[Mesa-dev] [RFC PATCH 1/1] st/dri: add a new driconf option override_glsl_version for ARK games

Samuel Pitoiset samuel.pitoiset at gmail.com
Sun Feb 5 13:19:39 UTC 2017

On 02/03/2017 07:48 PM, Bas Nieuwenhuizen wrote:
> On Fri, Feb 3, 2017, at 19:24, Jason Ekstrand wrote:
>> On Fri, Feb 3, 2017 at 9:23 AM, Samuel Pitoiset
>> <samuel.pitoiset at gmail.com <mailto:samuel.pitoiset at gmail.com>> wrote:
>>     This is similar to the MESA_GLSL_VERSION_OVERRIDE envvar (mainly
>>     for developers). But this one has the advantage to be configured
>>     for specific apps which require a context with an explicit version.
>>     For example, when an app requires a 3.2 core context, RadeonSI
>>     will return a 4.5 context but this might fail (eg. ARK games).
>> Why is returning a 4.5 context a problem?  It's supposed to be
>> more-or-less backwards compatible.
>> I'm also a bit concerned about making this driconf option be global
>> across drivers without it being just a maximum.  Suppose someone sets
>> it to 4.3 for some app that doesn't like 4.5.  What happens if they
>> try to run that app on hardware that doesn't support 4.3?  Do they get
>> a 4.3 context that just doesn't work?
> As far as I can see[1], when the game detects GL 4.3+, the engine tries
> to load a different set of shaders from disk, but the game developers
> have not enabled the right flag during building, so the shaders for
> GL4.3+ are not actually distributed with the game, which results in a
> failure to load the game. From my POV this is entirely the fault of the
> game.

I can confirm that these ARK games fail to start with a 4.3+ context. 
But from my point of view, I'm not really *sure* it's a game fault 
because as you said in the related ticket, the game starts correctly 
with AMDGPU-pro because it returns the right context. Maybe we should 
just do the same?

> That said, the bug contains a report from someone else that it works on
> intel with GL4.3+. I never got to test that or look deeper into this, so
> it could very well be that I overlooked something during this analysis.

It would be interesting to get a confirmation, but I don't have any 
intel hw here.

> - Bas
> [1]: https://bugs.freedesktop.org/show_bug.cgi?id=95374
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev

More information about the mesa-dev mailing list