Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <3e38ebb1695d0b5df8c2ebbae37eccab2fcfdc39.camel@gmail.com>
Subject: Re: Non-robust apps and resets (was Re: [PATCH v5 1/1] drm/doc:
 Document DRM device reset expectations)
From:   Timur =?ISO-8859-1?Q?Krist=F3f?= <timur.kristof@gmail.com>
To:     Michel =?ISO-8859-1?Q?D=E4nzer?= <michel.daenzer@mailbox.org>,
        Marek =?UTF-8?Q?Ol=C5=A1=C3=A1k?= <maraeo@gmail.com>
Cc:     pierre-eric.pelloux-prayer@amd.com, amd-gfx@lists.freedesktop.org,
        Pekka Paalanen <pekka.paalanen@collabora.com>,
        Randy Dunlap <rdunlap@infradead.org>,
        linux-kernel@vger.kernel.org,
        Samuel Pitoiset <samuel.pitoiset@gmail.com>,
        Pekka Paalanen <ppaalanen@gmail.com>,
        dri-devel@lists.freedesktop.org, kernel-dev@igalia.com,
        alexander.deucher@amd.com,
        =?ISO-8859-1?Q?Andr=E9?= Almeida <andrealmeid@igalia.com>,
        christian.koenig@amd.com
Date:   Wed, 26 Jul 2023 09:55:08 +0200
In-Reply-To: <36bc5f30-dc72-10a3-c9cb-de9cb5400eb3@mailbox.org>
References: <20230627132323.115440-1-andrealmeid@igalia.com>
         <e292a30f-5cad-1968-de4f-0d43c9c1e943@igalia.com>
         <45a1e527-f5dc-aa6f-9482-8958566ecb96@mailbox.org>
         <CAAxE2A6OhFVKSm8VY1iNJx_FvDH-ojfgxvSjTmJUwtcmouoLMw@mail.gmail.com>
         <36bc5f30-dc72-10a3-c9cb-de9cb5400eb3@mailbox.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.48.4 (3.48.4-1.fc38) 
MIME-Version: 1.0
Precedence: bulk

On Tue, 2023-07-25 at 19:00 +0200, Michel D=C3=A4nzer wrote:
> On 7/25/23 17:05, Marek Ol=C5=A1=C3=A1k wrote:
> > On Tue, Jul 25, 2023 at 4:03=E2=80=AFAM Michel D=C3=A4nzer
> > <michel.daenzer@mailbox.org> wrote:
> > > On 7/25/23 04:55, Andr=C3=A9 Almeida wrote:
> > > > Hi everyone,
> > > >=20
> > > > It's not clear what we should do about non-robust OpenGL apps
> > > > after GPU resets, so I'll try to summarize the topic, show some
> > > > options and my proposal to move forward on that.
> > > >=20
> > > > Em 27/06/2023 10:23, Andr=C3=A9 Almeida escreveu:
> > > > > +Robustness
> > > > > +----------
> > > > > +
> > > > > +The only way to try to keep an application working after a
> > > > > reset is if it
> > > > > +complies with the robustness aspects of the graphical API
> > > > > that it is using.
> > > > > +
> > > > > +Graphical APIs provide ways to applications to deal with
> > > > > device resets. However,
> > > > > +there is no guarantee that the app will use such features
> > > > > correctly, and the
> > > > > +UMD can implement policies to close the app if it is a
> > > > > repeating offender,
> > > > > +likely in a broken loop. This is done to ensure that it does
> > > > > not keep blocking
> > > > > +the user interface from being correctly displayed. This
> > > > > should be done even if
> > > > > +the app is correct but happens to trigger some bug in the
> > > > > hardware/driver.
> > > > > +
> > > > Depending on the OpenGL version, there are different robustness
> > > > API available:
> > > >=20
> > > > - OpenGL ABR extension [0]
> > > > - OpenGL KHR extension [1]
> > > > - OpenGL ES extension=C2=A0 [2]
> > > >=20
> > > > Apps written in OpenGL should use whatever version is available
> > > > for them to make the app robust for GPU resets. That usually
> > > > means calling GetGraphicsResetStatusARB(), checking the status,
> > > > and if it encounter something different from NO_ERROR, that
> > > > means that a reset has happened, the context is considered lost
> > > > and should be recreated. If an app follow this, it will likely
> > > > succeed recovering a reset.
> > > >=20
> > > > What should non-robustness apps do then? They certainly will
> > > > not be notified if a reset happens, and thus can't recover if
> > > > their context is lost. OpenGL specification does not explicitly
> > > > define what should be done in such situations[3], and I believe
> > > > that usually when the spec mandates to close the app, it would
> > > > explicitly note it.
> > > >=20
> > > > However, in reality there are different types of device resets,
> > > > causing different results. A reset can be precise enough to
> > > > damage only the guilty context, and keep others alive.
> > > >=20
> > > > Given that, I believe drivers have the following options:
> > > >=20
> > > > a) Kill all non-robust apps after a reset. This may lead to
> > > > lose work from innocent applications.
> > > >=20
> > > > b) Ignore all non-robust apps OpenGL calls. That means that
> > > > applications would still be alive, but the user interface would
> > > > be freeze. The user would need to close it manually anyway, but
> > > > in some corner cases, the app could autosave some work or the
> > > > user might be able to interact with it using some alternative
> > > > method (command line?).
> > > >=20
> > > > c) Kill just the affected non-robust applications. To do that,
> > > > the driver need to be 100% sure on the impact of its resets.
> > > >=20
> > > > RadeonSI currently implements a), as can be seen at [4], while
> > > > Iris implements what I think it's c)[5].
> > > >=20
> > > > For the user experience point-of-view, c) is clearly the best
> > > > option, but it's the hardest to archive. There's not much gain
> > > > on having b) over a), perhaps it could be an optional env var
> > > > for such corner case applications.
> > >=20
> > > I disagree on these conclusions.
> > >=20
> > > c) is certainly better than a), but it's not "clearly the best"
> > > in all cases. The OpenGL UMD is not a privileged/special
> > > component and is in no position to decide whether or not the
> > > process as a whole (only some thread(s) of which may use OpenGL
> > > at all) gets to continue running or not.
> >=20
> > That's not true.
>=20
> Which part of what I wrote are you referring to?
>=20
>=20
> > I recommend that you enable b) with your driver and then hang the
> > GPU under different scenarios and see the result.
>=20
> I've been doing GPU driver development for over two decades, I'm
> perfectly aware what it means. It doesn't change what I wrote above.
>=20

Michel, I understand that you disagree with the proposed solutions in
this email thread but from your mails it is unclear to me what exactly
is the solution that you would actually recommend, can you please
clarify?

Thanks,
Timur