lkml.org 
[lkml]   [2014]   [May]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH] memcg: fix swapcache charge from kernel thread context
On Mon 19-05-14 08:49:47, Johannes Weiner wrote:
> On Mon, May 19, 2014 at 10:27:56AM +0200, Michal Hocko wrote:
[...]
> > [1] - http://marc.info/?l=linux-mm&m=139463617808941&w=2
> >
> > Fixes: 03583f1a631c (3.15-rc1)
>
> Shouldn't this be the same format as other commit references?
>
> Fixes: 03583f1a631c ("memcg: remove unnecessary !mm check from try_get_mem_cgroup_from_mm()")

No idea, I just wanted to make clear to which kernel this applies to.
But other users seem to use the long version as above. So I will change
it.

> > Reported-and-tested-by: Stephan Kulow <coolo@suse.com>
> > Reported-by: Branimir Maksimovic <branimir.maksimovic@gmail.com>
> > Signed-off-by: Michal Hocko <mhocko@suse.cz>
> > ---
> > mm/memcontrol.c | 26 +++++++++++++-------------
> > 1 file changed, 13 insertions(+), 13 deletions(-)
> >
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index 2cb81478d30c..2248a648a127 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -1061,9 +1061,17 @@ static struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm)
> >
> > rcu_read_lock();
> > do {
> > - memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
> > - if (unlikely(!memcg))
> > + /*
> > + * Page cache or loopback insertions can happen without an
> > + * actual mm context, e.g. during disk probing on boot
> > + */
>
> Please include the other usecases:
>
> /*
> * Page cache insertions can happen without an
> * actual mm context, e.g. during disk probing
> * on boot, loopback IO, acct() writes etc.
> */

OK.

> Otherwise,
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Thanks!

---
From cc47d47bf640a90c13dfe662c7749c76806b76e5 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.cz>
Date: Fri, 16 May 2014 14:26:40 +0200
Subject: [PATCH] memcg: fix swapcache charge from kernel thread context

284f39afeaa4 (mm: memcg: push !mm handling out to page cache charge
function) explicitly checks for page cache charges without any mm
context (from kernel thread context[1]).

This seemed to be the only possible case where memory could be charged
without mm context so 03583f1a631c (memcg: remove unnecessary !mm check
from try_get_mem_cgroup_from_mm()) removed the mm check from
get_mem_cgroup_from_mm. This however caused another NULL ptr dereference
during early boot when loopback kernel thread splices to tmpfs as reported
by Stephan Kulow:

BUG: unable to handle kernel NULL pointer dereference at 0000000000000360
IP: [<ffffffff81196aab>] get_mem_cgroup_from_mm.isra.42+0x2b/0x60
PGD 5082067 PUD 83c3067 PMD 0
Oops: 0000 [#1] SMP
Modules linked in: btrfs dm_multipath dm_mod scsi_dh multipath raid10 raid456 async_raid6_recov async_memcpy async_pq raid6_pq async_xor xor async_tx raid1 raid0 md_mod parport_pc parport nls_utf8 isofs usb_storage iscsi_ibft iscsi_boot_sysfs arc4 ecb fan thermal nfs lockd fscache nls_iso8859_1 nls_cp437 sg st hid_generic usbhid af_packet sunrpc sr_mod cdrom ata_generic uhci_hcd virtio_net virtio_blk ehci_hcd usbcore ata_piix floppy processor button usb_common virtio_pci virtio_ring virtio edd squashfs loop
ppa]
CPU: 0 PID: 97 Comm: loop1 Not tainted 3.15.0-rc5-5-default #1
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
task: ffff880039b7a390 ti: ffff880038efe000 task.ti: ffff880038efe000
RIP: 0010:[<ffffffff81196aab>] [<ffffffff81196aab>] get_mem_cgroup_from_mm.isra.42+0x2b/0x60
RSP: 0018:ffff880038effa40 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffffea00001e5140 RCX: 0000000000000020
RDX: ffff88003c365020 RSI: ffffea00001e5140 RDI: 0000000000000360
RBP: ffff880038effa78 R08: 0000000000000ab3 R09: ffff880039572248
R10: 0000000000002ace R11: 0000000000000000 R12: 0000000000000010
R13: 0000000000000000 R14: ffff880038c72448 R15: 00000000fffffffe
FS: 00007fb0042ed880(0000) GS:ffff88003c000000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000000000360 CR3: 0000000005e2b000 CR4: 00000000000006f0
Stack:
ffffffff8119bae0 0000000000000000 0000000000000000 ffffea00001e5140
0000000000000001 00000000ffffffef ffffffff8119c04b 0000000000000000
ffff880038c722f8 0000000000000ab3 ffffffff8115129b 00000000000001d7
Call Trace:
[<ffffffff8119bae0>] __mem_cgroup_try_charge_swapin+0x40/0xe0
[<ffffffff8119c04b>] mem_cgroup_charge_file+0x8b/0xd0
[<ffffffff8115129b>] shmem_getpage_gfp+0x66b/0x7b0
[<ffffffff811518cf>] shmem_file_splice_read+0x18f/0x430
[<ffffffff811ceff2>] splice_direct_to_actor+0xa2/0x1c0
[<ffffffffa00019ea>] do_lo_receive+0x5a/0x60 [loop]
[<ffffffffa0002158>] loop_thread+0x298/0x720 [loop]
[<ffffffff810778d6>] kthread+0xc6/0xe0
[<ffffffff815c0dbc>] ret_from_fork+0x7c/0xb0
Code: 66 66 66 66 90 eb 24 66 0f 1f 84 00 00 00 00 00 f6 40 48 01 75 3a 48 8b 50 18 f6 c2 03 75 32 65 ff 02 ba 01 00 00 00 84 d2 75 25 <48> 8b 07 48 85 c0 74 10 48 8b 80 70 08 00 00 48 8b 40 60 48 85
RIP [<ffffffff81196aab>] get_mem_cgroup_from_mm.isra.42+0x2b/0x60
RSP <ffff880038effa40>
CR2: 0000000000000360

Also Branimir Maksimovic reported the following oops which is tiggered
for the swapcache charge path from the accounting code for kernel threads:

[ 388.522494] CPU: 1 PID: 160 Comm: kworker/u8:5 Tainted: P OE 3.15.0-rc5-core2-custom #159
[ 388.522496] Hardware name: System manufacturer System Product Name/MAXIMUSV GENE, BIOS 1903 08/19/2013
[ 388.522498] task: ffff880404e349b0 ti: ffff88040486a000 task.ti: ffff88040486a000
[ 388.522500] RIP: 0010:[<ffffffff81185b0b>] [<ffffffff81185b0b>] get_mem_cgroup_from_mm.isra.42+0x2b/0x60
[ 388.522504] RSP: 0000:ffff88040486bab8 EFLAGS: 00010246
[ 388.522506] RAX: 0000000000000000 RBX: ffffea000a416340 RCX: 0000000000000a40
[ 388.522508] RDX: ffff88041efe8a40 RSI: ffffea000a416340 RDI: 0000000000000340
[ 388.522509] RBP: ffff88040486bab8 R08: 000000000001cb56 R09: 0000000000072d5a
[ 388.522511] R10: 0000000000000000 R11: 0000000000000005 R12: ffff88040486bb00
[ 388.522512] R13: 00000000000000d0 R14: 0000000000000000 R15: ffff8803f3fe82f8
[ 388.522515] FS: 0000000000000000(0000) GS:ffff88041ec80000(0000) knlGS:0000000000000000
[ 388.522517] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 388.522518] CR2: 0000000000000340 CR3: 00000003ee44d000 CR4: 00000000001407e0
[ 388.522520] Stack:
[ 388.522521] ffff88040486baf0 ffffffff8118abf5 ffffffff8112ce1a 0000000000000000
[ 388.522524] ffffea000a416340 0000000000000003 00000000ffffffef ffff88040486bb18
[ 388.522527] ffffffff8118b1cc ffff88040486baf8 000000000001cb56 0000000000000000
[ 388.522530] Call Trace:
[ 388.522536] [<ffffffff8118abf5>] __mem_cgroup_try_charge_swapin+0x45/0xf0
[ 388.522539] [<ffffffff8112ce1a>] ? __lock_page+0x6a/0x70
[ 388.522543] [<ffffffff8118b1cc>] mem_cgroup_charge_file+0x9c/0xe0
[ 388.522548] [<ffffffff8114599c>] shmem_getpage_gfp+0x62c/0x770
[ 388.522552] [<ffffffff81145b18>] shmem_write_begin+0x38/0x40
[ 388.522555] [<ffffffff8112d1c5>] generic_perform_write+0xc5/0x1c0
[ 388.522559] [<ffffffff811ad53a>] ? file_update_time+0x8a/0xd0
[ 388.522563] [<ffffffff8112f211>] __generic_file_aio_write+0x1d1/0x3f0
[ 388.522567] [<ffffffff81084fc1>] ? enqueue_entity+0x291/0xb90
[ 388.522570] [<ffffffff8112f47f>] generic_file_aio_write+0x4f/0xc0
[ 388.522574] [<ffffffff81192eaa>] do_sync_write+0x5a/0x90
[ 388.522578] [<ffffffff810c53c1>] do_acct_process+0x4b1/0x550
[ 388.522582] [<ffffffff810c5acd>] acct_process+0x6d/0xa0
[ 388.522587] [<ffffffff810667d0>] ? manage_workers.isra.25+0x2a0/0x2a0
[ 388.522590] [<ffffffff8104d937>] do_exit+0x827/0xa70
[ 388.522594] [<ffffffff8106699e>] ? worker_thread+0x1ce/0x3a0
[ 388.522597] [<ffffffff810667d0>] ? manage_workers.isra.25+0x2a0/0x2a0
[ 388.522600] [<ffffffff8106cad3>] kthread+0xc3/0xf0
[ 388.522604] [<ffffffff8106ca10>] ? kthread_create_on_node+0x180/0x180
[ 388.522608] [<ffffffff816bfe6c>] ret_from_fork+0x7c/0xb0
[ 388.522611] [<ffffffff8106ca10>] ? kthread_create_on_node+0x180/0x180

This patch fixes the issue by reintroducing mm check into get_mem_cgroup_from_mm.
We could do the same trick in __mem_cgroup_try_charge_swapin as we do
for the regular page cache path but it is not worth troubles. The check
is not that expensive and it is better to have get_mem_cgroup_from_mm
more robust.

[1] - http://marc.info/?l=linux-mm&m=139463617808941&w=2

Fixes: 03583f1a631c ("memcg: remove unnecessary !mm check from try_get_mem_cgroup_from_mm()")
Reported-and-tested-by: Stephan Kulow <coolo@suse.com>
Reported-by: Branimir Maksimovic <branimir.maksimovic@gmail.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
mm/memcontrol.c | 27 ++++++++++++++-------------
1 file changed, 14 insertions(+), 13 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 2cb81478d30c..4e89d6962d11 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1061,9 +1061,18 @@ static struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm)

rcu_read_lock();
do {
- memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
- if (unlikely(!memcg))
+ /*
+ * Page cache insertions can happen withou an
+ * actual mm context, e.g. during disk probing
+ * on boot, loopback IO, acct() writes etc.
+ */
+ if (unlikely(!mm))
memcg = root_mem_cgroup;
+ else {
+ memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
+ if (unlikely(!memcg))
+ memcg = root_mem_cgroup;
+ }
} while (!css_tryget(&memcg->css));
rcu_read_unlock();
return memcg;
@@ -3857,17 +3866,9 @@ int mem_cgroup_charge_file(struct page *page, struct mm_struct *mm,
return 0;
}

- /*
- * Page cache insertions can happen without an actual mm
- * context, e.g. during disk probing on boot.
- */
- if (unlikely(!mm))
- memcg = root_mem_cgroup;
- else {
- memcg = mem_cgroup_try_charge_mm(mm, gfp_mask, 1, true);
- if (!memcg)
- return -ENOMEM;
- }
+ memcg = mem_cgroup_try_charge_mm(mm, gfp_mask, 1, true);
+ if (!memcg)
+ return -ENOMEM;
__mem_cgroup_commit_charge(memcg, page, 1, type, false);
return 0;
}
--
2.0.0.rc2

--
Michal Hocko
SUSE Labs


\
 
 \ /
  Last update: 2014-05-19 18:21    [W:0.519 / U:0.016 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site