From: Ashish Mhetre <amhetre@nvidia.com>
To: Michal Hocko <mhocko@kernel.org>
Cc: linux-mm@kvack.org, akpm@linux-foundation.org, vdumpa@nvidia.com,
Snikam@nvidia.com
Subject: Re: [PATCH] mm: Avoid swapping in interrupt context
Date: Wed, 3 Oct 2018 16:18:37 +0530 [thread overview]
Message-ID: <988dfe01-6553-1e0a-1d98-1b3d3aa67517@nvidia.com> (raw)
In-Reply-To: <20181001122400.GF18290@dhcp22.suse.cz>
[-- Attachment #1: Type: text/plain, Size: 1833 bytes --]
>How? No allocation request from the interrupt context can use a
>sleepable allocation context and that means that no reclaim is allowed
>from the IRQ context.
Kernel Oops happened when ZRAM was used as swap with zsmalloc as alloctor
under memory pressure condition.
This is probably because of kmalloc() from IRQ as pointed out by Sergey.
>Could you provide the Oops message?
BUG_ON() got triggered at https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/mm/zsmalloc.c?h=next-20181002#n1324 with Oops message:
[ 264.082531] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
[ 264.088350] Modules linked in:
[ 264.091406] CPU: 0 PID: 3805 Comm: kworker/0:4 Tainted: G W
3.10.33-g990282b #1
[ 264.099572] Workqueue: events netstat_work_func
[ 264.104097] task: e7b12040 ti: dc7d4000 task.ti: dc7d4000
[ 264.109485] PC is at zs_map_object+0x180/0x18c
[ 264.113918] LR is at zram_bvec_rw.isra.15+0x304/0x88c
[ 264.118956] pc : [<c01581e8>] lr : [<c0456618>] psr: 200f0013
[ 264.118956] sp : dc7d5460 ip : fff00814 fp : 00000002
[ 264.130407] r10: ea8ec000 r9 : ebc93340 r8 : 00000000
[ 264.135618] r7 : c191502c r6 : dc7d4020 r5 : d25f5684 r4 : ec3158c0
[ 264.142128] r3 : 00000200 r2 : 00000002 r1 : c191502c r0 : ea8ec000
On Monday 01 October 2018 05:54 PM, Michal Hocko wrote:
> On Mon 01-10-18 15:15:15, Ashish Mhetre wrote:
>> From: Sri Krishna chowdary <schowdary@nvidia.com>
>>
>> Pages can be swapped out from interrupt context as well.
> How? No allocation request from the interrupt context can use a
> sleepable allocation context and that means that no reclaim is allowed
> from the IRQ context.
>
>> ZRAM uses zsmalloc allocator to make room for these pages.
>> But zsmalloc is not made to be used from interrupt context.
>> This can result in a kernel Oops.
> Could you provide the Oops message?
[-- Attachment #2: Type: text/html, Size: 9177 bytes --]
next prev parent reply other threads:[~2018-10-03 10:48 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-10-01 9:45 Ashish Mhetre
2018-10-01 12:24 ` Michal Hocko
2018-10-03 10:48 ` Ashish Mhetre [this message]
2018-10-03 11:01 ` Michal Hocko
2018-10-03 11:50 ` Ashish Mhetre
2018-10-03 11:53 ` Michal Hocko
2018-10-16 9:35 ` Ashish Mhetre
2018-10-02 7:54 ` Sergey Senozhatsky
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=988dfe01-6553-1e0a-1d98-1b3d3aa67517@nvidia.com \
--to=amhetre@nvidia.com \
--cc=Snikam@nvidia.com \
--cc=akpm@linux-foundation.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=vdumpa@nvidia.com \
/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