Comment 61 for bug 656279

Revision history for this message
Øyvind Stegard (oyvindstegard) wrote : Re: [Bug 656279] Re: NVRM: os_raise_smp_barrier(), invalid context!

Claudio Melis <email address hidden> writes:

> @ Øyvind
> have you try my script?
> it disble logical cpu. before 260.19.21 I was unable to resume my laptop but
> i I disable logical cpu before suspend and reneable after resume all works
> fine. have a try.

I am assuming you mean that disabling hyperthreading logical CPUs help
on these issues ? Well, it does not make much difference here.
Suspend/resume works, but console switch is broken, triggering these
rather scary scheduling errors (and it's still broken when taking
ht-CPUs offline first). And console switch issues can probably affect
suspend/resume in bad ways.

Also tried enabling message signalled interrupt for the nvidia device,
but that does not change anything, so went back to default. I'll be
heading back to 195.X and officially give up until some new nvidia
driver release appears, since the problem is the same across several
kernel releases, up to the very latest.

>
> 2010/11/18 Øyvind Stegard <email address hidden>

[...]

>> 3100M, Dell E6510). I get "NVRM: os_schedule: Attempted to yield the CPU
>> while in atomic or interrupt context" at suspend/resume + when switching
>> to console, which eventually leads to hanging X. I got a thread going at

[...]