I only recently started testing 2.5.x kernels with X, and have
immediately run into problems.
My new test system is a Dell GX110 single PIII, IDE (Intel 82801AA IDE),
256MB memory. The 2.5.38-bk4,5 and -mm3 kernels were SMP (due to
not having Rusty's cpu_possible fix), non-PREEMPT.
Running XFree86 4.2.1, I get a hard lockup with the box not responsive
to pings. Ssh sessions don't respond and the pointer does not respond
to mouse movements. This occurs very soon (2-3 minutes) after starting
X with 2.5.38-bk4 and -bk5. I was able to run 2.5.38-mm3 for about 30
minutes under load (kernel compiles) before a freeze occurred. I ran
2.5.38-mm3 again for about 20 minutes with X and under load before a
second lockup occurred. Sorry, nothing logged, and when the freezes
occur, there is no response to SYSRQ-H,S or B. SYSRQ-H works as
expected before the freeze.
Without running X on this system, no freezes have been observed with any
of these kernels with the system loaded by running kernel compiles and
dbench with up to 40 clients.
The kernel compiles and dbench load tests were done on an ext3 /home
partition. /, /usr, /var are all ext3 also.
Steven
Steven Cole wrote:
>
> I only recently started testing 2.5.x kernels with X, and have
> immediately run into problems.
>
> My new test system is a Dell GX110 single PIII, IDE (Intel 82801AA IDE),
> 256MB memory. The 2.5.38-bk4,5 and -mm3 kernels were SMP (due to
> not having Rusty's cpu_possible fix), non-PREEMPT.
>
> Running XFree86 4.2.1, I get a hard lockup with the box not responsive
> to pings. Ssh sessions don't respond and the pointer does not respond
> to mouse movements. This occurs very soon (2-3 minutes) after starting
> X with 2.5.38-bk4 and -bk5. I was able to run 2.5.38-mm3 for about 30
> minutes under load (kernel compiles) before a freeze occurred. I ran
> 2.5.38-mm3 again for about 20 minutes with X and under load before a
> second lockup occurred. Sorry, nothing logged, and when the freezes
> occur, there is no response to SYSRQ-H,S or B. SYSRQ-H works as
> expected before the freeze.
>
> Without running X on this system, no freezes have been observed with any
> of these kernels with the system loaded by running kernel compiles and
> dbench with up to 40 clients.
>
Me too. deadlocks on tasklist_lock. Some fixes went into Linus's
tree a few hours ago, so please retest.
On Thu, 2002-09-26 at 16:55, Andrew Morton wrote:
> Steven Cole wrote:
> >
> > I only recently started testing 2.5.x kernels with X, and have
> > immediately run into problems.
[snipped]
> Me too. deadlocks on tasklist_lock. Some fixes went into Linus's
> tree a few hours ago, so please retest.
Sorry for the long delay (long weekend), but 2.5.39 works fine now with
X on my new test system. Thanks.
Steven