Commit ace618cd authored by Ben Gamari's avatar Ben Gamari 🐢 Committed by Marge Bot

nonmoving-gc: Track time usage of nonmoving marking

parent cedd6f30
......@@ -151,6 +151,23 @@ typedef struct GCDetails_ {
Time cpu_ns;
// The time elapsed during GC itself
Time elapsed_ns;
//
// Concurrent garbage collector
//
// The CPU time used during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_sync_cpu_ns;
// The time elapsed during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_sync_elapsed_ns;
// The CPU time used during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_cpu_ns;
// The time elapsed during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_elapsed_ns;
} GCDetails;
//
......@@ -241,6 +258,28 @@ typedef struct _RTSStats {
// The number of times a GC thread has iterated it's outer loop across all
// parallel GCs
uint64_t scav_find_work;
// ----------------------------------
// Concurrent garbage collector
// The CPU time used during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_sync_cpu_ns;
// The time elapsed during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_sync_elapsed_ns;
// The maximum time elapsed during the post-mark pause phase of the
// concurrent nonmoving GC.
Time nonmoving_gc_sync_max_elapsed_ns;
// The CPU time used during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_cpu_ns;
// The time elapsed during the post-mark pause phase of the concurrent
// nonmoving GC.
Time nonmoving_gc_elapsed_ns;
// The maximum time elapsed during the post-mark pause phase of the
// concurrent nonmoving GC.
Time nonmoving_gc_max_elapsed_ns;
} RTSStats;
void getRTSStats (RTSStats *s);
......
......@@ -33,6 +33,7 @@ typedef int64_t Time;
#define SecondsToTime(t) ((Time)(t) * TIME_RESOLUTION)
#define TimeToSeconds(t) ((t) / TIME_RESOLUTION)
#define TimeToSecondsDbl(t) ((double)(t) / TIME_RESOLUTION)
// Use instead of SecondsToTime() when we have a floating-point
// seconds value, to avoid truncating it.
......
......@@ -103,6 +103,25 @@ data RTSStats = RTSStats {
-- | Total elapsed time (at the previous GC)
, elapsed_ns :: RtsTime
-- | The CPU time used during the post-mark pause phase of the concurrent
-- nonmoving GC.
, nonmoving_gc_sync_cpu_ns :: RtsTime
-- | The time elapsed during the post-mark pause phase of the concurrent
-- nonmoving GC.
, nonmoving_gc_sync_elapsed_ns :: RtsTime
-- | The maximum time elapsed during the post-mark pause phase of the
-- concurrent nonmoving GC.
, nonmoving_gc_sync_max_elapsed_ns :: RtsTime
-- | The CPU time used during the post-mark pause phase of the concurrent
-- nonmoving GC.
, nonmoving_gc_cpu_ns :: RtsTime
-- | The time elapsed during the post-mark pause phase of the concurrent
-- nonmoving GC.
, nonmoving_gc_elapsed_ns :: RtsTime
-- | The maximum time elapsed during the post-mark pause phase of the
-- concurrent nonmoving GC.
, nonmoving_gc_max_elapsed_ns :: RtsTime
-- | Details about the most recent GC
, gc :: GCDetails
} deriving ( Read -- ^ @since 4.10.0.0
......@@ -146,6 +165,13 @@ data GCDetails = GCDetails {
, gcdetails_cpu_ns :: RtsTime
-- | The time elapsed during GC itself
, gcdetails_elapsed_ns :: RtsTime
-- | The CPU time used during the post-mark pause phase of the concurrent
-- nonmoving GC.
, gcdetails_nonmoving_gc_sync_cpu_ns :: RtsTime
-- | The time elapsed during the post-mark pause phase of the concurrent
-- nonmoving GC.
, gcdetails_nonmoving_gc_sync_elapsed_ns :: RtsTime
} deriving ( Read -- ^ @since 4.10.0.0
, Show -- ^ @since 4.10.0.0
)
......@@ -192,6 +218,12 @@ getRTSStats = do
gc_elapsed_ns <- (# peek RTSStats, gc_elapsed_ns) p
cpu_ns <- (# peek RTSStats, cpu_ns) p
elapsed_ns <- (# peek RTSStats, elapsed_ns) p
nonmoving_gc_sync_cpu_ns <- (# peek RTSStats, nonmoving_gc_sync_cpu_ns) p
nonmoving_gc_sync_elapsed_ns <- (# peek RTSStats, nonmoving_gc_sync_elapsed_ns) p
nonmoving_gc_sync_max_elapsed_ns <- (# peek RTSStats, nonmoving_gc_sync_max_elapsed_ns) p
nonmoving_gc_cpu_ns <- (# peek RTSStats, nonmoving_gc_cpu_ns) p
nonmoving_gc_elapsed_ns <- (# peek RTSStats, nonmoving_gc_elapsed_ns) p
nonmoving_gc_max_elapsed_ns <- (# peek RTSStats, nonmoving_gc_max_elapsed_ns) p
let pgc = (# ptr RTSStats, gc) p
gc <- do
gcdetails_gen <- (# peek GCDetails, gen) pgc
......@@ -211,5 +243,7 @@ getRTSStats = do
gcdetails_sync_elapsed_ns <- (# peek GCDetails, sync_elapsed_ns) pgc
gcdetails_cpu_ns <- (# peek GCDetails, cpu_ns) pgc
gcdetails_elapsed_ns <- (# peek GCDetails, elapsed_ns) pgc
gcdetails_nonmoving_gc_sync_cpu_ns <- (# peek GCDetails, nonmoving_gc_sync_cpu_ns) pgc
gcdetails_nonmoving_gc_sync_elapsed_ns <- (# peek GCDetails, nonmoving_gc_sync_elapsed_ns) pgc
return GCDetails{..}
return RTSStats{..}
......@@ -1212,6 +1212,8 @@ heapCensusChain( Census *census, bdescr *bd )
}
}
// Time is process CPU time of beginning of current GC and is used as
// the mutator CPU time reported as the census timestamp.
void heapCensus (Time t)
{
uint32_t g, n;
......@@ -1219,7 +1221,7 @@ void heapCensus (Time t)
gen_workspace *ws;
census = &censuses[era];
census->time = mut_user_time_until(t);
census->time = TimeToSecondsDbl(t);
census->rtime = TimeToNS(stat_getElapsedTime());
......
This diff is collapsed.
......@@ -30,13 +30,21 @@ void stat_endInit(void);
void stat_startGCSync(struct gc_thread_ *_gct);
void stat_startGC(Capability *cap, struct gc_thread_ *_gct);
void stat_endGC (Capability *cap, struct gc_thread_ *_gct, W_ live,
W_ copied, W_ slop, uint32_t gen, uint32_t n_gc_threads,
void stat_startGCWorker (Capability *cap, struct gc_thread_ *_gct);
void stat_endGCWorker (Capability *cap, struct gc_thread_ *_gct);
void stat_endGC (Capability *cap, struct gc_thread_ *initiating_gct, W_ live,
W_ copied, W_ slop, uint32_t gen,
uint32_t n_gc_threads, struct gc_thread_ **gc_threads,
W_ par_max_copied, W_ par_balanced_copied,
W_ gc_spin_spin, W_ gc_spin_yield, W_ mut_spin_spin,
W_ mut_spin_yield, W_ any_work, W_ no_work,
W_ scav_find_work);
void stat_startNonmovingGcSync(void);
void stat_endNonmovingGcSync(void);
void stat_startNonmovingGc (void);
void stat_endNonmovingGc (void);
#if defined(PROFILING)
void stat_startRP(void);
void stat_endRP(uint32_t, int, double);
......
......@@ -209,6 +209,14 @@ GarbageCollect (uint32_t collect_gen,
gc_thread *saved_gct;
#endif
uint32_t g, n;
// The time we should report our heap census as occurring at, if necessary.
Time mut_time = 0;
if (do_heap_census) {
RTSStats stats;
getRTSStats(&stats);
mut_time = stats.mutator_cpu_ns;
}
// necessary if we stole a callee-saves register for gct:
#if defined(THREADED_RTS)
......@@ -846,7 +854,7 @@ GarbageCollect (uint32_t collect_gen,
if (do_heap_census) {
debugTrace(DEBUG_sched, "performing heap census");
RELEASE_SM_LOCK;
heapCensus(gct->gc_start_cpu);
heapCensus(mut_time);
ACQUIRE_SM_LOCK;
}
......@@ -927,9 +935,11 @@ GarbageCollect (uint32_t collect_gen,
#endif
// ok, GC over: tell the stats department what happened.
stat_endGCWorker(cap, gct);
stat_endGC(cap, gct, live_words, copied,
live_blocks * BLOCK_SIZE_W - live_words /* slop */,
N, n_gc_threads, par_max_copied, par_balanced_copied,
N, n_gc_threads, gc_threads,
par_max_copied, par_balanced_copied,
gc_spin_spin, gc_spin_yield, mut_spin_spin, mut_spin_yield,
any_work, no_work, scav_find_work);
......@@ -1209,6 +1219,7 @@ gcWorkerThread (Capability *cap)
SET_GCT(gc_threads[cap->no]);
gct->id = osThreadId();
stat_startGCWorker (cap, gct);
// Wait until we're told to wake up
RELEASE_SPIN_LOCK(&gct->mut_spin);
......@@ -1247,6 +1258,7 @@ gcWorkerThread (Capability *cap)
gct->wakeup = GC_THREAD_WAITING_TO_CONTINUE;
debugTrace(DEBUG_gc, "GC thread %d waiting to continue...",
gct->thread_index);
stat_endGCWorker (cap, gct);
ACQUIRE_SPIN_LOCK(&gct->mut_spin);
debugTrace(DEBUG_gc, "GC thread %d on my way...", gct->thread_index);
......
......@@ -185,9 +185,11 @@ typedef struct gc_thread_ {
W_ no_work;
W_ scav_find_work;
Time gc_start_cpu; // process CPU time
Time gc_sync_start_elapsed; // start of GC sync
Time gc_start_elapsed; // process elapsed time
Time gc_start_cpu; // thread CPU time
Time gc_end_cpu; // thread CPU time
Time gc_sync_start_elapsed; // start of GC sync
Time gc_start_elapsed; // process elapsed time
Time gc_end_elapsed; // process elapsed time
W_ gc_start_faults;
// -------------------
......
......@@ -17,6 +17,7 @@
#include "GCThread.h"
#include "GCTDecl.h"
#include "Schedule.h"
#include "Stats.h"
#include "NonMoving.h"
#include "NonMovingMark.h"
......@@ -947,6 +948,7 @@ static void nonmovingMark_(MarkQueue *mark_queue, StgWeak **dead_weaks, StgTSO *
{
ACQUIRE_LOCK(&nonmoving_collection_mutex);
debugTrace(DEBUG_nonmoving_gc, "Starting mark...");
stat_startNonmovingGc();
// Do concurrent marking; most of the heap will get marked here.
nonmovingMarkThreadsWeaks(mark_queue);
......@@ -1098,6 +1100,7 @@ finish:
// We are done...
mark_thread = 0;
stat_endNonmovingGc();
// Signal that the concurrent collection is finished, allowing the next
// non-moving collection to proceed
......
......@@ -21,6 +21,7 @@
#include "Printer.h"
#include "Schedule.h"
#include "Weak.h"
#include "Stats.h"
#include "STM.h"
#include "MarkWeak.h"
#include "sm/Storage.h"
......@@ -316,6 +317,7 @@ void nonmovingBeginFlush(Task *task)
debugTrace(DEBUG_nonmoving_gc, "Starting update remembered set flush...");
traceConcSyncBegin();
upd_rem_set_flush_count = 0;
stat_startNonmovingGcSync();
stopAllCapabilitiesWith(NULL, task, SYNC_FLUSH_UPD_REM_SET);
// XXX: We may have been given a capability via releaseCapability (i.e. a
......@@ -407,6 +409,7 @@ void nonmovingFinishFlush(Task *task)
debugTrace(DEBUG_nonmoving_gc, "Finished update remembered set flush...");
traceConcSyncEnd();
stat_endNonmovingGcSync();
releaseAllCapabilities(n_capabilities, NULL, task);
}
#endif
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment