[PATCH] cgroup/kmemleak: add kmemleak_free() for cgroup deallocations.

Discussion:

Wang Nan

2014-09-18 01:38:05 UTC

Commit ff7ee93f4 introduces kmemleak_alloc() for alloc_page_cgroup(),
but corresponding kmemleak_free() is missing, which makes kmemleak be
wrongly disabled after memory offlining. Log is pasted at the end of
this commit message.

This patch add kmemleak_free() into free_page_cgroup(). During page
offlining, this patch removes corresponding entries in kmemleak rbtree.
After that, the freed memory can be allocated again by other subsystems
without killing kmemleak.

bash # for x in 1 2 3 4; do echo offline > /sys/devices/system/memory/memory$x/state ; sleep 1; done ; dmesg | grep leak
[ 45.537934] Offlined Pages 32768
[ 46.617892] kmemleak: Cannot insert 0xffff880016969000 into the object search tree (overlaps existing)
[ 46.617892] CPU: 0 PID: 412 Comm: sleep Not tainted 3.17.0-rc5+ #86
[ 46.617892] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 46.617892] ffff880016823d10 ffff880018bdfc38 ffffffff81725d2c ffff88001780e950
[ 46.617892] ffff880016969000 ffff880018bdfc88 ffffffff8117a9e6 ffff880018bdfc78
[ 46.617892] 0000000000000096 ffff880017812800 ffffffff81c2eda0 ffff880016969000
[ 46.617892] Call Trace:
[ 46.617892] [<ffffffff81725d2c>] dump_stack+0x46/0x58
[ 46.617892] [<ffffffff8117a9e6>] create_object+0x266/0x2c0
[ 46.617892] [<ffffffff8171d2f6>] kmemleak_alloc+0x26/0x50
[ 46.617892] [<ffffffff8116a3a3>] kmem_cache_alloc+0xd3/0x160
[ 46.617892] [<ffffffff81058e59>] __sigqueue_alloc+0x49/0xd0
[ 46.617892] [<ffffffff8105a41b>] __send_signal+0xcb/0x410
[ 46.617892] [<ffffffff8105a7a5>] send_signal+0x45/0x90
[ 46.617892] [<ffffffff8105a803>] __group_send_sig_info+0x13/0x20
[ 46.617892] [<ffffffff8105bd0b>] do_notify_parent+0x1bb/0x260
[ 46.617892] [<ffffffff81077e7a>] ? sched_move_task+0xaa/0x130
[ 46.617892] [<ffffffff81050917>] do_exit+0x767/0xa40
[ 46.617892] [<ffffffff81050c84>] do_group_exit+0x44/0xa0
[ 46.617892] [<ffffffff81050cf7>] SyS_exit_group+0x17/0x20
[ 46.617892] [<ffffffff8172cd12>] system_call_fastpath+0x16/0x1b
[ 46.617892] kmemleak: Kernel memory leak detector disabled
[ 46.617892] kmemleak: Object 0xffff880016900000 (size 524288):
[ 46.617892] kmemleak: comm "swapper/0", pid 0, jiffies 4294667296
[ 46.617892] kmemleak: min_count = 0
[ 46.617892] kmemleak: count = 0
[ 46.617892] kmemleak: flags = 0x1
[ 46.617892] kmemleak: checksum = 0
[ 46.617892] kmemleak: backtrace:
[ 46.617892] [<ffffffff81d0a7f0>] log_early+0x63/0x77
[ 46.617892] [<ffffffff8171d31b>] kmemleak_alloc+0x4b/0x50
[ 46.617892] [<ffffffff81720e4f>] init_section_page_cgroup+0x7f/0xf5
[ 46.617892] [<ffffffff81d0a6f0>] page_cgroup_init+0xc5/0xd0
[ 46.617892] [<ffffffff81ce4ed9>] start_kernel+0x333/0x408
[ 46.617892] [<ffffffff81ce45b2>] x86_64_start_reservations+0x2a/0x2c
[ 46.617892] [<ffffffff81ce46a9>] x86_64_start_kernel+0xf5/0xfc
[ 46.617892] [<ffffffffffffffff>] 0xffffffffffffffff

Signed-off-by: Wang Nan <***@huawei.com>
Cc: Steven Rostedt <***@goodmis.org>
---
mm/page_cgroup.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
index 3708264..5331c2b 100644
--- a/mm/page_cgroup.c
+++ b/mm/page_cgroup.c
@@ -171,6 +171,7 @@ static void free_page_cgroup(void *addr)
sizeof(struct page_cgroup) * PAGES_PER_SECTION;

BUG_ON(PageReserved(page));
+ kmemleak_free(addr);
free_pages_exact(addr, table_size);
}
}

--
1.8.4

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to ***@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"***@kvack.org"> ***@kvack.org </a>

Johannes Weiner

2014-09-18 14:16:39 UTC

Permalink

Post by Wang Nan
Commit ff7ee93f4 introduces kmemleak_alloc() for alloc_page_cgroup(),
but corresponding kmemleak_free() is missing, which makes kmemleak be
wrongly disabled after memory offlining. Log is pasted at the end of
this commit message.
This patch add kmemleak_free() into free_page_cgroup(). During page
offlining, this patch removes corresponding entries in kmemleak rbtree.
After that, the freed memory can be allocated again by other subsystems
without killing kmemleak.
bash # for x in 1 2 3 4; do echo offline > /sys/devices/system/memory/memory$x/state ; sleep 1; done ; dmesg | grep leak
[ 45.537934] Offlined Pages 32768
[ 46.617892] kmemleak: Cannot insert 0xffff880016969000 into the object search tree (overlaps existing)
[ 46.617892] CPU: 0 PID: 412 Comm: sleep Not tainted 3.17.0-rc5+ #86
[ 46.617892] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 46.617892] ffff880016823d10 ffff880018bdfc38 ffffffff81725d2c ffff88001780e950
[ 46.617892] ffff880016969000 ffff880018bdfc88 ffffffff8117a9e6 ffff880018bdfc78
[ 46.617892] 0000000000000096 ffff880017812800 ffffffff81c2eda0 ffff880016969000
[ 46.617892] [<ffffffff81725d2c>] dump_stack+0x46/0x58
[ 46.617892] [<ffffffff8117a9e6>] create_object+0x266/0x2c0
[ 46.617892] [<ffffffff8171d2f6>] kmemleak_alloc+0x26/0x50
[ 46.617892] [<ffffffff8116a3a3>] kmem_cache_alloc+0xd3/0x160
[ 46.617892] [<ffffffff81058e59>] __sigqueue_alloc+0x49/0xd0
[ 46.617892] [<ffffffff8105a41b>] __send_signal+0xcb/0x410
[ 46.617892] [<ffffffff8105a7a5>] send_signal+0x45/0x90
[ 46.617892] [<ffffffff8105a803>] __group_send_sig_info+0x13/0x20
[ 46.617892] [<ffffffff8105bd0b>] do_notify_parent+0x1bb/0x260
[ 46.617892] [<ffffffff81077e7a>] ? sched_move_task+0xaa/0x130
[ 46.617892] [<ffffffff81050917>] do_exit+0x767/0xa40
[ 46.617892] [<ffffffff81050c84>] do_group_exit+0x44/0xa0
[ 46.617892] [<ffffffff81050cf7>] SyS_exit_group+0x17/0x20
[ 46.617892] [<ffffffff8172cd12>] system_call_fastpath+0x16/0x1b
[ 46.617892] kmemleak: Kernel memory leak detector disabled
[ 46.617892] kmemleak: comm "swapper/0", pid 0, jiffies 4294667296
[ 46.617892] kmemleak: min_count = 0
[ 46.617892] kmemleak: count = 0
[ 46.617892] kmemleak: flags = 0x1
[ 46.617892] kmemleak: checksum = 0
[ 46.617892] [<ffffffff81d0a7f0>] log_early+0x63/0x77
[ 46.617892] [<ffffffff8171d31b>] kmemleak_alloc+0x4b/0x50
[ 46.617892] [<ffffffff81720e4f>] init_section_page_cgroup+0x7f/0xf5
[ 46.617892] [<ffffffff81d0a6f0>] page_cgroup_init+0xc5/0xd0
[ 46.617892] [<ffffffff81ce4ed9>] start_kernel+0x333/0x408
[ 46.617892] [<ffffffff81ce45b2>] x86_64_start_reservations+0x2a/0x2c
[ 46.617892] [<ffffffff81ce46a9>] x86_64_start_kernel+0xf5/0xfc
[ 46.617892] [<ffffffffffffffff>] 0xffffffffffffffff

Acked-by: Johannes Weiner <***@cmpxchg.org>

Should this go into -stable? I'm inclined to say no, this has been
busted since Steve's other kmemleak fix since 2011, and that change
also didn't go into -stable.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to ***@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"***@kvack.org"> ***@kvack.org </a>

Steven Rostedt

2014-09-18 15:16:34 UTC

Permalink

On Thu, 18 Sep 2014 10:16:39 -0400

Post by Johannes Weiner
Should this go into -stable? I'm inclined to say no, this has been
busted since Steve's other kmemleak fix since 2011, and that change
also didn't go into -stable.

It only breaks kmem tests, and since nobody noticed recently, I don't
think it needs to go into stable.

On the other hand, it's a very non intrusive fix that I highly doubt
will cause other regressions, so it may not be bad to add a stable tag
to it.

-- Steve

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to ***@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"***@kvack.org"> ***@kvack.org </a>

Michal Hocko

2014-09-22 13:48:27 UTC

Permalink

Acked-by: Michal Hocko <mhocko-***@public.gmane.org>

And sorry, I should have noticed that when reviewing the original patch
which added kmemleak_alloc.

I think this is worth going into stable. The risk is marginal and
somebody might be relying on kmemleak even on older kernels when
debugging a bug during memory hotplug.

Thanks!

Post by Wang Nan
---
mm/page_cgroup.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
index 3708264..5331c2b 100644
--- a/mm/page_cgroup.c
+++ b/mm/page_cgroup.c
@@ -171,6 +171,7 @@ static void free_page_cgroup(void *addr)
sizeof(struct page_cgroup) * PAGES_PER_SECTION;
BUG_ON(PageReserved(page));
+ kmemleak_free(addr);
free_pages_exact(addr, table_size);
}
}
--
1.8.4

--
Michal Hocko
SUSE Labs

Wang Nan

2014-10-14 11:49:44 UTC

Permalink

Hi,

I can't find this patch appear in any tree, is there any problem?

As Michal Hocko's suggestion, I resent this patch with stable tag added. Please review.

Thanks.