2.5.34 randomly freezes under X (seems input related)

2.5.34 randomly freezes under X (seems input related)

[Michel Eyckmans (MCE)]

Greetz,

On 2.5.34, moving the mouse around (just moving, not clicking) under 
an (almost) idle X "almost" reliably freezes my machine if I just do 
it long enough. I write "almost" because:

 1) It happened on 14 out of 15 sessions, sometimes immediately
    after logging in. I gave up on the one exception after trying
    for more than 5 minutes.

 2) It might be related to whether there is some background activity. 
    I have a shell script that forks a lot of stuff in quick succession 
    and it seems that a lockup is guaranteed if I move my mouse while 
    that script is running. The script is not a requirement, though.

Sadly, only the reset button helps and of course nothing useful shows 
up in the logs. :-( I tried very hard to reproduce it on a virtual 
console, but failed no matter how hard I stressed the box. So X "must" 
be part of the problem somehow.

This is 2.5.34 on a dual Pentium, gcc 2.95, no preempt, AT keyboard 
& serial mouse. I never got 2.5.32 and 2.5.33 to compile/boot, so the 
problem may not be all that new. 2.5.31 works OK (not fine, but OK).

Any ideas? It looks a bit like some input layer problem to me.
Feel free to ask for more configuration details if you're interested.

   MCE

PS: Thanks for the working floppies! 

   MCE
-- 
========================================================================
M. Eyckmans (MCE)          Code of the Geeks v3.1       mce-at-pi-dot-be
GCS d+ s+:- a37 C+++$ UHLUASO+++$ P+ L+++ E--- W++ N+++ !o K w--- !O M--
 V-- PS+ PE+ Y+ PGP- t--- !5 !X R- tv- b+ DI++ D-- G++ e+++ h+(*) !r y?
========================================================================

Re: 2.5.34 randomly freezes under X (seems input related)

[Andrew Morton]

"Michel Eyckmans (MCE)" wrote:
> 
> Greetz,
> 
> On 2.5.34, moving the mouse around (just moving, not clicking) under
> an (almost) idle X "almost" reliably freezes my machine if I just do
> it long enough. I write "almost" because:
> 
>  1) It happened on 14 out of 15 sessions, sometimes immediately
>     after logging in. I gave up on the one exception after trying
>     for more than 5 minutes.
> 
>  2) It might be related to whether there is some background activity.
>     I have a shell script that forks a lot of stuff in quick succession
>     and it seems that a lockup is guaranteed if I move my mouse while
>     that script is running. The script is not a requirement, though.
> 
> Sadly, only the reset button helps and of course nothing useful shows
> up in the logs. :-( I tried very hard to reproduce it on a virtual
> console, but failed no matter how hard I stressed the box. So X "must"
> be part of the problem somehow.
> 
> This is 2.5.34 on a dual Pentium, gcc 2.95, no preempt, AT keyboard
> & serial mouse. I never got 2.5.32 and 2.5.33 to compile/boot, so the
> problem may not be all that new. 2.5.31 works OK (not fine, but OK).
> 

Me too.  Dual PIII, 2.5.34+localhacks.  It's happened a couple
of times.  The system locks up and the CPU fans go nuts.  In X, so
no info available.

There are a few locking problems in 2.5.34 which _may_ be fixed
now, but they only seem to affect CLONE_THREAD.  It would be worth
grabbing the latest `gzipped full patch' from 
http://www.kernel.org/pub/linux/kernel/people/dwmw2/bk-2.5/
though.