Messages in this thread | | | Subject | Re: [PATCH 2/9] perf_counter: fix update_userpage() | From | Peter Zijlstra <> | Date | Thu, 02 Apr 2009 11:28:37 +0200 |
| |
On Thu, 2009-04-02 at 20:21 +1100, Paul Mackerras wrote: > Peter Zijlstra writes: > > > The below might work: > > > > u32 seq; > > s64 count; > > > > again: > > seq = pc->lock; > > > > if (unlikely(seq & 1)) { > > I don't believe we can ever see this condition, since pc->lock is > updated in the kernel either at interrupt level on the cpu this task > is running on, or in the kernel in the context of this task. So this > userspace code can never run in the middle of the kernel updating > things.
Colour me paranoid ;-)
| |