bpf: Add lookup_and_delete_elem for BPF_MAP_STACK_TRACE
Subsystems (5)
PATCH v4 1/2 bpf: Add lookup_and_delete_elem for BPF_MAP_STACK_TRACE
The stacktrace map can be easily full, which will lead to failure in obtaining the stack. In addition to increasing the size of the map, another solution is to delete the stack_id after looking it up from the user, so extend the existing bpf_map_lookup_and_delete_elem() functionality to stacktrace map types.
--- include/linux/bpf.h | 2 +- kernel/bpf/stackmap.c | 16 ++++++++++++++-- kernel/bpf/syscall.c | 8 +++++--- 3 files changed, 20 insertions(+), 6 deletions(-)
Change list:
v1 -> v2:
- typo s/detele/delete/.(Jiri)
- make sure following lookup fails after deleting the stack_id with NOENT.(Jiri)
- use '&key' directly as the update value.(Jiri)
v1: https://lore.kernel.org/bpf/[email protected]
v2 -> v3:
- rename bpf_stackmap_copy_and_delete with bpf_stackmap_extract.(Andrii)
v2: https://lore.kernel.org/bpf/[email protected]
v3 -> v4:
- describe what the test is for in commit message.(Leon)
- 'val_buf' should stay on a single line.(Leon)
v3: https://lore.kernel.org/bpf/[email protected]
include/linux/bpf.h +1 -1
@@ -2703,7 +2703,7 @@ int bpf_percpu_hash_update(struct bpf_map *map, void *key, void *value,int bpf_percpu_array_update (struct bpf_map *map,void *key,void *value,u64 flags); -int bpf_stackmap_copy (struct bpf_map *map,void *key,void *value); +int bpf_stackmap_extract (struct bpf_map *map,void *key,void *value,bool delete);int bpf_fd_array_map_update_elem (struct bpf_map *map,struct file *map_file,void *key,void *value,u64 map_flags);
kernel/bpf/stackmap.c +14 -2
@@ -646,7 +646,15 @@ static void *stack_map_lookup_elem(struct bpf_map *map, void *key) }/* Called from syscall */ -int bpf_stackmap_copy (struct bpf_map *map,void *key,void *value) +static int stack_map_lookup_and_delete_elem (struct bpf_map *map,void *key, +void *value,u64 flags) +{ +return bpf_stackmap_extract (map, key, value, true); +} + +/* Called from syscall */ +int bpf_stackmap_extract (struct bpf_map *map,void *key,void *value, +bool delete) { struct bpf_stack_map *smap = container_of(map, struct bpf_stack_map, map);struct stack_map_bucket *bucket, *old_bucket;
@@ -663,7 +671,10 @@ int bpf_stackmap_copy(struct bpf_map *map, void *key, void *value)memcpy (value, bucket->data, trace_len);memset (value + trace_len,0 , map->value_size - trace_len); - old_bucket =xchg (&smap->buckets[id], bucket); +if (delete) + old_bucket = bucket; +else + old_bucket =xchg (&smap->buckets[id], bucket);if (old_bucket)pcpu_freelist_push (&smap->freelist, &old_bucket->fnode);return 0 ;
@@ -754,6 +765,7 @@ const struct bpf_map_ops stack_trace_map_ops = { .map_free = stack_map_free, .map_get_next_key = stack_map_get_next_key, .map_lookup_elem = stack_map_lookup_elem, + .map_lookup_and_delete_elem = stack_map_lookup_and_delete_elem, .map_update_elem = stack_map_update_elem, .map_delete_elem = stack_map_delete_elem, .map_check_btf = map_check_no_btf,
kernel/bpf/syscall.c +5 -3
@@ -318,7 +318,7 @@ static int bpf_map_copy_value(struct bpf_map *map, void *key, void *value, }else if (map->map_type ==BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE ) { err =bpf_percpu_cgroup_storage_copy (map, key, value); }else if (map->map_type ==BPF_MAP_TYPE_STACK_TRACE ) { - err =bpf_stackmap_copy (map, key, value); + err =bpf_stackmap_extract (map, key, value, false); }else if (IS_FD_ARRAY (map) ||IS_FD_PROG_ARRAY (map)) { err =bpf_fd_array_map_lookup_elem (map, key, value); }else if (IS_FD_HASH (map)) {
@@ -1627,7 +1627,8 @@ struct bpf_map *bpf_map_inc_not_zero(struct bpf_map *map) }EXPORT_SYMBOL_GPL (bpf_map_inc_not_zero); -int __weak bpf_stackmap_copy (struct bpf_map *map,void *key,void *value) +int __weak bpf_stackmap_extract (struct bpf_map *map,void *key,void *value, +bool delete) {return -ENOTSUPP ; }
@@ -2158,7 +2159,8 @@ static int map_lookup_and_delete_elem(union bpf_attr *attr) }else if (map->map_type ==BPF_MAP_TYPE_HASH || map->map_type ==BPF_MAP_TYPE_PERCPU_HASH || map->map_type ==BPF_MAP_TYPE_LRU_HASH || - map->map_type ==BPF_MAP_TYPE_LRU_PERCPU_HASH ) { + map->map_type ==BPF_MAP_TYPE_LRU_PERCPU_HASH || + map->map_type ==BPF_MAP_TYPE_STACK_TRACE ) {if (!bpf_map_is_offloaded (map)) {bpf_disable_instrumentation ();rcu_read_lock ();
Signature · 1 line
2.48.1
PATCH v4 2/2 selftests/bpf: Add stacktrace map lookup_and_delete_elem test case
Add tests for stacktrace map lookup and delete:
1. use bpf_map_lookup_and_delete_elem to lookup and delete the target stack_id,
2. lookup the deleted stack_id again to double check.
--- .../selftests/bpf/prog_tests/stacktrace_map.c | 21 ++++++++++++++++++- .../selftests/bpf/progs/test_stacktrace_map.c | 8 +++++++ 2 files changed, 28 insertions(+), 1 deletion(-)
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +20 -1
@@ -3,7 +3,7 @@void test_stacktrace_map (void ) { -int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd; +int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd, stack_key_map_fd;const char *prog_name ="oncpu" ;int err, prog_fd, stack_trace_len;const char *file ="./test_stacktrace_map.bpf.o" ;
@@ -11,6 +11,8 @@ void test_stacktrace_map(void)struct bpf_program *prog;struct bpf_object *obj;struct bpf_link *link; +__u32 stack_id; +char val_buf[PERF_MAX_STACK_DEPTH *sizeof (struct bpf_stack_build_id )]; err =bpf_prog_test_load (file,BPF_PROG_TYPE_TRACEPOINT , &obj, &prog_fd);if (CHECK (err,"prog_load" ,"err %d errno %d\n" , err, errno))
@@ -41,6 +43,10 @@ void test_stacktrace_map(void)if (CHECK_FAIL (stack_amap_fd <0 )) goto disable_pmu; + stack_key_map_fd =bpf_find_map (__func__, obj,"stack_key_map" ); +if (CHECK_FAIL (stack_key_map_fd <0 )) + goto disable_pmu; +/* give some time for bpf program run */ sleep (1 );
@@ -68,6 +74,19 @@ void test_stacktrace_map(void) "err %d errno %d\n", err, errno)) goto disable_pmu; + err =bpf_map_lookup_elem (stack_key_map_fd, &key, &stack_id); +if (CHECK (err,"stack_key_map lookup" ,"err %d errno %d\n" , err, errno)) + goto disable_pmu; + + err =bpf_map_lookup_and_delete_elem (stackmap_fd, &stack_id, &val_buf); +if (CHECK (err,"stackmap lookup and delete" , +"err %d errno %d\n" , err, errno)) + goto disable_pmu; + + err =bpf_map_lookup_elem (stackmap_fd, &stack_id, &val_buf); +CHECK ((!err || errno !=ENOENT ),"stackmap lookup deleted stack_id" , +"err %d errno %d\n" , err, errno); + disable_pmu:bpf_link__destroy (link); close_prog:
tools/testing/selftests/bpf/progs/test_stacktrace_map.c +8 -0
@@ -38,6 +38,13 @@ struct {__type (value, stack_trace_t); } stack_amap SEC(".maps"); +struct { + __uint(type, BPF_MAP_TYPE_ARRAY); + __uint(max_entries, 1); + __type(key, __u32); + __type(value, __u32); +} stack_key_map SEC(".maps"); +/* taken from /sys/kernel/tracing/events/sched/sched_switch/format */ struct sched_switch_args {unsigned long long pad;
@@ -64,6 +71,7 @@ int oncpu(struct sched_switch_args *ctx)/* The size of stackmap and stackid_hmap should be the same */ key =bpf_get_stackid (ctx, &stackmap,0 );if ((int )key >=0 ) { +bpf_map_update_elem (&stack_key_map, &val, &key,0 );bpf_map_update_elem (&stackid_hmap, &key, &val,0 ); stack_p =bpf_map_lookup_elem (&stack_amap, &key);if (stack_p)
Signature · 1 line
2.48.1
Re: [PATCH bpf-next v4 2/2] selftests/bpf: Add stacktrace map lookup_and_delete_elem test case
Quoted 35 lines · Tao Chen
Add tests for stacktrace map lookup and delete:
1. usebpf_map_lookup_and_delete_elemto lookup and delete the targetstack_id,
2. lookup the deletedstack_idagain to double check.Signed-off-by Tao Chen <[email protected]>--- .../selftests/bpf/prog_tests/stacktrace_map.c | 21 ++++++++++++++++++- .../selftests/bpf/progs/test_stacktrace_map.c | 8 +++++++ 2 files changed, 28 insertions(+), 1 deletion(-)tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +3
-1@@ -3,7 +3,7 @@void test_stacktrace_map (void ) { -int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd; +int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd, stack_key_map_fd;const char *prog_name ="oncpu" ;int err, prog_fd, stack_trace_len;const char *file ="./test_stacktrace_map.bpf.o" ;@@ -11,6 +11,8 @@ void test_stacktrace_map(void)struct bpf_program *prog;struct bpf_object *obj;struct bpf_link *link; +__u32 stack_id; +char val_buf[PERF_MAX_STACK_DEPTH *sizeof (struct bpf_stack_build_id )]; err =bpf_prog_test_load (file,BPF_PROG_TYPE_TRACEPOINT , &obj, &prog_fd);if (CHECK (err,"prog_load" ,"err %d errno %d\n" , err, errno))
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c void test_stacktrace_map(void) · line 47
Quoted 5 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +2
-0@@ -41,6 +43,10 @@ void test_stacktrace_map(void)if (CHECK_FAIL (stack_amap_fd <0 )) goto disable_pmu; + stack_key_map_fd =bpf_find_map (__func__, obj,"stack_key_map" ); +if (CHECK_FAIL (stack_key_map_fd <0 ))
please don't use CHECK*() macros, they are superseded by more targeted ASSERT_xxx() ones
pw-bot: cr
Quoted 11 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +4
-0+ goto disable_pmu; +/* give some time for bpf program run */ sleep (1 );@@ -68,6 +74,19 @@ void test_stacktrace_map(void) "err %d errno %d\n", err, errno)) goto disable_pmu; + err =bpf_map_lookup_elem (stack_key_map_fd, &key, &stack_id); +if (CHECK (err,"stack_key_map lookup" ,"err %d errno %d\n" , err, errno))
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c void test_stacktrace_map(void) · line 88
Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +10
-0+ goto disable_pmu; + + err =bpf_map_lookup_and_delete_elem (stackmap_fd, &stack_id, &val_buf); +if (CHECK (err,"stackmap lookup and delete" , +"err %d errno %d\n" , err, errno)) + goto disable_pmu; + + err =bpf_map_lookup_elem (stackmap_fd, &stack_id, &val_buf); +CHECK ((!err || errno !=ENOENT ),"stackmap lookup deleted stack_id" , +"err %d errno %d\n" , err, errno);
bpf_map_lookup_elem() returns error code directly, no need to use errno, just check that err == -ENOENT
Quoted 23 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +1
-0+ disable_pmu:bpf_link__destroy (link); close_prog:tools/testing/selftests/bpf/progs/test_stacktrace_map.c +7
-0@@ -38,6 +38,13 @@ struct {__type (value, stack_trace_t); } stack_amap SEC(".maps"); +struct { + __uint(type, BPF_MAP_TYPE_ARRAY); + __uint(max_entries, 1); + __type(key, __u32); + __type(value, __u32); +} stack_key_map SEC(".maps"); +/* taken from /sys/kernel/tracing/events/sched/sched_switch/format */ struct sched_switch_args {unsigned long long pad;
tools/testing/selftests/bpf/progs/test_stacktrace_map.c int oncpu(struct sched_switch_args *ctx) · line 74
Quoted 4 lines
tools/testing/selftests/bpf/progs/test_stacktrace_map.c +1
-0@@ -64,6 +71,7 @@ int oncpu(struct sched_switch_args *ctx)/* The size of stackmap and stackid_hmap should be the same */ key =bpf_get_stackid (ctx, &stackmap,0 );if ((int )key >=0 ) { +bpf_map_update_elem (&stack_key_map, &val, &key,0 );
ugh... you'd just use a global variable if this test was used through skeleton... maybe convert the test to skeleton and get rid of all those unnecessary bpf_find_map() calls as well?
Quoted 6 lines
tools/testing/selftests/bpf/progs/test_stacktrace_map.c +0
-1bpf_map_update_elem (&stackid_hmap, &key, &val,0 ); stack_p =bpf_map_lookup_elem (&stack_amap, &key);if (stack_p) --2.48.1
Re: [PATCH bpf-next v4 2/2] selftests/bpf: Add stacktrace map lookup_and_delete_elem test case
Quoted 39 lines · Andrii Nakryiko
Quoted 38 lines · Tao Chen
Add tests for stacktrace map lookup and delete:
1. usebpf_map_lookup_and_delete_elemto lookup and delete the targetstack_id,
2. lookup the deletedstack_idagain to double check.Signed-off-by Tao Chen <[email protected]>--- .../selftests/bpf/prog_tests/stacktrace_map.c | 21 ++++++++++++++++++- .../selftests/bpf/progs/test_stacktrace_map.c | 8 +++++++ 2 files changed, 28 insertions(+), 1 deletion(-)tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +3
-1@@ -3,7 +3,7 @@void test_stacktrace_map (void ) { -int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd; +int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd, stack_key_map_fd;const char *prog_name ="oncpu" ;int err, prog_fd, stack_trace_len;const char *file ="./test_stacktrace_map.bpf.o" ;@@ -11,6 +11,8 @@ void test_stacktrace_map(void)struct bpf_program *prog;struct bpf_object *obj;struct bpf_link *link; +__u32 stack_id; +char val_buf[PERF_MAX_STACK_DEPTH *sizeof (struct bpf_stack_build_id )]; err =bpf_prog_test_load (file,BPF_PROG_TYPE_TRACEPOINT , &obj, &prog_fd);if (CHECK (err,"prog_load" ,"err %d errno %d\n" , err, errno))@@ -41,6 +43,10 @@ void test_stacktrace_map(void)if (CHECK_FAIL (stack_amap_fd <0 )) goto disable_pmu;
Quoted 6 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c void test_stacktrace_map(void) · line 47
Quoted 2 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +2
-0+ stack_key_map_fd =bpf_find_map (__func__, obj,"stack_key_map" ); +if (CHECK_FAIL (stack_key_map_fd <0 ))please don't use CHECK*() macros, they are superseded by more targeted
ASSERT_xxx()ones
got it, thanks.
Quoted 12 lines
pw-bot: cr
Quoted 9 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +2
-0+ goto disable_pmu; +/* give some time for bpf program run */ sleep (1 );@@ -68,6 +74,19 @@ void test_stacktrace_map(void) "err %d errno %d\n", err, errno)) goto disable_pmu;
Quoted 16 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c void test_stacktrace_map(void) · line 88
Quoted 12 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +12
-0+ err =bpf_map_lookup_elem (stack_key_map_fd, &key, &stack_id); +if (CHECK (err,"stack_key_map lookup" ,"err %d errno %d\n" , err, errno)) + goto disable_pmu; + + err =bpf_map_lookup_and_delete_elem (stackmap_fd, &stack_id, &val_buf); +if (CHECK (err,"stackmap lookup and delete" , +"err %d errno %d\n" , err, errno)) + goto disable_pmu; + + err =bpf_map_lookup_elem (stackmap_fd, &stack_id, &val_buf); +CHECK ((!err || errno !=ENOENT ),"stackmap lookup deleted stack_id" , +"err %d errno %d\n" , err, errno);
bpf_map_lookup_elem()returns error code directly, no need to use errno, just check that err ==-ENOENT
ok.
Quoted 23 lines
Quoted 23 lines
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +1
-0+ disable_pmu:bpf_link__destroy (link); close_prog:tools/testing/selftests/bpf/progs/test_stacktrace_map.c +7
-0@@ -38,6 +38,13 @@ struct {__type (value, stack_trace_t); } stack_amap SEC(".maps"); +struct { + __uint(type, BPF_MAP_TYPE_ARRAY); + __uint(max_entries, 1); + __type(key, __u32); + __type(value, __u32); +} stack_key_map SEC(".maps"); +/* taken from /sys/kernel/tracing/events/sched/sched_switch/format */ struct sched_switch_args {unsigned long long pad;@@ -64,6 +71,7 @@ int oncpu(struct sched_switch_args *ctx)
Quoted 9 lines
tools/testing/selftests/bpf/progs/test_stacktrace_map.c int oncpu(struct sched_switch_args *ctx) · line 74
Quoted 4 lines
tools/testing/selftests/bpf/progs/test_stacktrace_map.c +1
-0/* The size of stackmap and stackid_hmap should be the same */ key =bpf_get_stackid (ctx, &stackmap,0 );if ((int )key >=0 ) { +bpf_map_update_elem (&stack_key_map, &val, &key,0 );ugh... you'd just use a global variable if this test was used through skeleton... maybe convert the test to skeleton and get rid of all those unnecessary
bpf_find_map()calls as well?
Yes, the test case seems too old, i will refactor it with skeleton.
Quoted 6 lines
Quoted 6 lines
tools/testing/selftests/bpf/progs/test_stacktrace_map.c +0
-1bpf_map_update_elem (&stackid_hmap, &key, &val,0 ); stack_p =bpf_map_lookup_elem (&stack_amap, &key);if (stack_p) --2.48.1
Signature · 2 lines
Best Regards Tao Chen