linux-kernel - Re: [V9fs-developer] KCSAN BUG report on p9_client_cb / p9_client

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]

Message-ID: <CAJfuBxywD3QrsoGszMnVbF2RYcCF7r3h7sCOg6hK7K60E+4qKA@mail.gmail.com>
Date:   Tue, 22 Jun 2021 12:36:29 -0600
From:   jim.cromie@...il.com
To:     Dominique Martinet <asmadeus@...ewreck.org>
Cc:     kasan-dev@...glegroups.com, v9fs-developer@...ts.sourceforge.net,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [V9fs-developer] KCSAN BUG report on p9_client_cb / p9_client_rpc

On Tue, Jun 22, 2021 at 11:13 AM Dominique Martinet
<asmadeus@...ewreck.org> wrote:
>
> jim.cromie@...il.com wrote on Tue, Jun 22, 2021 at 10:42:58AM -0600:
> > I got this on rc7 + my hacks ( not near p9 )
> > ISTM someone here will know what it means.
> > If theres anything else i can do to help,
> > (configs, drop my patches and retry)
> >  please let me know
>
> Thanks for the report!
>
> > [   14.904783] ==================================================================
> > [   14.905848] BUG: KCSAN: data-race in p9_client_cb / p9_client_rpc
>
> hm, this code hasn't changed in ages (unless someone merged code behind
> my back :D)
>
> I had assumed the p9_req_put() in p9_client_cb would protect the tag,
> but that doesn't appear to be true -- could you try this patch if this
> is reproductible to you?
>

I applied your patch on top of my triggering case, it fixes the report  !
you have my tested-by

> The tag is actually reclaimed in the woken up p9_client_rpc thread so
> that would be a good match (reset in the other thread vs. read here),
> caching the value is good enough but that is definitely not obvious...
>
> --
> Dominique