[Bug 84104] Ring hang with geometry shaders on SNB

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=84104

--- Comment #5 from Iago Toral <itoral at igalia.com> ---
For reference, this is all that I could find above this hang so far:
http://lists.freedesktop.org/archives/mesa-dev/2014-August/065726.html

In summary:
  - glDrawElements is used
  - There are repeating indices (at least 8 consecutive repeating indices, or 8
repeating vertices in any subset of 9 consecutive indices)
  - primitive type is GL_TRIANGLE_STRIP_ADJACENCY
  - The hang persists even in the case that the GS is empty and we clip all the
vertices (so it really looks like this is unrelated to the code we generate)

I can reproduce the problem with a standalone program that can switch between
glDrawArrays and glDrawElements (for glDrawArrays I just pass the same vertex
data for all vertices). Both executions generate the exact same code for the VS
and GS, but only the version with glDrawElements version hangs...

So based on this, it would really look like the problem is not in the geometry
shader implementation, but something else.

Kenneth, here are two AUB traces I obtained using this program. One for
glDrawElements and one for  glDrawArrays. Could you run both through the
simulator and see if you spot any differences? Maybe that can give us a
clue....

http://people.igalia.com/itoral/intel.hangs_with_elements.aub
http://people.igalia.com/itoral/intel.works_with_arrays.aub

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20140922/d68ad338/attachment-0001.html>