Re: [PATCH 5/9] perf tools: Set maximum precise value for event 'p' modifier

2013-05-10 Thread Jiri Olsa
On Fri, May 10, 2013 at 10:53:41AM +0900, Namhyung Kim wrote: > On Thu, 9 May 2013 15:32:20 +0200, Jiri Olsa wrote: > > If single 'p' modifier is specified for event, set the > > system precise value for perf_events_attr::precise_ip. > > > > If more than a single 'p' is specified keep the intended

Re: [PATCH 5/9] perf tools: Set maximum precise value for event 'p' modifier

2013-05-10 Thread Jiri Olsa
On Thu, May 09, 2013 at 01:43:52PM -0600, David Ahern wrote: > On 5/9/13 7:32 AM, Jiri Olsa wrote: > >+static int precise_default(void) > >+{ > >+int precise = perf_precise__get(); > >+static int warned; > >+ > >+/* > >+ * Precise info not supported by by this kernel, > >+ * set

Re: [PATCH 5/9] perf tools: Set maximum precise value for event 'p' modifier

2013-05-09 Thread Namhyung Kim
On Thu, 9 May 2013 15:32:20 +0200, Jiri Olsa wrote: > If single 'p' modifier is specified for event, set the > system precise value for perf_events_attr::precise_ip. > > If more than a single 'p' is specified keep the intended > value. So there's no way to set precise=1 on a system suppports prec

Re: [PATCH 5/9] perf tools: Set maximum precise value for event 'p' modifier

2013-05-09 Thread David Ahern
On 5/9/13 7:32 AM, Jiri Olsa wrote: +static int precise_default(void) +{ + int precise = perf_precise__get(); + static int warned; + + /* +* Precise info not supported by by this kernel, +* set 1 as the precise value. +*/ + if (precise == -1) +

[PATCH 5/9] perf tools: Set maximum precise value for event 'p' modifier

2013-05-09 Thread Jiri Olsa
If single 'p' modifier is specified for event, set the system precise value for perf_events_attr::precise_ip. If more than a single 'p' is specified keep the intended value. If precise is not supported by system, warning is disaplyed. Signed-off-by: Jiri Olsa Cc: Corey Ashford Cc: Frederic Wei