WaitForSomething(): select: Invalid argument, part 2

2013-09-30 Thread Felipe Contreras
Hi,

I've been investigating the bug regarding WaitForSomething() invalid
arguments[1], and I've made some progress, but could use some help.

This has happened sporadically in my machine, it was happening
reliably with Chromium but the issue went away before I could track it
down, and now it's happening with Dota 2.

It happens always in the same part of the game, so at least I could
try to reproduce, but it doesn't happen all the time.

I've compiled my own X server and added debugging to find out the
culprit, and so far this is what I've found.

The issue seems to happen in dix/dixutils.c::BlockHandler(), after
this is run, waittime,tv_sec = 0, and waittime.tv_usec = -1000.
Clearly the arguments are wrong, so it's no wonder Select() complains.
It's clear that an integer -1 is being converted wrongly, but I've no
idea where.

I've tried adding debugging to all the BlockHandler() functions I
could see, as I don't know how that works, so far with no luck. It
seems to be one of the screens/gpuscreens BlockHandler(), but I don't
know where those are coming from.

I'll keep trying to reproduce the issue, and hopefully the next time I
do I would have more information.

In the meantime, do you have any pointers?

[1] http://thread.gmane.org/gmane.comp.freedesktop.xorg/49065

-- 
Felipe Contreras
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel


Re: WaitForSomething(): select: Invalid argument

2013-05-09 Thread Alan Coopersmith

On 05/ 9/13 12:32 PM, Grant wrote:

  WaitForSomething(): select: Invalid argument



https://lkml.org/lkml/2012/8/31/1 ?


Thanks, Google didn't bring that up for me.  So it's a kernel issue.
I'm on 3.9-rc7.  I'll pursue this further with the kernel guys if it
happens on 3.9-rc8.  I've only seen it once so far.


That is wrong, the patch mentioned in that mail thread was applied long
time ago, in v3.6[1].

And the issue is most definitely in X:

---
Also, letting the X people know that they are doing something crazy
sounds like a good idea. -- Linus
---

This happens on my Arch Linux 3.8.11 and I've seen it happen also on
3.9.

X crashing is not good.


Agreed.  What should we do?


To start with, xorg-devel is how you reach the X developers, but with no
detail about what argument is invalid or why, I don't know what we can do.

--
-Alan Coopersmith-  alan.coopersm...@oracle.com
 Oracle Solaris Engineering - http://blogs.oracle.com/alanc
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel