linux-kernel - Re: regression: nfs mount (even idle) eventually hangs server

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 22 Dec 2022 14:21:33 +0000
From:   Chuck Lever III <chuck.lever@...cle.com>
To:     Mike Galbraith <efault@....de>
CC:     Dai Ngo <dai.ngo@...cle.com>, LKML <linux-kernel@...r.kernel.org>,
        Linux NFS Mailing List <linux-nfs@...r.kernel.org>
Subject: Re: regression: nfs mount (even idle) eventually hangs server



> On Dec 22, 2022, at 8:30 AM, Mike Galbraith <efault@....de> wrote:
> 
> Quoting previous mail for context.
> 
> On Thu, 2022-12-22 at 05:14 +0100, Mike Galbraith wrote:
>> On Thu, 2022-12-22 at 04:42 +0100, Mike Galbraith wrote:
>>> On Wed, 2022-12-21 at 10:56 +0100, Mike Galbraith wrote:
>>>>> 6.1 didn't reproduce either, so it would appear to be a merge window bug.
>>> 
>>> Ah, not true, turning evolution loose in nfs mounted home and letting
>>> it refresh mailboxes while desktop box was kept busy jammed up 6.1.0 in
>>> fairly short order.
>> 
>> Well crap.  That was _not_ virgin 6.1.0 after all, it was 6.1.0 with...
>> 
>> 44df6f439a17 NFSD: add delegation reaper to react to low memory condition
>> 3959066b697b NFSD: add support for sending CB_RECALL_ANY
>> a1049eb47f20 NFSD: refactoring courtesy_client_reaper to a generic low memory shrinker
>> 
>> ...applied from poking about yesterday.  I had given up on those as
>> culprit, and intended to pop them off and rebuild, but they were in
>> fact in the booted kernel.  Oh well, booboo could have a bright side.
> 
> I let my desktop box play server/space-heater for a long test session
> of 6.1 with and without the above series: with, box bricked 3 times in
> ~5 hours, without, 0 bricks in ~6 hours.  Box says woof -> duck ;-)

Thank you for testing!


> Given the similarity you mentioned, I wonder if my bug is perhaps your
> sneaky bug rendered somewhat less sneaky by that series? One can hope.

Not likely: we've seen problems since 5.19, and these patches were
merged in 6.2-rc.

So my question is whether you see a problem when only a1049eb47f20
is applied, or only when all three are applied?


--
Chuck Lever