x86/perf: Exclude the debug stack from the callchains
authorFrederic Weisbecker <fweisbec@gmail.com>
Sat, 5 Dec 2009 11:01:17 +0000 (12:01 +0100)
committerFrederic Weisbecker <fweisbec@gmail.com>
Sun, 6 Dec 2009 07:27:21 +0000 (08:27 +0100)
commit7f33f9c5cc3c99aeaf4d266a7ed502b828115a53
treef37f964798815c3946c7334e598ba3600a0539d0
parentc0dfb2feb632537cf0a9d2ce3c29bcf5778fec59
x86/perf: Exclude the debug stack from the callchains

Dumping the callchains from breakpoint events with perf gives strange
results:

3.75%             perf  [kernel]           [k] _raw_read_unlock
                       |
                       --- _raw_read_unlock
                           perf_callchain
                           perf_prepare_sample
                           __perf_event_overflow
                           perf_swevent_overflow
                           perf_swevent_add
                           perf_bp_event
                           hw_breakpoint_exceptions_notify
                           notifier_call_chain
                           __atomic_notifier_call_chain
                           atomic_notifier_call_chain
                           notify_die
                           do_debug
                           debug
                           munmap

We are infected with all the debug stack. Like the nmi stack, the debug
stack is undesired as it is part of the profiling path, not helpful for
the user.

Ignore it.

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: "K. Prasad" <prasad@linux.vnet.ibm.com>
arch/x86/kernel/cpu/perf_event.c