Skip to contentSkip to conversation

bpf: Add lookup_and_delete_elem for BPF_MAP_STACK_TRACE

PATCH Superseded Tao Chen 2 participants 5 messages Sep 23, 2025 – Sep 25, 2025

Subsystems (5)
Kselftest BPF BPF selftests BPF core +1 more subsystems

v5 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Tao Chen

author
#0 lore raw

PATCH v5 1/3 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]
v4 -> v5:
- refactor stacktrace_map with skeleton in patch2.(Andrii)
- replace CHECK*() with ASSERT_xx() and just check err from bpf_map_lookup_elem only.(Andrii)
v4: https://lore.kernel.org/bpf/[email protected]

include/linux/bpf.h +1 -1
index 8f6e87f0f3a..9d6f7671ba1 100644
@@ -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
index 3615c06b7df..2e182a3ac4c 100644
@@ -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
index 0fbfa8532c3..c380a2576fc 100644
@@ -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
author
#1 lore raw

PATCH v5 2/3 selftests/bpf: Refactor stacktrace_map case with skeleton

The loading method of the stacktrace_map test case looks too outdated, refactor it with skeleton and replace control_map with control global variable.

---
 .../selftests/bpf/prog_tests/stacktrace_map.c | 60 +++++++------------
 ...test_stacktrace_map.c => stacktrace_map.c} | 15 ++---
 2 files changed, 26 insertions(+), 49 deletions(-)
 rename tools/testing/selftests/bpf/progs/{test_stacktrace_map.c => stacktrace_map.c} (83%)
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +22 -38
index 84a7e405e91..26a2bd25a6f 100644
@@ -1,53 +1,39 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <test_progs.h>
+#include "stacktrace_map.skel.h"
 
 void test_stacktrace_map(void)
 {
-	int control_map_fd, stackid_hmap_fd, stackmap_fd, stack_amap_fd;
-	const char *prog_name = "oncpu";
-	int err, prog_fd, stack_trace_len;
-	const char *file = "./test_stacktrace_map.bpf.o";
-	__u32 key, val, duration = 0;
-	struct bpf_program *prog;
-	struct bpf_object *obj;
-	struct bpf_link *link;
+	struct stacktrace_map *skel;
+	int stackid_hmap_fd, stackmap_fd, stack_amap_fd;
+	int err, stack_trace_len;
+	__u32 duration = 0;
 
-	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))
+	skel = stacktrace_map__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
 		return;
 
-	prog = bpf_object__find_program_by_name(obj, prog_name);
-	if (CHECK(!prog, "find_prog", "prog '%s' not found\n", prog_name))
-		goto close_prog;
-
-	link = bpf_program__attach_tracepoint(prog, "sched", "sched_switch");
-	if (!ASSERT_OK_PTR(link, "attach_tp"))
-		goto close_prog;
-
 	/* find map fds */
-	control_map_fd = bpf_find_map(__func__, obj, "control_map");
-	if (CHECK_FAIL(control_map_fd < 0))
-		goto disable_pmu;
-
-	stackid_hmap_fd = bpf_find_map(__func__, obj, "stackid_hmap");
+	stackid_hmap_fd = bpf_map__fd(skel->maps.stackid_hmap);
 	if (CHECK_FAIL(stackid_hmap_fd < 0))
-		goto disable_pmu;
+		goto out;
 
-	stackmap_fd = bpf_find_map(__func__, obj, "stackmap");
+	stackmap_fd = bpf_map__fd(skel->maps.stackmap);
 	if (CHECK_FAIL(stackmap_fd < 0))
-		goto disable_pmu;
+		goto out;
 
-	stack_amap_fd = bpf_find_map(__func__, obj, "stack_amap");
+	stack_amap_fd = bpf_map__fd(skel->maps.stack_amap);
 	if (CHECK_FAIL(stack_amap_fd < 0))
-		goto disable_pmu;
+		goto out;
 
+	err = stacktrace_map__attach(skel);
+	if (!ASSERT_OK(err, "skel_attach"))
+		goto out;
 	/* give some time for bpf program run */
 	sleep(1);
 
 	/* disable stack trace collection */
-	key = 0;
-	val = 1;
-	bpf_map_update_elem(control_map_fd, &key, &val, 0);
+	skel->bss->control = 1;
 
 	/* for every element in stackid_hmap, we can find a corresponding one
 	 * in stackmap, and vice versa.
@@ -55,21 +41,19 @@ void test_stacktrace_map(void)
 	err = compare_map_keys(stackid_hmap_fd, stackmap_fd);
 	if (CHECK(err, "compare_map_keys stackid_hmap vs. stackmap",
 		  "err %d errno %d\n", err, errno))
-		goto disable_pmu;
+		goto out;
 
 	err = compare_map_keys(stackmap_fd, stackid_hmap_fd);
 	if (CHECK(err, "compare_map_keys stackmap vs. stackid_hmap",
 		  "err %d errno %d\n", err, errno))
-		goto disable_pmu;
+		goto out;
 
 	stack_trace_len = PERF_MAX_STACK_DEPTH * sizeof(__u64);
 	err = compare_stack_ips(stackmap_fd, stack_amap_fd, stack_trace_len);
 	if (CHECK(err, "compare_stack_ips stackmap vs. stack_amap",
 		  "err %d errno %d\n", err, errno))
-		goto disable_pmu;
+		goto out;
 
-disable_pmu:
-	bpf_link__destroy(link);
-close_prog:
-	bpf_object__close(obj);
+out:
+	stacktrace_map__destroy(skel);
 }
tools/testing/selftests/bpf/progs/stacktrace_map.c renamed from tools/testing/selftests/bpf/progs/test_stacktrace_map.c +4 -11
similarity index 83%
rename from tools/testing/selftests/bpf/progs/test_stacktrace_map.c
rename to tools/testing/selftests/bpf/progs/stacktrace_map.c
index 47568007b66..9090d561312 100644
@@ -8,13 +8,6 @@
 #define PERF_MAX_STACK_DEPTH         127
 #endif
 
-struct {
-	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, 1);
-	__type(key, __u32);
-	__type(value, __u32);
-} control_map SEC(".maps");
-
 struct {
 	__uint(type, BPF_MAP_TYPE_HASH);
 	__uint(max_entries, 16384);
@@ -50,16 +43,16 @@ struct sched_switch_args {
 	int next_prio;
 };
 
+int control = 0;
 SEC("tracepoint/sched/sched_switch")
 int oncpu(struct sched_switch_args *ctx)
 {
 	__u32 max_len = PERF_MAX_STACK_DEPTH * sizeof(__u64);
-	__u32 key = 0, val = 0, *value_p;
+	__u32 key = 0, val = 0;
 	void *stack_p;
 
-	value_p = bpf_map_lookup_elem(&control_map, &key);
-	if (value_p && *value_p)
-		return 0; /* skip if non-zero *value_p */
+	if (control)
+		return 0;
 
 	/* The size of stackmap and stackid_hmap should be the same */
 	key = bpf_get_stackid(ctx, &stackmap, 0);
Signature · 1 line
2.48.1
author
#2 lore raw

PATCH v5 3/3 selftests/bpf: Add stacktrace map lookup_and_delete_elem test case

In reply to Tao Chen

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.

---
 .../testing/selftests/bpf/prog_tests/stacktrace_map.c | 11 ++++++++++-
 tools/testing/selftests/bpf/progs/stacktrace_map.c    |  2 ++
 2 files changed, 12 insertions(+), 1 deletion(-)
tools/testing/selftests/bpf/prog_tests/stacktrace_map.c +10 -1
index 26a2bd25a6f..f06bfef0bc8 100644
@@ -7,7 +7,8 @@ void test_stacktrace_map(void)
 	struct stacktrace_map *skel;
 	int stackid_hmap_fd, stackmap_fd, stack_amap_fd;
 	int err, stack_trace_len;
-	__u32 duration = 0;
+	__u32 stack_id, duration = 0;
+	__u64 val[PERF_MAX_STACK_DEPTH];
 
 	skel = stacktrace_map__open_and_load();
 	if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
@@ -54,6 +55,14 @@ void test_stacktrace_map(void)
 		  "err %d errno %d\n", err, errno))
 		goto out;
 
+	stack_id = skel->bss->stack_id;
+	err = bpf_map_lookup_and_delete_elem(stackmap_fd, &stack_id,  val);
+	if (!ASSERT_OK(err, "lookup and delete target stack_id"))
+		goto out;
+
+	err = bpf_map_lookup_elem(stackmap_fd, &stack_id, val);
+	if (!ASSERT_EQ(err, -ENOENT, "lookup deleted stack_id"))
+		goto out;
 out:
 	stacktrace_map__destroy(skel);
 }
tools/testing/selftests/bpf/progs/stacktrace_map.c +2 -0
index 9090d561312..7abf702a6ce 100644
@@ -44,6 +44,7 @@ struct sched_switch_args {
 };
 
 int control = 0;
+__u32 stack_id;
 SEC("tracepoint/sched/sched_switch")
 int oncpu(struct sched_switch_args *ctx)
 {
@@ -57,6 +58,7 @@ 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) {
+		stack_id = key;
 		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
maintainer
#3 lore raw

Re: [PATCH bpf-next v5 3/3] selftests/bpf: Add stacktrace map lookup_and_delete_elem test case

Quoted 10 lines · Tao Chen

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.

---
  .../testing/selftests/bpf/prog_tests/stacktrace_map.c | 11 ++++++++++-
  tools/testing/selftests/bpf/progs/stacktrace_map.c    |  2 ++
  2 files changed, 12 insertions(+), 1 deletion(-)

BPF CI fails with:

Notice: Success: 644/5645, Skipped: 82, Failed: 1
Error: #402 stacktrace_map_raw_tp
Error: #402 stacktrace_map_raw_tp
libbpf: elf: failed to open ./test_stacktrace_map.bpf.o: -ENOENT test_stacktrace_map_raw_tp:FAIL:prog_load raw tp err -2 errno 2
Test Results:
bpftool: PASS
test_progs: FAIL (returned 1)
Error: Process completed with exit code 1.

https://github.com/kernel-patches/bpf/actions/runs/17993777572

author
#4 lore raw

Re: [PATCH bpf-next v5 3/3] selftests/bpf: Add stacktrace map lookup_and_delete_elem test case

Quoted 23 lines · Daniel Borkmann
Quoted 10 lines · Tao Chen

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.

---
  .../testing/selftests/bpf/prog_tests/stacktrace_map.c | 11 ++++++++++- tools/testing/selftests/bpf/progs/stacktrace_map.c    |  2 ++

  2 files changed, 12 insertions(+), 1 deletion(-)

BPF CI fails with:

Notice: Success: 644/5645, Skipped: 82, Failed: 1
Error: #402 stacktrace_map_raw_tp
  Error: #402 stacktrace_map_raw_tp
  libbpf: elf: failed to open ./test_stacktrace_map.bpf.o: -ENOENT test_stacktrace_map_raw_tp:FAIL:prog_load raw tp err -2 errno 2
Test Results:
             bpftool: PASS
          test_progs: FAIL (returned 1)
Error: Process completed with exit code 1.

Quoted 1 line

https://github.com/kernel-patches/bpf/actions/runs/17993777572

Hi Daniel,

It seems that stacktrace_map_raw_tp will use test_stacktrace_map.bpf.c, which is renamed to stacktrace_map.c in my patch, i will fix it, thanks.

tools/testing/selftests/bpf/prog_tests/stacktrace_map_raw_tp.c:8: const char *file = "./test_stacktrace_map.bpf.o";

Signature · 2 lines
Best Regards
Tao Chen