lk/linux-kernel-test
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

perf_counter: update mmap() counter read

Browse Source 
Paul noted that we don't need SMP barriers for the mmap() counter read
because its always on the same cpu (otherwise you can't access the hw
counter anyway).

So remove the SMP barriers and replace them with regular compiler
barriers.

Further, update the comment to include a race free method of reading
said hardware counter. The primary change is putting the pmc_read
inside the seq-loop, otherwise we can still race and read rubbish.

Noticed-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Orig-LKML-Reference: <20090402091319.577951445@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This commit is contained in:
Peter Zijlstra
2009-04-02 11:12:04 +02:00
committed by Ingo Molnar
parent 5872bdb88a
commit 92f22a3865
2 changed files with 12 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
include/linux/perf_counter.h
View File

@@ -167,30 +167,28 @@ struct perf_counter_mmap_page {
/* /*
* Bits needed to read the hw counters in user-space. * Bits needed to read the hw counters in user-space.
* *
* The index and offset should be read atomically using the seqlock: * u32 seq;
* * s64 count;
* __u32 seq, index;
* __s64 offset;
* *
* again: * again:
* rmb();
* seq = pc->lock; * seq = pc->lock;
*
* if (unlikely(seq & 1)) { * if (unlikely(seq & 1)) {
* cpu_relax(); * cpu_relax();
* goto again; * goto again;
* } * }
* *
* index = pc->index; * if (pc->index) {
* offset = pc->offset; * count = pmc_read(pc->index - 1);
* count += pc->offset;
* } else
* goto regular_read;
* *
* rmb(); * barrier();
* if (pc->lock != seq) * if (pc->lock != seq)
* goto again; * goto again;
* *
* After this, index contains architecture specific counter index + 1, * NOTE: for obvious reason this only works on self-monitoring
* so that 0 means unavailable, offset contains the value to be added * processes.
* to the result of the raw timer read to obtain this counter's value.
*/ */
__u32 lock; /* seqlock for synchronization */ __u32 lock; /* seqlock for synchronization */
__u32 index; /* hardware counter identifier */ __u32 index; /* hardware counter identifier */

4
kernel/perf_counter.c
View File

@@ -1340,13 +1340,13 @@ void perf_counter_update_userpage(struct perf_counter *counter)
*/ */
preempt_disable(); preempt_disable();
++userpg->lock; ++userpg->lock;
smp_wmb(); barrier();
userpg->index = counter->hw.idx; userpg->index = counter->hw.idx;
userpg->offset = atomic64_read(&counter->count); userpg->offset = atomic64_read(&counter->count);
if (counter->state == PERF_COUNTER_STATE_ACTIVE) if (counter->state == PERF_COUNTER_STATE_ACTIVE)
userpg->offset -= atomic64_read(&counter->hw.prev_count); userpg->offset -= atomic64_read(&counter->hw.prev_count);
smp_wmb(); barrier();
++userpg->lock; ++userpg->lock;
preempt_enable(); preempt_enable();
unlock: unlock: