Messages in this thread | | | Date | Thu, 17 Oct 2013 17:09:44 +0200 | From | Peter Zijlstra <> | Subject | Re: [PATCH] perf, x86: Optimize intel_pmu_pebs_fixup_ip() |
| |
On Thu, Oct 17, 2013 at 11:03:58AM -0400, Don Zickus wrote: > On Thu, Oct 17, 2013 at 04:51:31PM +0200, Peter Zijlstra wrote: > > On Thu, Oct 17, 2013 at 10:49:13AM -0400, Don Zickus wrote: > > > For some reason this patch is page faulting at an invalid address inside > > > __intel_pmu_pebs_event(). > > > > Ah yes, I lost a refresh, but read on; I've send a gazillion new emails > > since ;-) > > Yes, I have noticed, but I got worried when you labeled some of them with > 'untested'. :-) Thought I would stick to something a little more close to > working. > > > > > I think it was something like: s/this_cpu_ptr/this_cpu_read/ to make it > > work again. > > Thanks. I will try some of the other pieces today (unlike yesterday).
The patches you find in:
http://programming.kicks-ass.net/sekrit/patches.tar.bz2
are actually running on my machine now.
One of the things I was considering was further shrinking the max basic block size from 4k to maybe 1k or 512 bytes. Not sure what a sane basic block length limit would be.
I did try 1k earlier today and I seemed to still get near 100% rewind success rates.
Stephane, Andi, any clues?
| |