Messages in this thread | | | Date | Mon, 14 Mar 2011 17:54:58 +0100 | From | Marcin Slusarz <> | Subject | Re: [PATCH] perf lock: fix sorting by wait_min |
| |
On Tue, Feb 22, 2011 at 06:47:15PM +0100, Marcin Slusarz wrote: > if lock was uncontended, wait_time_min == ULLONG_MAX, so we need to handle > this case differently to show high wait times first > > Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com> > --- > tools/perf/builtin-lock.c | 13 ++++++++++++- > 1 files changed, 12 insertions(+), 1 deletions(-) > > diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c > index 0305a40..812ad26 100644 > --- a/tools/perf/builtin-lock.c > +++ b/tools/perf/builtin-lock.c > @@ -202,9 +202,20 @@ static struct thread_stat *thread_stat_findnew_first(u32 tid) > SINGLE_KEY(nr_acquired) > SINGLE_KEY(nr_contended) > SINGLE_KEY(wait_time_total) > -SINGLE_KEY(wait_time_min) > SINGLE_KEY(wait_time_max) > > +static int lock_stat_key_wait_time_min(struct lock_stat *one, > + struct lock_stat *two) > +{ > + u64 s1 = one->wait_time_min; > + u64 s2 = two->wait_time_min; > + if (s1 == ULLONG_MAX) > + s1 = 0; > + if (s2 == ULLONG_MAX) > + s2 = 0; > + return s1 > s2; > +} > + > struct lock_key { > /* > * name: the value for specify by user > --
Anything wrong with this patch? I can't find it in linux-next.
Marcin
| |