[Bug 78679] Gen4-5 code lost: runtime_check_aads_emit

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=78679

--- Comment #5 from Iago Toral <itoral at igalia.com> ---
Created attachment 99623
  --> https://bugs.freedesktop.org/attachment.cgi?id=99623&action=edit
Patch for testing

This patch provides two different solutions for the problem that can be
selected via an environment variable. One uses the JMPI instruction as in the
original code but at least in my IronLake hardware it does not produce correct
rendering of the GL_FILL face of the polygon. The other one uses IF/ELSE/ENDIF
instructions and does produce correct results in IronLake. I don't know why
only the latter works, maybe someone can look at the code and point out the
problem.

To test, run the sample program (line_aa) attached to this bug report like
this:

# Test solution with IF/END/ELSE
./line_aa

# Test solution with JMPI
INTEL_TRY_AA_JMPI=1 ./line_aa

The correct result should show the front face of the triangle rendered in
GL_LINE mode with antialiased lines and the back face rendered in GL_FILL mode
with smooth interpolation of the red, green and blue vertices.

I am particularly interested in having this tested in Gen < 5 since I don't
have that hardware.

-- 
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/20140523/91755724/attachment-0001.html>