[Bug 97532] Regression: GLB 2.7 & Glmark-2 GLES versions segfault due to linker precision error (259fc505) on dead variable

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=97532

Eero Tamminen <eero.t.tamminen at intel.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |VERIFIED

--- Comment #25 from Eero Tamminen <eero.t.tamminen at intel.com> ---
(In reply to Kenneth Graunke from comment #22)
> Just discovered another reason I didn't notice this bug: this shader is only
> used for the loading screen.  If you run it with -skip_load_frames, it works
> fine.

Thanks, I didn't know about that, it indeed works fine.


(In reply to Kenneth Graunke from comment #23)
> Created attachment 127853 [details] [review]
> A patch that implements a drirc option
> 
> Here's a patch that implements a drirc option.

Thanks, this works fine too, and the patch is much smaller than I expected!

Comment added at the end of patch would need to be updated before commiting.  

This same override would be needed also for currently available glmark2-es2
releases (fixes for the bugs are only in glmark2 git repo).


(In reply to Ian Romanick from comment #24)
> (In reply to Eero Tamminen from comment #19)
> > [1] Linux desktop has only handful of GLES programs (and most of them
> > compositors i.e. not really using complex shaders).  There could be other
> > regressing cases on ChromeOS (WebGL www-pages) or Android.
> 
> I suspect this will not be a problem for WebGL.  WebGL conformance is even
> more picky about such restrictions that the native API CTS.
>
> There have been no reports of Android applications failing due to this
> change.  Should there be any problems with Android applications, they will
> be handled on a case-by-case basis.

With the GLB workaround option available, this is fine for me.  If there are
other buggy proprietary apps, Kenneth's patch can be applied. -> Verified

---
We just don't know whether there are other problematic apps:
* Different WebGL scripts on random www-pages aren't conformance tested.  They
can have this bug if they're developed against Browsers & 3D driver versions
that haven't (had) this check
* AFAIK there aren't Android devices shipping with Mesa 13.0 (which is the
first version with this issue), so there cannot be any reports of issues yet
* I didn't find (by googling) any reports of GLB 2.x failing with any other
driver (and it works fine with GLES version of Intel Windows driver)
* Glmark2 had the same issue and nobody had noticed anything on Linux or
Android with other drivers (at least I was the first one to file a bug about
it)

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20161109/f85b9c0a/attachment.html>