From: Greg KH <gregkh@suse.de>
To: linux-kernel@vger.kernel.org, stable@kernel.org,
akpm@linux-foundation.org, torvalds@linux-foundation.org,
stable-review@kernel.org, Greg KH <greg@kroah.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>,
linux-mm <linux-mm@kvack.org>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
David Rientjes <rientjes@google.com>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Balbir Singh <balbir@linux.vnet.ibm.com>
Subject: [33/39] memcg: avoid oom-killing innocent task in case of use_hierarchy
Date: Tue, 05 Jan 2010 12:02:29 -0800 [thread overview]
Message-ID: <20100105200303.273164182@mini.kroah.org> (raw)
In-Reply-To: <20100105195007.GA23952@kroah.com>
2.6.31-stable review patch. If anyone has any objections, please let us know.
------------------
From: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
commit d31f56dbf8bafaacb0c617f9a6f137498d5c7aed upstream
task_in_mem_cgroup(), which is called by select_bad_process() to check whether
a task can be a candidate for being oom-killed from memcg's limit, checks
"curr->use_hierarchy"("curr" is the mem_cgroup the task belongs to).
But this check return true(it's false positive) when:
<some path>/00 use_hierarchy == 0 <- hitting limit
<some path>/00/aa use_hierarchy == 1 <- "curr"
This leads to killing an innocent task in 00/aa. This patch is a fix for this
bug. And this patch also fixes the arg for mem_cgroup_print_oom_info(). We
should print information of mem_cgroup which the task being killed, not current,
belongs to.
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
mm/memcontrol.c | 8 +++++++-
mm/oom_kill.c | 2 +-
2 files changed, 8 insertions(+), 2 deletions(-)
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -496,7 +496,13 @@ int task_in_mem_cgroup(struct task_struc
task_unlock(task);
if (!curr)
return 0;
- if (curr->use_hierarchy)
+ /*
+ * We should check use_hierarchy of "mem" not "curr". Because checking
+ * use_hierarchy of "curr" here make this function true if hierarchy is
+ * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
+ * hierarchy(even if use_hierarchy is disabled in "mem").
+ */
+ if (mem->use_hierarchy)
ret = css_is_ancestor(&curr->css, &mem->css);
else
ret = (curr == mem);
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -400,7 +400,7 @@ static int oom_kill_process(struct task_
cpuset_print_task_mems_allowed(current);
task_unlock(current);
dump_stack();
- mem_cgroup_print_oom_info(mem, current);
+ mem_cgroup_print_oom_info(mem, p);
show_mem();
if (sysctl_oom_dump_tasks)
dump_tasks(mem);
WARNING: multiple messages have this Message-ID (diff)
From: Greg KH <gregkh@suse.de>
To: linux-kernel@vger.kernel.org, stable@kernel.org,
akpm@linux-foundation.org, torvalds@linux-foundation.org,
stable-review@kernel.org, Greg KH <greg@kroah.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>,
linux-mm <linux-mm@kvack.org>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
David Rientjes <rientjes@google.com>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Balbir Singh <balbir@linux.vnet.ibm.com>
Subject: [33/39] memcg: avoid oom-killing innocent task in case of use_hierarchy
Date: Tue, 05 Jan 2010 12:02:29 -0800 [thread overview]
Message-ID: <20100105200303.273164182@mini.kroah.org> (raw)
In-Reply-To: <20100105195007.GA23952@kroah.com>
2.6.31-stable review patch. If anyone has any objections, please let us know.
------------------
From: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
commit d31f56dbf8bafaacb0c617f9a6f137498d5c7aed upstream
task_in_mem_cgroup(), which is called by select_bad_process() to check whether
a task can be a candidate for being oom-killed from memcg's limit, checks
"curr->use_hierarchy"("curr" is the mem_cgroup the task belongs to).
But this check return true(it's false positive) when:
<some path>/00 use_hierarchy == 0 <- hitting limit
<some path>/00/aa use_hierarchy == 1 <- "curr"
This leads to killing an innocent task in 00/aa. This patch is a fix for this
bug. And this patch also fixes the arg for mem_cgroup_print_oom_info(). We
should print information of mem_cgroup which the task being killed, not current,
belongs to.
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
mm/memcontrol.c | 8 +++++++-
mm/oom_kill.c | 2 +-
2 files changed, 8 insertions(+), 2 deletions(-)
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -496,7 +496,13 @@ int task_in_mem_cgroup(struct task_struc
task_unlock(task);
if (!curr)
return 0;
- if (curr->use_hierarchy)
+ /*
+ * We should check use_hierarchy of "mem" not "curr". Because checking
+ * use_hierarchy of "curr" here make this function true if hierarchy is
+ * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
+ * hierarchy(even if use_hierarchy is disabled in "mem").
+ */
+ if (mem->use_hierarchy)
ret = css_is_ancestor(&curr->css, &mem->css);
else
ret = (curr == mem);
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -400,7 +400,7 @@ static int oom_kill_process(struct task_
cpuset_print_task_mems_allowed(current);
task_unlock(current);
dump_stack();
- mem_cgroup_print_oom_info(mem, current);
+ mem_cgroup_print_oom_info(mem, p);
show_mem();
if (sysctl_oom_dump_tasks)
dump_tasks(mem);
--
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:[~2010-01-05 20:06 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-01-05 19:50 [00/10] 2.6.27.43 stable review Greg KH
2010-01-05 19:47 ` [01/10] Libertas: fix buffer overflow in lbs_get_essid() Greg KH
2010-01-05 19:47 ` [02/10] pata_cmd64x: fix overclocking of UDMA0-2 modes Greg KH
2010-01-05 19:47 ` [03/10] sound: sgio2audio/pdaudiocf/usb-audio: initialize PCM buffer Greg KH
2010-01-05 19:47 ` [04/10] i2c/tsl2550: Fix lux value in extended mode Greg KH
2010-01-05 19:47 ` [05/10] ipv6: reassembly: use seperate reassembly queues for conntrack and local delivery Greg KH
2010-01-05 19:47 ` [06/10] S390: dasd: support DIAG access for read-only devices Greg KH
2010-01-05 19:47 ` [07/10] x86/ptrace: make genregs[32]_get/set more robust Greg KH
2010-01-05 19:47 ` [08/10] rt2x00: Disable powersaving for rt61pci and rt2800pci Greg KH
2010-01-05 20:58 ` Gertjan van Wingerde
2010-01-05 21:21 ` Greg KH
2010-01-05 19:48 ` [09/10] generic_permission: MAY_OPEN is not write access Greg KH
2010-01-05 19:48 ` [10/10] Revert: KVM: MMU: do not free active mmu pages in free_mmu_pages() Greg KH
2010-01-05 20:01 ` [01/39] acerhdf: limit modalias matching to supported Greg KH
2010-01-05 20:01 ` [02/39] ASoC: Do not write to invalid registers on the wm9712 Greg KH
2010-01-05 20:01 ` [03/39] cifs: NULL out tcon, pSesInfo, and srvTcp pointers when chasing DFS referrals Greg KH
2010-01-05 20:02 ` [04/39] clockevents: Prevent clockevent_devices list corruption on cpu hotplug Greg KH
2010-01-05 20:02 ` [05/39] dma: at_hdmac: correct incompatible type for argument 1 of spin_lock_bh Greg KH
2010-01-05 20:02 ` [06/39] drivers/net/usb: Correct code taking the size of a pointer Greg KH
2010-01-05 20:02 ` [07/39] iwmc3200wifi: fix array out-of-boundary access Greg KH
2010-01-06 2:52 ` Zhu Yi
2010-01-06 18:01 ` Greg KH
2010-01-06 18:27 ` [stable] " Greg KH
2010-01-05 20:02 ` [08/39] Libertas: fix buffer overflow in lbs_get_essid() Greg KH
2010-01-05 20:02 ` [09/39] md: Fix unfortunate interaction with evms Greg KH
2010-01-05 20:02 ` [10/39] pata_cmd64x: fix overclocking of UDMA0-2 modes Greg KH
2010-01-05 20:02 ` [11/39] pata_hpt3x2n: fix clock turnaround Greg KH
2010-01-05 20:02 ` [12/39] SCSI: fc class: fix fc_transport_init error handling Greg KH
2010-01-05 20:02 ` [13/39] sound: sgio2audio/pdaudiocf/usb-audio: initialize PCM buffer Greg KH
2010-01-05 20:02 ` [14/39] USB: emi62: fix crash when trying to load EMI 6|2 firmware Greg KH
2010-01-05 20:02 ` [15/39] USB: Fix a bug on appledisplay.c regarding signedness Greg KH
2010-01-05 20:02 ` [16/39] USB: musb: gadget_ep0: avoid SetupEnd interrupt Greg KH
2010-01-05 20:02 ` [17/39] USB: option: support hi speed for modem Haier CE100 Greg KH
2010-01-05 20:02 ` [18/39] x86, cpuid: Add "volatile" to asm in native_cpuid() Greg KH
2010-01-05 20:02 ` [19/39] e100: Use pci pool to work around GFP_ATOMIC order 5 memory allocation failure Greg KH
2010-03-15 21:29 ` [Stable-review] " Stephen Hemminger
2010-03-15 21:32 ` David Miller
2010-03-15 21:36 ` Stephen Hemminger
2010-03-15 21:39 ` David Miller
2010-03-15 22:20 ` David Miller
2010-03-15 22:25 ` Stephen Hemminger
2010-01-05 20:02 ` [20/39] e100: Fix broken cbs accounting due to missing memset Greg KH
2010-01-05 20:02 ` [21/39] hostap: Revert a toxic part of the conversion to net_device_ops Greg KH
2010-01-05 20:02 ` [22/39] hwmon: (fschmd) Fix check on unsigned in watchdog_write() Greg KH
2010-01-05 20:02 ` [23/39] hwmon: (sht15) Off-by-one error in array index + incorrect constants Greg KH
2010-01-05 20:02 ` [24/39] i2c/tsl2550: Fix lux value in extended mode Greg KH
2010-01-05 20:02 ` [25/39] ipv6: reassembly: use seperate reassembly queues for conntrack and local delivery Greg KH
2010-01-05 20:02 ` [26/39] S390: dasd: support DIAG access for read-only devices Greg KH
2010-01-05 20:02 ` [27/39] udf: Try harder when looking for VAT inode Greg KH
2010-01-05 20:02 ` [28/39] V4L/DVB (13596): ov511.c typo: lock => unlock Greg KH
2010-01-05 20:02 ` [29/39] x86/ptrace: make genregs[32]_get/set more robust Greg KH
2010-01-05 20:02 ` [30/39] XFS bug in log recover with quota (bugzilla id 855) Greg KH
2010-01-05 20:02 ` [31/39] generic_permission: MAY_OPEN is not write access Greg KH
2010-01-05 20:02 ` [32/39] rt2x00: Disable powersaving for rt61pci and rt2800pci Greg KH
2010-01-05 20:59 ` Gertjan van Wingerde
2010-01-05 21:21 ` Greg KH
2010-01-05 20:02 ` Greg KH [this message]
2010-01-05 20:02 ` [33/39] memcg: avoid oom-killing innocent task in case of use_hierarchy Greg KH
2010-01-05 20:02 ` [34/39] Input: atkbd - add force relese key quirk for Samsung R59P/R60P/R61P Greg KH
2010-01-05 20:02 ` [35/39] Add unlocked version of inode_add_bytes() function Greg KH
2010-01-05 20:02 ` [36/39] quota: decouple fs reserved space from quota reservation Greg KH
2010-01-05 20:02 ` [37/39] ext4: Convert to generic reserved quotas space management Greg KH
2010-01-05 20:02 ` [38/39] ext4: Fix potential quota deadlock Greg KH
2010-01-05 20:02 ` [39/39] ext4: fix sleep inside spinlock issue with quota and dealloc (#14739) Greg KH
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=20100105200303.273164182@mini.kroah.org \
--to=gregkh@suse.de \
--cc=akpm@linux-foundation.org \
--cc=balbir@linux.vnet.ibm.com \
--cc=greg@kroah.com \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=nishimura@mxp.nes.nec.co.jp \
--cc=rientjes@google.com \
--cc=stable-review@kernel.org \
--cc=stable@kernel.org \
--cc=torvalds@linux-foundation.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.