Bug 44963 - [bisected ] oglc depth-stencil(basic.read.ds) regressed on SNB DT GT2
Summary: [bisected ] oglc depth-stencil(basic.read.ds) regressed on SNB DT GT2
Status: CLOSED DUPLICATE of bug 49772
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: git
Hardware: All Linux (All)
: high major
Assignee: Ian Romanick
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 42993
  Show dependency treegraph
 
Reported: 2012-01-19 20:37 UTC by fangxun
Modified: 2012-10-26 08:30 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments

Description fangxun 2012-01-19 20:37:55 UTC
System Environment:
--------------------------
Arch:           i386
Platform:       Sugarbay
Libdrm:         (master)2.4.30-1-g66518ab5653cfdc840cd69e7b653ec05df060584
Mesa:           (8.0)c85402aba91755808729fadf57a9f92285f4e61a
Xserver:                (server-1.11-branch)xorg-server-1.11.3
Xf86_video_intel:(master)2.17.0-478-g35f81005f91d294e61bb4ced7cbddd1a76ccb324
Kernel:         (drm-intel-fixes) 8109021313c7a3d8947677391ce6ab9cd0bb1d28

Bug detailed description:
------------------------- 
It failed on SugarBay(GT2) and MahoBay. It passed on huronriver and sugarbay(GT1).
Bisect shows c60ac7b17993d28af65b04f9bbbf3ee74c35358c is the first bad commit.
commit c60ac7b17993d28af65b04f9bbbf3ee74c35358c
Author:     Brian Paul <brianp@vmware.com>
AuthorDate: Sat Dec 24 08:54:27 2011 -0700
Commit:     Brian Paul <brianp@vmware.com>
CommitDate: Sat Dec 24 09:25:41 2011 -0700

    swrast: rewrite glDrawPixels(GL_DEPTH) with zoom

    This gets rid of another renderbuffer->PutRow() call and _DepthBuffer
    usage.  We always work with 32-bit uint Z values now.

    Reviewed-by: Eric Anholt <eric@anholt.net>


Reproduce steps:
----------------
1. start X
2. ./oglconform -z -s -suite all -v 2 -test depth-stencil basic.read.ds
Comment 1 Ian Romanick 2012-01-28 12:18:34 UTC
With Brian's renderbuffer-cleanups-v2 branch merged, this test passes on master and 8.0-rc2 on my SNB GT2 mobile (8086:0126 (rev 09)).
Comment 2 Ouping Zhang 2012-01-31 00:08:57 UTC
System Environment:
--------------------------
Libdrm:         (master)2.4.30-18-gb643b0713aefdc0611e47654e88263b53b0de6f5
Mesa:           (8.0)caebd7929dca802ece8ef36b0f85094d66133b57
Xserver:                (server-1.11-branch)xorg-server-1.11.3
Xf86_video_intel:               (master)2.17.0-599-gca252e5b51d7b2f5a7b2c2e0d8fdb024b08096db

verify with mesa 8.0 branch, this case passed on huronriver(GT2, mobile), but still failed on Sugarbay(GT2, desktop)
(In reply to comment #1)
> With Brian's renderbuffer-cleanups-v2 branch merged, this test passes on master
> and 8.0-rc2 on my SNB GT2 mobile (8086:0126 (rev 09)).
Comment 3 Gordon Jin 2012-01-31 00:20:24 UTC
reopening
Comment 4 Gordon Jin 2012-03-31 18:19:09 UTC
(In reply to comment #2)
> System Environment:
> --------------------------
> Libdrm:         (master)2.4.30-18-gb643b0713aefdc0611e47654e88263b53b0de6f5
> Mesa:           (8.0)caebd7929dca802ece8ef36b0f85094d66133b57
> Xserver:                (server-1.11-branch)xorg-server-1.11.3
> Xf86_video_intel:              
> (master)2.17.0-599-gca252e5b51d7b2f5a7b2c2e0d8fdb024b08096db
> verify with mesa 8.0 branch, this case passed on huronriver(GT2, mobile), but
> still failed on Sugarbay(GT2, desktop)

Xun, can you confirm this is pure hw difference, with the same disk?
Comment 5 lu hua 2012-05-11 02:01:17 UTC
This issue happens on Sugarbay GT2, it doesn't happen on huronriver GT2.
This case has Bug 49772 on huronriver.
Comment 6 Eric Anholt 2012-08-06 15:56:42 UTC
I can't make any sense of this bug report.  Comment 5 says "This issue happens on Sugarbay GT2, it doesn't happen on huronriver GT2.", but also links to a bug report saying that it *does* happen on huronriver.

Of course, there is no excerpt of the testcase's failure report, which might be a way that comment 5 could make sense -- if there is a difference in the failure mode between the two.  But a developer has no way to tell from the bug report given.
Comment 7 Gordon Jin 2012-08-07 03:40:04 UTC
(In reply to comment #6)
> I can't make any sense of this bug report.  Comment 5 says "This issue happens
> on Sugarbay GT2, it doesn't happen on huronriver GT2.", but also links to a bug
> report saying that it *does* happen on huronriver.

They are two issues, at least from the happening time.
- when this bug (#44963) filed, huronriver got pass.
- later huronriver regressed, so we tracked that at bug#49772.

> Of course, there is no excerpt of the testcase's failure report, which might be
> a way that comment 5 could make sense -- if there is a difference in the
> failure mode between the two.  But a developer has no way to tell from the bug
> report given.

It's a gray area if we are allowed to publish oglconform output. I hope developers could also try reproducing. If you can't reproduce, we may communicate the output in another way.
Comment 8 Kenneth Graunke 2012-08-13 00:27:02 UTC
I don't believe in a distinction between Huron River or Sugar Bay.  I've never seen a real bug that occurred on one but not the other.  In my experience, they behave identically: the only distinction that matters is GT1/GT2.

Regardless, the test is broken on master on Sandybridge GT2.  Bug #49772 tracks the failing test case, and is much less confusing---and therefore more useful.  Marking this as a duplicate.  We don't need two parallel bugs for the same test case failing on Sandybridge platforms.

*** This bug has been marked as a duplicate of bug 49772 ***


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.