linux-kernel - Re: [PATCH 3/3] drm/vc4: Don't liberate the binner BO at runtime suspend

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] [day] [month] [year] [list]

Message-ID: <8d6f1e9191638e243616ac7832b1b650a80d85ad.camel@bootlin.com>
Date:   Wed, 20 Mar 2019 16:17:32 +0100
From:   Paul Kocialkowski <paul.kocialkowski@...tlin.com>
To:     Eric Anholt <eric@...olt.net>, dri-devel@...ts.freedesktop.org,
        linux-kernel@...r.kernel.org
Cc:     Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
        Maxime Ripard <maxime.ripard@...tlin.com>,
        Sean Paul <sean@...rly.run>, David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel@...ll.ch>,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        Eben Upton <eben@...pberrypi.org>
Subject: Re: [PATCH 3/3] drm/vc4: Don't liberate the binner BO at runtime
 suspend

Hi,

Le samedi 16 mars 2019 à 11:58 -0700, Eric Anholt a écrit :
> Paul Kocialkowski <paul.kocialkowski@...tlin.com> writes:
> 
> > The binner BO is a pre-requisite to GPU operations, so we must ensure
> > that it is always allocated when the GPU is in use.
> > 
> > Because the buffer is allocated from the same pool as other GPU buffers,
> > we might run into a situation where we are out of memory at runtime
> > resume. This causes the binner BO allocation to fail and results in all
> > subsequent operations to fail, resulting in a major hang in userspace.
> > 
> > Now that we allocate the buffer at firstopen and liberate it at
> > lastclose, we can just keep it alive during runtime suspend.
> 
> I think this needs to be squashed into the previous patch, as otherwise
> coming from suspended, a firstopen -> resume -> render will leak a copy
> of the bin BO.

Woops, you're definitely right: vc4_allocate_bin_bo won't check whether
we already have allocated it or not. I'll send a new version with both
patches squashed.

Cheers,

Paul

-- 
Paul Kocialkowski, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com