Message-ID: <2d8d962e-9fa3-7671-de05-8401a81b1356@mailbox.org>
Date:   Tue, 7 Feb 2023 15:58:33 +0100
MIME-Version: 1.0
Subject: Re: [RFC PATCH] drm: Create documentation about device resets
Content-Language: en-CA
To:     Pekka Paalanen <ppaalanen@gmail.com>,
        =?UTF-8?Q?Andr=c3=a9_Almeida?= <andrealmeid@igalia.com>
Cc:     pierre-eric.pelloux-prayer@amd.com,
        =?UTF-8?B?J01hcmVrIE9sxaHDoWsn?= <maraeo@gmail.com>,
        amaranath.somalapuram@amd.com, linux-kernel@vger.kernel.org,
        amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org,
        kernel-dev@igalia.com, alexander.deucher@amd.com,
        contactshashanksharma@gmail.com,
        =?UTF-8?Q?Christian_K=c3=b6nig?= <christian.koenig@amd.com>,
        "Pierre-Loup A . Griffais" <pgriffais@valvesoftware.com>
References: <20230123202646.356592-1-andrealmeid@igalia.com>
 <20230123202646.356592-2-andrealmeid@igalia.com>
 <37e78013-5ed6-e80f-4a1c-1e61bed59735@amd.com>
 <20230207153048.78697e7b@eldfell>
From:   =?UTF-8?Q?Michel_D=c3=a4nzer?= <michel.daenzer@mailbox.org>
In-Reply-To: <20230207153048.78697e7b@eldfell>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 2/7/23 14:30, Pekka Paalanen wrote:
> On Mon, 23 Jan 2023 21:38:11 +0100
> Christian König <christian.koenig@amd.com> wrote:
>> Am 23.01.23 um 21:26 schrieb André Almeida:
>>>
>>> diff --git a/Documentation/gpu/drm-reset.rst b/Documentation/gpu/drm-reset.rst
>>> new file mode 100644
>>> index 000000000000..0dd11a469cf9
>>> --- /dev/null
>>> +++ b/Documentation/gpu/drm-reset.rst
>>> @@ -0,0 +1,51 @@
>>> +================
>>> +DRM Device Reset
>>> +================
>>> +
>>> +The GPU stack is really complex and is prone to errors, from hardware bugs,
>>> +faulty applications and everything in the many layers in between. To recover
>>> +from this kind of state, sometimes is needed to reset the GPU. Unproper handling
>>> +of GPU resets can lead to an unstable userspace. This page describes what's the
>>> +expected behaviour from DRM drivers to do in those situations, from usermode
>>> +drivers and compositors as well.
>>> +
>>> +Robustness
>>> +----------
>>> +
>>> +First of all, application robust APIs, when available, should be used. This
>>> +allows the application to correctly recover and continue to run after a reset.
>>> +Apps that doesn't use this should be promptly killed when the kernel driver
>>> +detects that it's in broken state. Specifically guidelines for some APIs:
>>> +  
>>
>>> +- OpenGL: During a reset, KMD kill processes that haven't ARB Robustness
>>> +  enabled, assuming they can't recover.  
>>
>> This is a pretty clear NAK from my side to this approach. The KMD should 
>> never mess with an userspace process directly in such a way.
>>
>> Instead use something like this "OpenGL: KMD signals the abortion of 
>> submitted commands and the UMD should then react accordingly and abort 
>> the application.".
> 
> what Christian said, plus I would not assume that robust programs will
> always respond by creating a new context. They could also switch
> to a software renderer, [...]

That is indeed what Firefox does.


-- 
Earthling Michel Dänzer            |                  https://redhat.com
Libre software enthusiast          |         Mesa and Xwayland developer