From: Wu Fengguang <fengguang.wu@intel.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: LKML <linux-kernel@vger.kernel.org>,
Wu Fengguang <fengguang.wu@intel.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Andi Kleen <andi@firstfloor.org>,
"riel@redhat.com" <riel@redhat.com>,
"chris.mason@oracle.com" <chris.mason@oracle.com>,
"linux-mm@kvack.org" <linux-mm@kvack.org>
Subject: [PATCH 4/5] HWPOISON: report sticky EIO for poisoned file
Date: Thu, 11 Jun 2009 22:22:43 +0800 [thread overview]
Message-ID: <20090611144430.813191526@intel.com> (raw)
In-Reply-To: 20090611142239.192891591@intel.com
[-- Attachment #1: hwpoison-more-sticky-eio.patch --]
[-- Type: text/plain, Size: 2562 bytes --]
This makes the EIO reports on write(), fsync(), or the NFS close()
sticky enough. The only way to get rid of it may be
echo 3 > /proc/sys/vm/drop_caches
Note that the impacted process will only be killed if it mapped the page.
XXX
via read()/write()/fsync() instead of memory mapped reads/writes, simply
because it's very hard to find them.
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
---
include/linux/pagemap.h | 13 +++++++++++++
mm/filemap.c | 11 +++++++++++
mm/memory-failure.c | 2 +-
3 files changed, 25 insertions(+), 1 deletion(-)
--- sound-2.6.orig/include/linux/pagemap.h
+++ sound-2.6/include/linux/pagemap.h
@@ -23,6 +23,7 @@ enum mapping_flags {
AS_ENOSPC = __GFP_BITS_SHIFT + 1, /* ENOSPC on async write */
AS_MM_ALL_LOCKS = __GFP_BITS_SHIFT + 2, /* under mm_take_all_locks() */
AS_UNEVICTABLE = __GFP_BITS_SHIFT + 3, /* e.g., ramdisk, SHM_LOCK */
+ AS_HWPOISON = __GFP_BITS_SHIFT + 4, /* hardware memory corruption */
};
static inline void mapping_set_error(struct address_space *mapping, int error)
@@ -52,6 +53,18 @@ static inline int mapping_unevictable(st
return !!mapping;
}
+#ifdef CONFIG_MEMORY_FAILURE
+static inline int mapping_hwpoison(struct address_space *mapping)
+{
+ return test_bit(AS_HWPOISON, &mapping->flags);
+}
+#else
+static inline int mapping_hwpoison(struct address_space *mapping)
+{
+ return 0;
+}
+#endif
+
static inline gfp_t mapping_gfp_mask(struct address_space * mapping)
{
return (__force gfp_t)mapping->flags & __GFP_BITS_MASK;
--- sound-2.6.orig/mm/filemap.c
+++ sound-2.6/mm/filemap.c
@@ -302,6 +302,8 @@ int wait_on_page_writeback_range(struct
ret = -ENOSPC;
if (test_and_clear_bit(AS_EIO, &mapping->flags))
ret = -EIO;
+ if (mapping_hwpoison(mapping))
+ ret = -EIO;
return ret;
}
@@ -460,6 +462,15 @@ int add_to_page_cache_locked(struct page
VM_BUG_ON(!PageLocked(page));
+ /*
+ * Hardware corrupted page will be removed from mapping,
+ * so we want to deny (possibly) reloading the old data.
+ */
+ if (unlikely(mapping_hwpoison(mapping))) {
+ error = -EIO;
+ goto out;
+ }
+
error = mem_cgroup_cache_charge(page, current->mm,
gfp_mask & GFP_RECLAIM_MASK);
if (error)
--- sound-2.6.orig/mm/memory-failure.c
+++ sound-2.6/mm/memory-failure.c
@@ -184,7 +184,7 @@ static int me_pagecache_dirty(struct pag
* the first EIO, but we're not worse than other parts
* of the kernel.
*/
- mapping_set_error(mapping, EIO);
+ set_bit(AS_HWPOISON, &mapping->flags);
}
return me_pagecache_clean(p, pfn);
--
WARNING: multiple messages have this Message-ID (diff)
From: Wu Fengguang <fengguang.wu@intel.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: LKML <linux-kernel@vger.kernel.org>,
Wu Fengguang <fengguang.wu@intel.com>,
Hugh Dickins <hugh.dickins@tiscali.co.uk>,
Nick Piggin <npiggin@suse.de>, Andi Kleen <andi@firstfloor.org>,
"riel@redhat.com" <riel@redhat.com>,
"chris.mason@oracle.com" <chris.mason@oracle.com>,
"linux-mm@kvack.org" <linux-mm@kvack.org>
Subject: [PATCH 4/5] HWPOISON: report sticky EIO for poisoned file
Date: Thu, 11 Jun 2009 22:22:43 +0800 [thread overview]
Message-ID: <20090611144430.813191526@intel.com> (raw)
In-Reply-To: 20090611142239.192891591@intel.com
[-- Attachment #1: hwpoison-more-sticky-eio.patch --]
[-- Type: text/plain, Size: 2787 bytes --]
This makes the EIO reports on write(), fsync(), or the NFS close()
sticky enough. The only way to get rid of it may be
echo 3 > /proc/sys/vm/drop_caches
Note that the impacted process will only be killed if it mapped the page.
XXX
via read()/write()/fsync() instead of memory mapped reads/writes, simply
because it's very hard to find them.
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
---
include/linux/pagemap.h | 13 +++++++++++++
mm/filemap.c | 11 +++++++++++
mm/memory-failure.c | 2 +-
3 files changed, 25 insertions(+), 1 deletion(-)
--- sound-2.6.orig/include/linux/pagemap.h
+++ sound-2.6/include/linux/pagemap.h
@@ -23,6 +23,7 @@ enum mapping_flags {
AS_ENOSPC = __GFP_BITS_SHIFT + 1, /* ENOSPC on async write */
AS_MM_ALL_LOCKS = __GFP_BITS_SHIFT + 2, /* under mm_take_all_locks() */
AS_UNEVICTABLE = __GFP_BITS_SHIFT + 3, /* e.g., ramdisk, SHM_LOCK */
+ AS_HWPOISON = __GFP_BITS_SHIFT + 4, /* hardware memory corruption */
};
static inline void mapping_set_error(struct address_space *mapping, int error)
@@ -52,6 +53,18 @@ static inline int mapping_unevictable(st
return !!mapping;
}
+#ifdef CONFIG_MEMORY_FAILURE
+static inline int mapping_hwpoison(struct address_space *mapping)
+{
+ return test_bit(AS_HWPOISON, &mapping->flags);
+}
+#else
+static inline int mapping_hwpoison(struct address_space *mapping)
+{
+ return 0;
+}
+#endif
+
static inline gfp_t mapping_gfp_mask(struct address_space * mapping)
{
return (__force gfp_t)mapping->flags & __GFP_BITS_MASK;
--- sound-2.6.orig/mm/filemap.c
+++ sound-2.6/mm/filemap.c
@@ -302,6 +302,8 @@ int wait_on_page_writeback_range(struct
ret = -ENOSPC;
if (test_and_clear_bit(AS_EIO, &mapping->flags))
ret = -EIO;
+ if (mapping_hwpoison(mapping))
+ ret = -EIO;
return ret;
}
@@ -460,6 +462,15 @@ int add_to_page_cache_locked(struct page
VM_BUG_ON(!PageLocked(page));
+ /*
+ * Hardware corrupted page will be removed from mapping,
+ * so we want to deny (possibly) reloading the old data.
+ */
+ if (unlikely(mapping_hwpoison(mapping))) {
+ error = -EIO;
+ goto out;
+ }
+
error = mem_cgroup_cache_charge(page, current->mm,
gfp_mask & GFP_RECLAIM_MASK);
if (error)
--- sound-2.6.orig/mm/memory-failure.c
+++ sound-2.6/mm/memory-failure.c
@@ -184,7 +184,7 @@ static int me_pagecache_dirty(struct pag
* the first EIO, but we're not worse than other parts
* of the kernel.
*/
- mapping_set_error(mapping, EIO);
+ set_bit(AS_HWPOISON, &mapping->flags);
}
return me_pagecache_clean(p, pfn);
--
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2009-06-11 14:54 UTC|newest]
Thread overview: 84+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-06-11 14:22 [PATCH 0/5] [RFC] HWPOISON incremental fixes Wu Fengguang
2009-06-11 14:22 ` Wu Fengguang
2009-06-11 14:22 ` [PATCH 1/5] HWPOISON: define VM_FAULT_HWPOISON to 0 when feature is disabled Wu Fengguang
2009-06-11 14:22 ` Wu Fengguang
2009-06-11 15:44 ` Rik van Riel
2009-06-11 15:44 ` Rik van Riel
2009-06-12 10:00 ` Andi Kleen
2009-06-12 10:00 ` Andi Kleen
2009-06-12 13:15 ` Wu Fengguang
2009-06-12 13:15 ` Wu Fengguang
2009-06-12 11:22 ` Ingo Molnar
2009-06-12 11:22 ` Ingo Molnar
2009-06-12 12:57 ` Wu Fengguang
2009-06-12 12:57 ` Wu Fengguang
2009-06-12 13:17 ` Ingo Molnar
2009-06-12 13:17 ` Ingo Molnar
2009-06-12 13:33 ` Wu Fengguang
2009-06-12 13:33 ` Wu Fengguang
2009-06-12 15:36 ` Ingo Molnar
2009-06-12 15:36 ` Ingo Molnar
2009-06-12 16:14 ` Wu Fengguang
2009-06-12 16:14 ` Wu Fengguang
2009-06-12 18:07 ` Alan Cox
2009-06-12 18:07 ` Alan Cox
2009-06-12 17:55 ` Theodore Tso
2009-06-12 17:55 ` Theodore Tso
2009-06-12 13:58 ` Andi Kleen
2009-06-12 13:58 ` Andi Kleen
2009-06-12 15:28 ` Linus Torvalds
2009-06-12 15:28 ` Linus Torvalds
2009-06-12 15:35 ` Ingo Molnar
2009-06-12 15:35 ` Ingo Molnar
2009-06-12 16:05 ` Rik van Riel
2009-06-12 16:05 ` Rik van Riel
2009-06-12 16:37 ` H. Peter Anvin
2009-06-12 16:37 ` H. Peter Anvin
2009-06-12 16:48 ` Ingo Molnar
2009-06-12 16:48 ` Ingo Molnar
2009-06-15 7:04 ` Nick Piggin
2009-06-15 7:04 ` Nick Piggin
2009-06-15 6:52 ` Nick Piggin
2009-06-15 6:52 ` Nick Piggin
2009-06-16 20:27 ` Russ Anderson
2009-06-16 20:27 ` Russ Anderson
2009-06-17 7:51 ` Nick Piggin
2009-06-17 7:51 ` Nick Piggin
2009-06-12 15:45 ` Ingo Molnar
2009-06-12 15:45 ` Ingo Molnar
2009-06-12 16:12 ` Linus Torvalds
2009-06-12 16:12 ` Linus Torvalds
2009-06-11 14:22 ` [PATCH 2/5] HWPOISON: fix tasklist_lock/anon_vma locking order Wu Fengguang
2009-06-11 14:22 ` Wu Fengguang
2009-06-11 15:59 ` Rik van Riel
2009-06-11 15:59 ` Rik van Riel
2009-06-12 10:03 ` Andi Kleen
2009-06-12 10:03 ` Andi Kleen
2009-06-12 10:07 ` Nick Piggin
2009-06-12 10:07 ` Nick Piggin
2009-06-12 13:27 ` Wu Fengguang
2009-06-12 13:27 ` Wu Fengguang
2009-06-12 14:04 ` Wu Fengguang
2009-06-12 14:04 ` Wu Fengguang
2009-06-11 14:22 ` [PATCH 3/5] HWPOISON: remove early kill option for now Wu Fengguang
2009-06-11 14:22 ` Wu Fengguang
2009-06-11 16:06 ` Rik van Riel
2009-06-11 16:06 ` Rik van Riel
2009-06-12 9:59 ` Andi Kleen
2009-06-12 9:59 ` Andi Kleen
2009-06-11 14:22 ` Wu Fengguang [this message]
2009-06-11 14:22 ` [PATCH 4/5] HWPOISON: report sticky EIO for poisoned file Wu Fengguang
2009-06-11 16:31 ` Rik van Riel
2009-06-11 16:31 ` Rik van Riel
2009-06-12 10:07 ` Andi Kleen
2009-06-12 10:07 ` Andi Kleen
2009-06-12 13:41 ` Wu Fengguang
2009-06-12 13:41 ` Wu Fengguang
2009-06-11 14:22 ` [PATCH 5/5] HWPOISON: use the safer invalidate page for possible metadata pages Wu Fengguang
2009-06-11 14:22 ` Wu Fengguang
2009-06-11 16:36 ` Rik van Riel
2009-06-11 16:36 ` Rik van Riel
2009-06-12 10:56 ` [PATCH 0/5] [RFC] HWPOISON incremental fixes Andi Kleen
2009-06-12 10:56 ` Andi Kleen
2009-06-12 13:59 ` Wu Fengguang
2009-06-12 13:59 ` Wu Fengguang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090611144430.813191526@intel.com \
--to=fengguang.wu@intel.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.