From bd1ce5f91f545730df4af492f774d9d32f5da3cb Mon Sep 17 00:00:00 2001 From: Wu Fengguang Date: Wed, 16 Dec 2009 12:19:57 +0100 Subject: HWPOISON: avoid grabbing the page count multiple times during madvise injection If page is double referenced in madvise_hwpoison() and __memory_failure(), remove_mapping() will fail because it expects page_count=2. Fix it by not grabbing extra page count in __memory_failure(). Signed-off-by: Wu Fengguang Signed-off-by: Andi Kleen --- mm/madvise.c | 1 - 1 file changed, 1 deletion(-) (limited to 'mm/madvise.c') diff --git a/mm/madvise.c b/mm/madvise.c index 35b1479b7c9d..18970aec0d2f 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -238,7 +238,6 @@ static int madvise_hwpoison(unsigned long start, unsigned long end) page_to_pfn(p), start); /* Ignore return value for now */ __memory_failure(page_to_pfn(p), 0, 1); - put_page(p); } return ret; } -- cgit v1.2.3 From 82ba011b9041dd31c15e4f63797b08aa0a288e61 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 16 Dec 2009 12:19:57 +0100 Subject: HWPOISON: Turn ref argument into flags argument Now that "ref" is just a boolean turn it into a flags argument. First step is only a single flag that makes the code's intention more clear, but more may follow. Signed-off-by: Andi Kleen --- include/linux/mm.h | 5 ++++- mm/madvise.c | 2 +- mm/memory-failure.c | 5 +++-- 3 files changed, 8 insertions(+), 4 deletions(-) (limited to 'mm/madvise.c') diff --git a/include/linux/mm.h b/include/linux/mm.h index 68c84bb2ad3f..135e19198cd3 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1331,8 +1331,11 @@ extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim, size_t size); extern void refund_locked_memory(struct mm_struct *mm, size_t size); +enum mf_flags { + MF_COUNT_INCREASED = 1 << 0, +}; extern void memory_failure(unsigned long pfn, int trapno); -extern int __memory_failure(unsigned long pfn, int trapno, int ref); +extern int __memory_failure(unsigned long pfn, int trapno, int flags); extern int sysctl_memory_failure_early_kill; extern int sysctl_memory_failure_recovery; extern void shake_page(struct page *p); diff --git a/mm/madvise.c b/mm/madvise.c index 18970aec0d2f..6ca34f0cd4aa 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -237,7 +237,7 @@ static int madvise_hwpoison(unsigned long start, unsigned long end) 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); + __memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED); } return ret; } diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 4253e14fa709..3338c443272c 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -737,7 +737,7 @@ static void hwpoison_user_mappings(struct page *p, unsigned long pfn, ret != SWAP_SUCCESS, pfn); } -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; @@ -773,7 +773,8 @@ 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 (!ref && !get_page_unless_zero(compound_head(p))) { + if (!(flags & MF_COUNT_INCREASED) && + !get_page_unless_zero(compound_head(p))) { action_result(pfn, "free or high order kernel", IGNORED); return PageBuddy(compound_head(p)) ? 0 : -EBUSY; } -- cgit v1.2.3 From d15f107d97bd74c74d8f5144843d372666ddbdac Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 16 Dec 2009 12:20:00 +0100 Subject: HWPOISON: Use get_user_page_fast in hwpoison madvise The previous version didn't take the mmap_sem before calling gup(), which is racy. Use get_user_pages_fast() instead which doesn't need any locks. This is also faster of course, but then it doesn't really matter because this is just a testing path. Based on report from Nick Piggin. Cc: npiggin@suse.de Signed-off-by: Andi Kleen --- mm/madvise.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'mm/madvise.c') diff --git a/mm/madvise.c b/mm/madvise.c index 6ca34f0cd4aa..7964e36ba915 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -230,8 +230,7 @@ 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; printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n", -- cgit v1.2.3 From afcf938ee0aac4ef95b1a23bac704c6fbeb26de6 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 16 Dec 2009 12:20:00 +0100 Subject: HWPOISON: Add a madvise() injector for soft page offlining Process based injection is much easier to handle for test programs, who can first bring a page into a specific state and then test. So add a new MADV_SOFT_OFFLINE to soft offline a page, similar to the existing hard offline injector. Signed-off-by: Andi Kleen --- include/asm-generic/mman-common.h | 1 + mm/madvise.c | 15 ++++++++++++--- 2 files changed, 13 insertions(+), 3 deletions(-) (limited to 'mm/madvise.c') diff --git a/include/asm-generic/mman-common.h b/include/asm-generic/mman-common.h index 20111265afd8..3da9e2742fa0 100644 --- a/include/asm-generic/mman-common.h +++ b/include/asm-generic/mman-common.h @@ -40,6 +40,7 @@ #define MADV_DONTFORK 10 /* don't inherit across fork */ #define MADV_DOFORK 11 /* do inherit across fork */ #define MADV_HWPOISON 100 /* poison a page for testing */ +#define MADV_SOFT_OFFLINE 101 /* soft offline page for testing */ #define MADV_MERGEABLE 12 /* KSM may merge identical pages */ #define MADV_UNMERGEABLE 13 /* KSM may not merge identical pages */ diff --git a/mm/madvise.c b/mm/madvise.c index 7964e36ba915..319528b8db74 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -9,6 +9,7 @@ #include #include #include +#include #include #include #include @@ -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; @@ -233,6 +234,14 @@ static int madvise_hwpoison(unsigned long start, unsigned long end) 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 */ @@ -333,8 +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; -- cgit v1.2.3