On Mon, Jan 17, 2005 at 03:40:16PM +0100, Ingo Molnar wrote:
> no ... normally you should only use __get_cpu_var() if you know that
> you are in a non-preempt case. It's a __ internal function for a
> reason. Where did it trigger?
XFS has statistics which are 'per cpu' but doesn't use per_cpu
* Chris Wedgwood <[EMAIL PROTECTED]> wrote:
> It seems logical that __get_cpu_var should use __smp_processor_id()
> rather than smp_processor_id(). Noticed when __get_cpu_var was making
> lots of noise with CONFIG_DEBUG_PREEMPT=y
no ... normally you should only use __get_cpu_var() if you know
* Chris Wedgwood [EMAIL PROTECTED] wrote:
It seems logical that __get_cpu_var should use __smp_processor_id()
rather than smp_processor_id(). Noticed when __get_cpu_var was making
lots of noise with CONFIG_DEBUG_PREEMPT=y
no ... normally you should only use __get_cpu_var() if you know that
On Mon, Jan 17, 2005 at 03:40:16PM +0100, Ingo Molnar wrote:
no ... normally you should only use __get_cpu_var() if you know that
you are in a non-preempt case. It's a __ internal function for a
reason. Where did it trigger?
XFS has statistics which are 'per cpu' but doesn't use per_cpu
On Sun, Jan 16, 2005 at 09:50:44PM -0800, Chris Wedgwood wrote:
> Note, even with this removed I'm still seeing a few (many actually)
> "BUG: using smp_processor_id() in preemptible [0001] code: xxx"
> messages which I've not seen before --- that might be unrelated but
> I do see *many* such
On Sun, Jan 16, 2005 at 09:50:44PM -0800, Chris Wedgwood wrote:
Note, even with this removed I'm still seeing a few (many actually)
BUG: using smp_processor_id() in preemptible [0001] code: xxx
messages which I've not seen before --- that might be unrelated but
I do see *many* such
6 matches
Mail list logo