Re: [Mesa3d-dev] Segfault on glClear of non-existent stencil buffer caused by bd1ce874728c06d08a1f9881f51edbdd2f1c9db0

[Chia-I Wu <ol...@gm...>]

On Wed, Mar 24, 2010 at 12:56 AM, Brian Paul <br...@vm...> wrote:
>> The problem seems to be in st_manager.c:
>>   if (visual->depth_stencil_format != PIPE_FORMAT_NONE) {
>>      mode->haveDepthBuffer = GL_TRUE;
>>      mode->haveStencilBuffer = GL_TRUE;
>>
>>      mode->depthBits =
>>         util_format_get_component_bits(visual->depth_stencil_format,
>>               UTIL_FORMAT_COLORSPACE_ZS, 0);
>>      mode->stencilBits =
>>         util_format_get_component_bits(visual->depth_stencil_format,
>>               UTIL_FORMAT_COLORSPACE_ZS, 1);
>>   }
>>
>> This sets haveStencilBuffer even for depth-only buffers.
Yes, I think this is the root cause.  Thanks for looking into and fixing it.
>> How about fixing this to set haveDepthBuffer and haveStencilBuffer
>> only if bits > 0, and later removing haveStencilBuffer,
>> haveDepthBuffer and haveAccumBuffer in favor of just testing the *bits
>> counterparts?
> The haveDepth/Stencil fields come from the original SGI GLX.  We should
> probably just test the number of bits in Mesa, as you say. Want to make a
> patch for that?
>> BTW, what if we have a floating-point depth buffer, or, say, a shared
>> exponent floating-point color buffer? How do we represent that with
>> the visual structure?
> Those things came along after the SGI GLX release.  They'd have to be added
> to __GLcontextModes.
>> Shouldn't we be using the actual formats instead of this *bits stuff,
>> maybe by having Mesa look at its internal structures instead of a
>> GLXVisual-like struct?
> yeah, probably.  I'd have to study it for a while.
That sounds good.  Many fields of __GLcontextModes are of no interest in Mesa
core.  It might make some sense not to store __GLcontextModes in the contexts
and the framebuffers after creation.

-- 
ol...@Lu...