Merge branch 'hwpoison' of git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-mce-2.6

* 'hwpoison' of git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-mce-2.6: (34 commits)
  HWPOISON: Remove stray phrase in a comment
  HWPOISON: Try to allocate migration page on the same node
  HWPOISON: Don't do early filtering if filter is disabled
  HWPOISON: Add a madvise() injector for soft page offlining
  HWPOISON: Add soft page offline support
  HWPOISON: Undefine short-hand macros after use to avoid namespace conflict
  HWPOISON: Use new shake_page in memory_failure
  HWPOISON: Use correct name for MADV_HWPOISON in documentation
  HWPOISON: mention HWPoison in Kconfig entry
  HWPOISON: Use get_user_page_fast in hwpoison madvise
  HWPOISON: add an interface to switch off/on all the page filters
  HWPOISON: add memory cgroup filter
  memcg: add accessor to mem_cgroup.css
  memcg: rename and export try_get_mem_cgroup_from_page()
  HWPOISON: add page flags filter
  mm: export stable page flags
  HWPOISON: limit hwpoison injector to known page types
  HWPOISON: add fs/device filters
  HWPOISON: return 0 to indicate success reliably
  HWPOISON: make semantics of IGNORED/DELAYED clear
  ...

8 files changed, 673 insertions, 77 deletions

diff --git a/mm/Kconfig b/mm/Kconfig
index 2310984591ed..43ea8c3a2bbf 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -251,8 +251,9 @@ config MEMORY_FAILURE
 	  special hardware support and typically ECC memory.
 
 config HWPOISON_INJECT
-	tristate "Poison pages injector"
+	tristate "HWPoison pages injector"
 	depends on MEMORY_FAILURE && DEBUG_KERNEL
+	select PROC_PAGE_MONITOR
 
 config NOMMU_INITIAL_TRIM_EXCESS
 	int "Turn on mmap() excess space trimming before booting"
diff --git a/mm/hwpoison-inject.c b/mm/hwpoison-inject.c
index e1d85137f086..10ea71905c1f 100644
--- a/mm/hwpoison-inject.c
+++ b/mm/hwpoison-inject.c
@@ -3,18 +3,68 @@
 #include <linux/debugfs.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+#include "internal.h"
 
-static struct dentry *hwpoison_dir, *corrupt_pfn;
+static struct dentry *hwpoison_dir;
 
 static int hwpoison_inject(void *data, u64 val)
 {
+	unsigned long pfn = val;
+	struct page *p;
+	int err;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (!hwpoison_filter_enable)
+		goto inject;
+	if (!pfn_valid(pfn))
+		return -ENXIO;
+
+	p = pfn_to_page(pfn);
+	/*
+	 * This implies unable to support free buddy pages.
+	 */
+	if (!get_page_unless_zero(p))
+		return 0;
+
+	if (!PageLRU(p))
+		shake_page(p, 0);
+	/*
+	 * This implies unable to support non-LRU pages.
+	 */
+	if (!PageLRU(p))
+		return 0;
+
+	/*
+	 * do a racy check with elevated page count, to make sure PG_hwpoison
+	 * will only be set for the targeted owner (or on a free page).
+	 * We temporarily take page lock for try_get_mem_cgroup_from_page().
+	 * __memory_failure() will redo the check reliably inside page lock.
+	 */
+	lock_page(p);
+	err = hwpoison_filter(p);
+	unlock_page(p);
+	if (err)
+		return 0;
+
+inject:
+	printk(KERN_INFO "Injecting memory failure at pfn %lx\n", pfn);
+	return __memory_failure(pfn, 18, MF_COUNT_INCREASED);
+}
+
+static int hwpoison_unpoison(void *data, u64 val)
+{
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
-	printk(KERN_INFO "Injecting memory failure at pfn %Lx\n", val);
-	return __memory_failure(val, 18, 0);
+
+	return unpoison_memory(val);
 }
 
 DEFINE_SIMPLE_ATTRIBUTE(hwpoison_fops, NULL, hwpoison_inject, "%lli\n");
+DEFINE_SIMPLE_ATTRIBUTE(unpoison_fops, NULL, hwpoison_unpoison, "%lli\n");
 
 static void pfn_inject_exit(void)
 {
@@ -24,16 +74,63 @@ static void pfn_inject_exit(void)
 
 static int pfn_inject_init(void)
 {
+	struct dentry *dentry;
+
 	hwpoison_dir = debugfs_create_dir("hwpoison", NULL);
 	if (hwpoison_dir == NULL)
 		return -ENOMEM;
-	corrupt_pfn = debugfs_create_file("corrupt-pfn", 0600, hwpoison_dir,
+
+	/*
+	 * Note that the below poison/unpoison interfaces do not involve
+	 * hardware status change, hence do not require hardware support.
+	 * They are mainly for testing hwpoison in software level.
+	 */
+	dentry = debugfs_create_file("corrupt-pfn", 0600, hwpoison_dir,
 					  NULL, &hwpoison_fops);
-	if (corrupt_pfn == NULL) {
-		pfn_inject_exit();
-		return -ENOMEM;
-	}
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_file("unpoison-pfn", 0600, hwpoison_dir,
+				     NULL, &unpoison_fops);
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_u32("corrupt-filter-enable", 0600,
+				    hwpoison_dir, &hwpoison_filter_enable);
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_u32("corrupt-filter-dev-major", 0600,
+				    hwpoison_dir, &hwpoison_filter_dev_major);
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_u32("corrupt-filter-dev-minor", 0600,
+				    hwpoison_dir, &hwpoison_filter_dev_minor);
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_u64("corrupt-filter-flags-mask", 0600,
+				    hwpoison_dir, &hwpoison_filter_flags_mask);
+	if (!dentry)
+		goto fail;
+
+	dentry = debugfs_create_u64("corrupt-filter-flags-value", 0600,
+				    hwpoison_dir, &hwpoison_filter_flags_value);
+	if (!dentry)
+		goto fail;
+
+#ifdef	CONFIG_CGROUP_MEM_RES_CTLR_SWAP
+	dentry = debugfs_create_u64("corrupt-filter-memcg", 0600,
+				    hwpoison_dir, &hwpoison_filter_memcg);
+	if (!dentry)
+		goto fail;
+#endif
+
 	return 0;
+fail:
+	pfn_inject_exit();
+	return -ENOMEM;
 }
 
 module_init(pfn_inject_init);
diff --git a/mm/internal.h b/mm/internal.h
index 4fe67a162cb4..6a697bb97fc5 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -50,6 +50,9 @@ extern void putback_lru_page(struct page *page);
  */
 extern void __free_pages_bootmem(struct page *page, unsigned int order);
 extern void prep_compound_page(struct page *page, unsigned long order);
+#ifdef CONFIG_MEMORY_FAILURE
+extern bool is_free_buddy_page(struct page *page);
+#endif
 
 
 /*
@@ -247,3 +250,12 @@ int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
 #define ZONE_RECLAIM_SOME	0
 #define ZONE_RECLAIM_SUCCESS	1
 #endif
+
+extern int hwpoison_filter(struct page *p);
+
+extern u32 hwpoison_filter_dev_major;
+extern u32 hwpoison_filter_dev_minor;
+extern u64 hwpoison_filter_flags_mask;
+extern u64 hwpoison_filter_flags_value;
+extern u64 hwpoison_filter_memcg;
+extern u32 hwpoison_filter_enable;
diff --git a/mm/madvise.c b/mm/madvise.c
index 35b1479b7c9d..319528b8db74 100644
--- a/mm/madvise.c
+++ b/mm/madvise.c
@@ -9,6 +9,7 @@
 #include <linux/pagemap.h>
 #include <linux/syscalls.h>
 #include <linux/mempolicy.h>
+#include <linux/page-isolation.h>
 #include <linux/hugetlb.h>
 #include <linux/sched.h>
 #include <linux/ksm.h>
@@ -222,7 +223,7 @@ static long madvise_remove(struct vm_area_struct *vma,
 /*
  * Error injection support for memory error handling.
  */
-static int madvise_hwpoison(unsigned long start, unsigned long end)
+static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
 {
 	int ret = 0;
 
@@ -230,15 +231,21 @@ static int madvise_hwpoison(unsigned long start, unsigned long end)
 		return -EPERM;
 	for (; start < end; start += PAGE_SIZE) {
 		struct page *p;
-		int ret = get_user_pages(current, current->mm, start, 1,
-						0, 0, &p, NULL);
+		int ret = get_user_pages_fast(start, 1, 0, &p);
 		if (ret != 1)
 			return ret;
+		if (bhv == MADV_SOFT_OFFLINE) {
+			printk(KERN_INFO "Soft offlining page %lx at %lx\n",
+				page_to_pfn(p), start);
+			ret = soft_offline_page(p, MF_COUNT_INCREASED);
+			if (ret)
+				break;
+			continue;
+		}
 		printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n",
 		       page_to_pfn(p), start);
 		/* Ignore return value for now */
-		__memory_failure(page_to_pfn(p), 0, 1);
-		put_page(p);
+		__memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
 	}
 	return ret;
 }
@@ -335,8 +342,8 @@ SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
 	size_t len;
 
 #ifdef CONFIG_MEMORY_FAILURE
-	if (behavior == MADV_HWPOISON)
-		return madvise_hwpoison(start, start+len_in);
+	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
+		return madvise_hwpoison(behavior, start, start+len_in);
 #endif
 	if (!madvise_behavior_valid(behavior))
 		return error;
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 878808c4fcbe..488b644e0e8e 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -283,6 +283,11 @@ mem_cgroup_zoneinfo(struct mem_cgroup *mem, int nid, int zid)
 	return &mem->info.nodeinfo[nid]->zoneinfo[zid];
 }
 
+struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem)
+{
+	return &mem->css;
+}
+
 static struct mem_cgroup_per_zone *
 page_cgroup_zoneinfo(struct page_cgroup *pc)
 {
@@ -1536,25 +1541,22 @@ static struct mem_cgroup *mem_cgroup_lookup(unsigned short id)
 	return container_of(css, struct mem_cgroup, css);
 }
 
-static struct mem_cgroup *try_get_mem_cgroup_from_swapcache(struct page *page)
+struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page)
 {
-	struct mem_cgroup *mem;
+	struct mem_cgroup *mem = NULL;
 	struct page_cgroup *pc;
 	unsigned short id;
 	swp_entry_t ent;
 
 	VM_BUG_ON(!PageLocked(page));
 
-	if (!PageSwapCache(page))
-		return NULL;
-
 	pc = lookup_page_cgroup(page);
 	lock_page_cgroup(pc);
 	if (PageCgroupUsed(pc)) {
 		mem = pc->mem_cgroup;
 		if (mem && !css_tryget(&mem->css))
 			mem = NULL;
-	} else {
+	} else if (PageSwapCache(page)) {
 		ent.val = page_private(page);
 		id = lookup_swap_cgroup(ent);
 		rcu_read_lock();
@@ -1874,7 +1876,7 @@ int mem_cgroup_try_charge_swapin(struct mm_struct *mm,
 	 */
 	if (!PageSwapCache(page))
 		goto charge_cur_mm;
-	mem = try_get_mem_cgroup_from_swapcache(page);
+	mem = try_get_mem_cgroup_from_page(page);
 	if (!mem)
 		goto charge_cur_mm;
 	*ptr = mem;
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 50d4f8d7024a..6a0466ed5bfd 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -34,12 +34,16 @@
 #include <linux/kernel.h>
 #include <linux/mm.h>
 #include <linux/page-flags.h>
+#include <linux/kernel-page-flags.h>
 #include <linux/sched.h>
 #include <linux/ksm.h>
 #include <linux/rmap.h>
 #include <linux/pagemap.h>
 #include <linux/swap.h>
 #include <linux/backing-dev.h>
+#include <linux/migrate.h>
+#include <linux/page-isolation.h>
+#include <linux/suspend.h>
 #include "internal.h"
 
 int sysctl_memory_failure_early_kill __read_mostly = 0;
@@ -48,6 +52,120 @@ int sysctl_memory_failure_recovery __read_mostly = 1;
 
 atomic_long_t mce_bad_pages __read_mostly = ATOMIC_LONG_INIT(0);
 
+u32 hwpoison_filter_enable = 0;
+u32 hwpoison_filter_dev_major = ~0U;
+u32 hwpoison_filter_dev_minor = ~0U;
+u64 hwpoison_filter_flags_mask;
+u64 hwpoison_filter_flags_value;
+EXPORT_SYMBOL_GPL(hwpoison_filter_enable);
+EXPORT_SYMBOL_GPL(hwpoison_filter_dev_major);
+EXPORT_SYMBOL_GPL(hwpoison_filter_dev_minor);
+EXPORT_SYMBOL_GPL(hwpoison_filter_flags_mask);
+EXPORT_SYMBOL_GPL(hwpoison_filter_flags_value);
+
+static int hwpoison_filter_dev(struct page *p)
+{
+	struct address_space *mapping;
+	dev_t dev;
+
+	if (hwpoison_filter_dev_major == ~0U &&
+	    hwpoison_filter_dev_minor == ~0U)
+		return 0;
+
+	/*
+	 * page_mapping() does not accept slab page
+	 */
+	if (PageSlab(p))
+		return -EINVAL;
+
+	mapping = page_mapping(p);
+	if (mapping == NULL || mapping->host == NULL)
+		return -EINVAL;
+
+	dev = mapping->host->i_sb->s_dev;
+	if (hwpoison_filter_dev_major != ~0U &&
+	    hwpoison_filter_dev_major != MAJOR(dev))
+		return -EINVAL;
+	if (hwpoison_filter_dev_minor != ~0U &&
+	    hwpoison_filter_dev_minor != MINOR(dev))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int hwpoison_filter_flags(struct page *p)
+{
+	if (!hwpoison_filter_flags_mask)
+		return 0;
+
+	if ((stable_page_flags(p) & hwpoison_filter_flags_mask) ==
+				    hwpoison_filter_flags_value)
+		return 0;
+	else
+		return -EINVAL;
+}
+
+/*
+ * This allows stress tests to limit test scope to a collection of tasks
+ * by putting them under some memcg. This prevents killing unrelated/important
+ * processes such as /sbin/init. Note that the target task may share clean
+ * pages with init (eg. libc text), which is harmless. If the target task
+ * share _dirty_ pages with another task B, the test scheme must make sure B
+ * is also included in the memcg. At last, due to race conditions this filter
+ * can only guarantee that the page either belongs to the memcg tasks, or is
+ * a freed page.
+ */
+#ifdef	CONFIG_CGROUP_MEM_RES_CTLR_SWAP
+u64 hwpoison_filter_memcg;
+EXPORT_SYMBOL_GPL(hwpoison_filter_memcg);
+static int hwpoison_filter_task(struct page *p)
+{
+	struct mem_cgroup *mem;
+	struct cgroup_subsys_state *css;
+	unsigned long ino;
+
+	if (!hwpoison_filter_memcg)
+		return 0;
+
+	mem = try_get_mem_cgroup_from_page(p);
+	if (!mem)
+		return -EINVAL;
+
+	css = mem_cgroup_css(mem);
+	/* root_mem_cgroup has NULL dentries */
+	if (!css->cgroup->dentry)
+		return -EINVAL;
+
+	ino = css->cgroup->dentry->d_inode->i_ino;
+	css_put(css);
+
+	if (ino != hwpoison_filter_memcg)
+		return -EINVAL;
+
+	return 0;
+}
+#else
+static int hwpoison_filter_task(struct page *p) { return 0; }
+#endif
+
+int hwpoison_filter(struct page *p)
+{
+	if (!hwpoison_filter_enable)
+		return 0;
+
+	if (hwpoison_filter_dev(p))
+		return -EINVAL;
+
+	if (hwpoison_filter_flags(p))
+		return -EINVAL;
+
+	if (hwpoison_filter_task(p))
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(hwpoison_filter);
+
 /*
  * Send all the processes who have the page mapped an ``action optional''
  * signal.
@@ -83,6 +201,36 @@ static int kill_proc_ao(struct task_struct *t, unsigned long addr, int trapno,
 }
 
 /*
+ * When a unknown page type is encountered drain as many buffers as possible
+ * in the hope to turn the page into a LRU or free page, which we can handle.
+ */
+void shake_page(struct page *p, int access)
+{
+	if (!PageSlab(p)) {
+		lru_add_drain_all();
+		if (PageLRU(p))
+			return;
+		drain_all_pages();
+		if (PageLRU(p) || is_free_buddy_page(p))
+			return;
+	}
+
+	/*
+	 * Only all shrink_slab here (which would also
+	 * shrink other caches) if access is not potentially fatal.
+	 */
+	if (access) {
+		int nr;
+		do {
+			nr = shrink_slab(1000, GFP_KERNEL, 1000);
+			if (page_count(p) == 0)
+				break;
+		} while (nr > 10);
+	}
+}
+EXPORT_SYMBOL_GPL(shake_page);
+
+/*
  * Kill all processes that have a poisoned page mapped and then isolate
  * the page.
  *
@@ -177,7 +325,6 @@ static void kill_procs_ao(struct list_head *to_kill, int doit, int trapno,
 			 * In case something went wrong with munmapping
 			 * make sure the process doesn't catch the
 			 * signal and then access the memory. Just kill it.
-			 * the signal handlers
 			 */
 			if (fail || tk->addr_valid == 0) {
 				printk(KERN_ERR
@@ -314,33 +461,49 @@ static void collect_procs(struct page *page, struct list_head *tokill)
  */
 
 enum outcome {
-	FAILED,		/* Error handling failed */
+	IGNORED,	/* Error: cannot be handled */
+	FAILED,		/* Error: handling failed */
 	DELAYED,	/* Will be handled later */
-	IGNORED,	/* Error safely ignored */
 	RECOVERED,	/* Successfully recovered */
 };
 
 static const char *action_name[] = {
+	[IGNORED] = "Ignored",
 	[FAILED] = "Failed",
 	[DELAYED] = "Delayed",
-	[IGNORED] = "Ignored",
 	[RECOVERED] = "Recovered",
 };
 
 /*
- * Error hit kernel page.
- * Do nothing, try to be lucky and not touch this instead. For a few cases we
- * could be more sophisticated.
+ * XXX: It is possible that a page is isolated from LRU cache,
+ * and then kept in swap cache or failed to remove from page cache.
+ * The page count will stop it from being freed by unpoison.
+ * Stress tests should be aware of this memory leak problem.
  */
-static int me_kernel(struct page *p, unsigned long pfn)
+static int delete_from_lru_cache(struct page *p)
 {
-	return DELAYED;
+	if (!isolate_lru_page(p)) {
+		/*
+		 * Clear sensible page flags, so that the buddy system won't
+		 * complain when the page is unpoison-and-freed.
+		 */
+		ClearPageActive(p);
+		ClearPageUnevictable(p);
+		/*
+		 * drop the page count elevated by isolate_lru_page()
+		 */
+		page_cache_release(p);
+		return 0;
+	}
+	return -EIO;
 }
 
 /*
- * Already poisoned page.
+ * Error hit kernel page.
+ * Do nothing, try to be lucky and not touch this instead. For a few cases we
+ * could be more sophisticated.
  */
-static int me_ignore(struct page *p, unsigned long pfn)
+static int me_kernel(struct page *p, unsigned long pfn)
 {
 	return IGNORED;
 }
@@ -355,14 +518,6 @@ static int me_unknown(struct page *p, unsigned long pfn)
 }
 
 /*
- * Free memory
- */
-static int me_free(struct page *p, unsigned long pfn)
-{
-	return DELAYED;
-}
-
-/*
  * Clean (or cleaned) page cache page.
  */
 static int me_pagecache_clean(struct page *p, unsigned long pfn)
@@ -371,6 +526,8 @@ static int me_pagecache_clean(struct page *p, unsigned long pfn)
 	int ret = FAILED;
 	struct address_space *mapping;
 
+	delete_from_lru_cache(p);
+
 	/*
 	 * For anonymous pages we're done the only reference left
 	 * should be the one m_f() holds.
@@ -500,14 +657,20 @@ static int me_swapcache_dirty(struct page *p, unsigned long pfn)
 	/* Trigger EIO in shmem: */
 	ClearPageUptodate(p);
 
-	return DELAYED;
+	if (!delete_from_lru_cache(p))
+		return DELAYED;
+	else
+		return FAILED;
 }
 
 static int me_swapcache_clean(struct page *p, unsigned long pfn)
 {
 	delete_from_swap_cache(p);
 
-	return RECOVERED;
+	if (!delete_from_lru_cache(p))
+		return RECOVERED;
+	else
+		return FAILED;
 }
 
 /*
@@ -550,7 +713,6 @@ static int me_huge_page(struct page *p, unsigned long pfn)
 #define tail		(1UL << PG_tail)
 #define compound	(1UL << PG_compound)
 #define slab		(1UL << PG_slab)
-#define buddy		(1UL << PG_buddy)
 #define reserved	(1UL << PG_reserved)
 
 static struct page_state {
@@ -559,8 +721,11 @@ static struct page_state {
 	char *msg;
 	int (*action)(struct page *p, unsigned long pfn);
 } error_states[] = {
-	{ reserved,	reserved,	"reserved kernel",	me_ignore },
-	{ buddy,	buddy,		"free kernel",	me_free },
+	{ reserved,	reserved,	"reserved kernel",	me_kernel },
+	/*
+	 * free pages are specially detected outside this table:
+	 * PG_buddy pages only make a small fraction of all free pages.
+	 */
 
 	/*
 	 * Could in theory check if slab page is free or if we can drop
@@ -587,7 +752,6 @@ static struct page_state {
 
 	{ lru|dirty,	lru|dirty,	"LRU",		me_pagecache_dirty },
 	{ lru|dirty,	lru,		"clean LRU",	me_pagecache_clean },
-	{ swapbacked,	swapbacked,	"anonymous",	me_pagecache_clean },
 
 	/*
 	 * Catchall entry: must be at end.
@@ -595,20 +759,31 @@ static struct page_state {
 	{ 0,		0,		"unknown page state",	me_unknown },
 };
 
+#undef dirty
+#undef sc
+#undef unevict
+#undef mlock
+#undef writeback
+#undef lru
+#undef swapbacked
+#undef head
+#undef tail
+#undef compound
+#undef slab
+#undef reserved
+
 static void action_result(unsigned long pfn, char *msg, int result)
 {
-	struct page *page = NULL;
-	if (pfn_valid(pfn))
-		page = pfn_to_page(pfn);
+	struct page *page = pfn_to_page(pfn);
 
 	printk(KERN_ERR "MCE %#lx: %s%s page recovery: %s\n",
 		pfn,
-		page && PageDirty(page) ? "dirty " : "",
+		PageDirty(page) ? "dirty " : "",
 		msg, action_name[result]);
 }
 
 static int page_action(struct page_state *ps, struct page *p,
-			unsigned long pfn, int ref)
+			unsigned long pfn)
 {
 	int result;
 	int count;
@@ -616,18 +791,22 @@ static int page_action(struct page_state *ps, struct page *p,
 	result = ps->action(p, pfn);
 	action_result(pfn, ps->msg, result);
 
-	count = page_count(p) - 1 - ref;
-	if (count != 0)
+	count = page_count(p) - 1;
+	if (ps->action == me_swapcache_dirty && result == DELAYED)
+		count--;
+	if (count != 0) {
 		printk(KERN_ERR
 		       "MCE %#lx: %s page still referenced by %d users\n",
 		       pfn, ps->msg, count);
+		result = FAILED;
+	}
 
 	/* Could do more checks here if page looks ok */
 	/*
 	 * Could adjust zone counters here to correct for the missing page.
 	 */
 
-	return result == RECOVERED ? 0 : -EBUSY;
+	return (result == RECOVERED || result == DELAYED) ? 0 : -EBUSY;
 }
 
 #define N_UNMAP_TRIES 5
@@ -636,7 +815,7 @@ static int page_action(struct page_state *ps, struct page *p,
  * Do all that is necessary to remove user space mappings. Unmap
  * the pages and send SIGBUS to the processes if the data was dirty.
  */
-static void hwpoison_user_mappings(struct page *p, unsigned long pfn,
+static int hwpoison_user_mappings(struct page *p, unsigned long pfn,
 				  int trapno)
 {
 	enum ttu_flags ttu = TTU_UNMAP | TTU_IGNORE_MLOCK | TTU_IGNORE_ACCESS;
@@ -646,15 +825,18 @@ static void hwpoison_user_mappings(struct page *p, unsigned long pfn,
 	int i;
 	int kill = 1;
 
-	if (PageReserved(p) || PageCompound(p) || PageSlab(p) || PageKsm(p))
-		return;
+	if (PageReserved(p) || PageSlab(p))
+		return SWAP_SUCCESS;
 
 	/*
 	 * This check implies we don't kill processes if their pages
 	 * are in the swap cache early. Those are always late kills.
 	 */
 	if (!page_mapped(p))
-		return;
+		return SWAP_SUCCESS;
+
+	if (PageCompound(p) || PageKsm(p))
+		return SWAP_FAIL;
 
 	if (PageSwapCache(p)) {
 		printk(KERN_ERR
@@ -665,6 +847,8 @@ static void hwpoison_user_mappings(struct page *p, unsigned long pfn,
 	/*
 	 * Propagate the dirty bit from PTEs to struct page first, because we
 	 * need this to decide if we should kill or just drop the page.
+	 * XXX: the dirty test could be racy: set_page_dirty() may not always
+	 * be called inside page lock (it's recommended but not enforced).
 	 */
 	mapping = page_mapping(p);
 	if (!PageDirty(p) && mapping && mapping_cap_writeback_dirty(mapping)) {
@@ -716,11 +900,12 @@ static void hwpoison_user_mappings(struct page *p, unsigned long pfn,
 	 */
 	kill_procs_ao(&tokill, !!PageDirty(p), trapno,
 		      ret != SWAP_SUCCESS, pfn);
+
+	return ret;
 }
 
-int __memory_failure(unsigned long pfn, int trapno, int ref)
+int __memory_failure(unsigned long pfn, int trapno, int flags)
 {
-	unsigned long lru_flag;
 	struct page_state *ps;
 	struct page *p;
 	int res;
@@ -729,13 +914,15 @@ int __memory_failure(unsigned long pfn, int trapno, int ref)
 		panic("Memory failure from trap %d on page %lx", trapno, pfn);
 
 	if (!pfn_valid(pfn)) {
-		action_result(pfn, "memory outside kernel control", IGNORED);
-		return -EIO;
+		printk(KERN_ERR
+		       "MCE %#lx: memory outside kernel control\n",
+		       pfn);
+		return -ENXIO;
 	}
 
 	p = pfn_to_page(pfn);
 	if (TestSetPageHWPoison(p)) {
-		action_result(pfn, "already hardware poisoned", IGNORED);
+		printk(KERN_ERR "MCE %#lx: already hardware poisoned\n", pfn);
 		return 0;
 	}
 
@@ -752,9 +939,15 @@ int __memory_failure(unsigned long pfn, int trapno, int ref)
 	 * In fact it's dangerous to directly bump up page count from 0,
 	 * that may make page_freeze_refs()/page_unfreeze_refs() mismatch.
 	 */
-	if (!get_page_unless_zero(compound_head(p))) {
-		action_result(pfn, "free or high order kernel", IGNORED);
-		return PageBuddy(compound_head(p)) ? 0 : -EBUSY;
+	if (!(flags & MF_COUNT_INCREASED) &&
+		!get_page_unless_zero(compound_head(p))) {
+		if (is_free_buddy_page(p)) {
+			action_result(pfn, "free buddy", DELAYED);
+			return 0;
+		} else {
+			action_result(pfn, "high order kernel", IGNORED);
+			return -EBUSY;
+		}
 	}
 
 	/*
@@ -766,14 +959,19 @@ int __memory_failure(unsigned long pfn, int trapno, int ref)
 	 * walked by the page reclaim code, however that's not a big loss.
 	 */
 	if (!PageLRU(p))
-		lru_add_drain_all();
-	lru_flag = p->flags & lru;
-	if (isolate_lru_page(p)) {
+		shake_page(p, 0);
+	if (!PageLRU(p)) {
+		/*
+		 * shake_page could have turned it free.
+		 */
+		if (is_free_buddy_page(p)) {
+			action_result(pfn, "free buddy, 2nd try", DELAYED);
+			return 0;
+		}
 		action_result(pfn, "non LRU", IGNORED);
 		put_page(p);
 		return -EBUSY;
 	}
-	page_cache_release(p);
 
 	/*
 	 * Lock the page and wait for writeback to finish.
@@ -781,26 +979,48 @@ int __memory_failure(unsigned long pfn, int trapno, int ref)
 	 * and in many cases impossible, so we just avoid it here.
 	 */
 	lock_page_nosync(p);
+
+	/*
+	 * unpoison always clear PG_hwpoison inside page lock
+	 */
+	if (!PageHWPoison(p)) {
+		printk(KERN_ERR "MCE %#lx: just unpoisoned\n", pfn);
+		res = 0;
+		goto out;
+	}
+	if (hwpoison_filter(p)) {
+		if (TestClearPageHWPoison(p))
+			atomic_long_dec(&mce_bad_pages);
+		unlock_page(p);
+		put_page(p);
+		return 0;
+	}
+
 	wait_on_page_writeback(p);
 
 	/*
 	 * Now take care of user space mappings.
+	 * Abort on fail: __remove_from_page_cache() assumes unmapped page.
 	 */
-	hwpoison_user_mappings(p, pfn, trapno);
+	if (hwpoison_user_mappings(p, pfn, trapno) != SWAP_SUCCESS) {
+		printk(KERN_ERR "MCE %#lx: cannot unmap page, give up\n", pfn);
+		res = -EBUSY;
+		goto out;
+	}
 
 	/*
 	 * Torn down by someone else?
 	 */
-	if ((lru_flag & lru) && !PageSwapCache(p) && p->mapping == NULL) {
+	if (PageLRU(p) && !PageSwapCache(p) && p->mapping == NULL) {
 		action_result(pfn, "already truncated LRU", IGNORED);
-		res = 0;
+		res = -EBUSY;
 		goto out;
 	}
 
 	res = -EBUSY;
 	for (ps = error_states;; ps++) {
-		if (((p->flags | lru_flag)& ps->mask) == ps->res) {
-			res = page_action(ps, p, pfn, ref);
+		if ((p->flags & ps->mask) == ps->res) {
+			res = page_action(ps, p, pfn);
 			break;
 		}
 	}
@@ -831,3 +1051,235 @@ void memory_failure(unsigned long pfn, int trapno)
 {
 	__memory_failure(pfn, trapno, 0);
 }
+
+/**
+ * unpoison_memory - Unpoison a previously poisoned page
+ * @pfn: Page number of the to be unpoisoned page
+ *
+ * Software-unpoison a page that has been poisoned by
+ * memory_failure() earlier.
+ *
+ * This is only done on the software-level, so it only works
+ * for linux injected failures, not real hardware failures
+ *
+ * Returns 0 for success, otherwise -errno.
+ */
+int unpoison_memory(unsigned long pfn)
+{
+	struct page *page;
+	struct page *p;
+	int freeit = 0;
+
+	if (!pfn_valid(pfn))
+		return -ENXIO;
+
+	p = pfn_to_page(pfn);
+	page = compound_head(p);
+
+	if (!PageHWPoison(p)) {
+		pr_debug("MCE: Page was already unpoisoned %#lx\n", pfn);
+		return 0;
+	}
+
+	if (!get_page_unless_zero(page)) {
+		if (TestClearPageHWPoison(p))
+			atomic_long_dec(&mce_bad_pages);
+		pr_debug("MCE: Software-unpoisoned free page %#lx\n", pfn);
+		return 0;
+	}
+
+	lock_page_nosync(page);
+	/*
+	 * This test is racy because PG_hwpoison is set outside of page lock.
+	 * That's acceptable because that won't trigger kernel panic. Instead,
+	 * the PG_hwpoison page will be caught and isolated on the entrance to
+	 * the free buddy page pool.
+	 */
+	if (TestClearPageHWPoison(p)) {
+		pr_debug("MCE: Software-unpoisoned page %#lx\n", pfn);
+		atomic_long_dec(&mce_bad_pages);
+		freeit = 1;
+	}
+	unlock_page(page);
+
+	put_page(page);
+	if (freeit)
+		put_page(page);
+
+	return 0;
+}
+EXPORT_SYMBOL(unpoison_memory);
+
+static struct page *new_page(struct page *p, unsigned long private, int **x)
+{
+	int nid = page_to_nid(p);
+	return alloc_pages_exact_node(nid, GFP_HIGHUSER_MOVABLE, 0);
+}
+
+/*
+ * Safely get reference count of an arbitrary page.
+ * Returns 0 for a free page, -EIO for a zero refcount page
+ * that is not free, and 1 for any other page type.
+ * For 1 the page is returned with increased page count, otherwise not.
+ */
+static int get_any_page(struct page *p, unsigned long pfn, int flags)
+{
+	int ret;
+
+	if (flags & MF_COUNT_INCREASED)
+		return 1;
+
+	/*
+	 * The lock_system_sleep prevents a race with memory hotplug,
+	 * because the isolation assumes there's only a single user.
+	 * This is a big hammer, a better would be nicer.
+	 */
+	lock_system_sleep();
+
+	/*
+	 * Isolate the page, so that it doesn't get reallocated if it
+	 * was free.
+	 */
+	set_migratetype_isolate(p);
+	if (!get_page_unless_zero(compound_head(p))) {
+		if (is_free_buddy_page(p)) {
+			pr_debug("get_any_page: %#lx free buddy page\n", pfn);
+			/* Set hwpoison bit while page is still isolated */
+			SetPageHWPoison(p);
+			ret = 0;
+		} else {
+			pr_debug("get_any_page: %#lx: unknown zero refcount page type %lx\n",
+				pfn, p->flags);
+			ret = -EIO;
+		}
+	} else {
+		/* Not a free page */
+		ret = 1;
+	}
+	unset_migratetype_isolate(p);
+	unlock_system_sleep();
+	return ret;
+}
+
+/**
+ * soft_offline_page - Soft offline a page.
+ * @page: page to offline
+ * @flags: flags. Same as memory_failure().
+ *
+ * Returns 0 on success, otherwise negated errno.
+ *
+ * Soft offline a page, by migration or invalidation,
+ * without killing anything. This is for the case when
+ * a page is not corrupted yet (so it's still valid to access),
+ * but has had a number of corrected errors and is better taken
+ * out.
+ *
+ * The actual policy on when to do that is maintained by
+ * user space.
+ *
+ * This should never impact any application or cause data loss,
+ * however it might take some time.
+ *
+ * This is not a 100% solution for all memory, but tries to be
+ * ``good enough'' for the majority of memory.
+ */
+int soft_offline_page(struct page *page, int flags)
+{
+	int ret;
+	unsigned long pfn = page_to_pfn(page);
+
+	ret = get_any_page(page, pfn, flags);
+	if (ret < 0)
+		return ret;
+	if (ret == 0)
+		goto done;
+
+	/*
+	 * Page cache page we can handle?
+	 */
+	if (!PageLRU(page)) {
+		/*
+		 * Try to free it.
+		 */
+		put_page(page);
+		shake_page(page, 1);
+
+		/*
+		 * Did it turn free?
+		 */
+		ret = get_any_page(page, pfn, 0);
+		if (ret < 0)
+			return ret;
+		if (ret == 0)
+			goto done;
+	}
+	if (!PageLRU(page)) {
+		pr_debug("soft_offline: %#lx: unknown non LRU page type %lx\n",
+				pfn, page->flags);
+		return -EIO;
+	}
+
+	lock_page(page);
+	wait_on_page_writeback(page);
+
+	/*
+	 * Synchronized using the page lock with memory_failure()
+	 */
+	if (PageHWPoison(page)) {
+		unlock_page(page);
+		put_page(page);
+		pr_debug("soft offline: %#lx page already poisoned\n", pfn);
+		return -EBUSY;
+	}
+
+	/*
+	 * Try to invalidate first. This should work for
+	 * non dirty unmapped page cache pages.
+	 */
+	ret = invalidate_inode_page(page);
+	unlock_page(page);
+
+	/*
+	 * Drop count because page migration doesn't like raised
+	 * counts. The page could get re-allocated, but if it becomes
+	 * LRU the isolation will just fail.
+	 * RED-PEN would be better to keep it isolated here, but we
+	 * would need to fix isolation locking first.
+	 */
+	put_page(page);
+	if (ret == 1) {
+		ret = 0;
+		pr_debug("soft_offline: %#lx: invalidated\n", pfn);
+		goto done;
+	}
+
+	/*
+	 * Simple invalidation didn't work.
+	 * Try to migrate to a new page instead. migrate.c
+	 * handles a large number of cases for us.
+	 */
+	ret = isolate_lru_page(page);
+	if (!ret) {
+		LIST_HEAD(pagelist);
+
+		list_add(&page->lru, &pagelist);
+		ret = migrate_pages(&pagelist, new_page, MPOL_MF_MOVE_ALL, 0);
+		if (ret) {
+			pr_debug("soft offline: %#lx: migration failed %d, type %lx\n",
+				pfn, ret, page->flags);
+			if (ret > 0)
+				ret = -EIO;
+		}
+	} else {
+		pr_debug("soft offline: %#lx: isolation failed: %d, page count %d, type %lx\n",
+				pfn, ret, page_count(page), page->flags);
+	}
+	if (ret)
+		return ret;
+
+done:
+	atomic_long_add(1, &mce_bad_pages);
+	SetPageHWPoison(page);
+	/* keep elevated page count for bad page */
+	return ret;
+}
diff --git a/mm/memory.c b/mm/memory.c
index aed45eaf8ac9..09e4b1be7b67 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2555,6 +2555,10 @@ static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
 		ret = VM_FAULT_MAJOR;
 		count_vm_event(PGMAJFAULT);
 	} else if (PageHWPoison(page)) {
+		/*
+		 * hwpoisoned dirty swapcache pages are kept for killing
+		 * owner processes (which may be unknown at hwpoison time)
+		 */
 		ret = VM_FAULT_HWPOISON;
 		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 		goto out_release;
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 850c4a7e2fe5..74af449b1f1d 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5091,3 +5091,24 @@ __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
 	spin_unlock_irqrestore(&zone->lock, flags);
 }
 #endif
+
+#ifdef CONFIG_MEMORY_FAILURE
+bool is_free_buddy_page(struct page *page)
+{
+	struct zone *zone = page_zone(page);
+	unsigned long pfn = page_to_pfn(page);
+	unsigned long flags;
+	int order;
+
+	spin_lock_irqsave(&zone->lock, flags);
+	for (order = 0; order < MAX_ORDER; order++) {
+		struct page *page_head = page - (pfn & ((1 << order) - 1));
+
+		if (PageBuddy(page_head) && page_order(page_head) >= order)
+			break;
+	}
+	spin_unlock_irqrestore(&zone->lock, flags);
+
+	return order < MAX_ORDER;
+}
+#endif