Re: Preempt Violation

[Lee Revell]

On Fri, 2004-07-16 at 20:06, Gabriel Devenyi wrote:
> This one looks particularly nasty.
> 
> 20ms non-preemptible critical section violated 4 ms preempt threshold starting 
> a                                                                                        
> t sys_ioctl+0x42/0x260 and ending at sys_ioctl+0xbd/0x260
>  [<c015881d>] sys_ioctl+0xbd/0x260
>  [<c0116510>] dec_preempt_count+0x110/0x120
>  [<c015881d>] sys_ioctl+0xbd/0x260
>  [<c0103e95>] sysenter_past_esp+0x52/0x71

Yes, it looks like there are serious issues with ioctl.

Are you using either of the recent patches to fully daemonize softirqs? 
This should help a lot.  I am using this one:

http://lkml.org/lkml/2004/7/13/125

It applied against 2.6.8-mm1, with only one PPC-specific reject, I use
i386 so it doesn't matter.

Here is another:

http://lkml.org/lkml/2004/7/13/152

Have not tested yet.

Lee

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