Bug 46086

Summary:	[IVB] Lightsmark fails to run with "Variable shadowMap0 disappeared from shader"
Product:	Mesa	Reporter:	libo <bo.c.li>
Component:	Drivers/DRI/i965	Assignee:	Ian Romanick <idr>
Status:	VERIFIED FIXED	QA Contact:
Severity:	major
Priority:	high	CC:	kenneth
Version:	git
Hardware:	All
OS:	Linux (All)
Whiteboard:
i915 platform:		i915 features:
Bug Depends on:
Bug Blocks:	42993

Description libo 2012-02-15 00:32:52 UTC

System Environment:
--------------------------
Arch:         i386
Platform:     IvyBridge
Libdrm:       (master)2.4.31-3-gced219ebbd3b266ac8326223bad62f994907ae6b
Mesa:         (8.0)65526d54aa2599b069bd443d3e6e9762e613042d
Xserver:      (server-1.11-branch)xorg-server-1.11.4
Xf86_video_intel: (master)2.17.0-650-gc0376b7f7b083ab2e87edc36e56fd8eb99c3cd05
Kernel:    3.2.4

Bug detailed description:
-------------------------
There are such error after run lightsmark: 
"Variable shadowMap0 disappeared from shader. This is usually graphics card driver error." and it can't acquire fps value.

Reproduce steps:
-----------------------
1.start x
2.gnome-session
3.vblank_mode=0 ./backend 1600x900

Comment 1 Gordon Jin 2012-02-15 00:56:06 UTC

Is this regression? If so, can you provide a good commit?

How about mesa master?

Comment 2 libo 2012-02-15 02:06:28 UTC

 It's a regression. and it's caused by mesa.
 such is a good mesa commit:
(8.0)9489ae8938b74551913337876656bcb9178a8905
 This bug also exists with the latest master mesa.

Comment 3 Gordon Jin 2012-02-15 21:38:40 UTC

Ian/Ken, do you have idea which commits may bring this? Otherwise we may have to bisect.

Comment 4 libo 2012-02-15 21:44:57 UTC

This bug only exists on IVB ,not exists on other platforms.

Comment 5 Kenneth Graunke 2012-02-15 22:47:52 UTC

There was a short time when Lightsmark was broken on 8.0.  The break started at 1b5e151ffae3ff50e9012243b7fcbbb60c3a0042, but was fixed with 99f9c9789a2d316d97ca6791343fb697f220929e.

I can verify that libo's original reported revision, 65526d54aa2599b069bd443d3e6e9762e613042d, failed to compile one of the shaders partway through the run.  This resulted in an assertion failure.  Perhaps if you aren't compiling in debug mode, you'd see a "shadowMap0 disappeared" message.  I don't know.  (You should use --enable-debug.)

However, Lightsmark works fine on IVB with the latest 8.0 branch (0aadb240e191538b07c45733dc2597235a52d274) and with master (709f50928e1d4df755ffb90ec9f33ba6c9605a32).  It ran to completion and produces a FPS value with no complaints in both cases.

Comment 6 libo 2012-02-16 19:22:14 UTC

yes.this bug doesn't exists on the latest master and  8.0 branch.

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.