* Ingo Molnar <mingo@elte.hu> wrote:

quoted text
> hardirq_count() is correct, but looking at PF_HARDIRQ's definition in sched.h:
> 
>  #define PF_EXITPIDONE  0x00000008      /* pi exit done on shut down */
>  #define PF_VCPU        0x00000010      /* I'm a virtual CPU */
>  #define PF_HARDIRQ     0x08000020      /* hardirq context */
>  #define PF_NOSCHED     0x00000020      /* Userspace does not expect scheduling */
>  #define PF_FORKNOEXEC  0x00000040      /* forked but didn't exec */
> 
> Reveals that due to a typo it not only overlaps the PF_NOSCHED bit, but
> also has a spurious 0x08000000 component.

The reason is that when we forward ported the definition, i first moved it
to the 0x08000000 slot - but that slot was already taken. (our PF_ task
flag space is really crowded ...)

Then i moved it to a free spot, 0x20. Or so i thought: a later -rt patch
in the queue introduced PF_NOSCHED which overlapped it.

But the bigger problem was the spurious 0x08000000 component, which overlaps
with:

 #define PF_SOFTIRQ      0x08000000      /* softirq context */

Explaining why the warning triggered in ksoftirqd ;-)

Anyway, my fix should solve this. Do you still see the lockup under X? (make
sure you also have the IPI fix applied, see the patch in this same thread.)

	Ingo
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/