Early versions of KNC chips have a bug where bits above 32 were not properly set. We worked around this by only using the bottom 32 bits (out of 40 that should be available).
It turns out this workaround breaks overflow handling. The buggy silicon will in theory never be used in production systems, so remove this workaround so we get proper overflow support. Signed-off-by: Vince Weaver <[email protected]> diff -ur linux-3.7-rc1.orig/arch/x86/kernel/cpu/perf_event_knc.c linux-3.7-rc1/arch/x86/kernel/cpu/perf_event_knc.c --- linux-3.7-rc1.orig/arch/x86/kernel/cpu/perf_event_knc.c 2012-10-14 17:41:04.000000000 -0400 +++ linux-3.7-rc1/arch/x86/kernel/cpu/perf_event_knc.c 2012-10-17 11:50:08.688394147 -0400 @@ -226,12 +226,11 @@ .event_map = knc_pmu_event_map, .max_events = ARRAY_SIZE(knc_perfmon_event_map), .apic = 1, - .max_period = (1ULL << 31) - 1, + .max_period = (1ULL << 39) - 1, .version = 0, .num_counters = 2, - /* in theory 40 bits, early silicon is buggy though */ - .cntval_bits = 32, - .cntval_mask = (1ULL << 32) - 1, + .cntval_bits = 40, + .cntval_mask = (1ULL << 40) - 1, .get_event_constraints = x86_get_event_constraints, .event_constraints = knc_event_constraints, .format_attrs = intel_knc_formats_attr, -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
