The thread active cycle counter TXTACTCYC is used in __delay so it shouldn't really be reset to zero by perf. Fix perf to just read the value, and instead of clearing it, record the prev_count value in enable_counter so that the delta calculations know about the previous value.
Signed-off-by: James Hogan <james.ho...@imgtec.com> Cc: Peter Zijlstra <a.p.zijls...@chello.nl> Cc: Paul Mackerras <pau...@samba.org> Cc: Ingo Molnar <mi...@redhat.com> Cc: Arnaldo Carvalho de Melo <a...@ghostprotocols.net> --- arch/metag/kernel/perf/perf_event.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/arch/metag/kernel/perf/perf_event.c b/arch/metag/kernel/perf/perf_event.c index 54fde35..a1eff36 100644 --- a/arch/metag/kernel/perf/perf_event.c +++ b/arch/metag/kernel/perf/perf_event.c @@ -617,9 +617,7 @@ static void metag_pmu_enable_counter(struct hw_perf_event *event, int idx) WARN_ONCE((config != 0x100), "invalid configuration (%d) for counter (%d)\n", config, idx); - - /* Reset the cycle count */ - __core_reg_set(TXTACTCYC, 0); + local64_set(&event->prev_count, __core_reg_get(TXTACTCYC)); goto unlock; } @@ -708,9 +706,8 @@ static u64 metag_pmu_read_counter(int idx) { u32 tmp = 0; - /* The act of reading the cycle counter also clears it */ if (METAG_INST_COUNTER == idx) { - __core_reg_swap(TXTACTCYC, tmp); + tmp = __core_reg_get(TXTACTCYC); goto out; } -- 1.8.1.2 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/