Messages in this thread | | | Subject | Re: [RFC PATCH 2/6] jump label v3 - x86: Introduce generic jump patching without stop_machine | From | Steven Rostedt <> | Date | Wed, 18 Nov 2009 20:22:47 -0500 |
| |
On Wed, 2009-11-18 at 16:58 -0800, Paul E. McKenney wrote:
> I do invoke smp_send_reschedule() with irqs disabled, which did arouse my > suspicions of late. But this seems to bypass the smp_call_function() > code that is most definitely illegal to invoke with irqs disabled, > so no smoking gun. All that aside, if invoking smp_send_reschedule() > with irqs disabled is in any way a bad idea, please let me know so I > can rearrange the code appropriately.
I don't think you have anything to worry about here. If calling smp_send_reschedule was bad with keeping interrupts disabled, then we would have a lot of problems with the scheduler. That is called with the rq lock held and with interrupts disabled.
-- Steve
| |