Created attachment 30976 [details]
Bug detailed description:
Glean case pointsprite.c fails on 965GM, which used to fail.
Do you mean "used to pass"? If so, bisect.
We find the culprit commit is commit 5340b6dff73a0a23531ce2a5f28fba8303adab6e.
Maybe its makefile has some problem then, the i965_dri.so is always not built, which makes it run with software raster. This complexed our bisecting. Finally we will build the dri in its directory seperately, and find the culprit commit is a merge commit, and its parents commit works well.
Merge: 9fd26da ee4c921
Author: Brian Paul <firstname.lastname@example.org>
Date: Tue Feb 10 16:44:02 2009 -0700
Merge commit 'origin/gallium-master-merge'
This is the big merge of the gallium-0.2 branch into master.
gallium-master-merge was just the staging area for it.
Both gallium-0.2 and gallium-master-merge are considered closed now.
Note: the failure is only in the GL_POLYGON mode.
*** Bug 39669 has been marked as a duplicate of this bug. ***
I've looked at this a bit more: the gen4 SF code doesn't do point handling for unfilled polygons.
Would you mind provide me a few pointers on the code section to look at, which you identified as problematic on gen4?
I've presumed this would be one of the following based on your comment:
I'm not familiar with the code, but I suspect it's in brw_clip_unfilled.c
on Oct 01, 2016 at 19:01:55.
(provided by the Example extension).