[commit: ghc] master: rts: Fix inconsistencies in how retainer and heap censuses are timed. (f376eba)
git at git.haskell.org
git at git.haskell.org
Wed Nov 22 02:11:41 UTC 2017
Repository : ssh://git@git.haskell.org/ghc
On branch : master
Link : http://ghc.haskell.org/trac/ghc/changeset/f376ebac29d8dab023482a21ecfbc4dc032ab23e/ghc
>---------------------------------------------------------------
commit f376ebac29d8dab023482a21ecfbc4dc032ab23e
Author: Douglas Wilson <douglas.wilson at gmail.com>
Date: Tue Nov 21 18:05:22 2017 -0500
rts: Fix inconsistencies in how retainer and heap censuses are timed.
mut_elapsed should deduct retainer profiling and heap censuses, just as
mut_cpu does.
mutator_cpu_ns should not deduct retainer profiling or heap censuses,
since those times are included in stats.gc_cpu_ns.
Reviewers: bgamari, erikd, simonmar
Reviewed By: simonmar
Subscribers: rwbarton, thomie
Differential Revision: https://phabricator.haskell.org/D4185
>---------------------------------------------------------------
f376ebac29d8dab023482a21ecfbc4dc032ab23e
rts/Stats.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/rts/Stats.c b/rts/Stats.c
index 8f7865b..fa85878 100644
--- a/rts/Stats.c
+++ b/rts/Stats.c
@@ -633,7 +633,8 @@ stat_exit (void)
exit_elapsed = end_exit_elapsed - start_exit_elapsed - exit_gc_elapsed;
mut_elapsed = start_exit_elapsed - end_init_elapsed -
- (gc_elapsed - exit_gc_elapsed);
+ (gc_elapsed - exit_gc_elapsed) -
+ PROF_VAL(RPe_tot_time + HCe_tot_time);
mut_cpu = start_exit_cpu - end_init_cpu - (gc_cpu - exit_gc_cpu)
- PROF_VAL(RP_tot_time + HC_tot_time);
@@ -1010,8 +1011,7 @@ void getRTSStats( RTSStats *s )
s->cpu_ns = current_cpu - end_init_cpu;
s->elapsed_ns = current_elapsed - end_init_elapsed;
- s->mutator_cpu_ns = current_cpu - end_init_cpu - stats.gc_cpu_ns -
- PROF_VAL(RP_tot_time + HC_tot_time);
+ s->mutator_cpu_ns = current_cpu - end_init_cpu - stats.gc_cpu_ns;
s->mutator_elapsed_ns = current_elapsed - end_init_elapsed -
stats.gc_elapsed_ns;
}
More information about the ghc-commits
mailing list