshifted
+  *         values appropriate for use in field <y> of register <x>.
+  *
+  *     <x>_<y>_<z>_v(void) : Returns the constant value for <z> defined for
+  *         field <y> of register <x>.  This value is suitable for direct
+  *         comparison with unshifted values appropriate for use in field <y>
+  *         of register <x>.
+  */
+
+#ifndef __hw_host1x_uclass_host1x_h__
+#define __hw_host1x_uclass_host1x_h__
+
+static inline u32 host1x_uclass_incr_syncpt_r(void)
+{
+	return 0x0;
+}
+#define HOST1X_UCLASS_INCR_SYNCPT \
+	host1x_uclass_incr_syncpt_r()
+static inline u32 host1x_uclass_incr_syncpt_cond_f(u32 v)
+{
+	return (v & 0xff) << 8;
+}
+#define HOST1X_UCLASS_INCR_SYNCPT_COND_F(v) \
+	host1x_uclass_incr_syncpt_cond_f(v)
+static inline u32 host1x_uclass_incr_syncpt_indx_f(u32 v)
+{
+	return (v & 0xff) << 0;
+}
+#define HOST1X_UCLASS_INCR_SYNCPT_INDX_F(v) \
+	host1x_uclass_incr_syncpt_indx_f(v)
+static inline u32 host1x_uclass_wait_syncpt_r(void)
+{
+	return 0x8;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT \
+	host1x_uclass_wait_syncpt_r()
+static inline u32 host1x_uclass_wait_syncpt_indx_f(u32 v)
+{
+	return (v & 0xff) << 24;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT_INDX_F(v) \
+	host1x_uclass_wait_syncpt_indx_f(v)
+static inline u32 host1x_uclass_wait_syncpt_thresh_f(u32 v)
+{
+	return (v & 0xffffff) << 0;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT_THRESH_F(v) \
+	host1x_uclass_wait_syncpt_thresh_f(v)
+static inline u32 host1x_uclass_wait_syncpt_base_indx_f(u32 v)
+{
+	return (v & 0xff) << 24;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT_BASE_INDX_F(v) \
+	host1x_uclass_wait_syncpt_base_indx_f(v)
+static inline u32 host1x_uclass_wait_syncpt_base_base_indx_f(u32 v)
+{
+	return (v & 0xff) << 16;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT_BASE_BASE_INDX_F(v) \
+	host1x_uclass_wait_syncpt_base_base_indx_f(v)
+static inline u32 host1x_uclass_wait_syncpt_base_offset_f(u32 v)
+{
+	return (v & 0xffff) << 0;
+}
+#define HOST1X_UCLASS_WAIT_SYNCPT_BASE_OFFSET_F(v) \
+	host1x_uclass_wait_syncpt_base_offset_f(v)
+static inline u32 host1x_uclass_load_syncpt_base_base_indx_f(u32 v)
+{
+	return (v & 0xff) << 24;
+}
+#define HOST1X_UCLASS_LOAD_SYNCPT_BASE_BASE_INDX_F(v) \
+	host1x_uclass_load_syncpt_base_base_indx_f(v)
+static inline u32 host1x_uclass_load_syncpt_base_value_f(u32 v)
+{
+	return (v & 0xffffff) << 0;
+}
+#define HOST1X_UCLASS_LOAD_SYNCPT_BASE_VALUE_F(v) \
+	host1x_uclass_load_syncpt_base_value_f(v)
+static inline u32 host1x_uclass_incr_syncpt_base_base_indx_f(u32 v)
+{
+	return (v & 0xff) << 24;
+}
+#define HOST1X_UCLASS_INCR_SYNCPT_BASE_BASE_INDX_F(v) \
+	host1x_uclass_incr_syncpt_base_base_indx_f(v)
+static inline u32 host1x_uclass_incr_syncpt_base_offset_f(u32 v)
+{
+	return (v & 0xffffff) << 0;
+}
+#define HOST1X_UCLASS_INCR_SYNCPT_BASE_OFFSET_F(v) \
+	host1x_uclass_incr_syncpt_base_offset_f(v)
+static inline u32 host1x_uclass_indoff_r(void)
+{
+	return 0x2d;
+}
+#define HOST1X_UCLASS_INDOFF \
+	host1x_uclass_indoff_r()
+static inline u32 host1x_uclass_indoff_indbe_f(u32 v)
+{
+	return (v & 0xf) << 28;
+}
+#define HOST1X_UCLASS_INDOFF_INDBE_F(v) \
+	host1x_uclass_indoff_indbe_f(v)
+static inline u32 host1x_uclass_indoff_autoinc_f(u32 v)
+{
+	return (v & 0x1) << 27;
+}
+#define HOST1X_UCLASS_INDOFF_AUTOINC_F(v) \
+	host1x_uclass_indoff_autoinc_f(v)
+static inline u32 host1x_uclass_indoff_indmodid_f(u32 v)
+{
+	return (v & 0xff) << 18;
+}
+#define HOST1X_UCLASS_INDOFF_INDMODID_F(v) \
+	host1x_uclass_indoff_indmodid_f(v)
+static inline u32 host1x_uclass_indoff_indroffset_f(u32 v)
+{
+	return (v & 0xffff) << 2;
+}
+#define HOST1X_UCLASS_INDOFF_INDROFFSET_F(v) \
+	host1x_uclass_indoff_indroffset_f(v)
+static inline u32 host1x_uclass_indoff_rwn_read_v(void)
+{
+	return 1;
+}
+#define HOST1X_UCLASS_INDOFF_INDROFFSET_F(v) \
+	host1x_uclass_indoff_indroffset_f(v)
+#endif
diff --git a/drivers/gpu/host1x/hw/syncpt_hw.c b/drivers/gpu/host1x/hw/syncpt_hw.c
index 885b2578dc86..2c1f4af1094c 100644
--- a/drivers/gpu/host1x/hw/syncpt_hw.c
+++ b/drivers/gpu/host1x/hw/syncpt_hw.c
@@ -93,10 +93,21 @@ static void syncpt_cpu_incr(struct host1x_syncpt *sp)
 	wmb();
 }
 
+/* remove a wait pointed to by patch_addr */
+static int syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr)
+{
+	u32 override = host1x_class_host_wait_syncpt(
+		HOST1X_SYNCPT_RESERVED, 0);
+
+	*((u32 *)patch_addr) = override;
+	return 0;
+}
+
 static const struct host1x_syncpt_ops host1x_syncpt_ops = {
 	.restore = syncpt_restore,
 	.restore_wait_base = syncpt_restore_wait_base,
 	.load_wait_base = syncpt_read_wait_base,
 	.load = syncpt_load,
 	.cpu_incr = syncpt_cpu_incr,
+	.patch_wait = syncpt_patch_wait,
 };
diff --git a/drivers/gpu/host1x/intr.c b/drivers/gpu/host1x/intr.c
index b1b5a80e3125..2491bf82e30c 100644
--- a/drivers/gpu/host1x/intr.c
+++ b/drivers/gpu/host1x/intr.c
@@ -21,6 +21,8 @@
 #include <linux/slab.h>
 #include <linux/irq.h>
 
+#include <trace/events/host1x.h>
+#include "channel.h"
 #include "dev.h"
 #include "intr.h"
 
@@ -66,7 +68,7 @@ static void remove_completed_waiters(struct list_head *head, u32 sync,
 			struct list_head completed[HOST1X_INTR_ACTION_COUNT])
 {
 	struct list_head *dest;
-	struct host1x_waitlist *waiter, *next;
+	struct host1x_waitlist *waiter, *next, *prev;
 
 	list_for_each_entry_safe(waiter, next, head, list) {
 		if ((s32)(waiter->thresh - sync) > 0)
@@ -74,6 +76,17 @@ static void remove_completed_waiters(struct list_head *head, u32 sync,
 
 		dest = completed + waiter->action;
 
+		/* consolidate submit cleanups */
+		if (waiter->action == HOST1X_INTR_ACTION_SUBMIT_COMPLETE &&
+		    !list_empty(dest)) {
+			prev = list_entry(dest->prev,
+					  struct host1x_waitlist, list);
+			if (prev->data == waiter->data) {
+				prev->count++;
+				dest = NULL;
+			}
+		}
+
 		/* PENDING->REMOVED or CANCELLED->HANDLED */
 		if (atomic_inc_return(&waiter->state) == WLS_HANDLED || !dest) {
 			list_del(&waiter->list);
@@ -94,6 +107,18 @@ static void reset_threshold_interrupt(struct host1x *host,
 	host1x_hw_intr_enable_syncpt_intr(host, id);
 }
 
+static void action_submit_complete(struct host1x_waitlist *waiter)
+{
+	struct host1x_channel *channel = waiter->data;
+
+	host1x_cdma_update(&channel->cdma);
+
+	/*  Add nr_completed to trace */
+	trace_host1x_channel_submit_complete(dev_name(channel->dev),
+					     waiter->count, waiter->thresh);
+
+}
+
 static void action_wakeup(struct host1x_waitlist *waiter)
 {
 	wait_queue_head_t *wq = waiter->data;
@@ -109,6 +134,7 @@ static void action_wakeup_interruptible(struct host1x_waitlist *waiter)
 typedef void (*action_handler)(struct host1x_waitlist *waiter);
 
 static action_handler action_handlers[HOST1X_INTR_ACTION_COUNT] = {
+	action_submit_complete,
 	action_wakeup,
 	action_wakeup_interruptible,
 };
diff --git a/drivers/gpu/host1x/intr.h b/drivers/gpu/host1x/intr.h
index a3f06abe94bb..2b8adf016a05 100644
--- a/drivers/gpu/host1x/intr.h
+++ b/drivers/gpu/host1x/intr.h
@@ -25,6 +25,12 @@
 struct host1x;
 
 enum host1x_intr_action {
+	/*
+	 * Perform cleanup after a submit has completed.
+	 * 'data' points to a channel
+	 */
+	HOST1X_INTR_ACTION_SUBMIT_COMPLETE = 0,
+
 	/*
 	 * Wake up a  task.
 	 * 'data' points to a wait_queue_head_t
diff --git a/drivers/gpu/host1x/job.c b/drivers/gpu/host1x/job.c
new file mode 100644
index 000000000000..f665d679031c
--- /dev/null
+++ b/drivers/gpu/host1x/job.c
@@ -0,0 +1,603 @@
+/*
+ * Tegra host1x Job
+ *
+ * Copyright (c) 2010-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/err.h>
+#include <linux/kref.h>
+#include <linux/module.h>
+#include <linux/scatterlist.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <trace/events/host1x.h>
+
+#include "channel.h"
+#include "dev.h"
+#include "host1x_bo.h"
+#include "job.h"
+#include "syncpt.h"
+
+struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
+				    u32 num_cmdbufs, u32 num_relocs,
+				    u32 num_waitchks)
+{
+	struct host1x_job *job = NULL;
+	unsigned int num_unpins = num_cmdbufs + num_relocs;
+	u64 total;
+	void *mem;
+
+	/* Check that we're not going to overflow */
+	total = sizeof(struct host1x_job) +
+		num_relocs * sizeof(struct host1x_reloc) +
+		num_unpins * sizeof(struct host1x_job_unpin_data) +
+		num_waitchks * sizeof(struct host1x_waitchk) +
+		num_cmdbufs * sizeof(struct host1x_job_gather) +
+		num_unpins * sizeof(dma_addr_t) +
+		num_unpins * sizeof(u32 *);
+	if (total > ULONG_MAX)
+		return NULL;
+
+	mem = job = kzalloc(total, GFP_KERNEL);
+	if (!job)
+		return NULL;
+
+	kref_init(&job->ref);
+	job->channel = ch;
+
+	/* Redistribute memory to the structs  */
+	mem += sizeof(struct host1x_job);
+	job->relocarray = num_relocs ? mem : NULL;
+	mem += num_relocs * sizeof(struct host1x_reloc);
+	job->unpins = num_unpins ? mem : NULL;
+	mem += num_unpins * sizeof(struct host1x_job_unpin_data);
+	job->waitchk = num_waitchks ? mem : NULL;
+	mem += num_waitchks * sizeof(struct host1x_waitchk);
+	job->gathers = num_cmdbufs ? mem : NULL;
+	mem += num_cmdbufs * sizeof(struct host1x_job_gather);
+	job->addr_phys = num_unpins ? mem : NULL;
+
+	job->reloc_addr_phys = job->addr_phys;
+	job->gather_addr_phys = &job->addr_phys[num_relocs];
+
+	return job;
+}
+
+struct host1x_job *host1x_job_get(struct host1x_job *job)
+{
+	kref_get(&job->ref);
+	return job;
+}
+
+static void job_free(struct kref *ref)
+{
+	struct host1x_job *job = container_of(ref, struct host1x_job, ref);
+
+	kfree(job);
+}
+
+void host1x_job_put(struct host1x_job *job)
+{
+	kref_put(&job->ref, job_free);
+}
+
+void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *bo,
+			   u32 words, u32 offset)
+{
+	struct host1x_job_gather *cur_gather = &job->gathers[job->num_gathers];
+
+	cur_gather->words = words;
+	cur_gather->bo = bo;
+	cur_gather->offset = offset;
+	job->num_gathers++;
+}
+
+/*
+ * NULL an already satisfied WAIT_SYNCPT host method, by patching its
+ * args in the command stream. The method data is changed to reference
+ * a reserved (never given out or incr) HOST1X_SYNCPT_RESERVED syncpt
+ * with a matching threshold value of 0, so is guaranteed to be popped
+ * by the host HW.
+ */
+static void host1x_syncpt_patch_offset(struct host1x_syncpt *sp,
+				       struct host1x_bo *h, u32 offset)
+{
+	void *patch_addr = NULL;
+
+	/* patch the wait */
+	patch_addr = host1x_bo_kmap(h, offset >> PAGE_SHIFT);
+	if (patch_addr) {
+		host1x_syncpt_patch_wait(sp,
+					 patch_addr + (offset & ~PAGE_MASK));
+		host1x_bo_kunmap(h, offset >> PAGE_SHIFT, patch_addr);
+	} else
+		pr_err("Could not map cmdbuf for wait check\n");
+}
+
+/*
+ * Check driver supplied waitchk structs for syncpt thresholds
+ * that have already been satisfied and NULL the comparison (to
+ * avoid a wrap condition in the HW).
+ */
+static int do_waitchks(struct host1x_job *job, struct host1x *host,
+		       struct host1x_bo *patch)
+{
+	int i;
+
+	/* compare syncpt vs wait threshold */
+	for (i = 0; i < job->num_waitchk; i++) {
+		struct host1x_waitchk *wait = &job->waitchk[i];
+		struct host1x_syncpt *sp =
+			host1x_syncpt_get(host, wait->syncpt_id);
+
+		/* validate syncpt id */
+		if (wait->syncpt_id > host1x_syncpt_nb_pts(host))
+			continue;
+
+		/* skip all other gathers */
+		if (patch != wait->bo)
+			continue;
+
+		trace_host1x_syncpt_wait_check(wait->bo, wait->offset,
+					       wait->syncpt_id, wait->thresh,
+					       host1x_syncpt_read_min(sp));
+
+		if (host1x_syncpt_is_expired(sp, wait->thresh)) {
+			dev_dbg(host->dev,
+				"drop WAIT id %d (%s) thresh 0x%x, min 0x%x\n",
+				wait->syncpt_id, sp->name, wait->thresh,
+				host1x_syncpt_read_min(sp));
+
+			host1x_syncpt_patch_offset(sp, patch, wait->offset);
+		}
+
+		wait->bo = NULL;
+	}
+
+	return 0;
+}
+
+static unsigned int pin_job(struct host1x_job *job)
+{
+	unsigned int i;
+
+	job->num_unpins = 0;
+
+	for (i = 0; i < job->num_relocs; i++) {
+		struct host1x_reloc *reloc = &job->relocarray[i];
+		struct sg_table *sgt;
+		dma_addr_t phys_addr;
+
+		reloc->target = host1x_bo_get(reloc->target);
+		if (!reloc->target)
+			goto unpin;
+
+		phys_addr = host1x_bo_pin(reloc->target, &sgt);
+		if (!phys_addr)
+			goto unpin;
+
+		job->addr_phys[job->num_unpins] = phys_addr;
+		job->unpins[job->num_unpins].bo = reloc->target;
+		job->unpins[job->num_unpins].sgt = sgt;
+		job->num_unpins++;
+	}
+
+	for (i = 0; i < job->num_gathers; i++) {
+		struct host1x_job_gather *g = &job->gathers[i];
+		struct sg_table *sgt;
+		dma_addr_t phys_addr;
+
+		g->bo = host1x_bo_get(g->bo);
+		if (!g->bo)
+			goto unpin;
+
+		phys_addr = host1x_bo_pin(g->bo, &sgt);
+		if (!phys_addr)
+			goto unpin;
+
+		job->addr_phys[job->num_unpins] = phys_addr;
+		job->unpins[job->num_unpins].bo = g->bo;
+		job->unpins[job->num_unpins].sgt = sgt;
+		job->num_unpins++;
+	}
+
+	return job->num_unpins;
+
+unpin:
+	host1x_job_unpin(job);
+	return 0;
+}
+
+static unsigned int do_relocs(struct host1x_job *job, struct host1x_bo *cmdbuf)
+{
+	int i = 0;
+	u32 last_page = ~0;
+	void *cmdbuf_page_addr = NULL;
+
+	/* pin & patch the relocs for one gather */
+	while (i < job->num_relocs) {
+		struct host1x_reloc *reloc = &job->relocarray[i];
+		u32 reloc_addr = (job->reloc_addr_phys[i] +
+			reloc->target_offset) >> reloc->shift;
+		u32 *target;
+
+		/* skip all other gathers */
+		if (!(reloc->cmdbuf && cmdbuf == reloc->cmdbuf)) {
+			i++;
+			continue;
+		}
+
+		if (last_page != reloc->cmdbuf_offset >> PAGE_SHIFT) {
+			if (cmdbuf_page_addr)
+				host1x_bo_kunmap(cmdbuf, last_page,
+						 cmdbuf_page_addr);
+
+			cmdbuf_page_addr = host1x_bo_kmap(cmdbuf,
+					reloc->cmdbuf_offset >> PAGE_SHIFT);
+			last_page = reloc->cmdbuf_offset >> PAGE_SHIFT;
+
+			if (unlikely(!cmdbuf_page_addr)) {
+				pr_err("Could not map cmdbuf for relocation\n");
+				return -ENOMEM;
+			}
+		}
+
+		target = cmdbuf_page_addr + (reloc->cmdbuf_offset & ~PAGE_MASK);
+		*target = reloc_addr;
+
+		/* mark this gather as handled */
+		reloc->cmdbuf = 0;
+	}
+
+	if (cmdbuf_page_addr)
+		host1x_bo_kunmap(cmdbuf, last_page, cmdbuf_page_addr);
+
+	return 0;
+}
+
+static int check_reloc(struct host1x_reloc *reloc, struct host1x_bo *cmdbuf,
+		       unsigned int offset)
+{
+	offset *= sizeof(u32);
+
+	if (reloc->cmdbuf != cmdbuf || reloc->cmdbuf_offset != offset)
+		return -EINVAL;
+
+	return 0;
+}
+
+struct host1x_firewall {
+	struct host1x_job *job;
+	struct device *dev;
+
+	unsigned int num_relocs;
+	struct host1x_reloc *reloc;
+
+	struct host1x_bo *cmdbuf_id;
+	unsigned int offset;
+
+	u32 words;
+	u32 class;
+	u32 reg;
+	u32 mask;
+	u32 count;
+};
+
+static int check_mask(struct host1x_firewall *fw)
+{
+	u32 mask = fw->mask;
+	u32 reg = fw->reg;
+
+	while (mask) {
+		if (fw->words == 0)
+			return -EINVAL;
+
+		if (mask & 1) {
+			if (fw->job->is_addr_reg(fw->dev, fw->class, reg)) {
+				bool bad_reloc = check_reloc(fw->reloc,
+							     fw->cmdbuf_id,
+							     fw->offset);
+				if (!fw->num_relocs || bad_reloc)
+					return -EINVAL;
+				fw->reloc++;
+				fw->num_relocs--;
+			}
+			fw->words--;
+			fw->offset++;
+		}
+		mask >>= 1;
+		reg++;
+	}
+
+	return 0;
+}
+
+static int check_incr(struct host1x_firewall *fw)
+{
+	u32 count = fw->count;
+	u32 reg = fw->reg;
+
+	while (fw) {
+		if (fw->words == 0)
+			return -EINVAL;
+
+		if (fw->job->is_addr_reg(fw->dev, fw->class, reg)) {
+			bool bad_reloc = check_reloc(fw->reloc, fw->cmdbuf_id,
+						     fw->offset);
+			if (!fw->num_relocs || bad_reloc)
+				return -EINVAL;
+			fw->reloc++;
+			fw->num_relocs--;
+		}
+		reg++;
+		fw->words--;
+		fw->offset++;
+		count--;
+	}
+
+	return 0;
+}
+
+static int check_nonincr(struct host1x_firewall *fw)
+{
+	int is_addr_reg = fw->job->is_addr_reg(fw->dev, fw->class, fw->reg);
+	u32 count = fw->count;
+
+	while (count) {
+		if (fw->words == 0)
+			return -EINVAL;
+
+		if (is_addr_reg) {
+			bool bad_reloc = check_reloc(fw->reloc, fw->cmdbuf_id,
+						     fw->offset);
+			if (!fw->num_relocs || bad_reloc)
+				return -EINVAL;
+			fw->reloc++;
+			fw->num_relocs--;
+		}
+		fw->words--;
+		fw->offset++;
+		count--;
+	}
+
+	return 0;
+}
+
+static int validate(struct host1x_job *job, struct device *dev,
+		    struct host1x_job_gather *g)
+{
+	u32 *cmdbuf_base;
+	int err = 0;
+	struct host1x_firewall fw;
+
+	fw.job = job;
+	fw.dev = dev;
+	fw.reloc = job->relocarray;
+	fw.num_relocs = job->num_relocs;
+	fw.cmdbuf_id = g->bo;
+
+	fw.offset = 0;
+	fw.class = 0;
+
+	if (!job->is_addr_reg)
+		return 0;
+
+	cmdbuf_base = host1x_bo_mmap(g->bo);
+	if (!cmdbuf_base)
+		return -ENOMEM;
+
+	fw.words = g->words;
+	while (fw.words && !err) {
+		u32 word = cmdbuf_base[fw.offset];
+		u32 opcode = (word & 0xf0000000) >> 28;
+
+		fw.mask = 0;
+		fw.reg = 0;
+		fw.count = 0;
+		fw.words--;
+		fw.offset++;
+
+		switch (opcode) {
+		case 0:
+			fw.class = word >> 6 & 0x3ff;
+			fw.mask = word & 0x3f;
+			fw.reg = word >> 16 & 0xfff;
+			err = check_mask(&fw);
+			if (err)
+				goto out;
+			break;
+		case 1:
+			fw.reg = word >> 16 & 0xfff;
+			fw.count = word & 0xffff;
+			err = check_incr(&fw);
+			if (err)
+				goto out;
+			break;
+
+		case 2:
+			fw.reg = word >> 16 & 0xfff;
+			fw.count = word & 0xffff;
+			err = check_nonincr(&fw);
+			if (err)
+				goto out;
+			break;
+
+		case 3:
+			fw.mask = word & 0xffff;
+			fw.reg = word >> 16 & 0xfff;
+			err = check_mask(&fw);
+			if (err)
+				goto out;
+			break;
+		case 4:
+		case 5:
+		case 14:
+			break;
+		default:
+			err = -EINVAL;
+			break;
+		}
+	}
+
+	/* No relocs should remain at this point */
+	if (fw.num_relocs)
+		err = -EINVAL;
+
+out:
+	host1x_bo_munmap(g->bo, cmdbuf_base);
+
+	return err;
+}
+
+static inline int copy_gathers(struct host1x_job *job, struct device *dev)
+{
+	size_t size = 0;
+	size_t offset = 0;
+	int i;
+
+	for (i = 0; i < job->num_gathers; i++) {
+		struct host1x_job_gather *g = &job->gathers[i];
+		size += g->words * sizeof(u32);
+	}
+
+	job->gather_copy_mapped = dma_alloc_writecombine(dev, size,
+							 &job->gather_copy,
+							 GFP_KERNEL);
+	if (!job->gather_copy_mapped) {
+		int err = PTR_ERR(job->gather_copy_mapped);
+		job->gather_copy_mapped = NULL;
+		return err;
+	}
+
+	job->gather_copy_size = size;
+
+	for (i = 0; i < job->num_gathers; i++) {
+		struct host1x_job_gather *g = &job->gathers[i];
+		void *gather;
+
+		gather = host1x_bo_mmap(g->bo);
+		memcpy(job->gather_copy_mapped + offset, gather + g->offset,
+		       g->words * sizeof(u32));
+		host1x_bo_munmap(g->bo, gather);
+
+		g->base = job->gather_copy;
+		g->offset = offset;
+		g->bo = NULL;
+
+		offset += g->words * sizeof(u32);
+	}
+
+	return 0;
+}
+
+int host1x_job_pin(struct host1x_job *job, struct device *dev)
+{
+	int err;
+	unsigned int i, j;
+	struct host1x *host = dev_get_drvdata(dev->parent);
+	DECLARE_BITMAP(waitchk_mask, host1x_syncpt_nb_pts(host));
+
+	bitmap_zero(waitchk_mask, host1x_syncpt_nb_pts(host));
+	for (i = 0; i < job->num_waitchk; i++) {
+		u32 syncpt_id = job->waitchk[i].syncpt_id;
+		if (syncpt_id < host1x_syncpt_nb_pts(host))
+			set_bit(syncpt_id, waitchk_mask);
+	}
+
+	/* get current syncpt values for waitchk */
+	for_each_set_bit(i, waitchk_mask, host1x_syncpt_nb_pts(host))
+		host1x_syncpt_load(host->syncpt + i);
+
+	/* pin memory */
+	err = pin_job(job);
+	if (!err)
+		goto out;
+
+	/* patch gathers */
+	for (i = 0; i < job->num_gathers; i++) {
+		struct host1x_job_gather *g = &job->gathers[i];
+
+		/* process each gather mem only once */
+		if (g->handled)
+			continue;
+
+		g->base = job->gather_addr_phys[i];
+
+		for (j = 0; j < job->num_gathers; j++)
+			if (job->gathers[j].bo == g->bo)
+				job->gathers[j].handled = true;
+
+		err = 0;
+
+		if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL))
+			err = validate(job, dev, g);
+
+		if (err)
+			dev_err(dev, "Job invalid (err=%d)\n", err);
+
+		if (!err)
+			err = do_relocs(job, g->bo);
+
+		if (!err)
+			err = do_waitchks(job, host, g->bo);
+
+		if (err)
+			break;
+	}
+
+	if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) && !err) {
+		err = copy_gathers(job, dev);
+		if (err) {
+			host1x_job_unpin(job);
+			return err;
+		}
+	}
+
+out:
+	wmb();
+
+	return err;
+}
+
+void host1x_job_unpin(struct host1x_job *job)
+{
+	unsigned int i;
+
+	for (i = 0; i < job->num_unpins; i++) {
+		struct host1x_job_unpin_data *unpin = &job->unpins[i];
+		host1x_bo_unpin(unpin->bo, unpin->sgt);
+		host1x_bo_put(unpin->bo);
+	}
+	job->num_unpins = 0;
+
+	if (job->gather_copy_size)
+		dma_free_writecombine(job->channel->dev, job->gather_copy_size,
+				      job->gather_copy_mapped,
+				      job->gather_copy);
+}
+
+/*
+ * Debug routine used to dump job entries
+ */
+void host1x_job_dump(struct device *dev, struct host1x_job *job)
+{
+	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt_id);
+	dev_dbg(dev, "    SYNCPT_VAL  %d\n", job->syncpt_end);
+	dev_dbg(dev, "    FIRST_GET   0x%x\n", job->first_get);
+	dev_dbg(dev, "    TIMEOUT     %d\n", job->timeout);
+	dev_dbg(dev, "    NUM_SLOTS   %d\n", job->num_slots);
+	dev_dbg(dev, "    NUM_HANDLES %d\n", job->num_unpins);
+}
diff --git a/drivers/gpu/host1x/job.h b/drivers/gpu/host1x/job.h
new file mode 100644
index 000000000000..fba45f20458e
--- /dev/null
+++ b/drivers/gpu/host1x/job.h
@@ -0,0 +1,162 @@
+/*
+ * Tegra host1x Job
+ *
+ * Copyright (c) 2011-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __HOST1X_JOB_H
+#define __HOST1X_JOB_H
+
+struct host1x_job_gather {
+	u32 words;
+	dma_addr_t base;
+	struct host1x_bo *bo;
+	int offset;
+	bool handled;
+};
+
+struct host1x_cmdbuf {
+	u32 handle;
+	u32 offset;
+	u32 words;
+	u32 pad;
+};
+
+struct host1x_reloc {
+	struct host1x_bo *cmdbuf;
+	u32 cmdbuf_offset;
+	struct host1x_bo *target;
+	u32 target_offset;
+	u32 shift;
+	u32 pad;
+};
+
+struct host1x_waitchk {
+	struct host1x_bo *bo;
+	u32 offset;
+	u32 syncpt_id;
+	u32 thresh;
+};
+
+struct host1x_job_unpin_data {
+	struct host1x_bo *bo;
+	struct sg_table *sgt;
+};
+
+/*
+ * Each submit is tracked as a host1x_job.
+ */
+struct host1x_job {
+	/* When refcount goes to zero, job can be freed */
+	struct kref ref;
+
+	/* List entry */
+	struct list_head list;
+
+	/* Channel where job is submitted to */
+	struct host1x_channel *channel;
+
+	u32 client;
+
+	/* Gathers and their memory */
+	struct host1x_job_gather *gathers;
+	unsigned int num_gathers;
+
+	/* Wait checks to be processed at submit time */
+	struct host1x_waitchk *waitchk;
+	unsigned int num_waitchk;
+	u32 waitchk_mask;
+
+	/* Array of handles to be pinned & unpinned */
+	struct host1x_reloc *relocarray;
+	unsigned int num_relocs;
+	struct host1x_job_unpin_data *unpins;
+	unsigned int num_unpins;
+
+	dma_addr_t *addr_phys;
+	dma_addr_t *gather_addr_phys;
+	dma_addr_t *reloc_addr_phys;
+
+	/* Sync point id, number of increments and end related to the submit */
+	u32 syncpt_id;
+	u32 syncpt_incrs;
+	u32 syncpt_end;
+
+	/* Maximum time to wait for this job */
+	unsigned int timeout;
+
+	/* Index and number of slots used in the push buffer */
+	unsigned int first_get;
+	unsigned int num_slots;
+
+	/* Copy of gathers */
+	size_t gather_copy_size;
+	dma_addr_t gather_copy;
+	u8 *gather_copy_mapped;
+
+	/* Check if register is marked as an address reg */
+	int (*is_addr_reg)(struct device *dev, u32 reg, u32 class);
+
+	/* Request a SETCLASS to this class */
+	u32 class;
+
+	/* Add a channel wait for previous ops to complete */
+	bool serialize;
+};
+/*
+ * Allocate memory for a job. Just enough memory will be allocated to
+ * accomodate the submit.
+ */
+struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
+				    u32 num_cmdbufs, u32 num_relocs,
+				    u32 num_waitchks);
+
+/*
+ * Add a gather to a job.
+ */
+void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *mem_id,
+			   u32 words, u32 offset);
+
+/*
+ * Increment reference going to host1x_job.
+ */
+struct host1x_job *host1x_job_get(struct host1x_job *job);
+
+/*
+ * Decrement reference job, free if goes to zero.
+ */
+void host1x_job_put(struct host1x_job *job);
+
+/*
+ * Pin memory related to job. This handles relocation of addresses to the
+ * host1x address space. Handles both the gather memory and any other memory
+ * referred to from the gather buffers.
+ *
+ * Handles also patching out host waits that would wait for an expired sync
+ * point value.
+ */
+int host1x_job_pin(struct host1x_job *job, struct device *dev);
+
+/*
+ * Unpin memory related to job.
+ */
+void host1x_job_unpin(struct host1x_job *job);
+
+/*
+ * Dump contents of job to debug output.
+ */
+void host1x_job_dump(struct device *dev, struct host1x_job *job);
+
+#endif
diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c
index 07fad1412bab..7e77e63da57b 100644
--- a/drivers/gpu/host1x/syncpt.c
+++ b/drivers/gpu/host1x/syncpt.c
@@ -300,6 +300,12 @@ bool host1x_syncpt_is_expired(struct host1x_syncpt *sp, u32 thresh)
 		return (s32)(current_val - thresh) >= 0;
 }
 
+/* remove a wait pointed to by patch_addr */
+int host1x_syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr)
+{
+	return host1x_hw_syncpt_patch_wait(sp->host, sp, patch_addr);
+}
+
 int host1x_syncpt_init(struct host1x *host)
 {
 	struct host1x_syncpt *syncpt;
@@ -319,6 +325,11 @@ int host1x_syncpt_init(struct host1x *host)
 
 	host1x_syncpt_restore(host);
 
+	/* Allocate sync point to use for clearing waits for expired fences */
+	host->nop_sp = _host1x_syncpt_alloc(host, NULL, 0);
+	if (!host->nop_sp)
+		return -ENOMEM;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/host1x/syncpt.h b/drivers/gpu/host1x/syncpt.h
index 17c1616de100..c99806130f2e 100644
--- a/drivers/gpu/host1x/syncpt.h
+++ b/drivers/gpu/host1x/syncpt.h
@@ -27,6 +27,9 @@
 
 struct host1x;
 
+/* Reserved for replacing an expired wait with a NOP */
+#define HOST1X_SYNCPT_RESERVED			0
+
 struct host1x_syncpt {
 	int id;
 	atomic_t min_val;
@@ -146,6 +149,9 @@ static inline int host1x_syncpt_is_valid(struct host1x_syncpt *sp)
 	return sp->id < host1x_syncpt_nb_pts(sp->host);
 }
 
+/* Patch a wait by replacing it with a wait for syncpt 0 value 0 */
+int host1x_syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr);
+
 /* Return id of the sync point */
 u32 host1x_syncpt_id(struct host1x_syncpt *sp);
 
diff --git a/include/trace/events/host1x.h b/include/trace/events/host1x.h
index 4bd2b0253d27..94db6a2c3540 100644
--- a/include/trace/events/host1x.h
+++ b/include/trace/events/host1x.h
@@ -37,6 +37,171 @@ DECLARE_EVENT_CLASS(host1x,
 	TP_printk("name=%s", __entry->name)
 );
 
+DEFINE_EVENT(host1x, host1x_channel_open,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
+DEFINE_EVENT(host1x, host1x_channel_release,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
+DEFINE_EVENT(host1x, host1x_cdma_begin,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
+DEFINE_EVENT(host1x, host1x_cdma_end,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
+TRACE_EVENT(host1x_cdma_push,
+	TP_PROTO(const char *name, u32 op1, u32 op2),
+
+	TP_ARGS(name, op1, op2),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(u32, op1)
+		__field(u32, op2)
+	),
+
+	TP_fast_assign(
+		__entry->name = name;
+		__entry->op1 = op1;
+		__entry->op2 = op2;
+	),
+
+	TP_printk("name=%s, op1=%08x, op2=%08x",
+		__entry->name, __entry->op1, __entry->op2)
+);
+
+TRACE_EVENT(host1x_cdma_push_gather,
+	TP_PROTO(const char *name, u32 mem_id,
+			u32 words, u32 offset, void *cmdbuf),
+
+	TP_ARGS(name, mem_id, words, offset, cmdbuf),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(u32, mem_id)
+		__field(u32, words)
+		__field(u32, offset)
+		__field(bool, cmdbuf)
+		__dynamic_array(u32, cmdbuf, words)
+	),
+
+	TP_fast_assign(
+		if (cmdbuf) {
+			memcpy(__get_dynamic_array(cmdbuf), cmdbuf+offset,
+					words * sizeof(u32));
+		}
+		__entry->cmdbuf = cmdbuf;
+		__entry->name = name;
+		__entry->mem_id = mem_id;
+		__entry->words = words;
+		__entry->offset = offset;
+	),
+
+	TP_printk("name=%s, mem_id=%08x, words=%u, offset=%d, contents=[%s]",
+	  __entry->name, __entry->mem_id,
+	  __entry->words, __entry->offset,
+	  __print_hex(__get_dynamic_array(cmdbuf),
+		  __entry->cmdbuf ? __entry->words * 4 : 0))
+);
+
+TRACE_EVENT(host1x_channel_submit,
+	TP_PROTO(const char *name, u32 cmdbufs, u32 relocs, u32 waitchks,
+			u32 syncpt_id, u32 syncpt_incrs),
+
+	TP_ARGS(name, cmdbufs, relocs, waitchks, syncpt_id, syncpt_incrs),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(u32, cmdbufs)
+		__field(u32, relocs)
+		__field(u32, waitchks)
+		__field(u32, syncpt_id)
+		__field(u32, syncpt_incrs)
+	),
+
+	TP_fast_assign(
+		__entry->name = name;
+		__entry->cmdbufs = cmdbufs;
+		__entry->relocs = relocs;
+		__entry->waitchks = waitchks;
+		__entry->syncpt_id = syncpt_id;
+		__entry->syncpt_incrs = syncpt_incrs;
+	),
+
+	TP_printk("name=%s, cmdbufs=%u, relocs=%u, waitchks=%d,"
+		"syncpt_id=%u, syncpt_incrs=%u",
+	  __entry->name, __entry->cmdbufs, __entry->relocs, __entry->waitchks,
+	  __entry->syncpt_id, __entry->syncpt_incrs)
+);
+
+TRACE_EVENT(host1x_channel_submitted,
+	TP_PROTO(const char *name, u32 syncpt_base, u32 syncpt_max),
+
+	TP_ARGS(name, syncpt_base, syncpt_max),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(u32, syncpt_base)
+		__field(u32, syncpt_max)
+	),
+
+	TP_fast_assign(
+		__entry->name = name;
+		__entry->syncpt_base = syncpt_base;
+		__entry->syncpt_max = syncpt_max;
+	),
+
+	TP_printk("name=%s, syncpt_base=%d, syncpt_max=%d",
+		__entry->name, __entry->syncpt_base, __entry->syncpt_max)
+);
+
+TRACE_EVENT(host1x_channel_submit_complete,
+	TP_PROTO(const char *name, int count, u32 thresh),
+
+	TP_ARGS(name, count, thresh),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(int, count)
+		__field(u32, thresh)
+	),
+
+	TP_fast_assign(
+		__entry->name = name;
+		__entry->count = count;
+		__entry->thresh = thresh;
+	),
+
+	TP_printk("name=%s, count=%d, thresh=%d",
+		__entry->name, __entry->count, __entry->thresh)
+);
+
+TRACE_EVENT(host1x_wait_cdma,
+	TP_PROTO(const char *name, u32 eventid),
+
+	TP_ARGS(name, eventid),
+
+	TP_STRUCT__entry(
+		__field(const char *, name)
+		__field(u32, eventid)
+	),
+
+	TP_fast_assign(
+		__entry->name = name;
+		__entry->eventid = eventid;
+	),
+
+	TP_printk("name=%s, event=%d", __entry->name, __entry->eventid)
+);
+
 TRACE_EVENT(host1x_syncpt_load_min,
 	TP_PROTO(u32 id, u32 val),
 
@@ -55,6 +220,33 @@ TRACE_EVENT(host1x_syncpt_load_min,
 	TP_printk("id=%d, val=%d", __entry->id, __entry->val)
 );
 
+TRACE_EVENT(host1x_syncpt_wait_check,
+	TP_PROTO(void *mem_id, u32 offset, u32 syncpt_id, u32 thresh, u32 min),
+
+	TP_ARGS(mem_id, offset, syncpt_id, thresh, min),
+
+	TP_STRUCT__entry(
+		__field(void *, mem_id)
+		__field(u32, offset)
+		__field(u32, syncpt_id)
+		__field(u32, thresh)
+		__field(u32, min)
+	),
+
+	TP_fast_assign(
+		__entry->mem_id = mem_id;
+		__entry->offset = offset;
+		__entry->syncpt_id = syncpt_id;
+		__entry->thresh = thresh;
+		__entry->min = min;
+	),
+
+	TP_printk("mem_id=%p, offset=%05x, id=%d, thresh=%d, current=%d",
+		__entry->mem_id, __entry->offset,
+		__entry->syncpt_id, __entry->thresh,
+		__entry->min)
+);
+
 #endif /*  _TRACE_HOST1X_H */
 
 /* This part must be outside protection */
-- 
cgit v1.2.3


From d43f81cbaf43531a977e8b4c4427f19acf8a5061 Mon Sep 17 00:00:00 2001
From: Terje Bergstrom <tbergstrom@nvidia.com>
Date: Fri, 22 Mar 2013 16:34:09 +0200
Subject: drm/tegra: Add gr2d device

Add client driver for 2D device, and IOCTLs to pass work to host1x
channel for 2D.

Also adds functions that can be called to access sync points from
DRM.

Signed-off-by: Arto Merilainen <amerilainen@nvidia.com>
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Erik Faye-Lund <kusmabite@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
---
 drivers/gpu/host1x/Makefile    |   1 +
 drivers/gpu/host1x/dev.c       |   7 +
 drivers/gpu/host1x/drm/Kconfig |   8 +
 drivers/gpu/host1x/drm/drm.c   | 212 +++++++++++++++++++++++++-
 drivers/gpu/host1x/drm/drm.h   |  27 +++-
 drivers/gpu/host1x/drm/gr2d.c  | 339 +++++++++++++++++++++++++++++++++++++++++
 drivers/gpu/host1x/host1x.h    |   4 +-
 include/uapi/drm/Kbuild        |   1 +
 include/uapi/drm/tegra_drm.h   | 136 +++++++++++++++++
 9 files changed, 732 insertions(+), 3 deletions(-)
 create mode 100644 drivers/gpu/host1x/drm/gr2d.c
 create mode 100644 include/uapi/drm/tegra_drm.h

(limited to 'include')

diff --git a/drivers/gpu/host1x/Makefile b/drivers/gpu/host1x/Makefile
index 3768dbc27184..3b037b6e0298 100644
--- a/drivers/gpu/host1x/Makefile
+++ b/drivers/gpu/host1x/Makefile
@@ -16,4 +16,5 @@ ccflags-$(CONFIG_DRM_TEGRA_DEBUG) += -DDEBUG
 host1x-$(CONFIG_DRM_TEGRA) += drm/drm.o drm/fb.o drm/dc.o
 host1x-$(CONFIG_DRM_TEGRA) += drm/output.o drm/rgb.o drm/hdmi.o
 host1x-$(CONFIG_DRM_TEGRA) += drm/gem.o
+host1x-$(CONFIG_DRM_TEGRA) += drm/gr2d.o
 obj-$(CONFIG_TEGRA_HOST1X) += host1x.o
diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c
index 8ce9889cefd5..28e28a23d444 100644
--- a/drivers/gpu/host1x/dev.c
+++ b/drivers/gpu/host1x/dev.c
@@ -209,11 +209,17 @@ static int __init tegra_host1x_init(void)
 	err = platform_driver_register(&tegra_hdmi_driver);
 	if (err < 0)
 		goto unregister_dc;
+
+	err = platform_driver_register(&tegra_gr2d_driver);
+	if (err < 0)
+		goto unregister_hdmi;
 #endif
 
 	return 0;
 
 #ifdef CONFIG_DRM_TEGRA
+unregister_hdmi:
+	platform_driver_unregister(&tegra_hdmi_driver);
 unregister_dc:
 	platform_driver_unregister(&tegra_dc_driver);
 unregister_host1x:
@@ -226,6 +232,7 @@ module_init(tegra_host1x_init);
 static void __exit tegra_host1x_exit(void)
 {
 #ifdef CONFIG_DRM_TEGRA
+	platform_driver_unregister(&tegra_gr2d_driver);
 	platform_driver_unregister(&tegra_hdmi_driver);
 	platform_driver_unregister(&tegra_dc_driver);
 #endif
diff --git a/drivers/gpu/host1x/drm/Kconfig b/drivers/gpu/host1x/drm/Kconfig
index 9a28901d07ce..aa9d0768d6eb 100644
--- a/drivers/gpu/host1x/drm/Kconfig
+++ b/drivers/gpu/host1x/drm/Kconfig
@@ -13,6 +13,14 @@ config DRM_TEGRA
 
 if DRM_TEGRA
 
+config DRM_TEGRA_STAGING
+	bool "Enable HOST1X interface"
+	depends on STAGING
+	help
+	  Say yes if HOST1X should be available for userspace DRM users.
+
+	  If unsure, choose N.
+
 config DRM_TEGRA_DEBUG
 	bool "NVIDIA Tegra DRM debug support"
 	help
diff --git a/drivers/gpu/host1x/drm/drm.c b/drivers/gpu/host1x/drm/drm.c
index c4e45c16f991..2b561c9118c6 100644
--- a/drivers/gpu/host1x/drm/drm.c
+++ b/drivers/gpu/host1x/drm/drm.c
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2012 Avionic Design GmbH
- * Copyright (C) 2012 NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (C) 2012-2013 NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 as
@@ -14,6 +14,9 @@
 #include <linux/dma-mapping.h>
 #include <asm/dma-iommu.h>
 
+#include <drm/drm.h>
+#include <drm/drmP.h>
+
 #include "host1x_client.h"
 #include "dev.h"
 #include "drm.h"
@@ -81,8 +84,10 @@ static int host1x_parse_dt(struct host1x_drm *host1x)
 	static const char * const compat[] = {
 		"nvidia,tegra20-dc",
 		"nvidia,tegra20-hdmi",
+		"nvidia,tegra20-gr2d",
 		"nvidia,tegra30-dc",
 		"nvidia,tegra30-hdmi",
+		"nvidia,tegra30-gr2d",
 	};
 	unsigned int i;
 	int err;
@@ -277,9 +282,24 @@ static int tegra_drm_unload(struct drm_device *drm)
 
 static int tegra_drm_open(struct drm_device *drm, struct drm_file *filp)
 {
+	struct host1x_drm_file *fpriv;
+
+	fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL);
+	if (!fpriv)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&fpriv->contexts);
+	filp->driver_priv = fpriv;
+
 	return 0;
 }
 
+static void host1x_drm_context_free(struct host1x_drm_context *context)
+{
+	context->client->ops->close_channel(context);
+	kfree(context);
+}
+
 static void tegra_drm_lastclose(struct drm_device *drm)
 {
 	struct host1x_drm *host1x = drm->dev_private;
@@ -287,7 +307,190 @@ static void tegra_drm_lastclose(struct drm_device *drm)
 	tegra_fbdev_restore_mode(host1x->fbdev);
 }
 
+#ifdef CONFIG_DRM_TEGRA_STAGING
+static bool host1x_drm_file_owns_context(struct host1x_drm_file *file,
+					 struct host1x_drm_context *context)
+{
+	struct host1x_drm_context *ctx;
+
+	list_for_each_entry(ctx, &file->contexts, list)
+		if (ctx == context)
+			return true;
+
+	return false;
+}
+
+static int tegra_gem_create(struct drm_device *drm, void *data,
+			    struct drm_file *file)
+{
+	struct drm_tegra_gem_create *args = data;
+	struct tegra_bo *bo;
+
+	bo = tegra_bo_create_with_handle(file, drm, args->size,
+					 &args->handle);
+	if (IS_ERR(bo))
+		return PTR_ERR(bo);
+
+	return 0;
+}
+
+static int tegra_gem_mmap(struct drm_device *drm, void *data,
+			  struct drm_file *file)
+{
+	struct drm_tegra_gem_mmap *args = data;
+	struct drm_gem_object *gem;
+	struct tegra_bo *bo;
+
+	gem = drm_gem_object_lookup(drm, file, args->handle);
+	if (!gem)
+		return -EINVAL;
+
+	bo = to_tegra_bo(gem);
+
+	args->offset = tegra_bo_get_mmap_offset(bo);
+
+	drm_gem_object_unreference(gem);
+
+	return 0;
+}
+
+static int tegra_syncpt_read(struct drm_device *drm, void *data,
+			     struct drm_file *file)
+{
+	struct drm_tegra_syncpt_read *args = data;
+	struct host1x *host = dev_get_drvdata(drm->dev);
+	struct host1x_syncpt *sp = host1x_syncpt_get(host, args->id);
+
+	if (!sp)
+		return -EINVAL;
+
+	args->value = host1x_syncpt_read_min(sp);
+	return 0;
+}
+
+static int tegra_syncpt_incr(struct drm_device *drm, void *data,
+			     struct drm_file *file)
+{
+	struct drm_tegra_syncpt_incr *args = data;
+	struct host1x *host = dev_get_drvdata(drm->dev);
+	struct host1x_syncpt *sp = host1x_syncpt_get(host, args->id);
+
+	if (!sp)
+		return -EINVAL;
+
+	host1x_syncpt_incr(sp);
+	return 0;
+}
+
+static int tegra_syncpt_wait(struct drm_device *drm, void *data,
+			     struct drm_file *file)
+{
+	struct drm_tegra_syncpt_wait *args = data;
+	struct host1x *host = dev_get_drvdata(drm->dev);
+	struct host1x_syncpt *sp = host1x_syncpt_get(host, args->id);
+
+	if (!sp)
+		return -EINVAL;
+
+	return host1x_syncpt_wait(sp, args->thresh, args->timeout,
+				  &args->value);
+}
+
+static int tegra_open_channel(struct drm_device *drm, void *data,
+			      struct drm_file *file)
+{
+	struct drm_tegra_open_channel *args = data;
+	struct host1x_client *client;
+	struct host1x_drm_context *context;
+	struct host1x_drm_file *fpriv = file->driver_priv;
+	struct host1x_drm *host1x = drm->dev_private;
+	int err = -ENODEV;
+
+	context = kzalloc(sizeof(*context), GFP_KERNEL);
+	if (!context)
+		return -ENOMEM;
+
+	list_for_each_entry(client, &host1x->clients, list)
+		if (client->class == args->client) {
+			err = client->ops->open_channel(client, context);
+			if (err)
+				break;
+
+			context->client = client;
+			list_add(&context->list, &fpriv->contexts);
+			args->context = (uintptr_t)context;
+			return 0;
+		}
+
+	kfree(context);
+	return err;
+}
+
+static int tegra_close_channel(struct drm_device *drm, void *data,
+			       struct drm_file *file)
+{
+	struct drm_tegra_close_channel *args = data;
+	struct host1x_drm_file *fpriv = file->driver_priv;
+	struct host1x_drm_context *context =
+		(struct host1x_drm_context *)(uintptr_t)args->context;
+
+	if (!host1x_drm_file_owns_context(fpriv, context))
+		return -EINVAL;
+
+	list_del(&context->list);
+	host1x_drm_context_free(context);
+
+	return 0;
+}
+
+static int tegra_get_syncpt(struct drm_device *drm, void *data,
+			    struct drm_file *file)
+{
+	struct drm_tegra_get_syncpt *args = data;
+	struct host1x_drm_file *fpriv = file->driver_priv;
+	struct host1x_drm_context *context =
+		(struct host1x_drm_context *)(uintptr_t)args->context;
+	struct host1x_syncpt *syncpt;
+
+	if (!host1x_drm_file_owns_context(fpriv, context))
+		return -ENODEV;
+
+	if (args->index >= context->client->num_syncpts)
+		return -EINVAL;
+
+	syncpt = context->client->syncpts[args->index];
+	args->id = host1x_syncpt_id(syncpt);
+
+	return 0;
+}
+
+static int tegra_submit(struct drm_device *drm, void *data,
+			struct drm_file *file)
+{
+	struct drm_tegra_submit *args = data;
+	struct host1x_drm_file *fpriv = file->driver_priv;
+	struct host1x_drm_context *context =
+		(struct host1x_drm_context *)(uintptr_t)args->context;
+
+	if (!host1x_drm_file_owns_context(fpriv, context))
+		return -ENODEV;
+
+	return context->client->ops->submit(context, args, drm, file);
+}
+#endif
+
 static struct drm_ioctl_desc tegra_drm_ioctls[] = {
+#ifdef CONFIG_DRM_TEGRA_STAGING
+	DRM_IOCTL_DEF_DRV(TEGRA_GEM_CREATE, tegra_gem_create, DRM_UNLOCKED | DRM_AUTH),
+	DRM_IOCTL_DEF_DRV(TEGRA_GEM_MMAP, tegra_gem_mmap, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_SYNCPT_READ, tegra_syncpt_read, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_SYNCPT_INCR, tegra_syncpt_incr, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_SYNCPT_WAIT, tegra_syncpt_wait, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_OPEN_CHANNEL, tegra_open_channel, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_CLOSE_CHANNEL, tegra_close_channel, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_GET_SYNCPT, tegra_get_syncpt, DRM_UNLOCKED),
+	DRM_IOCTL_DEF_DRV(TEGRA_SUBMIT, tegra_submit, DRM_UNLOCKED),
+#endif
 };
 
 static const struct file_operations tegra_drm_fops = {
@@ -349,10 +552,17 @@ static void tegra_drm_disable_vblank(struct drm_device *drm, int pipe)
 
 static void tegra_drm_preclose(struct drm_device *drm, struct drm_file *file)
 {
+	struct host1x_drm_file *fpriv = file->driver_priv;
+	struct host1x_drm_context *context, *tmp;
 	struct drm_crtc *crtc;
 
 	list_for_each_entry(crtc, &drm->mode_config.crtc_list, head)
 		tegra_dc_cancel_page_flip(crtc, file);
+
+	list_for_each_entry_safe(context, tmp, &fpriv->contexts, list)
+		host1x_drm_context_free(context);
+
+	kfree(fpriv);
 }
 
 #ifdef CONFIG_DEBUG_FS
diff --git a/drivers/gpu/host1x/drm/drm.h b/drivers/gpu/host1x/drm/drm.h
index 3864a39f8ad3..02ce020f2575 100644
--- a/drivers/gpu/host1x/drm/drm.h
+++ b/drivers/gpu/host1x/drm/drm.h
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2012 Avionic Design GmbH
- * Copyright (C) 2012 NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (C) 2012-2013 NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 as
@@ -15,6 +15,9 @@
 #include <drm/drm_edid.h>
 #include <drm/drm_fb_helper.h>
 #include <drm/drm_fixed.h>
+#include <uapi/drm/tegra_drm.h>
+
+#include "host1x.h"
 
 struct tegra_fb {
 	struct drm_framebuffer base;
@@ -47,9 +50,25 @@ struct host1x_drm {
 
 struct host1x_client;
 
+struct host1x_drm_context {
+	struct host1x_client *client;
+	struct host1x_channel *channel;
+	struct list_head list;
+};
+
 struct host1x_client_ops {
 	int (*drm_init)(struct host1x_client *client, struct drm_device *drm);
 	int (*drm_exit)(struct host1x_client *client);
+	int (*open_channel)(struct host1x_client *client,
+			    struct host1x_drm_context *context);
+	void (*close_channel)(struct host1x_drm_context *context);
+	int (*submit)(struct host1x_drm_context *context,
+		      struct drm_tegra_submit *args, struct drm_device *drm,
+		      struct drm_file *file);
+};
+
+struct host1x_drm_file {
+	struct list_head contexts;
 };
 
 struct host1x_client {
@@ -58,6 +77,12 @@ struct host1x_client {
 
 	const struct host1x_client_ops *ops;
 
+	enum host1x_class class;
+	struct host1x_channel *channel;
+
+	struct host1x_syncpt **syncpts;
+	unsigned int num_syncpts;
+
 	struct list_head list;
 };
 
diff --git a/drivers/gpu/host1x/drm/gr2d.c b/drivers/gpu/host1x/drm/gr2d.c
new file mode 100644
index 000000000000..6a45ae090ee7
--- /dev/null
+++ b/drivers/gpu/host1x/drm/gr2d.c
@@ -0,0 +1,339 @@
+/*
+ * drivers/video/tegra/host/gr2d/gr2d.c
+ *
+ * Tegra Graphics 2D
+ *
+ * Copyright (c) 2012-2013, NVIDIA Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/export.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/clk.h>
+
+#include "channel.h"
+#include "drm.h"
+#include "gem.h"
+#include "job.h"
+#include "host1x.h"
+#include "host1x_bo.h"
+#include "host1x_client.h"
+#include "syncpt.h"
+
+struct gr2d {
+	struct host1x_client client;
+	struct clk *clk;
+	struct host1x_channel *channel;
+	unsigned long *addr_regs;
+};
+
+static inline struct gr2d *to_gr2d(struct host1x_client *client)
+{
+	return container_of(client, struct gr2d, client);
+}
+
+static int gr2d_is_addr_reg(struct device *dev, u32 class, u32 reg);
+
+static int gr2d_client_init(struct host1x_client *client,
+			    struct drm_device *drm)
+{
+	return 0;
+}
+
+static int gr2d_client_exit(struct host1x_client *client)
+{
+	return 0;
+}
+
+static int gr2d_open_channel(struct host1x_client *client,
+			     struct host1x_drm_context *context)
+{
+	struct gr2d *gr2d = to_gr2d(client);
+
+	context->channel = host1x_channel_get(gr2d->channel);
+
+	if (!context->channel)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void gr2d_close_channel(struct host1x_drm_context *context)
+{
+	host1x_channel_put(context->channel);
+}
+
+static struct host1x_bo *host1x_bo_lookup(struct drm_device *drm,
+					  struct drm_file *file,
+					  u32 handle)
+{
+	struct drm_gem_object *gem;
+	struct tegra_bo *bo;
+
+	gem = drm_gem_object_lookup(drm, file, handle);
+	if (!gem)
+		return 0;
+
+	mutex_lock(&drm->struct_mutex);
+	drm_gem_object_unreference(gem);
+	mutex_unlock(&drm->struct_mutex);
+
+	bo = to_tegra_bo(gem);
+	return &bo->base;
+}
+
+static int gr2d_submit(struct host1x_drm_context *context,
+		       struct drm_tegra_submit *args, struct drm_device *drm,
+		       struct drm_file *file)
+{
+	struct host1x_job *job;
+	unsigned int num_cmdbufs = args->num_cmdbufs;
+	unsigned int num_relocs = args->num_relocs;
+	unsigned int num_waitchks = args->num_waitchks;
+	struct drm_tegra_cmdbuf __user *cmdbufs =
+		(void * __user)(uintptr_t)args->cmdbufs;
+	struct drm_tegra_reloc __user *relocs =
+		(void * __user)(uintptr_t)args->relocs;
+	struct drm_tegra_waitchk __user *waitchks =
+		(void * __user)(uintptr_t)args->waitchks;
+	struct drm_tegra_syncpt syncpt;
+	int err;
+
+	/* We don't yet support other than one syncpt_incr struct per submit */
+	if (args->num_syncpts != 1)
+		return -EINVAL;
+
+	job = host1x_job_alloc(context->channel, args->num_cmdbufs,
+			       args->num_relocs, args->num_waitchks);
+	if (!job)
+		return -ENOMEM;
+
+	job->num_relocs = args->num_relocs;
+	job->num_waitchk = args->num_waitchks;
+	job->client = (u32)args->context;
+	job->class = context->client->class;
+	job->serialize = true;
+
+	while (num_cmdbufs) {
+		struct drm_tegra_cmdbuf cmdbuf;
+		struct host1x_bo *bo;
+
+		err = copy_from_user(&cmdbuf, cmdbufs, sizeof(cmdbuf));
+		if (err)
+			goto fail;
+
+		bo = host1x_bo_lookup(drm, file, cmdbuf.handle);
+		if (!bo)
+			goto fail;
+
+		host1x_job_add_gather(job, bo, cmdbuf.words, cmdbuf.offset);
+		num_cmdbufs--;
+		cmdbufs++;
+	}
+
+	err = copy_from_user(job->relocarray, relocs,
+			     sizeof(*relocs) * num_relocs);
+	if (err)
+		goto fail;
+
+	while (num_relocs--) {
+		struct host1x_reloc *reloc = &job->relocarray[num_relocs];
+		struct host1x_bo *cmdbuf, *target;
+
+		cmdbuf = host1x_bo_lookup(drm, file, (u32)reloc->cmdbuf);
+		target = host1x_bo_lookup(drm, file, (u32)reloc->target);
+
+		reloc->cmdbuf = cmdbuf;
+		reloc->target = target;
+
+		if (!reloc->target || !reloc->cmdbuf)
+			goto fail;
+	}
+
+	err = copy_from_user(job->waitchk, waitchks,
+			     sizeof(*waitchks) * num_waitchks);
+	if (err)
+		goto fail;
+
+	err = copy_from_user(&syncpt, (void * __user)(uintptr_t)args->syncpts,
+			     sizeof(syncpt));
+	if (err)
+		goto fail;
+
+	job->syncpt_id = syncpt.id;
+	job->syncpt_incrs = syncpt.incrs;
+	job->timeout = 10000;
+	job->is_addr_reg = gr2d_is_addr_reg;
+
+	if (args->timeout && args->timeout < 10000)
+		job->timeout = args->timeout;
+
+	err = host1x_job_pin(job, context->client->dev);
+	if (err)
+		goto fail;
+
+	err = host1x_job_submit(job);
+	if (err)
+		goto fail_submit;
+
+	args->fence = job->syncpt_end;
+
+	host1x_job_put(job);
+	return 0;
+
+fail_submit:
+	host1x_job_unpin(job);
+fail:
+	host1x_job_put(job);
+	return err;
+}
+
+static struct host1x_client_ops gr2d_client_ops = {
+	.drm_init = gr2d_client_init,
+	.drm_exit = gr2d_client_exit,
+	.open_channel = gr2d_open_channel,
+	.close_channel = gr2d_close_channel,
+	.submit = gr2d_submit,
+};
+
+static void gr2d_init_addr_reg_map(struct device *dev, struct gr2d *gr2d)
+{
+	const u32 gr2d_addr_regs[] = {0x1a, 0x1b, 0x26, 0x2b, 0x2c, 0x2d, 0x31,
+				      0x32, 0x48, 0x49, 0x4a, 0x4b, 0x4c};
+	unsigned long *bitmap;
+	int i;
+
+	bitmap = devm_kzalloc(dev, DIV_ROUND_UP(256, BITS_PER_BYTE),
+			      GFP_KERNEL);
+
+	for (i = 0; i < ARRAY_SIZE(gr2d_addr_regs); ++i) {
+		u32 reg = gr2d_addr_regs[i];
+		bitmap[BIT_WORD(reg)] |= BIT_MASK(reg);
+	}
+
+	gr2d->addr_regs = bitmap;
+}
+
+static int gr2d_is_addr_reg(struct device *dev, u32 class, u32 reg)
+{
+	struct gr2d *gr2d = dev_get_drvdata(dev);
+
+	switch (class) {
+	case HOST1X_CLASS_HOST1X:
+		return reg == 0x2b;
+	case HOST1X_CLASS_GR2D:
+	case HOST1X_CLASS_GR2D_SB:
+		reg &= 0xff;
+		if (gr2d->addr_regs[BIT_WORD(reg)] & BIT_MASK(reg))
+			return 1;
+	default:
+		return 0;
+	}
+}
+
+static const struct of_device_id gr2d_match[] = {
+	{ .compatible = "nvidia,tegra30-gr2d" },
+	{ .compatible = "nvidia,tegra20-gr2d" },
+	{ },
+};
+
+static int gr2d_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct host1x_drm *host1x = host1x_get_drm_data(dev->parent);
+	int err;
+	struct gr2d *gr2d = NULL;
+	struct host1x_syncpt **syncpts;
+
+	gr2d = devm_kzalloc(dev, sizeof(*gr2d), GFP_KERNEL);
+	if (!gr2d)
+		return -ENOMEM;
+
+	syncpts = devm_kzalloc(dev, sizeof(*syncpts), GFP_KERNEL);
+	if (!syncpts)
+		return -ENOMEM;
+
+	gr2d->clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(gr2d->clk)) {
+		dev_err(dev, "cannot get clock\n");
+		return PTR_ERR(gr2d->clk);
+	}
+
+	err = clk_prepare_enable(gr2d->clk);
+	if (err) {
+		dev_err(dev, "cannot turn on clock\n");
+		return err;
+	}
+
+	gr2d->channel = host1x_channel_request(dev);
+	if (!gr2d->channel)
+		return -ENOMEM;
+
+	*syncpts = host1x_syncpt_request(dev, 0);
+	if (!(*syncpts)) {
+		host1x_channel_free(gr2d->channel);
+		return -ENOMEM;
+	}
+
+	gr2d->client.ops = &gr2d_client_ops;
+	gr2d->client.dev = dev;
+	gr2d->client.class = HOST1X_CLASS_GR2D;
+	gr2d->client.syncpts = syncpts;
+	gr2d->client.num_syncpts = 1;
+
+	err = host1x_register_client(host1x, &gr2d->client);
+	if (err < 0) {
+		dev_err(dev, "failed to register host1x client: %d\n", err);
+		return err;
+	}
+
+	gr2d_init_addr_reg_map(dev, gr2d);
+
+	platform_set_drvdata(pdev, gr2d);
+
+	return 0;
+}
+
+static int __exit gr2d_remove(struct platform_device *pdev)
+{
+	struct host1x_drm *host1x = host1x_get_drm_data(pdev->dev.parent);
+	struct gr2d *gr2d = platform_get_drvdata(pdev);
+	unsigned int i;
+	int err;
+
+	err = host1x_unregister_client(host1x, &gr2d->client);
+	if (err < 0) {
+		dev_err(&pdev->dev, "failed to unregister client: %d\n", err);
+		return err;
+	}
+
+	for (i = 0; i < gr2d->client.num_syncpts; i++)
+		host1x_syncpt_free(gr2d->client.syncpts[i]);
+
+	host1x_channel_free(gr2d->channel);
+	clk_disable_unprepare(gr2d->clk);
+
+	return 0;
+}
+
+struct platform_driver tegra_gr2d_driver = {
+	.probe = gr2d_probe,
+	.remove = __exit_p(gr2d_remove),
+	.driver = {
+		.owner = THIS_MODULE,
+		.name = "gr2d",
+		.of_match_table = gr2d_match,
+	}
+};
diff --git a/drivers/gpu/host1x/host1x.h b/drivers/gpu/host1x/host1x.h
index bca6563f6766..a2bc1e65e972 100644
--- a/drivers/gpu/host1x/host1x.h
+++ b/drivers/gpu/host1x/host1x.h
@@ -22,7 +22,9 @@
 #define __LINUX_HOST1X_H
 
 enum host1x_class {
-	HOST1X_CLASS_HOST1X	= 0x1
+	HOST1X_CLASS_HOST1X	= 0x1,
+	HOST1X_CLASS_GR2D	= 0x51,
+	HOST1X_CLASS_GR2D_SB    = 0x52
 };
 
 #endif
diff --git a/include/uapi/drm/Kbuild b/include/uapi/drm/Kbuild
index a042a957296d..119487e05e65 100644
--- a/include/uapi/drm/Kbuild
+++ b/include/uapi/drm/Kbuild
@@ -13,5 +13,6 @@ header-y += r128_drm.h
 header-y += radeon_drm.h
 header-y += savage_drm.h
 header-y += sis_drm.h
+header-y += tegra_drm.h
 header-y += via_drm.h
 header-y += vmwgfx_drm.h
diff --git a/include/uapi/drm/tegra_drm.h b/include/uapi/drm/tegra_drm.h
new file mode 100644
index 000000000000..6e132a2f7420
--- /dev/null
+++ b/include/uapi/drm/tegra_drm.h
@@ -0,0 +1,136 @@
+/*
+ * Copyright (c) 2012-2013, NVIDIA CORPORATION.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _UAPI_TEGRA_DRM_H_
+#define _UAPI_TEGRA_DRM_H_
+
+struct drm_tegra_gem_create {
+	__u64 size;
+	__u32 flags;
+	__u32 handle;
+};
+
+struct drm_tegra_gem_mmap {
+	__u32 handle;
+	__u32 offset;
+};
+
+struct drm_tegra_syncpt_read {
+	__u32 id;
+	__u32 value;
+};
+
+struct drm_tegra_syncpt_incr {
+	__u32 id;
+	__u32 pad;
+};
+
+struct drm_tegra_syncpt_wait {
+	__u32 id;
+	__u32 thresh;
+	__u32 timeout;
+	__u32 value;
+};
+
+#define DRM_TEGRA_NO_TIMEOUT	(0xffffffff)
+
+struct drm_tegra_open_channel {
+	__u32 client;
+	__u32 pad;
+	__u64 context;
+};
+
+struct drm_tegra_close_channel {
+	__u64 context;
+};
+
+struct drm_tegra_get_syncpt {
+	__u64 context;
+	__u32 index;
+	__u32 id;
+};
+
+struct drm_tegra_syncpt {
+	__u32 id;
+	__u32 incrs;
+};
+
+struct drm_tegra_cmdbuf {
+	__u32 handle;
+	__u32 offset;
+	__u32 words;
+	__u32 pad;
+};
+
+struct drm_tegra_reloc {
+	struct {
+		__u32 handle;
+		__u32 offset;
+	} cmdbuf;
+	struct {
+		__u32 handle;
+		__u32 offset;
+	} target;
+	__u32 shift;
+	__u32 pad;
+};
+
+struct drm_tegra_waitchk {
+	__u32 handle;
+	__u32 offset;
+	__u32 syncpt;
+	__u32 thresh;
+};
+
+struct drm_tegra_submit {
+	__u64 context;
+	__u32 num_syncpts;
+	__u32 num_cmdbufs;
+	__u32 num_relocs;
+	__u32 num_waitchks;
+	__u32 waitchk_mask;
+	__u32 timeout;
+	__u32 pad;
+	__u64 syncpts;
+	__u64 cmdbufs;
+	__u64 relocs;
+	__u64 waitchks;
+	__u32 fence;		/* Return value */
+
+	__u32 reserved[5];	/* future expansion */
+};
+
+#define DRM_TEGRA_GEM_CREATE	0x00
+#define DRM_TEGRA_GEM_MMAP	0x01
+#define DRM_TEGRA_SYNCPT_READ	0x02
+#define DRM_TEGRA_SYNCPT_INCR	0x03
+#define DRM_TEGRA_SYNCPT_WAIT	0x04
+#define DRM_TEGRA_OPEN_CHANNEL	0x05
+#define DRM_TEGRA_CLOSE_CHANNEL	0x06
+#define DRM_TEGRA_GET_SYNCPT	0x07
+#define DRM_TEGRA_SUBMIT	0x08
+
+#define DRM_IOCTL_TEGRA_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_GEM_CREATE, struct drm_tegra_gem_create)
+#define DRM_IOCTL_TEGRA_GEM_MMAP DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_GEM_MMAP, struct drm_tegra_gem_mmap)
+#define DRM_IOCTL_TEGRA_SYNCPT_READ DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_SYNCPT_READ, struct drm_tegra_syncpt_read)
+#define DRM_IOCTL_TEGRA_SYNCPT_INCR DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_SYNCPT_INCR, struct drm_tegra_syncpt_incr)
+#define DRM_IOCTL_TEGRA_SYNCPT_WAIT DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_SYNCPT_WAIT, struct drm_tegra_syncpt_wait)
+#define DRM_IOCTL_TEGRA_OPEN_CHANNEL DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_OPEN_CHANNEL, struct drm_tegra_open_channel)
+#define DRM_IOCTL_TEGRA_CLOSE_CHANNEL DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_CLOSE_CHANNEL, struct drm_tegra_open_channel)
+#define DRM_IOCTL_TEGRA_GET_SYNCPT DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_GET_SYNCPT, struct drm_tegra_get_syncpt)
+#define DRM_IOCTL_TEGRA_SUBMIT DRM_IOWR(DRM_COMMAND_BASE + DRM_TEGRA_SUBMIT, struct drm_tegra_submit)
+
+#endif
-- 
cgit v1.2.3


From fe214163cc6e6b33253b7ac14bc3dd616e76b513 Mon Sep 17 00:00:00 2001
From: Rafał Miłecki <zajec5@gmail.com>
Date: Fri, 19 Apr 2013 19:01:25 +0200
Subject: drm: add drm_edid_to_eld helper extracting SADs from EDID (v2)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Some devices (ATI/AMD cards) don't support passing ELD struct to the
hardware but just require filling specific registers and then the
hardware/firmware does the rest. In such cases we need to read the info
from SAD blocks and put them in the correct registers.

agd5f: note that the returned pointer needs to be kfreed as per
Christian's suggestion.

v2: fix warning

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/drm_edid.c | 59 ++++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  9 +++++++
 2 files changed, 68 insertions(+)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c
index e2acfdbf7d3c..b6807e881a60 100644
--- a/drivers/gpu/drm/drm_edid.c
+++ b/drivers/gpu/drm/drm_edid.c
@@ -2510,6 +2510,65 @@ void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid)
 }
 EXPORT_SYMBOL(drm_edid_to_eld);
 
+/**
+ * drm_edid_to_sad - extracts SADs from EDID
+ * @edid: EDID to parse
+ * @sads: pointer that will be set to the extracted SADs
+ *
+ * Looks for CEA EDID block and extracts SADs (Short Audio Descriptors) from it.
+ * Note: returned pointer needs to be kfreed
+ *
+ * Return number of found SADs or negative number on error.
+ */
+int drm_edid_to_sad(struct edid *edid, struct cea_sad **sads)
+{
+	int count = 0;
+	int i, start, end, dbl;
+	u8 *cea;
+
+	cea = drm_find_cea_extension(edid);
+	if (!cea) {
+		DRM_DEBUG_KMS("SAD: no CEA Extension found\n");
+		return -ENOENT;
+	}
+
+	if (cea_revision(cea) < 3) {
+		DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+		return -ENOTSUPP;
+	}
+
+	if (cea_db_offsets(cea, &start, &end)) {
+		DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+		return -EPROTO;
+	}
+
+	for_each_cea_db(cea, i, start, end) {
+		u8 *db = &cea[i];
+
+		if (cea_db_tag(db) == AUDIO_BLOCK) {
+			int j;
+			dbl = cea_db_payload_len(db);
+
+			count = dbl / 3; /* SAD is 3B */
+			*sads = kcalloc(count, sizeof(**sads), GFP_KERNEL);
+			if (!*sads)
+				return -ENOMEM;
+			for (j = 0; j < count; j++) {
+				u8 *sad = &db[1 + j * 3];
+
+				(*sads)[j].format = (sad[0] & 0x78) >> 3;
+				(*sads)[j].channels = sad[0] & 0x7;
+				(*sads)[j].freq = sad[1] & 0x7F;
+				(*sads)[j].byte2 = sad[2];
+			}
+			break;
+		}
+	}
+
+	return count;
+}
+EXPORT_SYMBOL(drm_edid_to_sad);
+
 /**
  * drm_av_sync_delay - HDMI/DP sink audio-video sync delay in millisecond
  * @connector: connector associated with the HDMI/DP sink
diff --git a/include/drm/drm_edid.h b/include/drm/drm_edid.h
index 5da1b4ae7d84..fc481fc17085 100644
--- a/include/drm/drm_edid.h
+++ b/include/drm/drm_edid.h
@@ -244,12 +244,21 @@ struct edid {
 
 #define EDID_PRODUCT_ID(e) ((e)->prod_code[0] | ((e)->prod_code[1] << 8))
 
+/* Short Audio Descriptor */
+struct cea_sad {
+	u8 format;
+	u8 channels; /* max number of channels - 1 */
+	u8 freq;
+	u8 byte2; /* meaning depends on format */
+};
+
 struct drm_encoder;
 struct drm_connector;
 struct drm_display_mode;
 struct hdmi_avi_infoframe;
 
 void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid);
+int drm_edid_to_sad(struct edid *edid, struct cea_sad **sads);
 int drm_av_sync_delay(struct drm_connector *connector,
 		      struct drm_display_mode *mode);
 struct drm_connector *drm_select_eld(struct drm_encoder *encoder,
-- 
cgit v1.2.3


From 67af116e4ac0ebbb3d7849a1d3e2344e289dd76d Mon Sep 17 00:00:00 2001
From: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Wed, 24 Apr 2013 19:07:15 +0300
Subject: drm: Remove explicit vrefresh initialization from DRM_MODE()
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

No need to zero initialize .vrefresh in DRM_MODE() since it's using
desgignated initializers.

This will also avoid some duplicate initialization warnings later.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 include/drm/drm_crtc.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index 8c7846bd74f4..b85575bb82e6 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -120,7 +120,7 @@ enum drm_mode_status {
 	.hdisplay = (hd), .hsync_start = (hss), .hsync_end = (hse), \
 	.htotal = (ht), .hskew = (hsk), .vdisplay = (vd), \
 	.vsync_start = (vss), .vsync_end = (vse), .vtotal = (vt), \
-	.vscan = (vs), .flags = (f), .vrefresh = 0, \
+	.vscan = (vs), .flags = (f), \
 	.base.type = DRM_MODE_OBJECT_MODE
 
 #define CRTC_INTERLACE_HALVE_V 0x1 /* halve V values for interlacing */
-- 
cgit v1.2.3


From 8cc3f23caf4066880f54f636beb3e30b4d383eca Mon Sep 17 00:00:00 2001
From: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Wed, 24 Apr 2013 19:07:16 +0300
Subject: drm: Add drm_mode_equal_no_clocks()
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

drm_mode_equal_no_clocks() is like drm_mode_equal() except it doesn't
compare the clock or vrefresh values. drm_mode_equal() is now
implemented by first doing the clock checks, and then calling
drm_mode_equal_no_clocks().

v2: Add missing EXPORT_SYMBOL()

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_modes.c | 22 +++++++++++++++++++++-
 include/drm/drm_crtc.h      |  1 +
 2 files changed, 22 insertions(+), 1 deletion(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c
index 04fa6f1808d1..f264d08062f0 100644
--- a/drivers/gpu/drm/drm_modes.c
+++ b/drivers/gpu/drm/drm_modes.c
@@ -848,6 +848,26 @@ bool drm_mode_equal(const struct drm_display_mode *mode1, const struct drm_displ
 	} else if (mode1->clock != mode2->clock)
 		return false;
 
+	return drm_mode_equal_no_clocks(mode1, mode2);
+}
+EXPORT_SYMBOL(drm_mode_equal);
+
+/**
+ * drm_mode_equal_no_clocks - test modes for equality
+ * @mode1: first mode
+ * @mode2: second mode
+ *
+ * LOCKING:
+ * None.
+ *
+ * Check to see if @mode1 and @mode2 are equivalent, but
+ * don't check the pixel clocks.
+ *
+ * RETURNS:
+ * True if the modes are equal, false otherwise.
+ */
+bool drm_mode_equal_no_clocks(const struct drm_display_mode *mode1, const struct drm_display_mode *mode2)
+{
 	if (mode1->hdisplay == mode2->hdisplay &&
 	    mode1->hsync_start == mode2->hsync_start &&
 	    mode1->hsync_end == mode2->hsync_end &&
@@ -863,7 +883,7 @@ bool drm_mode_equal(const struct drm_display_mode *mode1, const struct drm_displ
 
 	return false;
 }
-EXPORT_SYMBOL(drm_mode_equal);
+EXPORT_SYMBOL(drm_mode_equal_no_clocks);
 
 /**
  * drm_mode_validate_size - make sure modes adhere to size constraints
diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index b85575bb82e6..836438dfb331 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -922,6 +922,7 @@ extern void drm_mode_config_reset(struct drm_device *dev);
 extern void drm_mode_config_cleanup(struct drm_device *dev);
 extern void drm_mode_set_name(struct drm_display_mode *mode);
 extern bool drm_mode_equal(const struct drm_display_mode *mode1, const struct drm_display_mode *mode2);
+extern bool drm_mode_equal_no_clocks(const struct drm_display_mode *mode1, const struct drm_display_mode *mode2);
 extern int drm_mode_width(const struct drm_display_mode *mode);
 extern int drm_mode_height(const struct drm_display_mode *mode);
 
-- 
cgit v1.2.3


From ea9cbb063ce7509d98febdc756e77d6b69a10b56 Mon Sep 17 00:00:00 2001
From: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Thu, 25 Apr 2013 20:09:20 +0300
Subject: drm: Silence some sparse warnings
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

drivers/gpu/drm/drm_pci.c:155:5: warning: symbol 'drm_pci_set_busid' was not declared. Should it be static?
drivers/gpu/drm/drm_pci.c:197:5: warning: symbol 'drm_pci_set_unique' was not declared. Should it be static?
drivers/gpu/drm/drm_pci.c:269:5: warning: symbol 'drm_pci_agp_init' was not declared. Should it be static?

drivers/gpu/drm/drm_crtc.c:181:1: warning: symbol 'drm_get_dirty_info_name' was not declared. Should it be static?
drivers/gpu/drm/drm_crtc.c:1123:5: warning: symbol 'drm_mode_group_init' was not declared. Should it be static?

drivers/gpu/drm/drm_modes.c:918:6: warning: symbol 'drm_mode_validate_clocks' was not declared. Should it be static?

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_crtc.c |  5 +----
 drivers/gpu/drm/drm_pci.c  | 10 +++++-----
 include/drm/drm_crtc.h     |  3 +++
 3 files changed, 9 insertions(+), 9 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c
index c577baec416f..287cd82f9d75 100644
--- a/drivers/gpu/drm/drm_crtc.c
+++ b/drivers/gpu/drm/drm_crtc.c
@@ -178,9 +178,6 @@ static struct drm_prop_enum_list drm_dirty_info_enum_list[] = {
 	{ DRM_MODE_DIRTY_ANNOTATE, "Annotate" },
 };
 
-DRM_ENUM_NAME_FN(drm_get_dirty_info_name,
-		 drm_dirty_info_enum_list)
-
 struct drm_conn_prop_enum_list {
 	int type;
 	char *name;
@@ -1120,7 +1117,7 @@ int drm_mode_create_dirty_info_property(struct drm_device *dev)
 }
 EXPORT_SYMBOL(drm_mode_create_dirty_info_property);
 
-int drm_mode_group_init(struct drm_device *dev, struct drm_mode_group *group)
+static int drm_mode_group_init(struct drm_device *dev, struct drm_mode_group *group)
 {
 	uint32_t total_objects = 0;
 
diff --git a/drivers/gpu/drm/drm_pci.c b/drivers/gpu/drm/drm_pci.c
index bd719e936e13..14194b6ef644 100644
--- a/drivers/gpu/drm/drm_pci.c
+++ b/drivers/gpu/drm/drm_pci.c
@@ -152,7 +152,7 @@ static const char *drm_pci_get_name(struct drm_device *dev)
 	return pdriver->name;
 }
 
-int drm_pci_set_busid(struct drm_device *dev, struct drm_master *master)
+static int drm_pci_set_busid(struct drm_device *dev, struct drm_master *master)
 {
 	int len, ret;
 	struct pci_driver *pdriver = dev->driver->kdriver.pci;
@@ -194,9 +194,9 @@ err:
 	return ret;
 }
 
-int drm_pci_set_unique(struct drm_device *dev,
-		       struct drm_master *master,
-		       struct drm_unique *u)
+static int drm_pci_set_unique(struct drm_device *dev,
+			      struct drm_master *master,
+			      struct drm_unique *u)
 {
 	int domain, bus, slot, func, ret;
 	const char *bus_name;
@@ -266,7 +266,7 @@ static int drm_pci_irq_by_busid(struct drm_device *dev, struct drm_irq_busid *p)
 	return 0;
 }
 
-int drm_pci_agp_init(struct drm_device *dev)
+static int drm_pci_agp_init(struct drm_device *dev)
 {
 	if (drm_core_has_AGP(dev)) {
 		if (drm_pci_device_is_agp(dev))
diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index 836438dfb331..cf474498cee0 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -939,6 +939,9 @@ extern void drm_mode_list_concat(struct list_head *head,
 extern void drm_mode_validate_size(struct drm_device *dev,
 				   struct list_head *mode_list,
 				   int maxX, int maxY, int maxPitch);
+extern void drm_mode_validate_clocks(struct drm_device *dev,
+				     struct list_head *mode_list,
+				     int *min, int *max, int n_ranges);
 extern void drm_mode_prune_invalid(struct drm_device *dev,
 				   struct list_head *mode_list, bool verbose);
 extern void drm_mode_sort(struct list_head *mode_list);
-- 
cgit v1.2.3


From c55b6b3da25aa3af36ec51a13a4ed15fef0d7a73 Mon Sep 17 00:00:00 2001
From: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Fri, 26 Apr 2013 17:40:28 +0300
Subject: drm: Kill user_modes list and the associated ioctls
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

There is no way to use modes added to the user_modes list. We never
look at the contents of said list in the kernel, and the only operations
userspace can do are attach and detach. So the only "benefit" of this
interface is wasting kernel memory.

Fortunately it seems no real user space application ever used these
ioctls. So just kill them.

Also remove the prototypes for the non-existing drm_mode_addmode_ioctl()
and drm_mode_rmmode_ioctl() functions.

v2: Use drm_noop instead of completely removing the ioctls

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_crtc.c | 190 ---------------------------------------------
 drivers/gpu/drm/drm_drv.c  |   4 +-
 include/drm/drm_crtc.h     |  15 ----
 include/uapi/drm/drm.h     |   4 +-
 4 files changed, 4 insertions(+), 209 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c
index 287cd82f9d75..3a8f7e6db295 100644
--- a/drivers/gpu/drm/drm_crtc.c
+++ b/drivers/gpu/drm/drm_crtc.c
@@ -703,7 +703,6 @@ int drm_connector_init(struct drm_device *dev,
 	connector->connector_type = connector_type;
 	connector->connector_type_id =
 		++drm_connector_enum_list[connector_type].count; /* TODO */
-	INIT_LIST_HEAD(&connector->user_modes);
 	INIT_LIST_HEAD(&connector->probed_modes);
 	INIT_LIST_HEAD(&connector->modes);
 	connector->edid_blob_ptr = NULL;
@@ -744,9 +743,6 @@ void drm_connector_cleanup(struct drm_connector *connector)
 	list_for_each_entry_safe(mode, t, &connector->modes, head)
 		drm_mode_remove(connector, mode);
 
-	list_for_each_entry_safe(mode, t, &connector->user_modes, head)
-		drm_mode_remove(connector, mode);
-
 	drm_mode_object_put(dev, &connector->base);
 	list_del(&connector->head);
 	dev->mode_config.num_connector--;
@@ -2613,192 +2609,6 @@ void drm_fb_release(struct drm_file *priv)
 	mutex_unlock(&priv->fbs_lock);
 }
 
-/**
- * drm_mode_attachmode - add a mode to the user mode list
- * @dev: DRM device
- * @connector: connector to add the mode to
- * @mode: mode to add
- *
- * Add @mode to @connector's user mode list.
- */
-static void drm_mode_attachmode(struct drm_device *dev,
-				struct drm_connector *connector,
-				struct drm_display_mode *mode)
-{
-	list_add_tail(&mode->head, &connector->user_modes);
-}
-
-int drm_mode_attachmode_crtc(struct drm_device *dev, struct drm_crtc *crtc,
-			     const struct drm_display_mode *mode)
-{
-	struct drm_connector *connector;
-	int ret = 0;
-	struct drm_display_mode *dup_mode, *next;
-	LIST_HEAD(list);
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		if (!connector->encoder)
-			continue;
-		if (connector->encoder->crtc == crtc) {
-			dup_mode = drm_mode_duplicate(dev, mode);
-			if (!dup_mode) {
-				ret = -ENOMEM;
-				goto out;
-			}
-			list_add_tail(&dup_mode->head, &list);
-		}
-	}
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		if (!connector->encoder)
-			continue;
-		if (connector->encoder->crtc == crtc)
-			list_move_tail(list.next, &connector->user_modes);
-	}
-
-	WARN_ON(!list_empty(&list));
-
- out:
-	list_for_each_entry_safe(dup_mode, next, &list, head)
-		drm_mode_destroy(dev, dup_mode);
-
-	return ret;
-}
-EXPORT_SYMBOL(drm_mode_attachmode_crtc);
-
-static int drm_mode_detachmode(struct drm_device *dev,
-			       struct drm_connector *connector,
-			       struct drm_display_mode *mode)
-{
-	int found = 0;
-	int ret = 0;
-	struct drm_display_mode *match_mode, *t;
-
-	list_for_each_entry_safe(match_mode, t, &connector->user_modes, head) {
-		if (drm_mode_equal(match_mode, mode)) {
-			list_del(&match_mode->head);
-			drm_mode_destroy(dev, match_mode);
-			found = 1;
-			break;
-		}
-	}
-
-	if (!found)
-		ret = -EINVAL;
-
-	return ret;
-}
-
-int drm_mode_detachmode_crtc(struct drm_device *dev, struct drm_display_mode *mode)
-{
-	struct drm_connector *connector;
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		drm_mode_detachmode(dev, connector, mode);
-	}
-	return 0;
-}
-EXPORT_SYMBOL(drm_mode_detachmode_crtc);
-
-/**
- * drm_fb_attachmode - Attach a user mode to an connector
- * @dev: drm device for the ioctl
- * @data: data pointer for the ioctl
- * @file_priv: drm file for the ioctl call
- *
- * This attaches a user specified mode to an connector.
- * Called by the user via ioctl.
- *
- * RETURNS:
- * Zero on success, errno on failure.
- */
-int drm_mode_attachmode_ioctl(struct drm_device *dev,
-			      void *data, struct drm_file *file_priv)
-{
-	struct drm_mode_mode_cmd *mode_cmd = data;
-	struct drm_connector *connector;
-	struct drm_display_mode *mode;
-	struct drm_mode_object *obj;
-	struct drm_mode_modeinfo *umode = &mode_cmd->mode;
-	int ret;
-
-	if (!drm_core_check_feature(dev, DRIVER_MODESET))
-		return -EINVAL;
-
-	drm_modeset_lock_all(dev);
-
-	obj = drm_mode_object_find(dev, mode_cmd->connector_id, DRM_MODE_OBJECT_CONNECTOR);
-	if (!obj) {
-		ret = -EINVAL;
-		goto out;
-	}
-	connector = obj_to_connector(obj);
-
-	mode = drm_mode_create(dev);
-	if (!mode) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	ret = drm_crtc_convert_umode(mode, umode);
-	if (ret) {
-		DRM_DEBUG_KMS("Invalid mode\n");
-		drm_mode_destroy(dev, mode);
-		goto out;
-	}
-
-	drm_mode_attachmode(dev, connector, mode);
-out:
-	drm_modeset_unlock_all(dev);
-	return ret;
-}
-
-
-/**
- * drm_fb_detachmode - Detach a user specified mode from an connector
- * @dev: drm device for the ioctl
- * @data: data pointer for the ioctl
- * @file_priv: drm file for the ioctl call
- *
- * Called by the user via ioctl.
- *
- * RETURNS:
- * Zero on success, errno on failure.
- */
-int drm_mode_detachmode_ioctl(struct drm_device *dev,
-			      void *data, struct drm_file *file_priv)
-{
-	struct drm_mode_object *obj;
-	struct drm_mode_mode_cmd *mode_cmd = data;
-	struct drm_connector *connector;
-	struct drm_display_mode mode;
-	struct drm_mode_modeinfo *umode = &mode_cmd->mode;
-	int ret;
-
-	if (!drm_core_check_feature(dev, DRIVER_MODESET))
-		return -EINVAL;
-
-	drm_modeset_lock_all(dev);
-
-	obj = drm_mode_object_find(dev, mode_cmd->connector_id, DRM_MODE_OBJECT_CONNECTOR);
-	if (!obj) {
-		ret = -EINVAL;
-		goto out;
-	}
-	connector = obj_to_connector(obj);
-
-	ret = drm_crtc_convert_umode(&mode, umode);
-	if (ret) {
-		DRM_DEBUG_KMS("Invalid mode\n");
-		goto out;
-	}
-
-	ret = drm_mode_detachmode(dev, connector, &mode);
-out:
-	drm_modeset_unlock_all(dev);
-	return ret;
-}
-
 struct drm_property *drm_property_create(struct drm_device *dev, int flags,
 					 const char *name, int num_values)
 {
diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c
index 80c68957e7cf..8d4f29075af5 100644
--- a/drivers/gpu/drm/drm_drv.c
+++ b/drivers/gpu/drm/drm_drv.c
@@ -150,8 +150,8 @@ static const struct drm_ioctl_desc drm_ioctls[] = {
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETGAMMA, drm_mode_gamma_set_ioctl, DRM_MASTER|DRM_UNLOCKED),
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETENCODER, drm_mode_getencoder, DRM_CONTROL_ALLOW|DRM_UNLOCKED),
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCONNECTOR, drm_mode_getconnector, DRM_CONTROL_ALLOW|DRM_UNLOCKED),
-	DRM_IOCTL_DEF(DRM_IOCTL_MODE_ATTACHMODE, drm_mode_attachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED),
-	DRM_IOCTL_DEF(DRM_IOCTL_MODE_DETACHMODE, drm_mode_detachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED),
+	DRM_IOCTL_DEF(DRM_IOCTL_MODE_ATTACHMODE, drm_noop, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED),
+	DRM_IOCTL_DEF(DRM_IOCTL_MODE_DETACHMODE, drm_noop, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED),
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPERTY, drm_mode_getproperty_ioctl, DRM_CONTROL_ALLOW|DRM_UNLOCKED),
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETPROPERTY, drm_mode_connector_property_set_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED),
 	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPBLOB, drm_mode_getblob_ioctl, DRM_CONTROL_ALLOW|DRM_UNLOCKED),
diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index cf474498cee0..adb3f9b625f6 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -554,7 +554,6 @@ enum drm_connector_force {
  * @probed_modes: list of modes derived directly from the display
  * @display_info: information about attached display (e.g. from EDID)
  * @funcs: connector control functions
- * @user_modes: user added mode list
  * @edid_blob_ptr: DRM property containing EDID if present
  * @properties: property tracking for this connector
  * @polled: a %DRM_CONNECTOR_POLL_<foo> value for core driven polling
@@ -598,7 +597,6 @@ struct drm_connector {
 	struct drm_display_info display_info;
 	const struct drm_connector_funcs *funcs;
 
-	struct list_head user_modes;
 	struct drm_property_blob *edid_blob_ptr;
 	struct drm_object_properties properties;
 
@@ -927,11 +925,6 @@ extern int drm_mode_width(const struct drm_display_mode *mode);
 extern int drm_mode_height(const struct drm_display_mode *mode);
 
 /* for us by fb module */
-extern int drm_mode_attachmode_crtc(struct drm_device *dev,
-				    struct drm_crtc *crtc,
-				    const struct drm_display_mode *mode);
-extern int drm_mode_detachmode_crtc(struct drm_device *dev, struct drm_display_mode *mode);
-
 extern struct drm_display_mode *drm_mode_create(struct drm_device *dev);
 extern void drm_mode_destroy(struct drm_device *dev, struct drm_display_mode *mode);
 extern void drm_mode_list_concat(struct list_head *head,
@@ -1040,14 +1033,6 @@ extern int drm_mode_getfb(struct drm_device *dev,
 			  void *data, struct drm_file *file_priv);
 extern int drm_mode_dirtyfb_ioctl(struct drm_device *dev,
 				  void *data, struct drm_file *file_priv);
-extern int drm_mode_addmode_ioctl(struct drm_device *dev,
-				  void *data, struct drm_file *file_priv);
-extern int drm_mode_rmmode_ioctl(struct drm_device *dev,
-				 void *data, struct drm_file *file_priv);
-extern int drm_mode_attachmode_ioctl(struct drm_device *dev,
-				     void *data, struct drm_file *file_priv);
-extern int drm_mode_detachmode_ioctl(struct drm_device *dev,
-				     void *data, struct drm_file *file_priv);
 
 extern int drm_mode_getproperty_ioctl(struct drm_device *dev,
 				      void *data, struct drm_file *file_priv);
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index 73a99e4664be..5a57be68bab7 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -711,8 +711,8 @@ struct drm_prime_handle {
 #define DRM_IOCTL_MODE_SETGAMMA		DRM_IOWR(0xA5, struct drm_mode_crtc_lut)
 #define DRM_IOCTL_MODE_GETENCODER	DRM_IOWR(0xA6, struct drm_mode_get_encoder)
 #define DRM_IOCTL_MODE_GETCONNECTOR	DRM_IOWR(0xA7, struct drm_mode_get_connector)
-#define DRM_IOCTL_MODE_ATTACHMODE	DRM_IOWR(0xA8, struct drm_mode_mode_cmd)
-#define DRM_IOCTL_MODE_DETACHMODE	DRM_IOWR(0xA9, struct drm_mode_mode_cmd)
+#define DRM_IOCTL_MODE_ATTACHMODE	DRM_IOWR(0xA8, struct drm_mode_mode_cmd) /* deprecated (never worked) */
+#define DRM_IOCTL_MODE_DETACHMODE	DRM_IOWR(0xA9, struct drm_mode_mode_cmd) /* deprecated (never worked) */
 
 #define DRM_IOCTL_MODE_GETPROPERTY	DRM_IOWR(0xAA, struct drm_mode_get_property)
 #define DRM_IOCTL_MODE_SETPROPERTY	DRM_IOWR(0xAB, struct drm_mode_connector_set_property)
-- 
cgit v1.2.3


From 219b47339ced80ca580bb6ce7d1636166984afa7 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@gmail.com>
Date: Mon, 22 Apr 2013 09:54:36 +1000
Subject: drm/prime: keep a reference from the handle to exported dma-buf (v6)

Currently we have a problem with this:
1. i915: create gem object
2. i915: export gem object to prime
3. radeon: import gem object
4. close prime fd
5. radeon: unref object
6. i915: unref object

i915 has an imported object reference in its file priv, that isn't
cleaned up properly until fd close. The reference gets added at step 2,
but at step 6 we don't have enough info to clean it up.

The solution is to take a reference on the dma-buf when we export it,
and drop the reference when the gem handle goes away.

So when we export a dma_buf from a gem object, we keep track of it
with the handle, we take a reference to the dma_buf. When we close
the handle (i.e. userspace is finished with the buffer), we drop
the reference to the dma_buf, and it gets collected.

This patch isn't meant to fix any other problem or bikesheds, and it doesn't
fix any races with other scenarios.

v1.1: move export symbol line back up.

v2: okay I had to do a bit more, as the first patch showed a leak
on one of my tests, that I found using the dma-buf debugfs support,
the problem case is exporting a buffer twice with the same handle,
we'd add another export handle for it unnecessarily, however
we now fail if we try to export the same object with a different gem handle,
however I'm not sure if that is a case I want to support, and I've
gotten the code to WARN_ON if we hit something like that.

v2.1: rebase this patch, write better commit msg.
v3: cleanup error handling, track import vs export in linked list,
these two patches were separate previously, but seem to work better
like this.
v4: danvet is correct, this code is no longer useful, since the buffer
better exist, so remove it.
v5: always take a reference to the dma buf object, import or export.
(Imre Deak contributed this originally)
v6: square the circle, remove import vs export tracking now
that there is no difference

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_gem.c   |  4 +--
 drivers/gpu/drm/drm_prime.c | 76 ++++++++++++++++++++++++---------------------
 include/drm/drmP.h          |  5 ++-
 3 files changed, 44 insertions(+), 41 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
index af779ae19ebf..cf919e36e8ae 100644
--- a/drivers/gpu/drm/drm_gem.c
+++ b/drivers/gpu/drm/drm_gem.c
@@ -205,11 +205,11 @@ static void
 drm_gem_remove_prime_handles(struct drm_gem_object *obj, struct drm_file *filp)
 {
 	if (obj->import_attach) {
-		drm_prime_remove_imported_buf_handle(&filp->prime,
+		drm_prime_remove_buf_handle(&filp->prime,
 				obj->import_attach->dmabuf);
 	}
 	if (obj->export_dma_buf) {
-		drm_prime_remove_imported_buf_handle(&filp->prime,
+		drm_prime_remove_buf_handle(&filp->prime,
 				obj->export_dma_buf);
 	}
 }
diff --git a/drivers/gpu/drm/drm_prime.c b/drivers/gpu/drm/drm_prime.c
index 25d02187067e..7830d8e1f212 100644
--- a/drivers/gpu/drm/drm_prime.c
+++ b/drivers/gpu/drm/drm_prime.c
@@ -62,6 +62,7 @@ struct drm_prime_member {
 	struct dma_buf *dma_buf;
 	uint32_t handle;
 };
+static int drm_prime_add_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t handle);
 
 static struct sg_table *drm_gem_map_dma_buf(struct dma_buf_attachment *attach,
 		enum dma_data_direction dir)
@@ -200,7 +201,8 @@ int drm_gem_prime_handle_to_fd(struct drm_device *dev,
 {
 	struct drm_gem_object *obj;
 	void *buf;
-	int ret;
+	int ret = 0;
+	struct dma_buf *dmabuf;
 
 	obj = drm_gem_object_lookup(dev, file_priv, handle);
 	if (!obj)
@@ -209,43 +211,44 @@ int drm_gem_prime_handle_to_fd(struct drm_device *dev,
 	mutex_lock(&file_priv->prime.lock);
 	/* re-export the original imported object */
 	if (obj->import_attach) {
-		get_dma_buf(obj->import_attach->dmabuf);
-		*prime_fd = dma_buf_fd(obj->import_attach->dmabuf, flags);
-		drm_gem_object_unreference_unlocked(obj);
-		mutex_unlock(&file_priv->prime.lock);
-		return 0;
+		dmabuf = obj->import_attach->dmabuf;
+		goto out_have_obj;
 	}
 
 	if (obj->export_dma_buf) {
-		get_dma_buf(obj->export_dma_buf);
-		*prime_fd = dma_buf_fd(obj->export_dma_buf, flags);
-		drm_gem_object_unreference_unlocked(obj);
-	} else {
-		buf = dev->driver->gem_prime_export(dev, obj, flags);
-		if (IS_ERR(buf)) {
-			/* normally the created dma-buf takes ownership of the ref,
-			 * but if that fails then drop the ref
-			 */
-			drm_gem_object_unreference_unlocked(obj);
-			mutex_unlock(&file_priv->prime.lock);
-			return PTR_ERR(buf);
-		}
-		obj->export_dma_buf = buf;
-		*prime_fd = dma_buf_fd(buf, flags);
+		dmabuf = obj->export_dma_buf;
+		goto out_have_obj;
 	}
+
+	buf = dev->driver->gem_prime_export(dev, obj, flags);
+	if (IS_ERR(buf)) {
+		/* normally the created dma-buf takes ownership of the ref,
+		 * but if that fails then drop the ref
+		 */
+		ret = PTR_ERR(buf);
+		goto out;
+	}
+	obj->export_dma_buf = buf;
+
 	/* if we've exported this buffer the cheat and add it to the import list
 	 * so we get the correct handle back
 	 */
-	ret = drm_prime_add_imported_buf_handle(&file_priv->prime,
-			obj->export_dma_buf, handle);
-	if (ret) {
-		drm_gem_object_unreference_unlocked(obj);
-		mutex_unlock(&file_priv->prime.lock);
-		return ret;
-	}
+	ret = drm_prime_add_buf_handle(&file_priv->prime,
+				       obj->export_dma_buf, handle);
+	if (ret)
+		goto out;
 
+	*prime_fd = dma_buf_fd(buf, flags);
 	mutex_unlock(&file_priv->prime.lock);
 	return 0;
+
+out_have_obj:
+	get_dma_buf(dmabuf);
+	*prime_fd = dma_buf_fd(dmabuf, flags);
+out:
+	drm_gem_object_unreference_unlocked(obj);
+	mutex_unlock(&file_priv->prime.lock);
+	return ret;
 }
 EXPORT_SYMBOL(drm_gem_prime_handle_to_fd);
 
@@ -314,7 +317,7 @@ int drm_gem_prime_fd_to_handle(struct drm_device *dev,
 
 	mutex_lock(&file_priv->prime.lock);
 
-	ret = drm_prime_lookup_imported_buf_handle(&file_priv->prime,
+	ret = drm_prime_lookup_buf_handle(&file_priv->prime,
 			dma_buf, handle);
 	if (!ret) {
 		ret = 0;
@@ -333,7 +336,7 @@ int drm_gem_prime_fd_to_handle(struct drm_device *dev,
 	if (ret)
 		goto out_put;
 
-	ret = drm_prime_add_imported_buf_handle(&file_priv->prime,
+	ret = drm_prime_add_buf_handle(&file_priv->prime,
 			dma_buf, *handle);
 	if (ret)
 		goto fail;
@@ -487,7 +490,7 @@ void drm_prime_destroy_file_private(struct drm_prime_file_private *prime_fpriv)
 }
 EXPORT_SYMBOL(drm_prime_destroy_file_private);
 
-int drm_prime_add_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t handle)
+static int drm_prime_add_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t handle)
 {
 	struct drm_prime_member *member;
 
@@ -495,14 +498,14 @@ int drm_prime_add_imported_buf_handle(struct drm_prime_file_private *prime_fpriv
 	if (!member)
 		return -ENOMEM;
 
+	get_dma_buf(dma_buf);
 	member->dma_buf = dma_buf;
 	member->handle = handle;
 	list_add(&member->entry, &prime_fpriv->head);
 	return 0;
 }
-EXPORT_SYMBOL(drm_prime_add_imported_buf_handle);
 
-int drm_prime_lookup_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t *handle)
+int drm_prime_lookup_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t *handle)
 {
 	struct drm_prime_member *member;
 
@@ -514,19 +517,20 @@ int drm_prime_lookup_imported_buf_handle(struct drm_prime_file_private *prime_fp
 	}
 	return -ENOENT;
 }
-EXPORT_SYMBOL(drm_prime_lookup_imported_buf_handle);
+EXPORT_SYMBOL(drm_prime_lookup_buf_handle);
 
-void drm_prime_remove_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf)
+void drm_prime_remove_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf)
 {
 	struct drm_prime_member *member, *safe;
 
 	mutex_lock(&prime_fpriv->lock);
 	list_for_each_entry_safe(member, safe, &prime_fpriv->head, entry) {
 		if (member->dma_buf == dma_buf) {
+			dma_buf_put(dma_buf);
 			list_del(&member->entry);
 			kfree(member);
 		}
 	}
 	mutex_unlock(&prime_fpriv->lock);
 }
-EXPORT_SYMBOL(drm_prime_remove_imported_buf_handle);
+EXPORT_SYMBOL(drm_prime_remove_buf_handle);
diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index 2d94d7413d71..f1ce786736e4 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1593,9 +1593,8 @@ extern void drm_prime_gem_destroy(struct drm_gem_object *obj, struct sg_table *s
 
 void drm_prime_init_file_private(struct drm_prime_file_private *prime_fpriv);
 void drm_prime_destroy_file_private(struct drm_prime_file_private *prime_fpriv);
-int drm_prime_add_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t handle);
-int drm_prime_lookup_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t *handle);
-void drm_prime_remove_imported_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf);
+int drm_prime_lookup_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf, uint32_t *handle);
+void drm_prime_remove_buf_handle(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf);
 
 int drm_prime_add_dma_buf(struct drm_device *dev, struct drm_gem_object *obj);
 int drm_prime_lookup_obj(struct drm_device *dev, struct dma_buf *buf,
-- 
cgit v1.2.3


From 18932a28419596bc9403770f5d8a108c5433fe59 Mon Sep 17 00:00:00 2001
From: Alex Deucher <alexander.deucher@amd.com>
Date: Thu, 25 Apr 2013 13:55:15 -0400
Subject: drm/radeon: add some new SI PCI ids

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
---
 include/drm/drm_pciids.h | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'include')

diff --git a/include/drm/drm_pciids.h b/include/drm/drm_pciids.h
index 918e8fe2f5e9..52300c7f2250 100644
--- a/include/drm/drm_pciids.h
+++ b/include/drm/drm_pciids.h
@@ -240,6 +240,7 @@
 	{0x1002, 0x6819, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6820, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6821, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
+	{0x1002, 0x6822, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6823, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6824, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
@@ -247,11 +248,13 @@
 	{0x1002, 0x6827, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6828, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6829, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
+	{0x1002, 0x682A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x682B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x682D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x682F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6830, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6831, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP}, \
+	{0x1002, 0x6835, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6837, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6838, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
 	{0x1002, 0x6839, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|RADEON_NEW_MEMMAP}, \
-- 
cgit v1.2.3


From 62d1f92e06aef9665d71ca7e986b3047ecf0b3c7 Mon Sep 17 00:00:00 2001
From: Alex Deucher <alexander.deucher@amd.com>
Date: Thu, 25 Apr 2013 14:06:05 -0400
Subject: drm/radeon: add new richland pci ids

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/radeon/ni.c | 6 ++++--
 include/drm/drm_pciids.h    | 2 ++
 2 files changed, 6 insertions(+), 2 deletions(-)

(limited to 'include')

diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c
index 7436b91699d0..7969c0c8ec20 100644
--- a/drivers/gpu/drm/radeon/ni.c
+++ b/drivers/gpu/drm/radeon/ni.c
@@ -749,7 +749,8 @@ static void cayman_gpu_init(struct radeon_device *rdev)
 		    (rdev->pdev->device == 0x990F) ||
 		    (rdev->pdev->device == 0x9910) ||
 		    (rdev->pdev->device == 0x9917) ||
-		    (rdev->pdev->device == 0x9999)) {
+		    (rdev->pdev->device == 0x9999) ||
+		    (rdev->pdev->device == 0x999C)) {
 			rdev->config.cayman.max_simds_per_se = 6;
 			rdev->config.cayman.max_backends_per_se = 2;
 		} else if ((rdev->pdev->device == 0x9903) ||
@@ -758,7 +759,8 @@ static void cayman_gpu_init(struct radeon_device *rdev)
 			   (rdev->pdev->device == 0x990D) ||
 			   (rdev->pdev->device == 0x990E) ||
 			   (rdev->pdev->device == 0x9913) ||
-			   (rdev->pdev->device == 0x9918)) {
+			   (rdev->pdev->device == 0x9918) ||
+			   (rdev->pdev->device == 0x999D)) {
 			rdev->config.cayman.max_simds_per_se = 4;
 			rdev->config.cayman.max_backends_per_se = 2;
 		} else if ((rdev->pdev->device == 0x9919) ||
diff --git a/include/drm/drm_pciids.h b/include/drm/drm_pciids.h
index 52300c7f2250..c2af598f701d 100644
--- a/include/drm/drm_pciids.h
+++ b/include/drm/drm_pciids.h
@@ -606,6 +606,8 @@
 	{0x1002, 0x9999, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
 	{0x1002, 0x999A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
 	{0x1002, 0x999B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
+	{0x1002, 0x999C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
+	{0x1002, 0x999D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
 	{0x1002, 0x99A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
 	{0x1002, 0x99A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
 	{0x1002, 0x99A4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARUBA|RADEON_NEW_MEMMAP|RADEON_IS_IGP}, \
-- 
cgit v1.2.3