perf lock contention: Handle error in a single place
Subsystems (2)
Namhyung Kim
maintainerPATCH 1/3 perf lock contention: Handle error in a single place
It has some duplicate codes to do the same job. Let's add a label and goto there to handle errors in a single place.
--- tools/perf/util/bpf_skel/lock_contention.bpf.c | 16 ++++------------ 1 file changed, 4 insertions(+), 12 deletions(-)
tools/perf/util/bpf_skel/lock_contention.bpf.c +4 -12
@@ -439,11 +439,8 @@ int contention_end(u64 *ctx) duration =bpf_ktime_get_ns () - pelem->timestamp;if ((__s64 )duration <0 ) { - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid);__sync_fetch_and_add (&time_fail,1 ); -return 0 ; + goto out; }switch (aggr_mode) {
@@ -477,11 +474,8 @@ int contention_end(u64 *ctx) data =bpf_map_lookup_elem (&lock_stat, &key);if (!data) {if (data_map_full) { - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid);__sync_fetch_and_add (&data_fail,1 ); -return 0 ; + goto out; }struct contention_data first = {
@@ -502,10 +496,7 @@ int contention_end(u64 *ctx) data_map_full =1 ;__sync_fetch_and_add (&data_fail,1 ); } - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid); -return 0 ; + goto out; }__sync_fetch_and_add (&data->total_time, duration);
@@ -517,6 +508,7 @@ int contention_end(u64 *ctx)if (data->min_time > duration) data->min_time = duration; +out: pelem->lock =0 ;if (need_delete)bpf_map_delete_elem (&tstamp, &pid);
Signature · 1 line
2.46.0.469.g59c65b2a67-goog
Namhyung Kim
maintainerPATCH 2/3 perf lock contention: Simplify spinlock check
The LCB_F_SPIN bit is used for spinlock, rwlock and optimistic spinning in mutex. In get_tstamp_elem() it needs to check spinlock and rwlock only. As mutex sets the LCB_F_MUTEX, it can check those two bits and reduce the number of operations.
--- tools/perf/util/bpf_skel/lock_contention.bpf.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
tools/perf/util/bpf_skel/lock_contention.bpf.c +1 -2
@@ -323,8 +323,7 @@ static inline struct tstamp_data *get_tstamp_elem(__u32 flags)struct tstamp_data *pelem;/* Use per-cpu array map for spinlock and rwlock */ -if (flags == (LCB_F_SPIN |LCB_F_READ ) || flags ==LCB_F_SPIN || - flags == (LCB_F_SPIN |LCB_F_WRITE )) { +if ((flags & (LCB_F_SPIN |LCB_F_MUTEX )) ==LCB_F_SPIN ) {__u32 idx =0 ; pelem =bpf_map_lookup_elem (&tstamp_cpu, &idx);
Signature · 1 line
2.46.0.469.g59c65b2a67-goog
Namhyung Kim
maintainerPATCH 3/3 perf lock contention: Do not fail EEXIST for update
In reply to Namhyung KimWhen it updates the lock stat for the first time, it needs to create an element in the BPF hash map. But if there's a concurrent thread waiting for the same lock (like for rwsem or rwlock), it might race with the thread and possibly failed to update with -EEXIST. In that case, it can lookup the map again and put the data there instead of failing.
--- tools/perf/util/bpf_skel/lock_contention.bpf.c | 7 +++++++ 1 file changed, 7 insertions(+)
tools/perf/util/bpf_skel/lock_contention.bpf.c +7 -0
@@ -491,6 +491,12 @@ int contention_end(u64 *ctx) err =bpf_map_update_elem (&lock_stat, &key, &first,BPF_NOEXIST );if (err <0 ) { +if (err == -EEXIST ) { +/* it lost the race, try to get it again */ + data =bpf_map_lookup_elem (&lock_stat, &key); +if (data !=NULL ) + goto found; + }if (err == -E2BIG ) data_map_full =1 ;__sync_fetch_and_add (&data_fail,1 );
@@ -498,6 +504,7 @@ int contention_end(u64 *ctx) goto out; } +found:__sync_fetch_and_add (&data->total_time, duration);__sync_fetch_and_add (&data->count,1 );
Signature · 1 line
2.46.0.469.g59c65b2a67-goog
Quoted 2 lines · Namhyung Kim
It has some duplicate codes to do the same job. Let's add a label and goto there to handle errors in a single place.
Thanks, applied to perf-tools-next,
- Arnaldo
Quoted 57 lines
Signed-off-by Namhyung Kim <[email protected]>--- tools/perf/util/bpf_skel/lock_contention.bpf.c | 16 ++++------------ 1 file changed, 4 insertions(+), 12 deletions(-)tools/perf/util/bpf_skel/lock_contention.bpf.c +4
-12@@ -439,11 +439,8 @@ int contention_end(u64 *ctx) duration =bpf_ktime_get_ns () - pelem->timestamp;if ((__s64 )duration <0 ) { - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid);__sync_fetch_and_add (&time_fail,1 ); -return 0 ; + goto out; }switch (aggr_mode) {@@ -477,11 +474,8 @@ int contention_end(u64 *ctx) data =bpf_map_lookup_elem (&lock_stat, &key);if (!data) {if (data_map_full) { - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid);__sync_fetch_and_add (&data_fail,1 ); -return 0 ; + goto out; }struct contention_data first = {@@ -502,10 +496,7 @@ int contention_end(u64 *ctx) data_map_full =1 ;__sync_fetch_and_add (&data_fail,1 ); } - pelem->lock =0 ; -if (need_delete) -bpf_map_delete_elem (&tstamp, &pid); -return 0 ; + goto out; }__sync_fetch_and_add (&data->total_time, duration);@@ -517,6 +508,7 @@ int contention_end(u64 *ctx)if (data->min_time > duration) data->min_time = duration; +out: pelem->lock =0 ;if (need_delete)bpf_map_delete_elem (&tstamp, &pid);Signature · 1 line
2.46.0.469.g59c65b2a67-goog