[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

--- Comment #38 from Ian Romanick <idr at freedesktop.org> ---
(In reply to Eero Tamminen from comment #36)
> (In reply to Tomasz Figa from comment #35)
> > By the way, it is not true that GLES 2.x didn't require this check, at least
> > according to following thread:
> > 
> > https://github.com/AnalyticalGraphicsInc/cesium/issues/817
> > 
> > It refers to GLSL ES 1.00 specification
> > (http://www.khronos.org/registry/OpenGL/specs/es/2.0/GLSL_ES_Specification_1.
> > 00.pdf) and quotes the following part:
> [...]
> 
> If you read it more carefully, it talks only about *used* uniforms.
> 
> GLES 3.x is more explicit and requires that *declared* uniforms have to
> match.
> 
> 
> Different drivers could interpret *used* in different ways, as the match
> requirement being for uniforms that are:
> * declared (same as GLES 3.x)
> * referred after post-processing, or
> * linked after all optimizations are done

For the most part, the GLSL solves this problem by requiring things that are
statically referenced to match.  This means rules are applied to some things
that may be unnecessary, but it avoids the problem of depending on how hard a
particular compiler will optimize things.

> (Last one is the only state that can cause real errors, but checking also
> declarations makes things easier for developers, things behave more
> consistently between drivers and different compiler versions.)
> 
> That's why I asked whether the problematic Android games actually use the
> uniforms or not.

-- 
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/20170925/adb96c22/attachment.html>