Bug 39254

Summary: [bisected ironlake]piglit fbo-readpixels-depth-formats fails
Product: Mesa Reporter: fangxun <xunx.fang>
Component: Drivers/DRI/i965Assignee: Eric Anholt <eric>
Status: VERIFIED FIXED QA Contact:
Severity: normal    
Priority: high CC: eric
Version: git   
Hardware: All   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Bug Depends on:    
Bug Blocks: 42993    

Description fangxun 2011-07-15 04:39:16 UTC 
System Environment:
--------------------------
Arch:        x86_64
Platform:    piketon
Libdrm:      (master)2.4.26-1-g8d055890d90c3d92647e3d8b98d32630ef87c2c8
Mesa:         (master)02c8ee202f5a159659a027deae4721ff05cd1530
Xserver:     (master)xorg-server-1.10.99.901-104-g01dbf2514aa0cc8c40a876a24e3cb0737c41da19
Xf86_video_intel: (master)2.15.0-203-g212fa9868767637e8f430485eeb522c99e63fd16
Kernel:  (drm-intel-next)99834ea446d5c0da3f6cfa355fe4670840d45f79

Bug detailed description:
-------------------------
This regression only happens on Ironlake. 
Bisect find commit 86e62b2357447b7c97f434 is the first bad commit.
commit 86e62b2357447b7c97f434be4834f4b50aa0764d
Author:     Eric Anholt <eric@anholt.net>
AuthorDate: Wed Jun 15 12:02:12 2011 -0700
Commit:     Eric Anholt <eric@anholt.net>
CommitDate: Thu Jul 7 14:05:20 2011 -0700

    intel: Mark MESA_FORMAT_X8_Z24 as always supported.

    This prevents developer surprise at seeing a GL_DEPTH_COMPONENT
    texture have stencil bits, and avoids the metaops path accidentally
    copying stencil bits around in glCopyTexImage(GL_DEPTH_COMPONENT) (and
    being broken because swrast's glReadPixels(GL_UNSIGNED_INT_24_8) is
    broken).

Test result output:
testing GL_DEPTH_COMPONENT:
    GL_FLOAT: passed
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: passed
testing GL_DEPTH_COMPONENT32:
    GL_FLOAT: passed
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: passed
testing GL_DEPTH_COMPONENT24:
    GL_FLOAT: passed
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: passed
testing GL_DEPTH_COMPONENT16:
    GL_FLOAT: Expected 0.000000 at (0,0), found 62.000008
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: Expected 0x00 at (2,2), found 0x41
testing GL_DEPTH_STENCIL_EXT:
    GL_FLOAT: passed
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: passed
testing GL_DEPTH24_STENCIL8_EXT:
    GL_FLOAT: passed
    GL_UNSIGNED_INT: passed
    GL_UNSIGNED_SHORT: passed
    GL_UNSIGNED_BYTE: passed
PIGLIT: {'result': 'fail' }


Reproduce steps:
----------------
1. xinit&
2. ./fbo-readpixels-depth-formats -auto
Comment 1 Eric Anholt 2011-07-19 09:13:22 UTC 
(Note that while this is a P1 regression, it doesn't apply to the 7.11 release)
Comment 2 Gordon Jin 2011-07-19 19:28:43 UTC 
(In reply to comment #1)
> (Note that while this is a P1 regression, it doesn't apply to the 7.11 release)

Right, hence it's marked as normal severity to unblock the release. (the release blocker is defined as P1 & severity>=major)
Comment 3 Eric Anholt 2011-08-16 10:45:52 UTC 
We're punting on this until the RB mapping rework.
Comment 4 fangxun 2012-01-04 01:04:56 UTC 
It passes on recent driver.
Comment 5 fangxun 2012-01-04 01:05:38 UTC 
Verified with mesa master commit 4535874c1a6da74d9130e494d514284302e4d08e.

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.