lockprof: add contested-only profiling

This allows tracking all wait times with much smaller runtime impact.

For example when doing -j 104 buildkernel on tmpfs:

no profiling:	2921.70s user 282.72s system 6598% cpu 48.562 total
all acquires:	2926.87s user 350.53s system 6656% cpu 49.237 total
contested only:	2919.64s user 290.31s system 6583% cpu 48.756 total
This commit is contained in:
Mateusz Guzik 2021-05-18 21:07:19 +02:00
parent fca5cfd584
commit a0842e69aa

View File

@ -272,6 +272,7 @@ DPCPU_DEFINE_STATIC(struct lock_prof_cpu, lp);
#define LP_CPU(cpu) (DPCPU_ID_PTR((cpu), lp)) #define LP_CPU(cpu) (DPCPU_ID_PTR((cpu), lp))
volatile int __read_mostly lock_prof_enable; volatile int __read_mostly lock_prof_enable;
int __read_mostly lock_contested_only;
static volatile int lock_prof_resetting; static volatile int lock_prof_resetting;
#define LPROF_SBUF_SIZE 256 #define LPROF_SBUF_SIZE 256
@ -605,6 +606,8 @@ lock_profile_obtain_lock_success(struct lock_object *lo, int contested,
/* don't reset the timer when/if recursing */ /* don't reset the timer when/if recursing */
if (!lock_prof_enable || (lo->lo_flags & LO_NOPROFILE)) if (!lock_prof_enable || (lo->lo_flags & LO_NOPROFILE))
return; return;
if (lock_contested_only && !contested)
return;
spin = (LOCK_CLASS(lo)->lc_flags & LC_SPINLOCK) ? 1 : 0; spin = (LOCK_CLASS(lo)->lc_flags & LC_SPINLOCK) ? 1 : 0;
if (spin && lock_prof_skipspin == 1) if (spin && lock_prof_skipspin == 1)
return; return;
@ -729,6 +732,8 @@ SYSCTL_INT(_debug_lock_prof, OID_AUTO, skipspin, CTLFLAG_RW,
&lock_prof_skipspin, 0, "Skip profiling on spinlocks."); &lock_prof_skipspin, 0, "Skip profiling on spinlocks.");
SYSCTL_INT(_debug_lock_prof, OID_AUTO, rejected, CTLFLAG_RD, SYSCTL_INT(_debug_lock_prof, OID_AUTO, rejected, CTLFLAG_RD,
&lock_prof_rejected, 0, "Number of rejected profiling records"); &lock_prof_rejected, 0, "Number of rejected profiling records");
SYSCTL_INT(_debug_lock_prof, OID_AUTO, contested_only, CTLFLAG_RW,
&lock_contested_only, 0, "Only profile contested acquires");
SYSCTL_PROC(_debug_lock_prof, OID_AUTO, stats, SYSCTL_PROC(_debug_lock_prof, OID_AUTO, stats,
CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0,
dump_lock_prof_stats, "A", dump_lock_prof_stats, "A",