mmotm-2013-05-22: Bad page state

DISCUSSION Dave Hansen 2 participants 3 messages May 28, 2013 – May 29, 2013 Memory management

Share & tools

Dave Hansen

author

2013-05-28 21:51 UTC #0 lore raw

I was rebasing my mapping->radix_tree lock batching patches on top of Mel's stuff. It looks like something is jumping the gun and freeing a page before it has been written out. Somebody probably did an extra put_page() or something.

I'm running 3.10.0-rc2-mm1-00322-g8d4c612 from

git://git.cmpxchg.org/linux-mmotm.git

This is pretty reproducible. I'll go try and test plain 3.10-rc2 next to make sure it's not coming from Linus's stuff.

Quoted 22 lines

[ 1628.443800] BUG: Bad page state in process kswapd0  pfn:1fc86c9
[ 1628.449781] page:ffffea007f21b240 count:0 mapcount:0 mapping:          (null) index:0x1c15
[ 1628.458214] page flags: 0x400000000c2008(uptodate|writeback|reclaim|swapbacked)
[ 1628.465624] Modules linked in:
[ 1628.468721] CPU: 5 PID: 68 Comm: kswapd0 Tainted: G        W    3.10.0-rc2-mm1-00322-g8d4c612-dirty #183
[ 1628.478400] Hardware name: FUJITSU-SV PRIMEQUEST 1800E2/SB, BIOS PRIMEQUEST 1000 Series BIOS Version 1.24 09/14/2011
[ 1628.489149]  0000000000000001 ffff881fd22498c8 ffffffff819163f2 0000000000005d42
[ 1628.496646]  ffffea007f21b240 ffff881fd22498e8 ffffffff811e2bd5 ffffea007f21b240
[ 1628.504159]  ffffea007f21b240 ffff881fd2249928 ffffffff811e3a7b ffff881fd2249948
[ 1628.511668] Call Trace:
[ 1628.514170]  [<ffffffff819163f2>] dump_stack+0x86/0xc3
[ 1628.519376]  [<ffffffff811e2bd5>] bad_page+0x145/0x180
[ 1628.524589]  [<ffffffff811e3a7b>] free_pages_prepare+0x1bb/0x1d0
[ 1628.530696]  [<ffffffff811e4658>] free_hot_cold_page+0x48/0x200
[ 1628.536701]  [<ffffffff811e4d8e>] free_hot_cold_page_list+0x6e/0x150
[ 1628.543154]  [<ffffffff811f32d5>] shrink_page_list+0x575/0xfc0
[ 1628.549065]  [<ffffffff811f3ef7>] shrink_inactive_list+0x1d7/0x730
[ 1628.555340]  [<ffffffff811f48a4>] shrink_lruvec+0x454/0x8b0
[ 1628.561002]  [<ffffffff8125b5d1>] ? __mem_cgroup_largest_soft_limit_node+0x31/0x160
[ 1628.568825]  [<ffffffff8125b5d1>] ? __mem_cgroup_largest_soft_limit_node+0x31/0x160
[ 1628.576659]  [<ffffffff811f4fe6>] shrink_zone+0x86/0x280
[ 1628.582038]  [<ffffffff811f5c67>] balance_pgdat+0x697/0x940

Quoted 8 lines

[ 1628.587694]  [<ffffffff811f6116>] kswapd+0x206/0x6e0
[ 1628.592732]  [<ffffffff8191de75>] ? __schedule+0x3e5/0xc10
[ 1628.598306]  [<ffffffff811015f0>] ? wake_up_bit+0x50/0x50
[ 1628.603779]  [<ffffffff811f5f10>] ? balance_pgdat+0x940/0x940
[ 1628.609614]  [<ffffffff8110033b>] kthread+0xcb/0xe0
[ 1628.614559]  [<ffffffff81100270>] ? __kthread_parkme+0xc0/0xc0
[ 1628.620481]  [<ffffffff8192cf5c>] ret_from_fork+0x7c/0xb0
[ 1628.625966]  [<ffffffff81100270>] ? __kthread_parkme+0xc0/0xc0

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

Dave Hansen

author

2013-05-28 21:53 UTC #1 lore raw

Quoted 8 lines · Dave Hansen

I was rebasing my mapping->radix_tree lock batching patches on top of Mel's stuff. It looks like something is jumping the gun and freeing a page before it has been written out. Somebody probably did an extra put_page() or something.
I'm running 3.10.0-rc2-mm1-00322-g8d4c612 from
git://git.cmpxchg.org/linux-mmotm.git

This is the config, btw:

	http://sr71.net/~dave/linux/config-3.10-mmotmbad1

Mel Gorman

2013-05-29 09:15 UTC #2 lore raw

In reply to Dave Hansen

Quoted 5 lines · Dave Hansen

I was rebasing my mapping->radix_tree lock batching patches on top of Mel's stuff. It looks like something is jumping the gun and freeing a page before it has been written out. Somebody probably did an extra put_page() or something.

Quoted 7 lines

I'm running 3.10.0-rc2-mm1-00322-g8d4c612 from
git://git.cmpxchg.org/linux-mmotm.git
This is pretty reproducible. I'll go try and test plain 3.10-rc2 next to make sure it's not coming from Linus's stuff.

Patch 1 from the follow-up series "mm: vmscan: Block kswapd if it is encountering pages under writeback -fix"

The rest of that follow-up series needs further work and I'm still working on it but patch 1 is what fixes this particular problem. Changelog says why.

Signature · 8 lines

Mel Gorman
SUSE Labs

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