[oi-dev] Osol bug id 5748

Alan Coopersmith alan.coopersmith at oracle.com
Thu Apr 14 18:19:38 UTC 2011


On 04/14/11 10:05 AM, Frank Middleton wrote:
> On 04/08/11 19:43, Alan Coopersmith wrote:
>> The nvidia driver obfuscates its symbols, only the nvidia developers can
>> do much if you're looping in there.
> 
> It could be, but the loops seem to be in user space, and the bug(s) seem to
> affect non-nvidia users too. Is there an easy way to tell if it is a driver
> problem or not?

I was actually referring to the nvidia_drv.so "driver" module that the Xorg
server loads in user space.   You'll see nv123456-style function names in
Xorg stack traces when it's in use.

> If it is a driver problem, what is the best way if any to get Nvidia to
> look at it?

Update the opensolaris bug, so that John Martin, our nvidia contact, has the
information he needs to reproduce it to either debug himself or take to nvidia.

> If the bug is in the Xorg signalling code, how would the putative fix(es)
> make it upstream?

You would mail suggested patches to the xorg-devel mailing list, following
their development process:
	http://www.x.org/wiki/Development/Documentation/SubmittingPatches
	http://www.x.org/wiki/XServer#DevelopmentProcess

> The interaction between hald and the hal code in Xorg, and the interaction
> between the hal code in Xorg and the mainline signalling code in Xorg is
> complex, too much so for my analytical skills. However it seems to me that
> the code that enables and disables signalling is partially duplicated and
> prone to the very race condition that seems to be occurring. It is quite
> possible that building Xorg with debug might slow it enough that the race
> no longer occurs, or occurs far more infrequently. I have a suitable build
> ready to try out this hypothesis but for a variety of reasons it will take
> a few more days for me to have the time to do it. In the interim, any
> thoughts you might have would be much appreciated. Subtle race
> conditions are notoriously difficult to debug...

I think I saw some signal handling fixes go by on xorg-devel since the 1.9
release series - but I wouldn't be surprised if there's more to find still.

-- 
	-Alan Coopersmith-        alan.coopersmith at oracle.com
	 Oracle Solaris Platform Engineering: X Window System





More information about the oi-dev mailing list