Summary: | [regression, bisected] Piglit spec/ext_packed_float/ getteximage-invalid-format-for-packed-type fails | ||
---|---|---|---|
Product: | Mesa | Reporter: | Samuel Iglesias Gonsálvez <siglesias> |
Component: | Mesa core | Assignee: | mesa-dev |
Status: | RESOLVED FIXED | QA Contact: | mesa-dev |
Severity: | normal | ||
Priority: | medium | CC: | brianp, imirkin |
Version: | git | ||
Hardware: | Other | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: |
Description
Samuel Iglesias Gonsálvez
2015-07-22 08:30:17 UTC
Hmm, I can't reproduce that. The test passes completely for me. What driver are you using? I'm testing llvmpipe/softpipe. FWIW, I got the same regression with radeonsi. (In reply to Brian Paul from comment #1) > Hmm, I can't reproduce that. The test passes completely for me. > > What driver are you using? I'm testing llvmpipe/softpipe. Odd. I repro with llvmpipe. I guess the issue is the if (depth != 1) { _mesa_error(ctx, GL_INVALID_VALUE, "%s(depth = %d)", caller, depth); return true; } check for 1d/2d targets. The piglit test does: glGetTexImage(GL_TEXTURE_2D, 0, format, type, pxBuffer); Since it's checking for, roughly speaking, invalid things, I assume that the texImage isn't there, and so get_texture_image_dims will end up setting a depth = 0. Should that be depth = 1 instead? OK, the problem is my local copy of getteximage-invalid-format-for-packed-type.c has the patch "ext_packed_float: fix getteximage-invalid-format-for-packed-type test" which I posted for review (and was R-b'd by Jose) but not yet pushed to master. I believe the piglit test was defective as-is, per my check-in comment: The GL spec doesn't explicitly say that glGetTexImage should generate GL_INVALID_OPERATION when attempting to retrieve a non-existant texture image, but that's what NVIDIA's driver does. The purpose of this test is to check the format/type parameters, so let's define a packed float texture to avoid the undefined texture situation. Test now passes with NVIDIA. I'll push that patch. (In reply to Brian Paul from comment #4) > OK, the problem is my local copy of > getteximage-invalid-format-for-packed-type.c has the patch > "ext_packed_float: fix getteximage-invalid-format-for-packed-type test" > which I posted for review (and was R-b'd by Jose) but not yet pushed to > master. > > I believe the piglit test was defective as-is, per my check-in comment: > > The GL spec doesn't explicitly say that glGetTexImage should generate > GL_INVALID_OPERATION when attempting to retrieve a non-existant texture > image, but that's what NVIDIA's driver does. > > The purpose of this test is to check the format/type parameters, so let's > define a packed float texture to avoid the undefined texture situation. > > Test now passes with NVIDIA. > > I'll push that patch. It passes now. Thanks Brian! |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.