Bug 29152

Summary:	[r300g] mempool makes ut2004 crash
Product:	Mesa	Reporter:	almos <aaalmosss>
Component:	Drivers/DRI/r300	Assignee:	Default DRI bug account <dri-devel>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	medium	CC:	pavel.ondracka
Version:	git
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:
Attachments:	backtrace Possible fix

Description almos 2010-07-19 11:19:01 UTC

It stops with

util/u_mempool.c:92:util_mempool_malloc_st: Assertion `block->magic == 0xcafe4321' failed.
debug_get_bool_option: GALLIUM_ABORT_ON_ASSERT = TRUE
Signal: SIGTRAP [trap]
Aborting.

upon loading a level.

Comment 1 Marek Olšák 2010-07-19 12:23:35 UTC

I can't reproduce this with current mesa master (3b189d888a99336a2df037e2b595844d4b9e05b5). Could you test it again after git pull?

Do you have an x86_64 system?

Does it crash if you set the environment variable GALLIUM_ABORT_ON_ASSERT to 0?

Comment 2 almos 2010-07-19 12:59:04 UTC

I tried with 3b189d888a99336a2df037e2b595844d4b9e05b5, and it asserts. If I start with GALLIUM_ABORT_ON_ASSERT=0, then lots of these

util/u_mempool.c:88:util_mempool_malloc_st: Assertion `block->magic == 0xcafe4321' failed.
debug_get_bool_option: GALLIUM_ABORT_ON_ASSERT = FALSE
continuing...
util/u_mempool.c:100:util_mempool_free_st: Assertion `block->magic == 0xcafe4321' failed.
debug_get_bool_option: GALLIUM_ABORT_ON_ASSERT = FALSE
continuing...

are printed, but it doesn't crash, and I didn't notice any visual corruption (other than the ones I already reported in #28993 and #28994). Maybe I'm not eagle-eyed enough...

It's an x86_32 system with kernel 2.6.33.1.

Comment 3 Pavel Ondračka 2010-07-19 13:15:19 UTC

Created attachment 37187 [details]
backtrace

I have the same problem with Heroes of Newerth, it crash at start. It works OK with GALLIUM_ABORT_ON_ASSERT=0 set.

Comment 4 Marek Olšák 2010-07-19 14:28:11 UTC

Created attachment 37188 [details] [review]
Possible fix

Could you try the attached patch?

Comment 5 Pavel Ondračka 2010-07-20 00:22:10 UTC

HoN works fine with your patch(In reply to comment #4)
> Created an attachment (id=37188) [details]
> Possible fix
> 
> Could you try the attached patch?
 
HoN works fine with your patch.

Comment 6 Marek Olšák 2010-07-20 01:27:50 UTC

I've committed the patch, closing..

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.