linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Andi Kleen <andi@firstfloor.org>
To: fengguang.wu@intel.com, haicheng.li@intel.com,
	npiggin@suse.defengguang.wu@intel.com,
	linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: [PATCH] [17/31] HWPOISON: add fs/device filters
Date: Tue,  8 Dec 2009 22:16:33 +0100 (CET)	[thread overview]
Message-ID: <20091208211633.71135B151F@basil.firstfloor.org> (raw)
In-Reply-To: <200912081016.198135742@firstfloor.org>


From: Wu Fengguang <fengguang.wu@intel.com>

Filesystem data/metadata present the most tricky-to-isolate pages.
It requires careful code review and stress testing to get them right.

The fs/device filter helps to target the stress tests to some specific
filesystem pages. The filter condition is block device's major/minor
numbers:
        - corrupt-filter-dev-major
        - corrupt-filter-dev-minor
When specified (non -1), only page cache pages that belong to that
device will be poisoned.

The filters are checked reliably on the locked and refcounted page.

Haicheng: clear PG_hwpoison and drop bad page count if filter not OK
AK: Add documentation

CC: Haicheng Li <haicheng.li@intel.com>
CC: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>

---
 Documentation/vm/hwpoison.txt |    7 +++++
 mm/hwpoison-inject.c          |   11 +++++++++
 mm/internal.h                 |    3 ++
 mm/memory-failure.c           |   51 ++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 72 insertions(+)

Index: linux/mm/memory-failure.c
===================================================================
--- linux.orig/mm/memory-failure.c
+++ linux/mm/memory-failure.c
@@ -48,6 +48,50 @@ int sysctl_memory_failure_recovery __rea
 
 atomic_long_t mce_bad_pages __read_mostly = ATOMIC_LONG_INIT(0);
 
+u32 hwpoison_filter_dev_major = ~0U;
+u32 hwpoison_filter_dev_minor = ~0U;
+EXPORT_SYMBOL_GPL(hwpoison_filter_dev_major);
+EXPORT_SYMBOL_GPL(hwpoison_filter_dev_minor);
+
+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;
+}
+
+int hwpoison_filter(struct page *p)
+{
+	if (hwpoison_filter_dev(p))
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(hwpoison_filter);
+
 /*
  * Send all the processes who have the page mapped an ``action optional''
  * signal.
@@ -845,6 +889,13 @@ int __memory_failure(unsigned long 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);
 
Index: linux/mm/hwpoison-inject.c
===================================================================
--- linux.orig/mm/hwpoison-inject.c
+++ linux/mm/hwpoison-inject.c
@@ -3,6 +3,7 @@
 #include <linux/debugfs.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
+#include "internal.h"
 
 static struct dentry *hwpoison_dir;
 
@@ -54,6 +55,16 @@ static int pfn_inject_init(void)
 	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;
+
 	return 0;
 fail:
 	pfn_inject_exit();
Index: linux/mm/internal.h
===================================================================
--- linux.orig/mm/internal.h
+++ linux/mm/internal.h
@@ -263,3 +263,6 @@ int __get_user_pages(struct task_struct
 #define ZONE_RECLAIM_SOME	0
 #define ZONE_RECLAIM_SUCCESS	1
 #endif
+
+extern u32 hwpoison_filter_dev_major;
+extern u32 hwpoison_filter_dev_minor;
Index: linux/Documentation/vm/hwpoison.txt
===================================================================
--- linux.orig/Documentation/vm/hwpoison.txt
+++ linux/Documentation/vm/hwpoison.txt
@@ -115,6 +115,13 @@ memory failures.
 Note these injection interfaces are not stable and might change between
 kernel versions
 
+corrupt-filter-dev-major
+corrupt-filter-dev-minor
+
+Only handle memory failures to pages associated with the file system defined
+by block device major/minor.  -1U is the wildcard value.
+This should be only used for testing with artificial injection.
+
 Architecture specific MCE injector
 
 x86 has mce-inject, mce-test

--
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>

  parent reply	other threads:[~2009-12-08 21:18 UTC|newest]

Thread overview: 61+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-12-08 21:16 [PATCH] [0/31] HWPOISON 2.6.33 pre-merge posting Andi Kleen
2009-12-08 21:16 ` [PATCH] [1/31] HWPOISON: Add Andi Kleen as hwpoison maintainer to MAINTAINERS Andi Kleen
2009-12-08 21:16 ` [PATCH] [2/31] HWPOISON: Be more aggressive at freeing non LRU caches Andi Kleen
2009-12-08 21:16 ` [PATCH] [3/31] page-types: add standard GPL license header Andi Kleen
2009-12-08 21:16 ` [PATCH] [4/31] HWPOISON: remove the anonymous entry Andi Kleen
2009-12-08 21:16 ` [PATCH] [5/31] HWPOISON: return ENXIO on invalid page number Andi Kleen
2009-12-08 21:16 ` [PATCH] [6/31] HWPOISON: avoid grabbing the page count multiple times during madvise injection Andi Kleen
2009-12-08 21:16 ` [PATCH] [7/31] HWPOISON: Turn ref argument into flags argument Andi Kleen
2009-12-08 21:16 ` [PATCH] [8/31] HWPOISON: abort on failed unmap Andi Kleen
2009-12-08 21:16 ` [PATCH] [9/31] HWPOISON: comment the possible set_page_dirty() race Andi Kleen
2009-12-08 21:16 ` [PATCH] [10/31] HWPOISON: comment dirty swapcache pages Andi Kleen
2009-12-08 21:16 ` [PATCH] [11/31] HWPOISON: introduce delete_from_lru_cache() Andi Kleen
2009-12-08 21:16 ` [PATCH] [12/31] HWPOISON: remove the free buddy page handler Andi Kleen
2009-12-08 21:16 ` [PATCH] [13/31] HWPOISON: detect free buddy pages explicitly Andi Kleen
2009-12-08 21:16 ` [PATCH] [14/31] HWPOISON: Add unpoisoning support Andi Kleen
2009-12-08 21:16 ` [PATCH] [15/31] HWPOISON: make semantics of IGNORED/DELAYED clear Andi Kleen
2009-12-08 21:16 ` [PATCH] [16/31] HWPOISON: return 0 to indicate success reliably Andi Kleen
2009-12-08 21:16 ` Andi Kleen [this message]
2009-12-08 21:16 ` [PATCH] [18/31] HWPOISON: limit hwpoison injector to known page types Andi Kleen
2009-12-08 21:16 ` [PATCH] [19/31] mm: export stable page flags Andi Kleen
2009-12-08 22:27   ` Matt Mackall
2009-12-09  2:00     ` Wu Fengguang
2009-12-09 21:38       ` Matt Mackall
2009-12-10  1:50       ` Andi Kleen
2009-12-10  2:09         ` Wu Fengguang
2009-12-10 13:42           ` Andi Kleen
2009-12-08 21:16 ` [PATCH] [20/31] HWPOISON: add page flags filter Andi Kleen
2009-12-08 21:16 ` [PATCH] [21/31] memcg: rename and export try_get_mem_cgroup_from_page() Andi Kleen
2009-12-08 21:16 ` [PATCH] [22/31] memcg: add accessor to mem_cgroup.css Andi Kleen
2009-12-08 21:16 ` [PATCH] [23/31] HWPOISON: add memory cgroup filter Andi Kleen
2009-12-09  5:04   ` Li Zefan
2009-12-09  5:06     ` KAMEZAWA Hiroyuki
2009-12-09  5:33       ` Balbir Singh
2009-12-09  9:15     ` Andi Kleen
2009-12-09 20:47   ` Paul Menage
2009-12-09 23:56     ` KAMEZAWA Hiroyuki
2009-12-10  1:42     ` Andi Kleen
2009-12-10  2:21       ` Balbir Singh
2009-12-11  2:14         ` Wu Fengguang
2009-12-14 12:53           ` Andi Kleen
2009-12-08 21:16 ` [PATCH] [24/31] HWPOISON: add an interface to switch off/on all the page filters Andi Kleen
2009-12-08 21:16 ` [PATCH] [25/31] HWPOISON: Don't do early filtering if filter is disabled Andi Kleen
2009-12-08 21:16 ` [PATCH] [26/31] HWPOISON: mention HWPoison in Kconfig entry Andi Kleen
2009-12-08 21:16 ` [PATCH] [27/31] HWPOISON: Use correct name for MADV_HWPOISON in documentation Andi Kleen
2009-12-08 21:16 ` [PATCH] [28/31] HWPOISON: Use new shake_page in memory_failure Andi Kleen
2009-12-08 21:16 ` [PATCH] [29/31] HWPOISON: Undefine short-hand macros after use to avoid namespace conflict Andi Kleen
2009-12-08 21:16 ` [PATCH] [30/31] HWPOISON: Add soft page offline support Andi Kleen
2009-12-08 21:16 ` [PATCH] [31/31] HWPOISON: Add a madvise() injector for soft page offlining Andi Kleen
2010-06-19 12:36   ` Michael Kerrisk
2010-06-19 13:20     ` Andi Kleen
2010-06-19 13:25       ` Michael Kerrisk
2010-06-19 13:30         ` Andi Kleen
2010-06-19 13:43           ` Michael Kerrisk
2010-06-19 14:09             ` Andi Kleen
2010-06-19 14:17               ` Michael Kerrisk
2010-06-19 19:52                 ` Andi Kleen
2010-06-20  6:19                   ` Michael Kerrisk
2010-06-20  7:14                     ` Wu Fengguang
2010-06-26 13:18                       ` Michael Kerrisk
2010-06-26 23:30                         ` Wu Fengguang
2010-06-27  4:38                           ` Michael Kerrisk

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=20091208211633.71135B151F@basil.firstfloor.org \
    --to=andi@firstfloor.org \
    --cc=fengguang.wu@intel.com \
    --cc=haicheng.li@intel.com \
    --cc=npiggin@suse.defengguang.wu \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).