[PATCH] avr32 architecture

This adds support for the Atmel AVR32 architecture as well as the AT32AP7000
CPU and the AT32STK1000 development board.

AVR32 is a new high-performance 32-bit RISC microprocessor core, designed for
cost-sensitive embedded applications, with particular emphasis on low power
consumption and high code density.  The AVR32 architecture is not binary
compatible with earlier 8-bit AVR architectures.

The AVR32 architecture, including the instruction set, is described by the
AVR32 Architecture Manual, available from

http://www.atmel.com/dyn/resources/prod_documents/doc32000.pdf

The Atmel AT32AP7000 is the first CPU implementing the AVR32 architecture.  It
features a 7-stage pipeline, 16KB instruction and data caches and a full
Memory Management Unit.  It also comes with a large set of integrated
peripherals, many of which are shared with the AT91 ARM-based controllers from
Atmel.

Full data sheet is available from

http://www.atmel.com/dyn/resources/prod_documents/doc32003.pdf

while the CPU core implementation including caches and MMU is documented by
the AVR32 AP Technical Reference, available from

http://www.atmel.com/dyn/resources/prod_documents/doc32001.pdf

Information about the AT32STK1000 development board can be found at

http://www.atmel.com/dyn/products/tools_card.asp?tool_id=3918

including a BSP CD image with an earlier version of this patch, development
tools (binaries and source/patches) and a root filesystem image suitable for
booting from SD card.

Alternatively, there's a preliminary "getting started" guide available at
http://avr32linux.org/twiki/bin/view/Main/GettingStarted which provides links
to the sources and patches you will need in order to set up a cross-compiling
environment for avr32-linux.

This patch, as well as the other patches included with the BSP and the
toolchain patches, is actively supported by Atmel Corporation.

[dmccr@us.ibm.com: Fix more pxx_page macro locations]
[bunk@stusta.de: fix `make defconfig']
Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dave McCracken <dmccr@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 139 insertions, 0 deletions

diff --git a/arch/avr32/mm/dma-coherent.c b/arch/avr32/mm/dma-coherent.c
new file mode 100644
index 000000000000..44ab8a7bdae2
--- /dev/null
+++ b/arch/avr32/mm/dma-coherent.c
@@ -0,0 +1,139 @@
+/*
+ *  Copyright (C) 2004-2006 Atmel Corporation
+ *
+ * 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.
+ */
+
+#include <linux/dma-mapping.h>
+
+#include <asm/addrspace.h>
+#include <asm/cacheflush.h>
+
+void dma_cache_sync(void *vaddr, size_t size, int direction)
+{
+	/*
+	 * No need to sync an uncached area
+	 */
+	if (PXSEG(vaddr) == P2SEG)
+		return;
+
+	switch (direction) {
+	case DMA_FROM_DEVICE:		/* invalidate only */
+		dma_cache_inv(vaddr, size);
+		break;
+	case DMA_TO_DEVICE:		/* writeback only */
+		dma_cache_wback(vaddr, size);
+		break;
+	case DMA_BIDIRECTIONAL:		/* writeback and invalidate */
+		dma_cache_wback_inv(vaddr, size);
+		break;
+	default:
+		BUG();
+	}
+}
+EXPORT_SYMBOL(dma_cache_sync);
+
+static struct page *__dma_alloc(struct device *dev, size_t size,
+				dma_addr_t *handle, gfp_t gfp)
+{
+	struct page *page, *free, *end;
+	int order;
+
+	size = PAGE_ALIGN(size);
+	order = get_order(size);
+
+	page = alloc_pages(gfp, order);
+	if (!page)
+		return NULL;
+	split_page(page, order);
+
+	/*
+	 * When accessing physical memory with valid cache data, we
+	 * get a cache hit even if the virtual memory region is marked
+	 * as uncached.
+	 *
+	 * Since the memory is newly allocated, there is no point in
+	 * doing a writeback. If the previous owner cares, he should
+	 * have flushed the cache before releasing the memory.
+	 */
+	invalidate_dcache_region(phys_to_virt(page_to_phys(page)), size);
+
+	*handle = page_to_bus(page);
+	free = page + (size >> PAGE_SHIFT);
+	end = page + (1 << order);
+
+	/*
+	 * Free any unused pages
+	 */
+	while (free < end) {
+		__free_page(free);
+		free++;
+	}
+
+	return page;
+}
+
+static void __dma_free(struct device *dev, size_t size,
+		       struct page *page, dma_addr_t handle)
+{
+	struct page *end = page + (PAGE_ALIGN(size) >> PAGE_SHIFT);
+
+	while (page < end)
+		__free_page(page++);
+}
+
+void *dma_alloc_coherent(struct device *dev, size_t size,
+			 dma_addr_t *handle, gfp_t gfp)
+{
+	struct page *page;
+	void *ret = NULL;
+
+	page = __dma_alloc(dev, size, handle, gfp);
+	if (page)
+		ret = phys_to_uncached(page_to_phys(page));
+
+	return ret;
+}
+EXPORT_SYMBOL(dma_alloc_coherent);
+
+void dma_free_coherent(struct device *dev, size_t size,
+		       void *cpu_addr, dma_addr_t handle)
+{
+	void *addr = phys_to_cached(uncached_to_phys(cpu_addr));
+	struct page *page;
+
+	pr_debug("dma_free_coherent addr %p (phys %08lx) size %u\n",
+		 cpu_addr, (unsigned long)handle, (unsigned)size);
+	BUG_ON(!virt_addr_valid(addr));
+	page = virt_to_page(addr);
+	__dma_free(dev, size, page, handle);
+}
+EXPORT_SYMBOL(dma_free_coherent);
+
+#if 0
+void *dma_alloc_writecombine(struct device *dev, size_t size,
+			     dma_addr_t *handle, gfp_t gfp)
+{
+	struct page *page;
+
+	page = __dma_alloc(dev, size, handle, gfp);
+
+	/* Now, map the page into P3 with write-combining turned on */
+	return __ioremap(page_to_phys(page), size, _PAGE_BUFFER);
+}
+EXPORT_SYMBOL(dma_alloc_writecombine);
+
+void dma_free_writecombine(struct device *dev, size_t size,
+			   void *cpu_addr, dma_addr_t handle)
+{
+	struct page *page;
+
+	iounmap(cpu_addr);
+
+	page = bus_to_page(handle);
+	__dma_free(dev, size, page, handle);
+}
+EXPORT_SYMBOL(dma_free_writecombine);
+#endif