* [PATCHv7 bpf-next 09/29] bpf: Factor fsession link to use struct bpf_tramp_node
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Now that we split trampoline attachment object (bpf_tramp_node) from
the link object (bpf_tramp_link) we can use bpf_tramp_node as fsession's
fexit attachment object and get rid of the bpf_fsession_link object.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf.h | 6 +-----
kernel/bpf/syscall.c | 21 ++++++---------------
kernel/bpf/trampoline.c | 12 ++++++------
3 files changed, 13 insertions(+), 26 deletions(-)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 27d13d6c14be..4764b4aa7081 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1908,15 +1908,11 @@ struct bpf_shim_tramp_link {
struct bpf_tracing_link {
struct bpf_tramp_link link;
+ struct bpf_tramp_node fexit;
struct bpf_trampoline *trampoline;
struct bpf_prog *tgt_prog;
};
-struct bpf_fsession_link {
- struct bpf_tracing_link link;
- struct bpf_tramp_link fexit;
-};
-
struct bpf_raw_tp_link {
struct bpf_link link;
struct bpf_raw_event_map *btp;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 7bb2271072e9..f308ebdab750 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3705,21 +3705,7 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
key = bpf_trampoline_compute_key(tgt_prog, NULL, btf_id);
}
- if (prog->expected_attach_type == BPF_TRACE_FSESSION) {
- struct bpf_fsession_link *fslink;
-
- fslink = kzalloc_obj(*fslink, GFP_USER);
- if (fslink) {
- bpf_tramp_link_init(&fslink->fexit, BPF_LINK_TYPE_TRACING,
- &bpf_tracing_link_lops, prog, attach_type,
- bpf_cookie);
- link = &fslink->link;
- } else {
- link = NULL;
- }
- } else {
- link = kzalloc_obj(*link, GFP_USER);
- }
+ link = kzalloc_obj(*link, GFP_USER);
if (!link) {
err = -ENOMEM;
goto out_put_prog;
@@ -3727,6 +3713,11 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
bpf_tramp_link_init(&link->link, BPF_LINK_TYPE_TRACING,
&bpf_tracing_link_lops, prog, attach_type, bpf_cookie);
+ if (prog->expected_attach_type == BPF_TRACE_FSESSION) {
+ link->fexit.link = &link->link.link;
+ link->fexit.cookie = bpf_cookie;
+ }
+
mutex_lock(&prog->aux->dst_mutex);
/* There are a few possible cases here:
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 6a45c09fc0d8..5776d2b8e36e 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -823,7 +823,7 @@ static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
struct bpf_tramp_node *node,
int cnt)
{
- struct bpf_fsession_link *fslink = NULL;
+ struct bpf_tracing_link *tr_link = NULL;
enum bpf_tramp_prog_type kind;
struct bpf_tramp_node *node_existing;
struct hlist_head *prog_list;
@@ -850,8 +850,8 @@ static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
hlist_add_head(&node->tramp_hlist, prog_list);
if (kind == BPF_TRAMP_FSESSION) {
tr->progs_cnt[BPF_TRAMP_FENTRY]++;
- fslink = container_of(node, struct bpf_fsession_link, link.link.node);
- hlist_add_head(&fslink->fexit.node.tramp_hlist, &tr->progs_hlist[BPF_TRAMP_FEXIT]);
+ tr_link = container_of(node, struct bpf_tracing_link, link.node);
+ hlist_add_head(&tr_link->fexit.tramp_hlist, &tr->progs_hlist[BPF_TRAMP_FEXIT]);
tr->progs_cnt[BPF_TRAMP_FEXIT]++;
} else {
tr->progs_cnt[kind]++;
@@ -862,13 +862,13 @@ static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
static void bpf_trampoline_remove_prog(struct bpf_trampoline *tr,
struct bpf_tramp_node *node)
{
- struct bpf_fsession_link *fslink;
+ struct bpf_tracing_link *tr_link;
enum bpf_tramp_prog_type kind;
kind = bpf_attach_type_to_tramp(node->link->prog);
if (kind == BPF_TRAMP_FSESSION) {
- fslink = container_of(node, struct bpf_fsession_link, link.link.node);
- hlist_del_init(&fslink->fexit.node.tramp_hlist);
+ tr_link = container_of(node, struct bpf_tracing_link, link.node);
+ hlist_del_init(&tr_link->fexit.tramp_hlist);
tr->progs_cnt[BPF_TRAMP_FEXIT]--;
kind = BPF_TRAMP_FENTRY;
}
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 10/29] bpf: Add multi tracing attach types
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding new program attach types multi tracing attachment:
BPF_TRACE_FENTRY_MULTI
BPF_TRACE_FEXIT_MULTI
and their base support in verifier code.
Programs with such attach type will use specific link attachment
interface coming in following changes.
This was suggested by Andrii some (long) time ago and turned out
to be easier than having special program flag for that.
Bpf programs with such types have 'bpf_multi_func' function set as
their attach_btf_id and keep module reference when it's specified
by attach_prog_fd.
They are also accepted as sleepable programs during verification,
and the real validation for specific BTF_IDs/functions will happen
during the multi link attachment in following changes.
Suggested-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf.h | 5 +++++
include/linux/btf_ids.h | 1 +
include/uapi/linux/bpf.h | 2 ++
kernel/bpf/fixups.c | 1 +
kernel/bpf/syscall.c | 28 ++++++++++++++++++++----
kernel/bpf/trampoline.c | 5 ++++-
kernel/bpf/verifier.c | 40 +++++++++++++++++++++++++++++++++-
net/bpf/test_run.c | 2 ++
tools/include/uapi/linux/bpf.h | 2 ++
tools/lib/bpf/libbpf.c | 2 ++
10 files changed, 82 insertions(+), 6 deletions(-)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 4764b4aa7081..bc5a57a9faee 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -2094,6 +2094,11 @@ static inline void bpf_prog_put_recursion_context(struct bpf_prog *prog)
#endif
}
+static inline bool is_tracing_multi(enum bpf_attach_type type)
+{
+ return type == BPF_TRACE_FENTRY_MULTI || type == BPF_TRACE_FEXIT_MULTI;
+}
+
#if defined(CONFIG_BPF_JIT) && defined(CONFIG_BPF_SYSCALL)
/* This macro helps developer to register a struct_ops type and generate
* type information correctly. Developers should use this macro to register
diff --git a/include/linux/btf_ids.h b/include/linux/btf_ids.h
index af011db39ab3..8b5a9ee92513 100644
--- a/include/linux/btf_ids.h
+++ b/include/linux/btf_ids.h
@@ -284,5 +284,6 @@ extern u32 bpf_cgroup_btf_id[];
extern u32 bpf_local_storage_map_btf_id[];
extern u32 btf_bpf_map_id[];
extern u32 bpf_kmem_cache_btf_id[];
+extern u32 bpf_multi_func_btf_id[];
#endif
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index aec171ccb6ef..0cde0e2f3d4b 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -1155,6 +1155,8 @@ enum bpf_attach_type {
BPF_TRACE_KPROBE_SESSION,
BPF_TRACE_UPROBE_SESSION,
BPF_TRACE_FSESSION,
+ BPF_TRACE_FENTRY_MULTI,
+ BPF_TRACE_FEXIT_MULTI,
__MAX_BPF_ATTACH_TYPE
};
diff --git a/kernel/bpf/fixups.c b/kernel/bpf/fixups.c
index 5aa3f7d99ac9..0cf9735929f5 100644
--- a/kernel/bpf/fixups.c
+++ b/kernel/bpf/fixups.c
@@ -2186,6 +2186,7 @@ int bpf_do_misc_fixups(struct bpf_verifier_env *env)
insn->imm == BPF_FUNC_get_func_ret) {
if (eatype == BPF_TRACE_FEXIT ||
eatype == BPF_TRACE_FSESSION ||
+ eatype == BPF_TRACE_FEXIT_MULTI ||
eatype == BPF_MODIFY_RETURN) {
/* Load nr_args from ctx - 8 */
insn_buf[0] = BPF_LDX_MEM(BPF_DW, BPF_REG_0, BPF_REG_1, -8);
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index f308ebdab750..f354e9f64e65 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -41,6 +41,7 @@
#include <linux/overflow.h>
#include <linux/cookie.h>
#include <linux/verification.h>
+#include <linux/btf_ids.h>
#include <net/netfilter/nf_bpf_link.h>
#include <net/netkit.h>
@@ -2713,7 +2714,8 @@ static int
bpf_prog_load_check_attach(enum bpf_prog_type prog_type,
enum bpf_attach_type expected_attach_type,
struct btf *attach_btf, u32 btf_id,
- struct bpf_prog *dst_prog)
+ struct bpf_prog *dst_prog,
+ bool multi_func)
{
if (btf_id) {
if (btf_id > BTF_MAX_TYPE)
@@ -2733,6 +2735,14 @@ bpf_prog_load_check_attach(enum bpf_prog_type prog_type,
}
}
+ if (multi_func) {
+ if (prog_type != BPF_PROG_TYPE_TRACING)
+ return -EINVAL;
+ if (!attach_btf || btf_id)
+ return -EINVAL;
+ return 0;
+ }
+
if (attach_btf && (!btf_id || dst_prog))
return -EINVAL;
@@ -2925,6 +2935,11 @@ static int bpf_prog_mark_insn_arrays_ready(struct bpf_prog *prog)
return 0;
}
+extern int bpf_multi_func(void);
+int __init __used bpf_multi_func(void) { return 0; }
+
+BTF_ID_LIST_GLOBAL_SINGLE(bpf_multi_func_btf_id, func, bpf_multi_func)
+
/* last field in 'union bpf_attr' used by this command */
#define BPF_PROG_LOAD_LAST_FIELD keyring_id
@@ -2937,6 +2952,7 @@ static int bpf_prog_load(union bpf_attr *attr, bpfptr_t uattr, struct bpf_log_at
bool bpf_cap;
int err;
char license[128];
+ bool multi_func;
if (CHECK_ATTR(BPF_PROG_LOAD))
return -EINVAL;
@@ -3003,6 +3019,8 @@ static int bpf_prog_load(union bpf_attr *attr, bpfptr_t uattr, struct bpf_log_at
if (is_perfmon_prog_type(type) && !bpf_token_capable(token, CAP_PERFMON))
goto put_token;
+ multi_func = is_tracing_multi(attr->expected_attach_type);
+
/* attach_prog_fd/attach_btf_obj_fd can specify fd of either bpf_prog
* or btf, we need to check which one it is
*/
@@ -3024,7 +3042,7 @@ static int bpf_prog_load(union bpf_attr *attr, bpfptr_t uattr, struct bpf_log_at
goto put_token;
}
}
- } else if (attr->attach_btf_id) {
+ } else if (attr->attach_btf_id || multi_func) {
/* fall back to vmlinux BTF, if BTF type ID is specified */
attach_btf = bpf_get_btf_vmlinux();
if (IS_ERR(attach_btf)) {
@@ -3040,7 +3058,7 @@ static int bpf_prog_load(union bpf_attr *attr, bpfptr_t uattr, struct bpf_log_at
if (bpf_prog_load_check_attach(type, attr->expected_attach_type,
attach_btf, attr->attach_btf_id,
- dst_prog)) {
+ dst_prog, multi_func)) {
if (dst_prog)
bpf_prog_put(dst_prog);
if (attach_btf)
@@ -3063,7 +3081,7 @@ static int bpf_prog_load(union bpf_attr *attr, bpfptr_t uattr, struct bpf_log_at
prog->expected_attach_type = attr->expected_attach_type;
prog->sleepable = !!(attr->prog_flags & BPF_F_SLEEPABLE);
prog->aux->attach_btf = attach_btf;
- prog->aux->attach_btf_id = attr->attach_btf_id;
+ prog->aux->attach_btf_id = multi_func ? bpf_multi_func_btf_id[0] : attr->attach_btf_id;
prog->aux->dst_prog = dst_prog;
prog->aux->dev_bound = !!attr->prog_ifindex;
prog->aux->xdp_has_frags = attr->prog_flags & BPF_F_XDP_HAS_FRAGS;
@@ -4455,6 +4473,8 @@ attach_type_to_prog_type(enum bpf_attach_type attach_type)
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
case BPF_MODIFY_RETURN:
return BPF_PROG_TYPE_TRACING;
case BPF_LSM_MAC:
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 5776d2b8e36e..ae7e4fdfe2a3 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -182,7 +182,8 @@ bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
switch (ptype) {
case BPF_PROG_TYPE_TRACING:
if (eatype == BPF_TRACE_FENTRY || eatype == BPF_TRACE_FEXIT ||
- eatype == BPF_MODIFY_RETURN || eatype == BPF_TRACE_FSESSION)
+ eatype == BPF_MODIFY_RETURN || eatype == BPF_TRACE_FSESSION ||
+ eatype == BPF_TRACE_FENTRY_MULTI || eatype == BPF_TRACE_FEXIT_MULTI)
return true;
return false;
case BPF_PROG_TYPE_LSM:
@@ -781,10 +782,12 @@ static enum bpf_tramp_prog_type bpf_attach_type_to_tramp(struct bpf_prog *prog)
{
switch (prog->expected_attach_type) {
case BPF_TRACE_FENTRY:
+ case BPF_TRACE_FENTRY_MULTI:
return BPF_TRAMP_FENTRY;
case BPF_MODIFY_RETURN:
return BPF_TRAMP_MODIFY_RETURN;
case BPF_TRACE_FEXIT:
+ case BPF_TRACE_FEXIT_MULTI:
return BPF_TRAMP_FEXIT;
case BPF_TRACE_FSESSION:
return BPF_TRAMP_FSESSION;
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 8ed484cb1a8a..521ec2a9051e 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -16314,6 +16314,8 @@ static bool return_retval_range(struct bpf_verifier_env *env, struct bpf_retval_
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
*range = retval_range(0, 0);
break;
case BPF_TRACE_RAW_TP:
@@ -18699,6 +18701,11 @@ static int check_attach_modify_return(unsigned long addr, const char *func_name)
#endif /* CONFIG_FUNCTION_ERROR_INJECTION */
+static bool is_tracing_multi_id(const struct bpf_prog *prog, u32 btf_id)
+{
+ return is_tracing_multi(prog->expected_attach_type) && bpf_multi_func_btf_id[0] == btf_id;
+}
+
int bpf_check_attach_target(struct bpf_verifier_log *log,
const struct bpf_prog *prog,
const struct bpf_prog *tgt_prog,
@@ -18821,6 +18828,8 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
prog_extension &&
(tgt_prog->expected_attach_type == BPF_TRACE_FENTRY ||
tgt_prog->expected_attach_type == BPF_TRACE_FEXIT ||
+ tgt_prog->expected_attach_type == BPF_TRACE_FENTRY_MULTI ||
+ tgt_prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI ||
tgt_prog->expected_attach_type == BPF_TRACE_FSESSION)) {
/* Program extensions can extend all program types
* except fentry/fexit. The reason is the following.
@@ -18927,6 +18936,8 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
if (prog->expected_attach_type == BPF_TRACE_FSESSION &&
!bpf_jit_supports_fsession()) {
bpf_log(log, "JIT does not support fsession\n");
@@ -18956,7 +18967,18 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
if (ret < 0)
return ret;
- if (tgt_prog) {
+ /*
+ * *.multi programs don't need an address during program
+ * verification, we just take the module ref if needed.
+ */
+ if (is_tracing_multi_id(prog, btf_id)) {
+ if (btf_is_module(btf)) {
+ mod = btf_try_get_module(btf);
+ if (!mod)
+ return -ENOENT;
+ }
+ addr = 0;
+ } else if (tgt_prog) {
if (subprog == 0)
addr = (long) tgt_prog->bpf_func;
else
@@ -18984,6 +19006,12 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
ret = -EINVAL;
switch (prog->type) {
case BPF_PROG_TYPE_TRACING:
+ /* *.multi sleepable programs will pass initial sleepable check,
+ * the actual attached btf ids are checked later during the link
+ * attachment.
+ */
+ if (is_tracing_multi_id(prog, btf_id))
+ ret = 0;
if (!check_attach_sleepable(btf_id, addr, tname))
ret = 0;
/* fentry/fexit/fmod_ret progs can also be sleepable if they are
@@ -19094,6 +19122,8 @@ static bool can_be_sleepable(struct bpf_prog *prog)
case BPF_TRACE_ITER:
case BPF_TRACE_FSESSION:
case BPF_TRACE_RAW_TP:
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
return true;
default:
return false;
@@ -19185,6 +19215,14 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
return -EINVAL;
}
+ /*
+ * We don't get trampoline for tracing_multi programs at this point,
+ * it's done when tracing_multi link is created.
+ */
+ if (prog->type == BPF_PROG_TYPE_TRACING &&
+ is_tracing_multi(prog->expected_attach_type))
+ return 0;
+
key = bpf_trampoline_compute_key(tgt_prog, prog->aux->attach_btf, btf_id);
tr = bpf_trampoline_get(key, &tgt_info);
if (!tr)
diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c
index c9aea7052ba7..67769c700cae 100644
--- a/net/bpf/test_run.c
+++ b/net/bpf/test_run.c
@@ -703,6 +703,8 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog,
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
if (bpf_fentry_test1(1) != 2 ||
bpf_fentry_test2(2, 3) != 5 ||
bpf_fentry_test3(4, 5, 6) != 15 ||
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 37142e6d911a..7b1a1552491e 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -1155,6 +1155,8 @@ enum bpf_attach_type {
BPF_TRACE_KPROBE_SESSION,
BPF_TRACE_UPROBE_SESSION,
BPF_TRACE_FSESSION,
+ BPF_TRACE_FENTRY_MULTI,
+ BPF_TRACE_FEXIT_MULTI,
__MAX_BPF_ATTACH_TYPE
};
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index ab2071fdd3e8..bfed9b076a98 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -136,6 +136,8 @@ static const char * const attach_type_name[] = {
[BPF_NETKIT_PEER] = "netkit_peer",
[BPF_TRACE_KPROBE_SESSION] = "trace_kprobe_session",
[BPF_TRACE_UPROBE_SESSION] = "trace_uprobe_session",
+ [BPF_TRACE_FENTRY_MULTI] = "trace_fentry_multi",
+ [BPF_TRACE_FEXIT_MULTI] = "trace_fexit_multi",
};
static const char * const link_type_name[] = {
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 11/29] bpf: Move sleepable verification code to btf_id_allow_sleepable
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Move sleepable verification code to btf_id_allow_sleepable function.
It will be used in following changes.
Adding code to retrieve type's name instead of passing it from
bpf_check_attach_target function, because this function will be
called from another place in following changes and it's easier
to retrieve the name directly in here.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
kernel/bpf/verifier.c | 82 ++++++++++++++++++++++++++-----------------
1 file changed, 50 insertions(+), 32 deletions(-)
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 521ec2a9051e..9aae972b3b51 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -18706,6 +18706,55 @@ static bool is_tracing_multi_id(const struct bpf_prog *prog, u32 btf_id)
return is_tracing_multi(prog->expected_attach_type) && bpf_multi_func_btf_id[0] == btf_id;
}
+static int btf_id_allow_sleepable(u32 btf_id, unsigned long addr, const struct bpf_prog *prog,
+ const struct btf *btf)
+{
+ const struct btf_type *t;
+ const char *tname;
+
+ switch (prog->type) {
+ case BPF_PROG_TYPE_TRACING:
+ t = btf_type_by_id(btf, btf_id);
+ if (!t)
+ return -EINVAL;
+ tname = btf_name_by_offset(btf, t->name_off);
+ if (!tname)
+ return -EINVAL;
+
+ /*
+ * *.multi sleepable programs will pass initial sleepable check,
+ * the actual attached btf ids are checked later during the link
+ * attachment.
+ */
+ if (is_tracing_multi_id(prog, btf_id))
+ return 0;
+ if (!check_attach_sleepable(btf_id, addr, tname))
+ return 0;
+ /*
+ * fentry/fexit/fmod_ret progs can also be sleepable if they are
+ * in the fmodret id set with the KF_SLEEPABLE flag.
+ */
+ else {
+ u32 *flags = btf_kfunc_is_modify_return(btf, btf_id, prog);
+
+ if (flags && (*flags & KF_SLEEPABLE))
+ return 0;
+ }
+ break;
+ case BPF_PROG_TYPE_LSM:
+ /*
+ * LSM progs check that they are attached to bpf_lsm_*() funcs.
+ * Only some of them are sleepable.
+ */
+ if (bpf_lsm_is_sleepable_hook(btf_id))
+ return 0;
+ break;
+ default:
+ break;
+ }
+ return -EINVAL;
+}
+
int bpf_check_attach_target(struct bpf_verifier_log *log,
const struct bpf_prog *prog,
const struct bpf_prog *tgt_prog,
@@ -19003,38 +19052,7 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
}
if (prog->sleepable) {
- ret = -EINVAL;
- switch (prog->type) {
- case BPF_PROG_TYPE_TRACING:
- /* *.multi sleepable programs will pass initial sleepable check,
- * the actual attached btf ids are checked later during the link
- * attachment.
- */
- if (is_tracing_multi_id(prog, btf_id))
- ret = 0;
- if (!check_attach_sleepable(btf_id, addr, tname))
- ret = 0;
- /* fentry/fexit/fmod_ret progs can also be sleepable if they are
- * in the fmodret id set with the KF_SLEEPABLE flag.
- */
- else {
- u32 *flags = btf_kfunc_is_modify_return(btf, btf_id,
- prog);
-
- if (flags && (*flags & KF_SLEEPABLE))
- ret = 0;
- }
- break;
- case BPF_PROG_TYPE_LSM:
- /* LSM progs check that they are attached to bpf_lsm_*() funcs.
- * Only some of them are sleepable.
- */
- if (bpf_lsm_is_sleepable_hook(btf_id))
- ret = 0;
- break;
- default:
- break;
- }
+ ret = btf_id_allow_sleepable(btf_id, addr, prog, btf);
if (ret) {
module_put(mod);
bpf_log(log, "%s is not sleepable\n", tname);
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 12/29] bpf: Add bpf_trampoline_multi_attach/detach functions
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding bpf_trampoline_multi_attach/detach functions that allows to
attach/detach tracing program to multiple functions/trampolines.
The attachment is defined with bpf_program and array of BTF ids of
functions to attach the bpf program to.
Adding bpf_tracing_multi_link object that holds all the attached
trampolines and is initialized in attach and used in detach.
The attachment allocates or uses currently existing trampoline
for each function to attach and links it with the bpf program.
The attach works as follows:
- we get all the needed trampolines
- lock them and add the bpf program to each (__bpf_trampoline_link_prog)
- the trampoline_multi_ops passed in __bpf_trampoline_link_prog gathers
ftrace_hash (ip -> trampoline) objects
- we call update_ftrace_direct_add/mod to update needed locations
- we unlock all the trampolines
The detach works as follows:
- we lock all the needed trampolines
- remove the program from each (__bpf_trampoline_unlink_prog)
- the trampoline_multi_ops passed in __bpf_trampoline_unlink_prog gathers
ftrace_hash (ip -> trampoline) objects
- we call update_ftrace_direct_del/mod to update needed locations
- we unlock and put all the trampolines
We store the old image/flags in the trampoline before the update
and use it in case we need to rollback the attachment.
We keep the ftrace_hash objects allocated during attach in the link
so they can be used for detach as well.
Adding trampoline_(un)lock_all functions to (un)lock all trampolines
to gate the tracing_multi attachment.
Note this is supported only for archs (x86_64) with ftrace direct and
have single ops support.
CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS &&
CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS
It also needs CONFIG_BPF_SYSCALL enabled.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf.h | 43 ++++++
include/linux/bpf_verifier.h | 4 +
kernel/bpf/trampoline.c | 271 +++++++++++++++++++++++++++++++++++
kernel/bpf/verifier.c | 55 +++++++
4 files changed, 373 insertions(+)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index bc5a57a9faee..ae0a4b255a35 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -32,6 +32,7 @@
#include <linux/static_call.h>
#include <linux/memcontrol.h>
#include <linux/cfi.h>
+#include <linux/ftrace.h>
#include <asm/rqspinlock.h>
struct bpf_verifier_env;
@@ -1372,6 +1373,11 @@ struct bpf_trampoline {
int progs_cnt[BPF_TRAMP_MAX];
/* Executable image of trampoline */
struct bpf_tramp_image *cur_image;
+ /* Used as temporary old image storage for multi_attach */
+ struct {
+ struct bpf_tramp_image *old_image;
+ u32 old_flags;
+ } multi_attach;
};
struct bpf_attach_target_info {
@@ -1469,6 +1475,8 @@ static inline int bpf_dynptr_check_off_len(const struct bpf_dynptr_kern *ptr, u6
return 0;
}
+struct bpf_tracing_multi_link;
+
#ifdef CONFIG_BPF_JIT
int bpf_trampoline_link_prog(struct bpf_tramp_node *node,
struct bpf_trampoline *tr,
@@ -1481,6 +1489,11 @@ struct bpf_trampoline *bpf_trampoline_get(u64 key,
void bpf_trampoline_put(struct bpf_trampoline *tr);
int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_funcs);
+int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
+ struct bpf_tracing_multi_link *link);
+int bpf_trampoline_multi_detach(struct bpf_prog *prog,
+ struct bpf_tracing_multi_link *link);
+
/*
* When the architecture supports STATIC_CALL replace the bpf_dispatcher_fn
* indirection with a direct call to the bpf program. If the architecture does
@@ -1593,6 +1606,16 @@ static inline bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
{
return false;
}
+static inline int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
+ struct bpf_tracing_multi_link *link)
+{
+ return -ENOTSUPP;
+}
+static inline int bpf_trampoline_multi_detach(struct bpf_prog *prog,
+ struct bpf_tracing_multi_link *link)
+{
+ return -ENOTSUPP;
+}
#endif
struct bpf_func_info_aux {
@@ -1913,6 +1936,26 @@ struct bpf_tracing_link {
struct bpf_prog *tgt_prog;
};
+struct bpf_tracing_multi_node {
+ struct bpf_tramp_node node;
+ struct bpf_trampoline *trampoline;
+ struct ftrace_func_entry entry;
+};
+
+struct bpf_tracing_multi_data {
+ struct ftrace_hash *unreg;
+ struct ftrace_hash *modify;
+ struct ftrace_hash *reg;
+ struct ftrace_func_entry *entry;
+};
+
+struct bpf_tracing_multi_link {
+ struct bpf_link link;
+ struct bpf_tracing_multi_data data;
+ int nodes_cnt;
+ struct bpf_tracing_multi_node nodes[] __counted_by(nodes_cnt);
+};
+
struct bpf_raw_tp_link {
struct bpf_link link;
struct bpf_raw_event_map *btp;
diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
index c248ff41f42a..d57b339a8cb8 100644
--- a/include/linux/bpf_verifier.h
+++ b/include/linux/bpf_verifier.h
@@ -1591,6 +1591,10 @@ int bpf_add_kfunc_call(struct bpf_verifier_env *env, u32 func_id, u16 offset);
int bpf_fixup_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
struct bpf_insn *insn_buf, int insn_idx, int *cnt);
+/* Functions exported from verifier.c, used by trampoline.c */
+int bpf_check_attach_btf_id_multi(struct btf *btf, struct bpf_prog *prog, u32 btf_id,
+ struct bpf_attach_target_info *tgt_info);
+
/* Functions in fixups.c, called from bpf_check() */
int bpf_remove_fastcall_spills_fills(struct bpf_verifier_env *env);
int bpf_optimize_bpf_loop(struct bpf_verifier_env *env);
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index ae7e4fdfe2a3..957e5d7f9554 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -1447,6 +1447,277 @@ int __weak arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
return -ENOTSUPP;
}
+#if defined(CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS) && \
+ defined(CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS) && \
+ defined(CONFIG_BPF_SYSCALL)
+
+static void trampoline_lock_all(void)
+{
+ int i;
+
+ for (i = 0; i < TRAMPOLINE_LOCKS_TABLE_SIZE; i++)
+ mutex_lock(&trampoline_locks[i].mutex);
+}
+
+static void trampoline_unlock_all(void)
+{
+ int i;
+
+ for (i = 0; i < TRAMPOLINE_LOCKS_TABLE_SIZE; i++)
+ mutex_unlock(&trampoline_locks[i].mutex);
+}
+
+static void remove_tracing_multi_data(struct bpf_tracing_multi_data *data)
+{
+ ftrace_hash_remove(data->reg);
+ ftrace_hash_remove(data->unreg);
+ ftrace_hash_remove(data->modify);
+}
+
+static void clear_tracing_multi_data(struct bpf_tracing_multi_data *data)
+{
+ remove_tracing_multi_data(data);
+
+ free_ftrace_hash(data->reg);
+ free_ftrace_hash(data->unreg);
+ free_ftrace_hash(data->modify);
+}
+
+static int init_tracing_multi_data(struct bpf_tracing_multi_data *data)
+{
+ data->reg = alloc_ftrace_hash(FTRACE_HASH_DEFAULT_BITS);
+ data->unreg = alloc_ftrace_hash(FTRACE_HASH_DEFAULT_BITS);
+ data->modify = alloc_ftrace_hash(FTRACE_HASH_DEFAULT_BITS);
+
+ if (!data->reg || !data->unreg || !data->modify) {
+ clear_tracing_multi_data(data);
+ return -ENOMEM;
+ }
+ return 0;
+}
+
+static void ftrace_hash_add(struct ftrace_hash *hash, struct ftrace_func_entry *entry,
+ unsigned long ip, unsigned long direct)
+{
+ entry->ip = ip;
+ entry->direct = direct;
+ add_ftrace_hash_entry(hash, entry);
+}
+
+static int register_fentry_multi(struct bpf_trampoline *tr, struct bpf_tramp_image *im, void *ptr)
+{
+ unsigned long addr = (unsigned long) im->image;
+ unsigned long ip = ftrace_location(tr->ip);
+ struct bpf_tracing_multi_data *data = ptr;
+
+ if (bpf_trampoline_use_jmp(tr->flags))
+ addr = ftrace_jmp_set(addr);
+
+ ftrace_hash_add(data->reg, data->entry, ip, addr);
+ tr->cur_image = im;
+ return 0;
+}
+
+static int unregister_fentry_multi(struct bpf_trampoline *tr, u32 orig_flags, void *ptr)
+{
+ unsigned long addr = (unsigned long) tr->cur_image->image;
+ unsigned long ip = ftrace_location(tr->ip);
+ struct bpf_tracing_multi_data *data = ptr;
+
+ if (bpf_trampoline_use_jmp(tr->flags))
+ addr = ftrace_jmp_set(addr);
+
+ ftrace_hash_add(data->unreg, data->entry, ip, addr);
+ tr->cur_image = NULL;
+ return 0;
+}
+
+static int modify_fentry_multi(struct bpf_trampoline *tr, u32 orig_flags, struct bpf_tramp_image *im,
+ bool lock_direct_mutex, void *ptr)
+{
+ unsigned long addr = (unsigned long) im->image;
+ unsigned long ip = ftrace_location(tr->ip);
+ struct bpf_tracing_multi_data *data = ptr;
+
+ if (bpf_trampoline_use_jmp(tr->flags))
+ addr = ftrace_jmp_set(addr);
+
+ ftrace_hash_add(data->modify, data->entry, ip, addr);
+ tr->cur_image = im;
+ return 0;
+}
+
+static const struct bpf_trampoline_ops trampoline_multi_ops = {
+ .register_fentry = register_fentry_multi,
+ .unregister_fentry = unregister_fentry_multi,
+ .modify_fentry = modify_fentry_multi,
+};
+
+static void bpf_trampoline_multi_attach_init(struct bpf_trampoline *tr)
+{
+ tr->multi_attach.old_image = tr->cur_image;
+ tr->multi_attach.old_flags = tr->flags;
+}
+
+static void bpf_trampoline_multi_attach_free(struct bpf_trampoline *tr)
+{
+ if (tr->multi_attach.old_image)
+ bpf_tramp_image_put(tr->multi_attach.old_image);
+
+ tr->multi_attach.old_image = NULL;
+ tr->multi_attach.old_flags = 0;
+}
+
+static void bpf_trampoline_multi_attach_rollback(struct bpf_trampoline *tr)
+{
+ if (tr->cur_image)
+ bpf_tramp_image_put(tr->cur_image);
+ tr->cur_image = tr->multi_attach.old_image;
+ tr->flags = tr->multi_attach.old_flags;
+
+ tr->multi_attach.old_image = NULL;
+ tr->multi_attach.old_flags = 0;
+}
+
+#define for_each_mnode_cnt(mnode, link, cnt) \
+ for (i = 0, mnode = &link->nodes[i]; i < cnt; i++, mnode = &link->nodes[i])
+
+#define for_each_mnode(mnode, link) \
+ for_each_mnode_cnt(mnode, link, link->nodes_cnt)
+
+int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
+ struct bpf_tracing_multi_link *link)
+{
+ struct bpf_tracing_multi_data *data = &link->data;
+ struct bpf_attach_target_info tgt_info = {};
+ struct btf *btf = prog->aux->attach_btf;
+ struct bpf_tracing_multi_node *mnode;
+ struct bpf_trampoline *tr;
+ int i, err, rollback_cnt;
+ u64 key;
+
+ for_each_mnode(mnode, link) {
+ rollback_cnt = i;
+
+ err = bpf_check_attach_btf_id_multi(btf, prog, ids[i], &tgt_info);
+ if (err)
+ goto rollback_put;
+
+ key = bpf_trampoline_compute_key(NULL, btf, ids[i]);
+
+ tr = bpf_trampoline_get(key, &tgt_info);
+ if (!tr) {
+ err = -ENOMEM;
+ goto rollback_put;
+ }
+
+ mnode->trampoline = tr;
+ mnode->node.link = &link->link;
+
+ cond_resched();
+ }
+
+ err = init_tracing_multi_data(data);
+ if (err) {
+ rollback_cnt = link->nodes_cnt;
+ goto rollback_put;
+ }
+
+ trampoline_lock_all();
+
+ for_each_mnode(mnode, link) {
+ bpf_trampoline_multi_attach_init(mnode->trampoline);
+
+ data->entry = &mnode->entry;
+ err = __bpf_trampoline_link_prog(&mnode->node, mnode->trampoline, NULL,
+ &trampoline_multi_ops, data);
+ if (err) {
+ rollback_cnt = i;
+ goto rollback_unlink;
+ }
+ }
+
+ rollback_cnt = link->nodes_cnt;
+ if (ftrace_hash_count(data->reg)) {
+ err = update_ftrace_direct_add(&direct_ops, data->reg);
+ if (err)
+ goto rollback_unlink;
+ }
+
+ if (ftrace_hash_count(data->modify)) {
+ err = update_ftrace_direct_mod(&direct_ops, data->modify, true);
+ if (err) {
+ if (ftrace_hash_count(data->reg))
+ WARN_ON_ONCE(update_ftrace_direct_del(&direct_ops, data->reg));
+ goto rollback_unlink;
+ }
+ }
+
+ for_each_mnode(mnode, link)
+ bpf_trampoline_multi_attach_free(mnode->trampoline);
+
+ trampoline_unlock_all();
+
+ remove_tracing_multi_data(data);
+ return 0;
+
+rollback_unlink:
+ for_each_mnode_cnt(mnode, link, rollback_cnt) {
+ bpf_trampoline_remove_prog(mnode->trampoline, &mnode->node);
+ bpf_trampoline_multi_attach_rollback(mnode->trampoline);
+ }
+
+ trampoline_unlock_all();
+
+ clear_tracing_multi_data(data);
+ rollback_cnt = link->nodes_cnt;
+
+rollback_put:
+ for_each_mnode_cnt(mnode, link, rollback_cnt)
+ bpf_trampoline_put(mnode->trampoline);
+
+ return err;
+}
+
+int bpf_trampoline_multi_detach(struct bpf_prog *prog, struct bpf_tracing_multi_link *link)
+{
+ struct bpf_tracing_multi_data *data = &link->data;
+ struct bpf_tracing_multi_node *mnode;
+ int i;
+
+ trampoline_lock_all();
+
+ for_each_mnode(mnode, link) {
+ data->entry = &mnode->entry;
+ bpf_trampoline_multi_attach_init(mnode->trampoline);
+ WARN_ON_ONCE(__bpf_trampoline_unlink_prog(&mnode->node, mnode->trampoline,
+ NULL, &trampoline_multi_ops, data));
+ }
+
+ if (ftrace_hash_count(data->unreg))
+ WARN_ON_ONCE(update_ftrace_direct_del(&direct_ops, data->unreg));
+ if (ftrace_hash_count(data->modify))
+ WARN_ON_ONCE(update_ftrace_direct_mod(&direct_ops, data->modify, true));
+
+ for_each_mnode(mnode, link)
+ bpf_trampoline_multi_attach_free(mnode->trampoline);
+
+ trampoline_unlock_all();
+
+ for_each_mnode(mnode, link)
+ bpf_trampoline_put(mnode->trampoline);
+
+ clear_tracing_multi_data(data);
+ return 0;
+}
+
+#undef for_each_mnode_cnt
+#undef for_each_mnode
+
+#endif /* CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS &&
+ CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS &&
+ CONFIG_BPF_SYSCALL */
+
static int __init init_trampolines(void)
{
int i;
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 9aae972b3b51..bf200508527c 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -19253,6 +19253,61 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
return 0;
}
+int bpf_check_attach_btf_id_multi(struct btf *btf, struct bpf_prog *prog, u32 btf_id,
+ struct bpf_attach_target_info *tgt_info)
+{
+ const struct btf_type *t;
+ unsigned long addr;
+ const char *tname;
+ int err;
+
+ if (!btf_id || !btf)
+ return -EINVAL;
+
+ /* Check noreturn attachment. */
+ if (prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI &&
+ btf_id_set_contains(&noreturn_deny, btf_id))
+ return -EINVAL;
+ /* Check denied attachment. */
+ if (btf_id_set_contains(&btf_id_deny, btf_id))
+ return -EINVAL;
+
+ /* Check and get function target data. */
+ t = btf_type_by_id(btf, btf_id);
+ if (!t)
+ return -EINVAL;
+ tname = btf_name_by_offset(btf, t->name_off);
+ if (!tname)
+ return -EINVAL;
+ if (!btf_type_is_func(t))
+ return -EINVAL;
+ t = btf_type_by_id(btf, t->type);
+ if (!btf_type_is_func_proto(t))
+ return -EINVAL;
+ err = btf_distill_func_proto(NULL, btf, t, tname, &tgt_info->fmodel);
+ if (err < 0)
+ return err;
+ if (btf_is_module(btf)) {
+ /* The bpf program already holds reference to module. */
+ if (WARN_ON_ONCE(!prog->aux->mod))
+ return -EINVAL;
+ addr = find_kallsyms_symbol_value(prog->aux->mod, tname);
+ } else {
+ addr = kallsyms_lookup_name(tname);
+ }
+ if (!addr || !ftrace_location(addr))
+ return -ENOENT;
+
+ /* Check sleepable program attachment. */
+ if (prog->sleepable) {
+ err = btf_id_allow_sleepable(btf_id, addr, prog, btf);
+ if (err)
+ return err;
+ }
+ tgt_info->tgt_addr = addr;
+ return 0;
+}
+
struct btf *bpf_get_btf_vmlinux(void)
{
if (!btf_vmlinux && IS_ENABLED(CONFIG_DEBUG_INFO_BTF)) {
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 13/29] bpf: Add support for tracing multi link
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding new link to allow to attach program to multiple function
BTF IDs. The link is represented by struct bpf_tracing_multi_link.
To configure the link, new fields are added to bpf_attr::link_create
to pass array of BTF IDs;
struct {
__aligned_u64 ids;
__u32 cnt;
} tracing_multi;
Each BTF ID represents function (BTF_KIND_FUNC) that the link will
attach bpf program to.
We use previously added bpf_trampoline_multi_attach/detach functions
to attach/detach the link.
The linkinfo/fdinfo callbacks will be implemented in following changes.
Note this is supported only for archs (x86_64) with ftrace direct and
have single ops support.
CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS &&
CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS
Note using sort_r (instead of plain sort) in check_dup_ids, because we
will use the swap callback in following changes.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf_types.h | 1 +
include/linux/trace_events.h | 6 ++
include/uapi/linux/bpf.h | 5 ++
kernel/bpf/syscall.c | 2 +
kernel/trace/bpf_trace.c | 130 +++++++++++++++++++++++++++++++++
tools/include/uapi/linux/bpf.h | 6 ++
tools/lib/bpf/libbpf.c | 1 +
7 files changed, 151 insertions(+)
diff --git a/include/linux/bpf_types.h b/include/linux/bpf_types.h
index b13de31e163f..96575b5b563e 100644
--- a/include/linux/bpf_types.h
+++ b/include/linux/bpf_types.h
@@ -155,3 +155,4 @@ BPF_LINK_TYPE(BPF_LINK_TYPE_PERF_EVENT, perf)
BPF_LINK_TYPE(BPF_LINK_TYPE_KPROBE_MULTI, kprobe_multi)
BPF_LINK_TYPE(BPF_LINK_TYPE_STRUCT_OPS, struct_ops)
BPF_LINK_TYPE(BPF_LINK_TYPE_UPROBE_MULTI, uprobe_multi)
+BPF_LINK_TYPE(BPF_LINK_TYPE_TRACING_MULTI, tracing_multi)
diff --git a/include/linux/trace_events.h b/include/linux/trace_events.h
index d49338c44014..308c76b57d13 100644
--- a/include/linux/trace_events.h
+++ b/include/linux/trace_events.h
@@ -787,6 +787,7 @@ int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id,
unsigned long *missed);
int bpf_kprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
int bpf_uprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
+int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr);
#else
static inline unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx)
{
@@ -844,6 +845,11 @@ bpf_uprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
{
return -EOPNOTSUPP;
}
+static inline int
+bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
+{
+ return -EOPNOTSUPP;
+}
#endif
enum {
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 0cde0e2f3d4b..951a0122baaf 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -1181,6 +1181,7 @@ enum bpf_link_type {
BPF_LINK_TYPE_UPROBE_MULTI = 12,
BPF_LINK_TYPE_NETKIT = 13,
BPF_LINK_TYPE_SOCKMAP = 14,
+ BPF_LINK_TYPE_TRACING_MULTI = 15,
__MAX_BPF_LINK_TYPE,
};
@@ -1871,6 +1872,10 @@ union bpf_attr {
};
__u64 expected_revision;
} cgroup;
+ struct {
+ __aligned_u64 ids;
+ __u32 cnt;
+ } tracing_multi;
};
} link_create;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index f354e9f64e65..82e982f1a0ac 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -5858,6 +5858,8 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
ret = bpf_iter_link_attach(attr, uattr, prog);
else if (prog->expected_attach_type == BPF_LSM_CGROUP)
ret = cgroup_bpf_link_attach(attr, prog);
+ else if (is_tracing_multi(prog->expected_attach_type))
+ ret = bpf_tracing_multi_attach(prog, attr);
else
ret = bpf_tracing_prog_attach(prog,
attr->link_create.target_fd,
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index d853f97bd154..9e3cb547651e 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -42,6 +42,7 @@
#define MAX_UPROBE_MULTI_CNT (1U << 20)
#define MAX_KPROBE_MULTI_CNT (1U << 20)
+#define MAX_TRACING_MULTI_CNT (1U << 20)
#ifdef CONFIG_MODULES
struct bpf_trace_module {
@@ -3641,3 +3642,132 @@ __bpf_kfunc int bpf_copy_from_user_task_str_dynptr(const struct bpf_dynptr *dptr
}
__bpf_kfunc_end_defs();
+
+#if defined(CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS) && \
+ defined(CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS)
+
+static void bpf_tracing_multi_link_release(struct bpf_link *link)
+{
+ struct bpf_tracing_multi_link *tr_link =
+ container_of(link, struct bpf_tracing_multi_link, link);
+
+ WARN_ON_ONCE(bpf_trampoline_multi_detach(link->prog, tr_link));
+}
+
+static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)
+{
+ struct bpf_tracing_multi_link *tr_link =
+ container_of(link, struct bpf_tracing_multi_link, link);
+
+ kvfree(tr_link);
+}
+
+static const struct bpf_link_ops bpf_tracing_multi_link_lops = {
+ .release = bpf_tracing_multi_link_release,
+ .dealloc_deferred = bpf_tracing_multi_link_dealloc,
+};
+
+static int ids_cmp_r(const void *pa, const void *pb, const void *priv __maybe_unused)
+{
+ u32 a = *(u32 *) pa;
+ u32 b = *(u32 *) pb;
+
+ return (a > b) - (a < b);
+}
+
+static void ids_swap_r(void *a, void *b, int size __maybe_unused,
+ const void *priv __maybe_unused)
+{
+ u32 *id_a = a, *id_b = b;
+
+ swap(*id_a, *id_b);
+}
+
+static int check_dup_ids(u32 *ids, u32 cnt)
+{
+ int err = 0;
+
+ /*
+ * Sort ids array (together with cookies array if defined)
+ * and check it for duplicates. The ids and cookies arrays
+ * are left sorted.
+ */
+ sort_r_nonatomic(ids, cnt, sizeof(ids[0]), ids_cmp_r, ids_swap_r, NULL);
+
+ for (int i = 1; i < cnt; i++) {
+ if (ids[i] == ids[i - 1]) {
+ err = -EINVAL;
+ break;
+ }
+ }
+ return err;
+}
+
+int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
+{
+ struct bpf_tracing_multi_link *link = NULL;
+ struct bpf_link_primer link_primer;
+ u32 cnt, *ids = NULL;
+ u32 __user *uids;
+ int err;
+
+ uids = u64_to_user_ptr(attr->link_create.tracing_multi.ids);
+ cnt = attr->link_create.tracing_multi.cnt;
+
+ if (!cnt || !uids)
+ return -EINVAL;
+ if (cnt > MAX_TRACING_MULTI_CNT)
+ return -E2BIG;
+ if (attr->link_create.flags || attr->link_create.target_fd)
+ return -EINVAL;
+
+ ids = kvmalloc_objs(*ids, cnt);
+ if (!ids)
+ return -ENOMEM;
+
+ if (copy_from_user(ids, uids, cnt * sizeof(*ids))) {
+ err = -EFAULT;
+ goto error;
+ }
+
+ err = check_dup_ids(ids, cnt);
+ if (err)
+ goto error;
+
+ link = kvzalloc_flex(*link, nodes, cnt);
+ if (!link) {
+ err = -ENOMEM;
+ goto error;
+ }
+
+ bpf_link_init(&link->link, BPF_LINK_TYPE_TRACING_MULTI,
+ &bpf_tracing_multi_link_lops, prog, prog->expected_attach_type);
+
+ err = bpf_link_prime(&link->link, &link_primer);
+ if (err)
+ goto error;
+
+ link->nodes_cnt = cnt;
+
+ err = bpf_trampoline_multi_attach(prog, ids, link);
+ kvfree(ids);
+ if (err) {
+ bpf_link_cleanup(&link_primer);
+ return err;
+ }
+ return bpf_link_settle(&link_primer);
+
+error:
+ kvfree(ids);
+ kvfree(link);
+ return err;
+}
+
+#else
+
+int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
+{
+ return -EOPNOTSUPP;
+}
+
+#endif /* CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS && CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS */
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 7b1a1552491e..951a0122baaf 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -1181,6 +1181,7 @@ enum bpf_link_type {
BPF_LINK_TYPE_UPROBE_MULTI = 12,
BPF_LINK_TYPE_NETKIT = 13,
BPF_LINK_TYPE_SOCKMAP = 14,
+ BPF_LINK_TYPE_TRACING_MULTI = 15,
__MAX_BPF_LINK_TYPE,
};
@@ -1871,6 +1872,10 @@ union bpf_attr {
};
__u64 expected_revision;
} cgroup;
+ struct {
+ __aligned_u64 ids;
+ __u32 cnt;
+ } tracing_multi;
};
} link_create;
@@ -7246,6 +7251,7 @@ enum {
TCP_BPF_SOCK_OPS_CB_FLAGS = 1008, /* Get or Set TCP sock ops flags */
SK_BPF_CB_FLAGS = 1009, /* Get or set sock ops flags in socket */
SK_BPF_BYPASS_PROT_MEM = 1010, /* Get or Set sk->sk_bypass_prot_mem */
+
};
enum {
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index bfed9b076a98..7b99737590d1 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -156,6 +156,7 @@ static const char * const link_type_name[] = {
[BPF_LINK_TYPE_UPROBE_MULTI] = "uprobe_multi",
[BPF_LINK_TYPE_NETKIT] = "netkit",
[BPF_LINK_TYPE_SOCKMAP] = "sockmap",
+ [BPF_LINK_TYPE_TRACING_MULTI] = "tracing_multi",
};
static const char * const map_type_name[] = {
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 14/29] bpf: Add support for tracing_multi link cookies
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Add support to specify cookies for tracing_multi link.
Cookies are provided in array where each value is paired with provided
BTF ID value with the same array index.
Such cookie can be retrieved by bpf program with bpf_get_attach_cookie
helper call.
We need to sort cookies array together with ids array in check_dup_ids,
to keep the id->cookie relation.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf.h | 1 +
include/uapi/linux/bpf.h | 1 +
kernel/bpf/trampoline.c | 1 +
kernel/trace/bpf_trace.c | 37 ++++++++++++++++++++++++++++++----
tools/include/uapi/linux/bpf.h | 1 +
5 files changed, 37 insertions(+), 4 deletions(-)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index ae0a4b255a35..2da0de17275a 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1952,6 +1952,7 @@ struct bpf_tracing_multi_data {
struct bpf_tracing_multi_link {
struct bpf_link link;
struct bpf_tracing_multi_data data;
+ u64 *cookies;
int nodes_cnt;
struct bpf_tracing_multi_node nodes[] __counted_by(nodes_cnt);
};
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 951a0122baaf..39407a77133b 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -1874,6 +1874,7 @@ union bpf_attr {
} cgroup;
struct {
__aligned_u64 ids;
+ __aligned_u64 cookies;
__u32 cnt;
} tracing_multi;
};
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 957e5d7f9554..a3537fda50cf 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -1613,6 +1613,7 @@ int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
mnode->trampoline = tr;
mnode->node.link = &link->link;
+ mnode->node.cookie = link->cookies ? link->cookies[i] : 0;
cond_resched();
}
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 9e3cb547651e..e33492739ed1 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -3659,6 +3659,7 @@ static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)
struct bpf_tracing_multi_link *tr_link =
container_of(link, struct bpf_tracing_multi_link, link);
+ kvfree(tr_link->cookies);
kvfree(tr_link);
}
@@ -3678,13 +3679,24 @@ static int ids_cmp_r(const void *pa, const void *pb, const void *priv __maybe_un
static void ids_swap_r(void *a, void *b, int size __maybe_unused,
const void *priv __maybe_unused)
{
- u32 *id_a = a, *id_b = b;
+ u64 *cookie_a, *cookie_b, *cookies;
+ u32 *id_a = a, *id_b = b, *ids;
+ void **data = (void **) priv;
+ ids = data[0];
+ cookies = data[1];
+
+ if (cookies) {
+ cookie_a = cookies + (id_a - ids);
+ cookie_b = cookies + (id_b - ids);
+ swap(*cookie_a, *cookie_b);
+ }
swap(*id_a, *id_b);
}
-static int check_dup_ids(u32 *ids, u32 cnt)
+static int check_dup_ids(u32 *ids, u64 *cookies, u32 cnt)
{
+ void *data[2] = { ids, cookies };
int err = 0;
/*
@@ -3692,7 +3704,7 @@ static int check_dup_ids(u32 *ids, u32 cnt)
* and check it for duplicates. The ids and cookies arrays
* are left sorted.
*/
- sort_r_nonatomic(ids, cnt, sizeof(ids[0]), ids_cmp_r, ids_swap_r, NULL);
+ sort_r_nonatomic(ids, cnt, sizeof(ids[0]), ids_cmp_r, ids_swap_r, data);
for (int i = 1; i < cnt; i++) {
if (ids[i] == ids[i - 1]) {
@@ -3708,6 +3720,8 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
struct bpf_tracing_multi_link *link = NULL;
struct bpf_link_primer link_primer;
u32 cnt, *ids = NULL;
+ u64 __user *ucookies;
+ u64 *cookies = NULL;
u32 __user *uids;
int err;
@@ -3730,7 +3744,20 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
goto error;
}
- err = check_dup_ids(ids, cnt);
+ ucookies = u64_to_user_ptr(attr->link_create.tracing_multi.cookies);
+ if (ucookies) {
+ cookies = kvmalloc_objs(*cookies, cnt);
+ if (!cookies) {
+ err = -ENOMEM;
+ goto error;
+ }
+ if (copy_from_user(cookies, ucookies, cnt * sizeof(*cookies))) {
+ err = -EFAULT;
+ goto error;
+ }
+ }
+
+ err = check_dup_ids(ids, cookies, cnt);
if (err)
goto error;
@@ -3748,6 +3775,7 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
goto error;
link->nodes_cnt = cnt;
+ link->cookies = cookies;
err = bpf_trampoline_multi_attach(prog, ids, link);
kvfree(ids);
@@ -3758,6 +3786,7 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
return bpf_link_settle(&link_primer);
error:
+ kvfree(cookies);
kvfree(ids);
kvfree(link);
return err;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 951a0122baaf..39407a77133b 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -1874,6 +1874,7 @@ union bpf_attr {
} cgroup;
struct {
__aligned_u64 ids;
+ __aligned_u64 cookies;
__u32 cnt;
} tracing_multi;
};
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 15/29] bpf: Add support for tracing_multi link session
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding support to use session attachment with tracing_multi link.
Adding new BPF_TRACE_FSESSION_MULTI program attach type, that follows
the BPF_TRACE_FSESSION behaviour but on the tracing_multi link.
Such program is called on entry and exit of the attached function
and allows to pass cookie value from entry to exit execution.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
include/linux/bpf.h | 6 ++++-
include/uapi/linux/bpf.h | 1 +
kernel/bpf/fixups.c | 1 +
kernel/bpf/syscall.c | 1 +
kernel/bpf/trampoline.c | 44 +++++++++++++++++++++++++++-------
kernel/bpf/verifier.c | 20 ++++++++++++----
kernel/trace/bpf_trace.c | 15 +++++++++++-
net/bpf/test_run.c | 1 +
tools/include/uapi/linux/bpf.h | 1 +
tools/lib/bpf/libbpf.c | 1 +
10 files changed, 76 insertions(+), 15 deletions(-)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 2da0de17275a..5bbd36e8f8c6 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1953,6 +1953,7 @@ struct bpf_tracing_multi_link {
struct bpf_link link;
struct bpf_tracing_multi_data data;
u64 *cookies;
+ struct bpf_tramp_node *fexits;
int nodes_cnt;
struct bpf_tracing_multi_node nodes[] __counted_by(nodes_cnt);
};
@@ -2140,7 +2141,8 @@ static inline void bpf_prog_put_recursion_context(struct bpf_prog *prog)
static inline bool is_tracing_multi(enum bpf_attach_type type)
{
- return type == BPF_TRACE_FENTRY_MULTI || type == BPF_TRACE_FEXIT_MULTI;
+ return type == BPF_TRACE_FENTRY_MULTI || type == BPF_TRACE_FEXIT_MULTI ||
+ type == BPF_TRACE_FSESSION_MULTI;
}
#if defined(CONFIG_BPF_JIT) && defined(CONFIG_BPF_SYSCALL)
@@ -2267,6 +2269,8 @@ static inline int bpf_fsession_cnt(struct bpf_tramp_nodes *nodes)
for (int i = 0; i < nodes[BPF_TRAMP_FENTRY].nr_nodes; i++) {
if (fentries.nodes[i]->link->prog->expected_attach_type == BPF_TRACE_FSESSION)
cnt++;
+ if (fentries.nodes[i]->link->prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI)
+ cnt++;
}
return cnt;
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 39407a77133b..b49bed575ce3 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -1157,6 +1157,7 @@ enum bpf_attach_type {
BPF_TRACE_FSESSION,
BPF_TRACE_FENTRY_MULTI,
BPF_TRACE_FEXIT_MULTI,
+ BPF_TRACE_FSESSION_MULTI,
__MAX_BPF_ATTACH_TYPE
};
diff --git a/kernel/bpf/fixups.c b/kernel/bpf/fixups.c
index 0cf9735929f5..3cf2cc6e3ab6 100644
--- a/kernel/bpf/fixups.c
+++ b/kernel/bpf/fixups.c
@@ -2187,6 +2187,7 @@ int bpf_do_misc_fixups(struct bpf_verifier_env *env)
if (eatype == BPF_TRACE_FEXIT ||
eatype == BPF_TRACE_FSESSION ||
eatype == BPF_TRACE_FEXIT_MULTI ||
+ eatype == BPF_TRACE_FSESSION_MULTI ||
eatype == BPF_MODIFY_RETURN) {
/* Load nr_args from ctx - 8 */
insn_buf[0] = BPF_LDX_MEM(BPF_DW, BPF_REG_0, BPF_REG_1, -8);
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 82e982f1a0ac..63f4854a3618 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -4473,6 +4473,7 @@ attach_type_to_prog_type(enum bpf_attach_type attach_type)
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FSESSION_MULTI:
case BPF_TRACE_FENTRY_MULTI:
case BPF_TRACE_FEXIT_MULTI:
case BPF_MODIFY_RETURN:
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index a3537fda50cf..1a721fc4bef5 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -183,7 +183,8 @@ bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
case BPF_PROG_TYPE_TRACING:
if (eatype == BPF_TRACE_FENTRY || eatype == BPF_TRACE_FEXIT ||
eatype == BPF_MODIFY_RETURN || eatype == BPF_TRACE_FSESSION ||
- eatype == BPF_TRACE_FENTRY_MULTI || eatype == BPF_TRACE_FEXIT_MULTI)
+ eatype == BPF_TRACE_FENTRY_MULTI || eatype == BPF_TRACE_FEXIT_MULTI ||
+ eatype == BPF_TRACE_FSESSION_MULTI)
return true;
return false;
case BPF_PROG_TYPE_LSM:
@@ -790,6 +791,7 @@ static enum bpf_tramp_prog_type bpf_attach_type_to_tramp(struct bpf_prog *prog)
case BPF_TRACE_FEXIT_MULTI:
return BPF_TRAMP_FEXIT;
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FSESSION_MULTI:
return BPF_TRAMP_FSESSION;
case BPF_LSM_MAC:
if (!prog->aux->attach_func_proto->type)
@@ -822,13 +824,30 @@ static int bpf_freplace_check_tgt_prog(struct bpf_prog *tgt_prog)
return 0;
}
+static struct bpf_tramp_node *fsession_exit(struct bpf_tramp_node *node)
+{
+ if (node->link->type == BPF_LINK_TYPE_TRACING) {
+ struct bpf_tracing_link *link;
+
+ link = container_of(node->link, struct bpf_tracing_link, link.link);
+ return &link->fexit;
+ } else if (node->link->type == BPF_LINK_TYPE_TRACING_MULTI) {
+ struct bpf_tracing_multi_link *link;
+ struct bpf_tracing_multi_node *mnode;
+
+ link = container_of(node->link, struct bpf_tracing_multi_link, link);
+ mnode = container_of(node, struct bpf_tracing_multi_node, node);
+ return &link->fexits[mnode - link->nodes];
+ }
+ return NULL;
+}
+
static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
struct bpf_tramp_node *node,
int cnt)
{
- struct bpf_tracing_link *tr_link = NULL;
enum bpf_tramp_prog_type kind;
- struct bpf_tramp_node *node_existing;
+ struct bpf_tramp_node *node_existing, *fexit;
struct hlist_head *prog_list;
kind = bpf_attach_type_to_tramp(node->link->prog);
@@ -853,8 +872,10 @@ static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
hlist_add_head(&node->tramp_hlist, prog_list);
if (kind == BPF_TRAMP_FSESSION) {
tr->progs_cnt[BPF_TRAMP_FENTRY]++;
- tr_link = container_of(node, struct bpf_tracing_link, link.node);
- hlist_add_head(&tr_link->fexit.tramp_hlist, &tr->progs_hlist[BPF_TRAMP_FEXIT]);
+ fexit = fsession_exit(node);
+ if (WARN_ON_ONCE(!fexit))
+ return -EINVAL;
+ hlist_add_head(&fexit->tramp_hlist, &tr->progs_hlist[BPF_TRAMP_FEXIT]);
tr->progs_cnt[BPF_TRAMP_FEXIT]++;
} else {
tr->progs_cnt[kind]++;
@@ -865,13 +886,15 @@ static int bpf_trampoline_add_prog(struct bpf_trampoline *tr,
static void bpf_trampoline_remove_prog(struct bpf_trampoline *tr,
struct bpf_tramp_node *node)
{
- struct bpf_tracing_link *tr_link;
enum bpf_tramp_prog_type kind;
+ struct bpf_tramp_node *fexit;
kind = bpf_attach_type_to_tramp(node->link->prog);
if (kind == BPF_TRAMP_FSESSION) {
- tr_link = container_of(node, struct bpf_tracing_link, link.node);
- hlist_del_init(&tr_link->fexit.tramp_hlist);
+ fexit = fsession_exit(node);
+ if (WARN_ON_ONCE(!fexit))
+ return;
+ hlist_del_init(&fexit->tramp_hlist);
tr->progs_cnt[BPF_TRAMP_FEXIT]--;
kind = BPF_TRAMP_FENTRY;
}
@@ -1615,6 +1638,11 @@ int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
mnode->node.link = &link->link;
mnode->node.cookie = link->cookies ? link->cookies[i] : 0;
+ if (prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI) {
+ link->fexits[i].link = &link->link;
+ link->fexits[i].cookie = link->cookies ? link->cookies[i] : 0;
+ }
+
cond_resched();
}
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index bf200508527c..8cb96fc37edf 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -16316,6 +16316,7 @@ static bool return_retval_range(struct bpf_verifier_env *env, struct bpf_retval_
case BPF_TRACE_FSESSION:
case BPF_TRACE_FENTRY_MULTI:
case BPF_TRACE_FEXIT_MULTI:
+ case BPF_TRACE_FSESSION_MULTI:
*range = retval_range(0, 0);
break;
case BPF_TRACE_RAW_TP:
@@ -18879,7 +18880,8 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
tgt_prog->expected_attach_type == BPF_TRACE_FEXIT ||
tgt_prog->expected_attach_type == BPF_TRACE_FENTRY_MULTI ||
tgt_prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI ||
- tgt_prog->expected_attach_type == BPF_TRACE_FSESSION)) {
+ tgt_prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ tgt_prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI)) {
/* Program extensions can extend all program types
* except fentry/fexit. The reason is the following.
* The fentry/fexit programs are used for performance
@@ -18985,9 +18987,11 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
case BPF_TRACE_FSESSION:
+ case BPF_TRACE_FSESSION_MULTI:
case BPF_TRACE_FENTRY_MULTI:
case BPF_TRACE_FEXIT_MULTI:
- if (prog->expected_attach_type == BPF_TRACE_FSESSION &&
+ if ((prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI) &&
!bpf_jit_supports_fsession()) {
bpf_log(log, "JIT does not support fsession\n");
return -EOPNOTSUPP;
@@ -19142,6 +19146,7 @@ static bool can_be_sleepable(struct bpf_prog *prog)
case BPF_TRACE_RAW_TP:
case BPF_TRACE_FENTRY_MULTI:
case BPF_TRACE_FEXIT_MULTI:
+ case BPF_TRACE_FSESSION_MULTI:
return true;
default:
return false;
@@ -19226,6 +19231,7 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
return -EINVAL;
} else if ((prog->expected_attach_type == BPF_TRACE_FEXIT ||
prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI ||
prog->expected_attach_type == BPF_MODIFY_RETURN) &&
btf_id_set_contains(&noreturn_deny, btf_id)) {
verbose(env, "Attaching fexit/fsession/fmod_ret to __noreturn function '%s' is rejected.\n",
@@ -19265,7 +19271,8 @@ int bpf_check_attach_btf_id_multi(struct btf *btf, struct bpf_prog *prog, u32 bt
return -EINVAL;
/* Check noreturn attachment. */
- if (prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI &&
+ if ((prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI ||
+ prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI) &&
btf_id_set_contains(&noreturn_deny, btf_id))
return -EINVAL;
/* Check denied attachment. */
@@ -19548,7 +19555,9 @@ int bpf_fixup_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
insn_buf[0] = BPF_MOV64_REG(BPF_REG_0, BPF_REG_1);
*cnt = 1;
} else if (desc->func_id == special_kfunc_list[KF_bpf_session_is_return] &&
- env->prog->expected_attach_type == BPF_TRACE_FSESSION) {
+ (env->prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ env->prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI)) {
+
/*
* inline the bpf_session_is_return() for fsession:
* bool bpf_session_is_return(void *ctx)
@@ -19561,7 +19570,8 @@ int bpf_fixup_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
insn_buf[2] = BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1);
*cnt = 3;
} else if (desc->func_id == special_kfunc_list[KF_bpf_session_cookie] &&
- env->prog->expected_attach_type == BPF_TRACE_FSESSION) {
+ (env->prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ env->prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI)) {
/*
* inline bpf_session_cookie() for fsession:
* __u64 *bpf_session_cookie(void *ctx)
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index e33492739ed1..a0d688fffc5a 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -1334,7 +1334,8 @@ static inline bool is_uprobe_session(const struct bpf_prog *prog)
static inline bool is_trace_fsession(const struct bpf_prog *prog)
{
return prog->type == BPF_PROG_TYPE_TRACING &&
- prog->expected_attach_type == BPF_TRACE_FSESSION;
+ (prog->expected_attach_type == BPF_TRACE_FSESSION ||
+ prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI);
}
static const struct bpf_func_proto *
@@ -3659,6 +3660,7 @@ static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)
struct bpf_tracing_multi_link *tr_link =
container_of(link, struct bpf_tracing_multi_link, link);
+ kvfree(tr_link->fexits);
kvfree(tr_link->cookies);
kvfree(tr_link);
}
@@ -3718,6 +3720,7 @@ static int check_dup_ids(u32 *ids, u64 *cookies, u32 cnt)
int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
{
struct bpf_tracing_multi_link *link = NULL;
+ struct bpf_tramp_node *fexits = NULL;
struct bpf_link_primer link_primer;
u32 cnt, *ids = NULL;
u64 __user *ucookies;
@@ -3761,6 +3764,14 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
if (err)
goto error;
+ if (prog->expected_attach_type == BPF_TRACE_FSESSION_MULTI) {
+ fexits = kvmalloc_objs(*fexits, cnt);
+ if (!fexits) {
+ err = -ENOMEM;
+ goto error;
+ }
+ }
+
link = kvzalloc_flex(*link, nodes, cnt);
if (!link) {
err = -ENOMEM;
@@ -3776,6 +3787,7 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
link->nodes_cnt = cnt;
link->cookies = cookies;
+ link->fexits = fexits;
err = bpf_trampoline_multi_attach(prog, ids, link);
kvfree(ids);
@@ -3786,6 +3798,7 @@ int bpf_tracing_multi_attach(struct bpf_prog *prog, const union bpf_attr *attr)
return bpf_link_settle(&link_primer);
error:
+ kvfree(fexits);
kvfree(cookies);
kvfree(ids);
kvfree(link);
diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c
index 67769c700cae..a831682ee982 100644
--- a/net/bpf/test_run.c
+++ b/net/bpf/test_run.c
@@ -705,6 +705,7 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog,
case BPF_TRACE_FSESSION:
case BPF_TRACE_FENTRY_MULTI:
case BPF_TRACE_FEXIT_MULTI:
+ case BPF_TRACE_FSESSION_MULTI:
if (bpf_fentry_test1(1) != 2 ||
bpf_fentry_test2(2, 3) != 5 ||
bpf_fentry_test3(4, 5, 6) != 15 ||
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 39407a77133b..b49bed575ce3 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -1157,6 +1157,7 @@ enum bpf_attach_type {
BPF_TRACE_FSESSION,
BPF_TRACE_FENTRY_MULTI,
BPF_TRACE_FEXIT_MULTI,
+ BPF_TRACE_FSESSION_MULTI,
__MAX_BPF_ATTACH_TYPE
};
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 7b99737590d1..2f5ac3e91d21 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -138,6 +138,7 @@ static const char * const attach_type_name[] = {
[BPF_TRACE_UPROBE_SESSION] = "trace_uprobe_session",
[BPF_TRACE_FENTRY_MULTI] = "trace_fentry_multi",
[BPF_TRACE_FEXIT_MULTI] = "trace_fexit_multi",
+ [BPF_TRACE_FSESSION_MULTI] = "trace_fsession_multi",
};
static const char * const link_type_name[] = {
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 16/29] bpf: Add support for tracing_multi link fdinfo
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tracing_multi link fdinfo support with following output:
pos: 0
flags: 02000000
mnt_id: 19
ino: 3087
link_type: tracing_multi
link_id: 9
prog_tag: 599ba0e317244f86
prog_id: 94
attach_type: 59
cnt: 10
obj-id btf-id cookie func
1 91593 8 bpf_fentry_test1+0x4/0x10
1 91595 9 bpf_fentry_test2+0x4/0x10
1 91596 7 bpf_fentry_test3+0x4/0x20
1 91597 5 bpf_fentry_test4+0x4/0x20
1 91598 4 bpf_fentry_test5+0x4/0x20
1 91599 2 bpf_fentry_test6+0x4/0x20
1 91600 3 bpf_fentry_test7+0x4/0x10
1 91601 1 bpf_fentry_test8+0x4/0x10
1 91602 10 bpf_fentry_test9+0x4/0x10
1 91594 6 bpf_fentry_test10+0x4/0x10
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
kernel/trace/bpf_trace.c | 30 ++++++++++++++++++++++++++++++
1 file changed, 30 insertions(+)
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index a0d688fffc5a..90432f0fc2a8 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -3665,9 +3665,39 @@ static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)
kvfree(tr_link);
}
+#ifdef CONFIG_PROC_FS
+static void bpf_tracing_multi_show_fdinfo(const struct bpf_link *link,
+ struct seq_file *seq)
+{
+ struct bpf_tracing_multi_link *tr_link =
+ container_of(link, struct bpf_tracing_multi_link, link);
+ bool has_cookies = !!tr_link->cookies;
+
+ seq_printf(seq, "attach_type:\t%u\n", tr_link->link.attach_type);
+ seq_printf(seq, "cnt:\t%u\n", tr_link->nodes_cnt);
+
+ seq_printf(seq, "%s\t %s\t %s\t %s\n", "obj-id", "btf-id", "cookie", "func");
+ for (int i = 0; i < tr_link->nodes_cnt; i++) {
+ struct bpf_tracing_multi_node *mnode = &tr_link->nodes[i];
+ u32 btf_id, obj_id;
+
+ bpf_trampoline_unpack_key(mnode->trampoline->key, &obj_id, &btf_id);
+ seq_printf(seq, "%u\t %u\t %llu\t %pS\n",
+ obj_id, btf_id,
+ has_cookies ? tr_link->cookies[i] : 0,
+ (void *) mnode->trampoline->ip);
+
+ cond_resched();
+ }
+}
+#endif
+
static const struct bpf_link_ops bpf_tracing_multi_link_lops = {
.release = bpf_tracing_multi_link_release,
.dealloc_deferred = bpf_tracing_multi_link_dealloc,
+#ifdef CONFIG_PROC_FS
+ .show_fdinfo = bpf_tracing_multi_show_fdinfo,
+#endif
};
static int ids_cmp_r(const void *pa, const void *pb, const void *priv __maybe_unused)
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 17/29] libbpf: Add bpf_object_cleanup_btf function
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding bpf_object_cleanup_btf function to cleanup btf objects.
It will be used in following changes.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/lib/bpf/libbpf.c | 16 ++++++++++++----
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 2f5ac3e91d21..240cc37fe621 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -8940,13 +8940,10 @@ static void bpf_object_unpin(struct bpf_object *obj)
bpf_map__unpin(&obj->maps[i], NULL);
}
-static void bpf_object_post_load_cleanup(struct bpf_object *obj)
+static void bpf_object_cleanup_btf(struct bpf_object *obj)
{
int i;
- /* clean up fd_array */
- zfree(&obj->fd_array);
-
/* clean up module BTFs */
for (i = 0; i < obj->btf_module_cnt; i++) {
close(obj->btf_modules[i].fd);
@@ -8954,6 +8951,8 @@ static void bpf_object_post_load_cleanup(struct bpf_object *obj)
free(obj->btf_modules[i].name);
}
obj->btf_module_cnt = 0;
+ obj->btf_module_cap = 0;
+ obj->btf_modules_loaded = false;
zfree(&obj->btf_modules);
/* clean up vmlinux BTF */
@@ -8961,6 +8960,15 @@ static void bpf_object_post_load_cleanup(struct bpf_object *obj)
obj->btf_vmlinux = NULL;
}
+static void bpf_object_post_load_cleanup(struct bpf_object *obj)
+{
+ /* clean up fd_array */
+ zfree(&obj->fd_array);
+
+ /* clean up BTF */
+ bpf_object_cleanup_btf(obj);
+}
+
static int bpf_object_prepare(struct bpf_object *obj, const char *target_btf_path)
{
int err;
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 18/29] libbpf: Add bpf_link_create support for tracing_multi link
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding bpf_link_create support for tracing_multi link with
new tracing_multi record in struct bpf_link_create_opts.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/lib/bpf/bpf.c | 9 +++++++++
tools/lib/bpf/bpf.h | 5 +++++
2 files changed, 14 insertions(+)
diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index bc513aa8f404..f37e3416f61a 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -845,6 +845,15 @@ int bpf_link_create(int prog_fd, int target_fd,
if (!OPTS_ZEROED(opts, uprobe_multi))
return libbpf_err(-EINVAL);
break;
+ case BPF_TRACE_FENTRY_MULTI:
+ case BPF_TRACE_FEXIT_MULTI:
+ case BPF_TRACE_FSESSION_MULTI:
+ attr.link_create.tracing_multi.ids = ptr_to_u64(OPTS_GET(opts, tracing_multi.ids, 0));
+ attr.link_create.tracing_multi.cookies = ptr_to_u64(OPTS_GET(opts, tracing_multi.cookies, 0));
+ attr.link_create.tracing_multi.cnt = OPTS_GET(opts, tracing_multi.cnt, 0);
+ if (!OPTS_ZEROED(opts, tracing_multi))
+ return libbpf_err(-EINVAL);
+ break;
case BPF_TRACE_RAW_TP:
case BPF_TRACE_FENTRY:
case BPF_TRACE_FEXIT:
diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
index 2312900a3263..012354131cf6 100644
--- a/tools/lib/bpf/bpf.h
+++ b/tools/lib/bpf/bpf.h
@@ -469,6 +469,11 @@ struct bpf_link_create_opts {
__u32 relative_id;
__u64 expected_revision;
} cgroup;
+ struct {
+ const __u32 *ids;
+ const __u64 *cookies;
+ __u32 cnt;
+ } tracing_multi;
};
size_t :0;
};
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 19/29] libbpf: Add btf_type_is_traceable_func function
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding btf_type_is_traceable_func function to perform same checks
as the kernel's btf_distill_func_proto function to prevent attachment
on some of the functions.
Exporting the function via libbpf_internal.h because it will be used
by benchmark test in following changes.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/lib/bpf/libbpf.c | 79 +++++++++++++++++++++++++++++++++
tools/lib/bpf/libbpf_internal.h | 1 +
2 files changed, 80 insertions(+)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 240cc37fe621..df3fe00f8b6f 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -12449,6 +12449,85 @@ static int attach_uprobe_multi(const struct bpf_program *prog, long cookie, stru
return ret;
}
+#define MAX_BPF_FUNC_ARGS 12
+
+static bool btf_type_is_modifier(const struct btf_type *t)
+{
+ switch (BTF_INFO_KIND(t->info)) {
+ case BTF_KIND_TYPEDEF:
+ case BTF_KIND_VOLATILE:
+ case BTF_KIND_CONST:
+ case BTF_KIND_RESTRICT:
+ case BTF_KIND_TYPE_TAG:
+ return true;
+ default:
+ return false;
+ }
+}
+
+#define MAX_RESOLVE_DEPTH 32
+
+static int btf_get_type_size(const struct btf *btf, __u32 type_id,
+ const struct btf_type **ret_type)
+{
+ const struct btf_type *t;
+ int i;
+
+ *ret_type = btf__type_by_id(btf, 0);
+ if (!type_id)
+ return 0;
+ t = btf__type_by_id(btf, type_id);
+ for (i = 0; i < MAX_RESOLVE_DEPTH && t && btf_type_is_modifier(t); i++)
+ t = btf__type_by_id(btf, t->type);
+ if (!t || i == MAX_RESOLVE_DEPTH)
+ return -EINVAL;
+ *ret_type = t;
+ if (btf_is_ptr(t))
+ return btf__pointer_size(btf);
+ if (btf_is_int(t) || btf_is_any_enum(t) || btf_is_struct(t))
+ return t->size;
+ return -EINVAL;
+}
+
+bool btf_type_is_traceable_func(const struct btf *btf, const struct btf_type *t)
+{
+ const struct btf_param *args;
+ const struct btf_type *proto;
+ __u32 i, nargs;
+ int ret;
+
+ if (!btf_is_func(t))
+ return false;
+ proto = btf__type_by_id(btf, t->type);
+ if (!proto || !btf_is_func_proto(proto))
+ return false;
+
+ args = (const struct btf_param *)(proto + 1);
+ nargs = btf_vlen(proto);
+ if (nargs > MAX_BPF_FUNC_ARGS)
+ return false;
+
+ /* No support for struct return type. */
+ ret = btf_get_type_size(btf, proto->type, &t);
+ if (ret < 0 || btf_is_struct(t) || btf_is_union(t))
+ return false;
+
+ for (i = 0; i < nargs; i++) {
+ /* No support for variable args. */
+ if (i == nargs - 1 && args[i].type == 0)
+ return false;
+ ret = btf_get_type_size(btf, args[i].type, &t);
+ /* No support of struct argument size greater than 16 bytes. */
+ if (ret < 0 || ret > 16)
+ return false;
+ /* No support for void argument. */
+ if (ret == 0)
+ return false;
+ }
+
+ return true;
+}
+
static inline int add_uprobe_event_legacy(const char *probe_name, bool retprobe,
const char *binary_path, size_t offset)
{
diff --git a/tools/lib/bpf/libbpf_internal.h b/tools/lib/bpf/libbpf_internal.h
index 7d93c6c01d60..04cd303fb5a8 100644
--- a/tools/lib/bpf/libbpf_internal.h
+++ b/tools/lib/bpf/libbpf_internal.h
@@ -250,6 +250,7 @@ const struct btf_type *skip_mods_and_typedefs(const struct btf *btf, __u32 id, _
const struct btf_header *btf_header(const struct btf *btf);
void btf_set_base_btf(struct btf *btf, const struct btf *base_btf);
int btf_relocate(struct btf *btf, const struct btf *base_btf, __u32 **id_map);
+bool btf_type_is_traceable_func(const struct btf *btf, const struct btf_type *t);
static inline enum btf_func_linkage btf_func_linkage(const struct btf_type *t)
{
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 20/29] libbpf: Add support to create tracing multi link
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding bpf_program__attach_tracing_multi function for attaching
tracing program to multiple functions.
struct bpf_link *
bpf_program__attach_tracing_multi(const struct bpf_program *prog,
const char *pattern,
const struct bpf_tracing_multi_opts *opts);
User can specify functions to attach with 'pattern' argument that
allows wildcards (*?' supported) or provide BTF ids of functions
in array directly via opts argument. These options are mutually
exclusive.
When using BTF ids, user can also provide cookie value for each
provided id/function, that can be retrieved later in bpf program
with bpf_get_attach_cookie helper. Each cookie value is paired with
provided BTF id with the same array index.
Adding support to auto attach programs with following sections:
fsession.multi/<pattern>
fsession.multi.s/<pattern>
fentry.multi/<pattern>
fexit.multi/<pattern>
fentry.multi.s/<pattern>
fexit.multi.s/<pattern>
The provided <pattern> is used as 'pattern' argument in
bpf_program__attach_kprobe_multi_opts function.
The <pattern> allows to specify optional kernel module name with
following syntax:
<module>:<function_pattern>
In order to attach tracing_multi link to a module functions:
- program must be loaded with 'module' btf fd
(in attr::attach_btf_obj_fd)
- bpf_program__attach_tracing_multi must either have
pattern with module spec or BTF ids from the module
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/lib/bpf/libbpf.c | 276 +++++++++++++++++++++++++++++++++++++++
tools/lib/bpf/libbpf.h | 15 +++
tools/lib/bpf/libbpf.map | 1 +
3 files changed, 292 insertions(+)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index df3fe00f8b6f..d92eb3e2975c 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -7771,6 +7771,69 @@ static int bpf_object__sanitize_prog(struct bpf_object *obj, struct bpf_program
static int libbpf_find_attach_btf_id(struct bpf_program *prog, const char *attach_name,
int *btf_obj_fd, int *btf_type_id);
+static inline bool is_tracing_multi(enum bpf_attach_type type)
+{
+ return type == BPF_TRACE_FENTRY_MULTI || type == BPF_TRACE_FEXIT_MULTI ||
+ type == BPF_TRACE_FSESSION_MULTI;
+}
+
+static const struct module_btf *find_attach_module(struct bpf_object *obj, const char *attach)
+{
+ const char *sep, *mod_name = NULL;
+ int i, mod_len, err;
+
+ /*
+ * We expect attach string in the form of either
+ * - function_pattern or
+ * - <module>:function_pattern
+ */
+ sep = strchr(attach, ':');
+ if (sep) {
+ mod_name = attach;
+ mod_len = sep - mod_name;
+ }
+ if (!mod_name)
+ return NULL;
+
+ err = load_module_btfs(obj);
+ if (err)
+ return NULL;
+
+ for (i = 0; i < obj->btf_module_cnt; i++) {
+ const struct module_btf *mod = &obj->btf_modules[i];
+
+ if (strncmp(mod->name, mod_name, mod_len) == 0 && mod->name[mod_len] == '\0')
+ return mod;
+ }
+ return NULL;
+}
+
+static int tracing_multi_mod_fd(struct bpf_program *prog, int *btf_obj_fd)
+{
+ const char *attach_name, *sep;
+ const struct module_btf *mod;
+
+ *btf_obj_fd = 0;
+ attach_name = strchr(prog->sec_name, '/');
+
+ /* Program with no details in spec, using kernel btf. */
+ if (!attach_name)
+ return 0;
+
+ /* Program with no module section, using kernel btf. */
+ sep = strchr(++attach_name, ':');
+ if (!sep)
+ return 0;
+
+ /* Program with module specified, get its btf fd. */
+ mod = find_attach_module(prog->obj, attach_name);
+ if (!mod)
+ return -EINVAL;
+
+ *btf_obj_fd = mod->fd;
+ return 0;
+}
+
/* this is called as prog->sec_def->prog_prepare_load_fn for libbpf-supported sec_defs */
static int libbpf_prepare_prog_load(struct bpf_program *prog,
struct bpf_prog_load_opts *opts, long cookie)
@@ -7834,6 +7897,18 @@ static int libbpf_prepare_prog_load(struct bpf_program *prog,
opts->attach_btf_obj_fd = btf_obj_fd;
opts->attach_btf_id = btf_type_id;
}
+
+ if (is_tracing_multi(prog->expected_attach_type)) {
+ int err, btf_obj_fd = 0;
+
+ err = tracing_multi_mod_fd(prog, &btf_obj_fd);
+ if (err < 0)
+ return err;
+
+ prog->attach_btf_obj_fd = btf_obj_fd;
+ opts->attach_btf_obj_fd = btf_obj_fd;
+ }
+
return 0;
}
@@ -9995,6 +10070,7 @@ static int attach_kprobe_session(const struct bpf_program *prog, long cookie, st
static int attach_uprobe_multi(const struct bpf_program *prog, long cookie, struct bpf_link **link);
static int attach_lsm(const struct bpf_program *prog, long cookie, struct bpf_link **link);
static int attach_iter(const struct bpf_program *prog, long cookie, struct bpf_link **link);
+static int attach_tracing_multi(const struct bpf_program *prog, long cookie, struct bpf_link **link);
static const struct bpf_sec_def section_defs[] = {
SEC_DEF("socket", SOCKET_FILTER, 0, SEC_NONE),
@@ -10048,6 +10124,12 @@ static const struct bpf_sec_def section_defs[] = {
SEC_DEF("fexit.s+", TRACING, BPF_TRACE_FEXIT, SEC_ATTACH_BTF | SEC_SLEEPABLE, attach_trace),
SEC_DEF("fsession+", TRACING, BPF_TRACE_FSESSION, SEC_ATTACH_BTF, attach_trace),
SEC_DEF("fsession.s+", TRACING, BPF_TRACE_FSESSION, SEC_ATTACH_BTF | SEC_SLEEPABLE, attach_trace),
+ SEC_DEF("fsession.multi+", TRACING, BPF_TRACE_FSESSION_MULTI, 0, attach_tracing_multi),
+ SEC_DEF("fsession.multi.s+", TRACING, BPF_TRACE_FSESSION_MULTI, SEC_SLEEPABLE, attach_tracing_multi),
+ SEC_DEF("fentry.multi+", TRACING, BPF_TRACE_FENTRY_MULTI, 0, attach_tracing_multi),
+ SEC_DEF("fexit.multi+", TRACING, BPF_TRACE_FEXIT_MULTI, 0, attach_tracing_multi),
+ SEC_DEF("fentry.multi.s+", TRACING, BPF_TRACE_FENTRY_MULTI, SEC_SLEEPABLE, attach_tracing_multi),
+ SEC_DEF("fexit.multi.s+", TRACING, BPF_TRACE_FEXIT_MULTI, SEC_SLEEPABLE, attach_tracing_multi),
SEC_DEF("freplace+", EXT, 0, SEC_ATTACH_BTF, attach_trace),
SEC_DEF("lsm+", LSM, BPF_LSM_MAC, SEC_ATTACH_BTF, attach_lsm),
SEC_DEF("lsm.s+", LSM, BPF_LSM_MAC, SEC_ATTACH_BTF | SEC_SLEEPABLE, attach_lsm),
@@ -12528,6 +12610,200 @@ bool btf_type_is_traceable_func(const struct btf *btf, const struct btf_type *t)
return true;
}
+static int
+collect_btf_func_ids_by_glob(const struct btf *btf, const char *pattern, __u32 **ids)
+{
+ __u32 type_id, nr_types = btf__type_cnt(btf);
+ size_t cap = 0, cnt = 0;
+
+ if (!pattern)
+ return -EINVAL;
+
+ for (type_id = 1; type_id < nr_types; type_id++) {
+ const struct btf_type *t = btf__type_by_id(btf, type_id);
+ const char *name;
+ int err;
+
+ if (btf_kind(t) != BTF_KIND_FUNC)
+ continue;
+ name = btf__name_by_offset(btf, t->name_off);
+ if (!name)
+ continue;
+
+ if (!glob_match(name, pattern))
+ continue;
+ if (!btf_type_is_traceable_func(btf, t))
+ continue;
+
+ err = libbpf_ensure_mem((void **) ids, &cap, sizeof(**ids), cnt + 1);
+ if (err) {
+ free(*ids);
+ return -ENOMEM;
+ }
+ (*ids)[cnt++] = type_id;
+ }
+
+ return cnt;
+}
+
+static int collect_func_ids_by_glob(const struct bpf_program *prog, const char *pattern, __u32 **ids)
+{
+ struct bpf_object *obj = prog->obj;
+ const struct module_btf *mod;
+ struct btf *btf = NULL;
+ const char *sep;
+ int err;
+
+ err = bpf_object__load_vmlinux_btf(obj, true);
+ if (err)
+ return err;
+
+ /* In case we have module specified, we will find its btf and use that. */
+ sep = strchr(pattern, ':');
+ if (sep) {
+ mod = find_attach_module(obj, pattern);
+ if (!mod) {
+ err = -EINVAL;
+ goto cleanup;
+ }
+ btf = mod->btf;
+ pattern = sep + 1;
+ } else {
+ /* Program is loaded for kernel module. */
+ if (prog->attach_btf_obj_fd) {
+ err = -EINVAL;
+ goto cleanup;
+ }
+ btf = obj->btf_vmlinux;
+ }
+
+ err = collect_btf_func_ids_by_glob(btf, pattern, ids);
+
+cleanup:
+ bpf_object_cleanup_btf(obj);
+ return err;
+}
+
+struct bpf_link *
+bpf_program__attach_tracing_multi(const struct bpf_program *prog, const char *pattern,
+ const struct bpf_tracing_multi_opts *opts)
+{
+ LIBBPF_OPTS(bpf_link_create_opts, lopts);
+ int prog_fd, link_fd, err, cnt;
+ __u32 *free_ids = NULL;
+ struct bpf_link *link;
+ const __u64 *cookies;
+ const __u32 *ids;
+
+ if (!OPTS_VALID(opts, bpf_tracing_multi_opts))
+ return libbpf_err_ptr(-EINVAL);
+
+ prog_fd = bpf_program__fd(prog);
+ if (prog_fd < 0) {
+ pr_warn("prog '%s': can't attach BPF program without FD (was it loaded?)\n",
+ prog->name);
+ return libbpf_err_ptr(-EINVAL);
+ }
+
+ cnt = OPTS_GET(opts, cnt, 0);
+ ids = OPTS_GET(opts, ids, NULL);
+ cookies = OPTS_GET(opts, cookies, NULL);
+
+ if (!!ids != !!cnt)
+ return libbpf_err_ptr(-EINVAL);
+ if (pattern && (ids || cookies))
+ return libbpf_err_ptr(-EINVAL);
+ if (!pattern && !ids)
+ return libbpf_err_ptr(-EINVAL);
+
+ if (pattern) {
+ cnt = collect_func_ids_by_glob(prog, pattern, &free_ids);
+ if (cnt < 0)
+ return libbpf_err_ptr(cnt);
+ if (cnt == 0)
+ return libbpf_err_ptr(-EINVAL);
+ ids = (const __u32 *) free_ids;
+ }
+
+ lopts.tracing_multi.ids = ids;
+ lopts.tracing_multi.cookies = cookies;
+ lopts.tracing_multi.cnt = cnt;
+
+ link = calloc(1, sizeof(*link));
+ if (!link) {
+ err = -ENOMEM;
+ goto error;
+ }
+ link->detach = &bpf_link__detach_fd;
+
+ link_fd = bpf_link_create(prog_fd, 0, prog->expected_attach_type, &lopts);
+ if (link_fd < 0) {
+ err = -errno;
+ pr_warn("prog '%s': failed to attach: %s\n", prog->name, errstr(err));
+ goto error;
+ }
+ link->fd = link_fd;
+ free(free_ids);
+ return link;
+
+error:
+ free(link);
+ free(free_ids);
+ return libbpf_err_ptr(err);
+}
+
+static int attach_tracing_multi(const struct bpf_program *prog, long cookie, struct bpf_link **link)
+{
+ static const char *const prefixes[] = {
+ "fentry.multi",
+ "fexit.multi",
+ "fsession.multi",
+ "fentry.multi.s",
+ "fexit.multi.s",
+ "fsession.multi.s",
+ };
+ const char *spec = NULL;
+ char *pattern;
+ size_t i;
+ int n;
+
+ *link = NULL;
+
+ for (i = 0; i < ARRAY_SIZE(prefixes); i++) {
+ size_t pfx_len;
+
+ if (!str_has_pfx(prog->sec_name, prefixes[i]))
+ continue;
+
+ pfx_len = strlen(prefixes[i]);
+ /* no auto-attach case of, e.g., SEC("fentry.multi") */
+ if (prog->sec_name[pfx_len] == '\0')
+ return 0;
+
+ if (prog->sec_name[pfx_len] != '/')
+ continue;
+
+ spec = prog->sec_name + pfx_len + 1;
+ break;
+ }
+
+ if (!spec) {
+ pr_warn("prog '%s': invalid section name '%s'\n",
+ prog->name, prog->sec_name);
+ return -EINVAL;
+ }
+
+ n = sscanf(spec, "%m[a-zA-Z0-9_.*?:]", &pattern);
+ if (n < 1) {
+ pr_warn("tracing multi pattern is invalid: %s\n", spec);
+ return -EINVAL;
+ }
+
+ *link = bpf_program__attach_tracing_multi(prog, pattern, NULL);
+ free(pattern);
+ return libbpf_get_error(*link);
+}
+
static inline int add_uprobe_event_legacy(const char *probe_name, bool retprobe,
const char *binary_path, size_t offset)
{
diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index bba4e8464396..b965ad571540 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h
@@ -726,6 +726,21 @@ bpf_program__attach_ksyscall(const struct bpf_program *prog,
const char *syscall_name,
const struct bpf_ksyscall_opts *opts);
+struct bpf_tracing_multi_opts {
+ /* size of this struct, for forward/backward compatibility */
+ size_t sz;
+ const __u32 *ids;
+ const __u64 *cookies;
+ size_t cnt;
+ size_t :0;
+};
+
+#define bpf_tracing_multi_opts__last_field cnt
+
+LIBBPF_API struct bpf_link *
+bpf_program__attach_tracing_multi(const struct bpf_program *prog, const char *pattern,
+ const struct bpf_tracing_multi_opts *opts);
+
struct bpf_uprobe_opts {
/* size of this struct, for forward/backward compatibility */
size_t sz;
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index dfed8d60af05..b731df19ae69 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -458,6 +458,7 @@ LIBBPF_1.7.0 {
LIBBPF_1.8.0 {
global:
+ bpf_program__attach_tracing_multi;
bpf_program__clone;
btf__new_empty_opts;
} LIBBPF_1.7.0;
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 21/29] selftests/bpf: Add tracing multi skel/pattern/ids attach tests
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for tracing_multi link attachment via all possible
libbpf apis - skeleton, function pattern and btf ids.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/testing/selftests/bpf/Makefile | 3 +-
.../selftests/bpf/prog_tests/tracing_multi.c | 258 ++++++++++++++++++
.../bpf/progs/tracing_multi_attach.c | 39 +++
.../selftests/bpf/progs/tracing_multi_check.c | 151 ++++++++++
4 files changed, 450 insertions(+), 1 deletion(-)
create mode 100644 tools/testing/selftests/bpf/prog_tests/tracing_multi.c
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_attach.c
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_check.c
diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index bc049620c774..396b68f862fe 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -500,7 +500,7 @@ SKEL_BLACKLIST := btf__% test_pinning_invalid.c test_sk_assign.c
LINKED_SKELS := test_static_linked.skel.h linked_funcs.skel.h \
linked_vars.skel.h linked_maps.skel.h \
test_subskeleton.skel.h test_subskeleton_lib.skel.h \
- test_usdt.skel.h
+ test_usdt.skel.h tracing_multi.skel.h
LSKELS := fexit_sleep.c trace_printk.c trace_vprintk.c map_ptr_kern.c \
core_kern.c core_kern_overflow.c test_ringbuf.c \
@@ -526,6 +526,7 @@ test_usdt.skel.h-deps := test_usdt.bpf.o test_usdt_multispec.bpf.o
xsk_xdp_progs.skel.h-deps := xsk_xdp_progs.bpf.o
xdp_hw_metadata.skel.h-deps := xdp_hw_metadata.bpf.o
xdp_features.skel.h-deps := xdp_features.bpf.o
+tracing_multi.skel.h-deps := tracing_multi_attach.bpf.o tracing_multi_check.bpf.o
LINKED_BPF_OBJS := $(foreach skel,$(LINKED_SKELS),$($(skel)-deps))
LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.c,$(LINKED_BPF_OBJS))
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
new file mode 100644
index 000000000000..f333b2514b34
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -0,0 +1,258 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <test_progs.h>
+#include <bpf/btf.h>
+#include <search.h>
+#include "bpf/libbpf_internal.h"
+#include "tracing_multi.skel.h"
+#include "trace_helpers.h"
+
+static const char * const bpf_fentry_test[] = {
+ "bpf_fentry_test1",
+ "bpf_fentry_test2",
+ "bpf_fentry_test3",
+ "bpf_fentry_test4",
+ "bpf_fentry_test5",
+ "bpf_fentry_test6",
+ "bpf_fentry_test7",
+ "bpf_fentry_test8",
+ "bpf_fentry_test9",
+ "bpf_fentry_test10",
+};
+
+#define FUNCS_CNT (ARRAY_SIZE(bpf_fentry_test))
+
+static int compare(const void *ppa, const void *ppb)
+{
+ const char *pa = *(const char **) ppa;
+ const char *pb = *(const char **) ppb;
+
+ return strcmp(pa, pb);
+}
+
+static void tdestroy_free_nop(void *ptr)
+{
+}
+
+static __u32 *get_ids(const char * const funcs[], int funcs_cnt, const char *mod)
+{
+ struct btf *btf, *vmlinux_btf = NULL;
+ __u32 nr, type_id, cnt = 0;
+ void *root = NULL;
+ __u32 *ids = NULL;
+ int i, err = 0;
+
+ btf = btf__load_vmlinux_btf();
+ if (!ASSERT_OK_PTR(btf, "btf__load_vmlinux_btf"))
+ return NULL;
+
+ if (mod) {
+ vmlinux_btf = btf;
+ btf = btf__load_module_btf(mod, vmlinux_btf);
+ if (!ASSERT_OK_PTR(btf, "btf__load_module_btf")) {
+ btf__free(vmlinux_btf);
+ return NULL;
+ }
+ }
+
+ ids = calloc(funcs_cnt, sizeof(ids[0]));
+ if (!ids)
+ goto out;
+
+ /*
+ * We sort function names by name and search them
+ * below for each function.
+ */
+ for (i = 0; i < funcs_cnt; i++) {
+ if (!tsearch(&funcs[i], &root, compare)) {
+ ASSERT_FAIL("tsearch failed");
+ err = -1;
+ goto error;
+ }
+ }
+
+ nr = btf__type_cnt(btf);
+ for (type_id = 1; type_id < nr && cnt < funcs_cnt; type_id++) {
+ const struct btf_type *type;
+ const char *str, ***val;
+ unsigned int idx;
+
+ type = btf__type_by_id(btf, type_id);
+ if (!type) {
+ err = -1;
+ break;
+ }
+
+ if (BTF_INFO_KIND(type->info) != BTF_KIND_FUNC)
+ continue;
+
+ str = btf__name_by_offset(btf, type->name_off);
+ if (!str) {
+ err = -1;
+ break;
+ }
+
+ val = tfind(&str, &root, compare);
+ if (!val)
+ continue;
+
+ /*
+ * We keep pointer for each function name so we can get the original
+ * array index and have the resulting ids array matching the original
+ * function array.
+ *
+ * Doing it this way allow us to easily test the cookies support,
+ * because each cookie is attached to particular function/id.
+ */
+ idx = *val - funcs;
+ ids[idx] = type_id;
+ cnt++;
+ }
+
+error:
+ if (err) {
+ free(ids);
+ ids = NULL;
+ }
+
+out:
+ tdestroy(root, tdestroy_free_nop);
+ btf__free(vmlinux_btf);
+ btf__free(btf);
+ return ids;
+}
+
+static void tracing_multi_test_run(struct tracing_multi *skel)
+{
+ LIBBPF_OPTS(bpf_test_run_opts, topts);
+ int err, prog_fd;
+
+ prog_fd = bpf_program__fd(skel->progs.test_fentry);
+ err = bpf_prog_test_run_opts(prog_fd, &topts);
+ ASSERT_OK(err, "test_run");
+
+ /* extra +1 count for sleepable programs */
+ ASSERT_EQ(skel->bss->test_result_fentry, FUNCS_CNT + 1, "test_result_fentry");
+ ASSERT_EQ(skel->bss->test_result_fexit, FUNCS_CNT + 1, "test_result_fexit");
+}
+
+static void test_skel_api(void)
+{
+ struct tracing_multi *skel;
+ int err;
+
+ skel = tracing_multi__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ err = tracing_multi__attach(skel);
+ if (!ASSERT_OK(err, "tracing_multi__attach"))
+ goto cleanup;
+
+ tracing_multi_test_run(skel);
+
+cleanup:
+ tracing_multi__destroy(skel);
+}
+
+static void test_link_api_pattern(void)
+{
+ struct tracing_multi *skel;
+
+ skel = tracing_multi__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ "bpf_fentry_test*", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit = bpf_program__attach_tracing_multi(skel->progs.test_fexit,
+ "bpf_fentry_test*", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fentry_s = bpf_program__attach_tracing_multi(skel->progs.test_fentry_s,
+ "bpf_fentry_test1", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry_s, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit_s = bpf_program__attach_tracing_multi(skel->progs.test_fexit_s,
+ "bpf_fentry_test1", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit_s, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ tracing_multi_test_run(skel);
+
+cleanup:
+ tracing_multi__destroy(skel);
+}
+
+static void test_link_api_ids(void)
+{
+ LIBBPF_OPTS(bpf_tracing_multi_opts, opts);
+ struct tracing_multi *skel;
+ size_t cnt = FUNCS_CNT;
+ __u32 *ids;
+
+ skel = tracing_multi__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ ids = get_ids(bpf_fentry_test, cnt, NULL);
+ if (!ASSERT_OK_PTR(ids, "get_ids"))
+ goto cleanup;
+
+ opts.ids = ids;
+ opts.cnt = cnt;
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit = bpf_program__attach_tracing_multi(skel->progs.test_fexit,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ /* Only bpf_fentry_test1 is allowed for sleepable programs. */
+ opts.cnt = 1;
+ skel->links.test_fentry_s = bpf_program__attach_tracing_multi(skel->progs.test_fentry_s,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry_s, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit_s = bpf_program__attach_tracing_multi(skel->progs.test_fexit_s,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit_s, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ tracing_multi_test_run(skel);
+
+cleanup:
+ tracing_multi__destroy(skel);
+ free(ids);
+}
+
+void test_tracing_multi_test(void)
+{
+#ifndef __x86_64__
+ test__skip();
+ return;
+#endif
+
+ if (test__start_subtest("skel_api"))
+ test_skel_api();
+ if (test__start_subtest("link_api_pattern"))
+ test_link_api_pattern();
+ if (test__start_subtest("link_api_ids"))
+ test_link_api_ids();
+}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_attach.c b/tools/testing/selftests/bpf/progs/tracing_multi_attach.c
new file mode 100644
index 000000000000..332d0a423a43
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_attach.c
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__hidden extern int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return);
+
+__u64 test_result_fentry = 0;
+__u64 test_result_fexit = 0;
+
+SEC("fentry.multi/bpf_fentry_test*")
+int BPF_PROG(test_fentry)
+{
+ tracing_multi_arg_check(ctx, &test_result_fentry, false);
+ return 0;
+}
+
+SEC("fexit.multi/bpf_fentry_test*")
+int BPF_PROG(test_fexit)
+{
+ tracing_multi_arg_check(ctx, &test_result_fexit, true);
+ return 0;
+}
+
+SEC("fentry.multi.s/bpf_fentry_test1")
+int BPF_PROG(test_fentry_s)
+{
+ tracing_multi_arg_check(ctx, &test_result_fentry, false);
+ return 0;
+}
+
+SEC("fexit.multi.s/bpf_fentry_test1")
+int BPF_PROG(test_fexit_s)
+{
+ tracing_multi_arg_check(ctx, &test_result_fexit, true);
+ return 0;
+}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_check.c b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
new file mode 100644
index 000000000000..333a3a7bae8a
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
@@ -0,0 +1,151 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+int pid = 0;
+
+/* bpf_fentry_test1 is exported as kfunc via vmlinux.h */
+extern const void bpf_fentry_test2 __ksym;
+extern const void bpf_fentry_test3 __ksym;
+extern const void bpf_fentry_test4 __ksym;
+extern const void bpf_fentry_test5 __ksym;
+extern const void bpf_fentry_test6 __ksym;
+extern const void bpf_fentry_test7 __ksym;
+extern const void bpf_fentry_test8 __ksym;
+extern const void bpf_fentry_test9 __ksym;
+extern const void bpf_fentry_test10 __ksym;
+
+int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
+{
+ void *ip = (void *) bpf_get_func_ip(ctx);
+ __u64 value = 0, ret = 0;
+ long err = 0;
+
+ if (bpf_get_current_pid_tgid() >> 32 != pid)
+ return 1;
+
+ if (is_return)
+ err |= bpf_get_func_ret(ctx, &ret);
+
+ if (ip == &bpf_fentry_test1) {
+ int a;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (int) value;
+
+ err |= is_return ? ret != 2 : 0;
+
+ *test_result += err == 0 && a == 1;
+ } else if (ip == &bpf_fentry_test2) {
+ __u64 b;
+ int a;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (int) value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = value;
+
+ err |= is_return ? ret != 5 : 0;
+
+ *test_result += err == 0 && a == 2 && b == 3;
+ } else if (ip == &bpf_fentry_test3) {
+ __u64 c;
+ char a;
+ int b;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (char) value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (int) value;
+ err |= bpf_get_func_arg(ctx, 2, &value);
+ c = value;
+
+ err |= is_return ? ret != 15 : 0;
+
+ *test_result += err == 0 && a == 4 && b == 5 && c == 6;
+ } else if (ip == &bpf_fentry_test4) {
+ void *a;
+ char b;
+ int c;
+ __u64 d;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (void *) value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (char) value;
+ err |= bpf_get_func_arg(ctx, 2, &value);
+ c = (int) value;
+ err |= bpf_get_func_arg(ctx, 3, &value);
+ d = value;
+
+ err |= is_return ? ret != 34 : 0;
+
+ *test_result += err == 0 && a == (void *) 7 && b == 8 && c == 9 && d == 10;
+ } else if (ip == &bpf_fentry_test5) {
+ __u64 a;
+ void *b;
+ short c;
+ int d;
+ __u64 e;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (void *) value;
+ err |= bpf_get_func_arg(ctx, 2, &value);
+ c = (short) value;
+ err |= bpf_get_func_arg(ctx, 3, &value);
+ d = (int) value;
+ err |= bpf_get_func_arg(ctx, 4, &value);
+ e = value;
+
+ err |= is_return ? ret != 65 : 0;
+
+ *test_result += err == 0 && a == 11 && b == (void *) 12 && c == 13 && d == 14 && e == 15;
+ } else if (ip == &bpf_fentry_test6) {
+ __u64 a;
+ void *b;
+ short c;
+ int d;
+ void *e;
+ __u64 f;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (void *) value;
+ err |= bpf_get_func_arg(ctx, 2, &value);
+ c = (short) value;
+ err |= bpf_get_func_arg(ctx, 3, &value);
+ d = (int) value;
+ err |= bpf_get_func_arg(ctx, 4, &value);
+ e = (void *) value;
+ err |= bpf_get_func_arg(ctx, 5, &value);
+ f = value;
+
+ err |= is_return ? ret != 111 : 0;
+
+ *test_result += err == 0 && a == 16 && b == (void *) 17 && c == 18 && d == 19 && e == (void *) 20 && f == 21;
+ } else if (ip == &bpf_fentry_test7) {
+ err |= is_return ? ret != 0 : 0;
+
+ *test_result += err == 0 ? 1 : 0;
+ } else if (ip == &bpf_fentry_test8) {
+ err |= is_return ? ret != 0 : 0;
+
+ *test_result += err == 0 ? 1 : 0;
+ } else if (ip == &bpf_fentry_test9) {
+ err |= is_return ? ret != 0 : 0;
+
+ *test_result += err == 0 ? 1 : 0;
+ } else if (ip == &bpf_fentry_test10) {
+ err |= is_return ? ret != 0 : 0;
+
+ *test_result += err == 0 ? 1 : 0;
+ }
+
+ return 0;
+}
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 22/29] selftests/bpf: Add tracing multi skel/pattern/ids module attach tests
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for tracing_multi link attachment via all possible
libbpf apis - skeleton, function pattern and btf ids on top of
bpf_testmod kernel module.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/testing/selftests/bpf/Makefile | 4 +-
.../selftests/bpf/prog_tests/tracing_multi.c | 105 ++++++++++++++++++
.../bpf/progs/tracing_multi_attach_module.c | 25 +++++
.../selftests/bpf/progs/tracing_multi_check.c | 50 +++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_attach_module.c
diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index 396b68f862fe..fcfeb0aeff11 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -500,7 +500,8 @@ SKEL_BLACKLIST := btf__% test_pinning_invalid.c test_sk_assign.c
LINKED_SKELS := test_static_linked.skel.h linked_funcs.skel.h \
linked_vars.skel.h linked_maps.skel.h \
test_subskeleton.skel.h test_subskeleton_lib.skel.h \
- test_usdt.skel.h tracing_multi.skel.h
+ test_usdt.skel.h tracing_multi.skel.h \
+ tracing_multi_module.skel.h
LSKELS := fexit_sleep.c trace_printk.c trace_vprintk.c map_ptr_kern.c \
core_kern.c core_kern_overflow.c test_ringbuf.c \
@@ -527,6 +528,7 @@ xsk_xdp_progs.skel.h-deps := xsk_xdp_progs.bpf.o
xdp_hw_metadata.skel.h-deps := xdp_hw_metadata.bpf.o
xdp_features.skel.h-deps := xdp_features.bpf.o
tracing_multi.skel.h-deps := tracing_multi_attach.bpf.o tracing_multi_check.bpf.o
+tracing_multi_module.skel.h-deps := tracing_multi_attach_module.bpf.o tracing_multi_check.bpf.o
LINKED_BPF_OBJS := $(foreach skel,$(LINKED_SKELS),$($(skel)-deps))
LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.c,$(LINKED_BPF_OBJS))
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index f333b2514b34..77134f1e2dc3 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -5,6 +5,7 @@
#include <search.h>
#include "bpf/libbpf_internal.h"
#include "tracing_multi.skel.h"
+#include "tracing_multi_module.skel.h"
#include "trace_helpers.h"
static const char * const bpf_fentry_test[] = {
@@ -20,6 +21,14 @@ static const char * const bpf_fentry_test[] = {
"bpf_fentry_test10",
};
+static const char * const bpf_testmod_fentry_test[] = {
+ "bpf_testmod_fentry_test1",
+ "bpf_testmod_fentry_test2",
+ "bpf_testmod_fentry_test3",
+ "bpf_testmod_fentry_test7",
+ "bpf_testmod_fentry_test11",
+};
+
#define FUNCS_CNT (ARRAY_SIZE(bpf_fentry_test))
static int compare(const void *ppa, const void *ppb)
@@ -242,6 +251,96 @@ static void test_link_api_ids(void)
free(ids);
}
+static void test_module_skel_api(void)
+{
+ struct tracing_multi_module *skel = NULL;
+ int err;
+
+ skel = tracing_multi_module__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ err = tracing_multi_module__attach(skel);
+ if (!ASSERT_OK(err, "tracing_multi__attach"))
+ goto cleanup;
+
+ ASSERT_OK(trigger_module_test_read(1), "trigger_read");
+ ASSERT_EQ(skel->bss->test_result_fentry, 5, "test_result_fentry");
+ ASSERT_EQ(skel->bss->test_result_fexit, 5, "test_result_fexit");
+
+cleanup:
+ tracing_multi_module__destroy(skel);
+}
+
+static void test_module_link_api_pattern(void)
+{
+ struct tracing_multi_module *skel = NULL;
+
+ skel = tracing_multi_module__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi_module__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ "bpf_testmod:bpf_testmod_fentry_test*", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit = bpf_program__attach_tracing_multi(skel->progs.test_fexit,
+ "bpf_testmod:bpf_testmod_fentry_test*", NULL);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ ASSERT_OK(trigger_module_test_read(1), "trigger_read");
+ ASSERT_EQ(skel->bss->test_result_fentry, 5, "test_result_fentry");
+ ASSERT_EQ(skel->bss->test_result_fexit, 5, "test_result_fexit");
+
+cleanup:
+ tracing_multi_module__destroy(skel);
+}
+
+static void test_module_link_api_ids(void)
+{
+ size_t cnt = ARRAY_SIZE(bpf_testmod_fentry_test);
+ LIBBPF_OPTS(bpf_tracing_multi_opts, opts);
+ struct tracing_multi_module *skel = NULL;
+ __u32 *ids;
+
+ skel = tracing_multi_module__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi_module__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ ids = get_ids(bpf_testmod_fentry_test, cnt, "bpf_testmod");
+ if (!ASSERT_OK_PTR(ids, "get_ids"))
+ goto cleanup;
+
+ opts.ids = ids;
+ opts.cnt = cnt;
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fentry, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ skel->links.test_fexit = bpf_program__attach_tracing_multi(skel->progs.test_fexit,
+ NULL, &opts);
+ if (!ASSERT_OK_PTR(skel->links.test_fexit, "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ ASSERT_OK(trigger_module_test_read(1), "trigger_read");
+ ASSERT_EQ(skel->bss->test_result_fentry, 5, "test_result_fentry");
+ ASSERT_EQ(skel->bss->test_result_fexit, 5, "test_result_fexit");
+
+cleanup:
+ tracing_multi_module__destroy(skel);
+ free(ids);
+}
+
void test_tracing_multi_test(void)
{
#ifndef __x86_64__
@@ -255,4 +354,10 @@ void test_tracing_multi_test(void)
test_link_api_pattern();
if (test__start_subtest("link_api_ids"))
test_link_api_ids();
+ if (test__start_subtest("module_skel_api"))
+ test_module_skel_api();
+ if (test__start_subtest("module_link_api_pattern"))
+ test_module_link_api_pattern();
+ if (test__start_subtest("module_link_api_ids"))
+ test_module_link_api_ids();
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_attach_module.c b/tools/testing/selftests/bpf/progs/tracing_multi_attach_module.c
new file mode 100644
index 000000000000..b3374f2db450
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_attach_module.c
@@ -0,0 +1,25 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__hidden extern int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return);
+
+__u64 test_result_fentry = 0;
+__u64 test_result_fexit = 0;
+
+SEC("fentry.multi/bpf_testmod:bpf_testmod_fentry_test*")
+int BPF_PROG(test_fentry)
+{
+ tracing_multi_arg_check(ctx, &test_result_fentry, false);
+ return 0;
+}
+
+SEC("fexit.multi/bpf_testmod:bpf_testmod_fentry_test*")
+int BPF_PROG(test_fexit)
+{
+ tracing_multi_arg_check(ctx, &test_result_fexit, true);
+ return 0;
+}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_check.c b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
index 333a3a7bae8a..7ede84c50cb6 100644
--- a/tools/testing/selftests/bpf/progs/tracing_multi_check.c
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
@@ -18,6 +18,12 @@ extern const void bpf_fentry_test8 __ksym;
extern const void bpf_fentry_test9 __ksym;
extern const void bpf_fentry_test10 __ksym;
+extern const void bpf_testmod_fentry_test1 __ksym;
+extern const void bpf_testmod_fentry_test2 __ksym;
+extern const void bpf_testmod_fentry_test3 __ksym;
+extern const void bpf_testmod_fentry_test7 __ksym;
+extern const void bpf_testmod_fentry_test11 __ksym;
+
int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
{
void *ip = (void *) bpf_get_func_ip(ctx);
@@ -145,6 +151,50 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
err |= is_return ? ret != 0 : 0;
*test_result += err == 0 ? 1 : 0;
+ } else if (ip == &bpf_testmod_fentry_test1) {
+ int a;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (int) value;
+
+ err |= is_return ? ret != 2 : 0;
+
+ *test_result += err == 0 && a == 1;
+ } else if (ip == &bpf_testmod_fentry_test2) {
+ int a;
+ __u64 b;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (int) value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (__u64) value;
+
+ err |= is_return ? ret != 5 : 0;
+
+ *test_result += err == 0 && a == 2 && b == 3;
+ } else if (ip == &bpf_testmod_fentry_test3) {
+ char a;
+ int b;
+ __u64 c;
+
+ err |= bpf_get_func_arg(ctx, 0, &value);
+ a = (char) value;
+ err |= bpf_get_func_arg(ctx, 1, &value);
+ b = (int) value;
+ err |= bpf_get_func_arg(ctx, 2, &value);
+ c = (__u64) value;
+
+ err |= is_return ? ret != 15 : 0;
+
+ *test_result += err == 0 && a == 4 && b == 5 && c == 6;
+ } else if (ip == &bpf_testmod_fentry_test7) {
+ err |= is_return ? ret != 133 : 0;
+
+ *test_result += err == 0;
+ } else if (ip == &bpf_testmod_fentry_test11) {
+ err |= is_return ? ret != 231 : 0;
+
+ *test_result += err == 0;
}
return 0;
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 23/29] selftests/bpf: Add tracing multi intersect tests
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tracing multi tests for intersecting attached functions.
Using bits from (from 1 to 16 values) to specify (up to 4) attached
programs, and randomly choosing bpf_fentry_test* functions they are
attached to.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/testing/selftests/bpf/Makefile | 4 +-
.../selftests/bpf/prog_tests/tracing_multi.c | 99 +++++++++++++++++++
.../progs/tracing_multi_intersect_attach.c | 41 ++++++++
3 files changed, 143 insertions(+), 1 deletion(-)
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_intersect_attach.c
diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index fcfeb0aeff11..95b87cba46bc 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -501,7 +501,8 @@ LINKED_SKELS := test_static_linked.skel.h linked_funcs.skel.h \
linked_vars.skel.h linked_maps.skel.h \
test_subskeleton.skel.h test_subskeleton_lib.skel.h \
test_usdt.skel.h tracing_multi.skel.h \
- tracing_multi_module.skel.h
+ tracing_multi_module.skel.h \
+ tracing_multi_intersect.skel.h
LSKELS := fexit_sleep.c trace_printk.c trace_vprintk.c map_ptr_kern.c \
core_kern.c core_kern_overflow.c test_ringbuf.c \
@@ -529,6 +530,7 @@ xdp_hw_metadata.skel.h-deps := xdp_hw_metadata.bpf.o
xdp_features.skel.h-deps := xdp_features.bpf.o
tracing_multi.skel.h-deps := tracing_multi_attach.bpf.o tracing_multi_check.bpf.o
tracing_multi_module.skel.h-deps := tracing_multi_attach_module.bpf.o tracing_multi_check.bpf.o
+tracing_multi_intersect.skel.h-deps := tracing_multi_intersect_attach.bpf.o tracing_multi_check.bpf.o
LINKED_BPF_OBJS := $(foreach skel,$(LINKED_SKELS),$($(skel)-deps))
LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.c,$(LINKED_BPF_OBJS))
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index 77134f1e2dc3..4dd610e74f9a 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -6,6 +6,7 @@
#include "bpf/libbpf_internal.h"
#include "tracing_multi.skel.h"
#include "tracing_multi_module.skel.h"
+#include "tracing_multi_intersect.skel.h"
#include "trace_helpers.h"
static const char * const bpf_fentry_test[] = {
@@ -31,6 +32,20 @@ static const char * const bpf_testmod_fentry_test[] = {
#define FUNCS_CNT (ARRAY_SIZE(bpf_fentry_test))
+static int get_random_funcs(const char **funcs)
+{
+ int i, cnt = 0;
+
+ for (i = 0; i < FUNCS_CNT; i++) {
+ if (rand() % 2)
+ funcs[cnt++] = bpf_fentry_test[i];
+ }
+ /* we always need at least one.. */
+ if (!cnt)
+ funcs[cnt++] = bpf_fentry_test[rand() % FUNCS_CNT];
+ return cnt;
+}
+
static int compare(const void *ppa, const void *ppb)
{
const char *pa = *(const char **) ppa;
@@ -341,6 +356,88 @@ static void test_module_link_api_ids(void)
free(ids);
}
+static bool is_set(__u32 mask, __u32 bit)
+{
+ return (1 << bit) & mask;
+}
+
+static void __test_intersect(__u32 mask, const struct bpf_program *progs[4], __u64 *test_results[4])
+{
+ LIBBPF_OPTS(bpf_tracing_multi_opts, opts);
+ LIBBPF_OPTS(bpf_test_run_opts, topts);
+ struct bpf_link *links[4] = { NULL };
+ const char *funcs[FUNCS_CNT];
+ __u64 expected[4];
+ __u32 *ids, i;
+ int err, cnt;
+
+ /*
+ * We have 4 programs in progs and the mask bits pick which
+ * of them gets attached to randomly chosen functions.
+ */
+ for (i = 0; i < 4; i++) {
+ if (!is_set(mask, i))
+ continue;
+
+ cnt = get_random_funcs(funcs);
+ ids = get_ids(funcs, cnt, NULL);
+ if (!ASSERT_OK_PTR(ids, "get_ids"))
+ goto cleanup;
+
+ opts.ids = ids;
+ opts.cnt = cnt;
+ links[i] = bpf_program__attach_tracing_multi(progs[i], NULL, &opts);
+ free(ids);
+
+ if (!ASSERT_OK_PTR(links[i], "bpf_program__attach_tracing_multi"))
+ goto cleanup;
+
+ expected[i] = *test_results[i] + cnt;
+ }
+
+ err = bpf_prog_test_run_opts(bpf_program__fd(progs[0]), &topts);
+ ASSERT_OK(err, "test_run");
+
+ for (i = 0; i < 4; i++) {
+ if (!is_set(mask, i))
+ continue;
+ ASSERT_EQ(*test_results[i], expected[i], "test_results");
+ }
+
+cleanup:
+ for (i = 0; i < 4; i++)
+ bpf_link__destroy(links[i]);
+}
+
+static void test_intersect(void)
+{
+ struct tracing_multi_intersect *skel;
+ const struct bpf_program *progs[4];
+ __u64 *test_results[4];
+ __u32 i;
+
+ skel = tracing_multi_intersect__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi_intersect__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ progs[0] = skel->progs.fentry_1;
+ progs[1] = skel->progs.fexit_1;
+ progs[2] = skel->progs.fentry_2;
+ progs[3] = skel->progs.fexit_2;
+
+ test_results[0] = &skel->bss->test_result_fentry_1;
+ test_results[1] = &skel->bss->test_result_fexit_1;
+ test_results[2] = &skel->bss->test_result_fentry_2;
+ test_results[3] = &skel->bss->test_result_fexit_2;
+
+ for (i = 1; i < 16; i++)
+ __test_intersect(i, progs, test_results);
+
+ tracing_multi_intersect__destroy(skel);
+}
+
void test_tracing_multi_test(void)
{
#ifndef __x86_64__
@@ -360,4 +457,6 @@ void test_tracing_multi_test(void)
test_module_link_api_pattern();
if (test__start_subtest("module_link_api_ids"))
test_module_link_api_ids();
+ if (test__start_subtest("intersect"))
+ test_intersect();
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_intersect_attach.c b/tools/testing/selftests/bpf/progs/tracing_multi_intersect_attach.c
new file mode 100644
index 000000000000..cd5be0bb6ffd
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_intersect_attach.c
@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__hidden extern int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return);
+
+__u64 test_result_fentry_1 = 0;
+__u64 test_result_fentry_2 = 0;
+__u64 test_result_fexit_1 = 0;
+__u64 test_result_fexit_2 = 0;
+
+SEC("fentry.multi")
+int BPF_PROG(fentry_1)
+{
+ tracing_multi_arg_check(ctx, &test_result_fentry_1, false);
+ return 0;
+}
+
+SEC("fentry.multi")
+int BPF_PROG(fentry_2)
+{
+ tracing_multi_arg_check(ctx, &test_result_fentry_2, false);
+ return 0;
+}
+
+SEC("fexit.multi")
+int BPF_PROG(fexit_1)
+{
+ tracing_multi_arg_check(ctx, &test_result_fexit_1, true);
+ return 0;
+}
+
+SEC("fexit.multi")
+int BPF_PROG(fexit_2)
+{
+ tracing_multi_arg_check(ctx, &test_result_fexit_2, true);
+ return 0;
+}
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 24/29] selftests/bpf: Add tracing multi cookies test
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for using cookies on tracing multi link.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/tracing_multi.c | 23 +++++++++++++++++--
.../selftests/bpf/progs/tracing_multi_check.c | 15 +++++++++++-
2 files changed, 35 insertions(+), 3 deletions(-)
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index 4dd610e74f9a..0f066063cb82 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -9,6 +9,19 @@
#include "tracing_multi_intersect.skel.h"
#include "trace_helpers.h"
+static __u64 bpf_fentry_test_cookies[] = {
+ 8, /* bpf_fentry_test1 */
+ 9, /* bpf_fentry_test2 */
+ 7, /* bpf_fentry_test3 */
+ 5, /* bpf_fentry_test4 */
+ 4, /* bpf_fentry_test5 */
+ 2, /* bpf_fentry_test6 */
+ 3, /* bpf_fentry_test7 */
+ 1, /* bpf_fentry_test8 */
+ 10, /* bpf_fentry_test9 */
+ 6, /* bpf_fentry_test10 */
+};
+
static const char * const bpf_fentry_test[] = {
"bpf_fentry_test1",
"bpf_fentry_test2",
@@ -217,7 +230,7 @@ static void test_link_api_pattern(void)
tracing_multi__destroy(skel);
}
-static void test_link_api_ids(void)
+static void test_link_api_ids(bool test_cookies)
{
LIBBPF_OPTS(bpf_tracing_multi_opts, opts);
struct tracing_multi *skel;
@@ -229,6 +242,7 @@ static void test_link_api_ids(void)
return;
skel->bss->pid = getpid();
+ skel->bss->test_cookies = test_cookies;
ids = get_ids(bpf_fentry_test, cnt, NULL);
if (!ASSERT_OK_PTR(ids, "get_ids"))
@@ -237,6 +251,9 @@ static void test_link_api_ids(void)
opts.ids = ids;
opts.cnt = cnt;
+ if (test_cookies)
+ opts.cookies = bpf_fentry_test_cookies;
+
skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
NULL, &opts);
if (!ASSERT_OK_PTR(skel->links.test_fentry, "bpf_program__attach_tracing_multi"))
@@ -450,7 +467,7 @@ void test_tracing_multi_test(void)
if (test__start_subtest("link_api_pattern"))
test_link_api_pattern();
if (test__start_subtest("link_api_ids"))
- test_link_api_ids();
+ test_link_api_ids(false);
if (test__start_subtest("module_skel_api"))
test_module_skel_api();
if (test__start_subtest("module_link_api_pattern"))
@@ -459,4 +476,6 @@ void test_tracing_multi_test(void)
test_module_link_api_ids();
if (test__start_subtest("intersect"))
test_intersect();
+ if (test__start_subtest("cookies"))
+ test_link_api_ids(true);
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_check.c b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
index 7ede84c50cb6..b2959ba71179 100644
--- a/tools/testing/selftests/bpf/progs/tracing_multi_check.c
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_check.c
@@ -6,6 +6,7 @@
char _license[] SEC("license") = "GPL";
int pid = 0;
+bool test_cookies = false;
/* bpf_fentry_test1 is exported as kfunc via vmlinux.h */
extern const void bpf_fentry_test2 __ksym;
@@ -27,7 +28,7 @@ extern const void bpf_testmod_fentry_test11 __ksym;
int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
{
void *ip = (void *) bpf_get_func_ip(ctx);
- __u64 value = 0, ret = 0;
+ __u64 value = 0, ret = 0, cookie = 0;
long err = 0;
if (bpf_get_current_pid_tgid() >> 32 != pid)
@@ -35,6 +36,8 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
if (is_return)
err |= bpf_get_func_ret(ctx, &ret);
+ if (test_cookies)
+ cookie = bpf_get_attach_cookie(ctx);
if (ip == &bpf_fentry_test1) {
int a;
@@ -43,6 +46,7 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
a = (int) value;
err |= is_return ? ret != 2 : 0;
+ err |= test_cookies ? cookie != 8 : 0;
*test_result += err == 0 && a == 1;
} else if (ip == &bpf_fentry_test2) {
@@ -55,6 +59,7 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
b = value;
err |= is_return ? ret != 5 : 0;
+ err |= test_cookies ? cookie != 9 : 0;
*test_result += err == 0 && a == 2 && b == 3;
} else if (ip == &bpf_fentry_test3) {
@@ -70,6 +75,7 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
c = value;
err |= is_return ? ret != 15 : 0;
+ err |= test_cookies ? cookie != 7 : 0;
*test_result += err == 0 && a == 4 && b == 5 && c == 6;
} else if (ip == &bpf_fentry_test4) {
@@ -88,6 +94,7 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
d = value;
err |= is_return ? ret != 34 : 0;
+ err |= test_cookies ? cookie != 5 : 0;
*test_result += err == 0 && a == (void *) 7 && b == 8 && c == 9 && d == 10;
} else if (ip == &bpf_fentry_test5) {
@@ -109,6 +116,7 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
e = value;
err |= is_return ? ret != 65 : 0;
+ err |= test_cookies ? cookie != 4 : 0;
*test_result += err == 0 && a == 11 && b == (void *) 12 && c == 13 && d == 14 && e == 15;
} else if (ip == &bpf_fentry_test6) {
@@ -133,22 +141,27 @@ int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return)
f = value;
err |= is_return ? ret != 111 : 0;
+ err |= test_cookies ? cookie != 2 : 0;
*test_result += err == 0 && a == 16 && b == (void *) 17 && c == 18 && d == 19 && e == (void *) 20 && f == 21;
} else if (ip == &bpf_fentry_test7) {
err |= is_return ? ret != 0 : 0;
+ err |= test_cookies ? cookie != 3 : 0;
*test_result += err == 0 ? 1 : 0;
} else if (ip == &bpf_fentry_test8) {
err |= is_return ? ret != 0 : 0;
+ err |= test_cookies ? cookie != 1 : 0;
*test_result += err == 0 ? 1 : 0;
} else if (ip == &bpf_fentry_test9) {
err |= is_return ? ret != 0 : 0;
+ err |= test_cookies ? cookie != 10 : 0;
*test_result += err == 0 ? 1 : 0;
} else if (ip == &bpf_fentry_test10) {
err |= is_return ? ret != 0 : 0;
+ err |= test_cookies ? cookie != 6 : 0;
*test_result += err == 0 ? 1 : 0;
} else if (ip == &bpf_testmod_fentry_test1) {
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 25/29] selftests/bpf: Add tracing multi session test
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for tracing multi link session.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
tools/testing/selftests/bpf/Makefile | 4 +-
.../selftests/bpf/prog_tests/tracing_multi.c | 45 +++++++++++++
.../bpf/progs/tracing_multi_session_attach.c | 65 +++++++++++++++++++
3 files changed, 113 insertions(+), 1 deletion(-)
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_session_attach.c
diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index 95b87cba46bc..30ad24bb7054 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -502,7 +502,8 @@ LINKED_SKELS := test_static_linked.skel.h linked_funcs.skel.h \
test_subskeleton.skel.h test_subskeleton_lib.skel.h \
test_usdt.skel.h tracing_multi.skel.h \
tracing_multi_module.skel.h \
- tracing_multi_intersect.skel.h
+ tracing_multi_intersect.skel.h \
+ tracing_multi_session.skel.h
LSKELS := fexit_sleep.c trace_printk.c trace_vprintk.c map_ptr_kern.c \
core_kern.c core_kern_overflow.c test_ringbuf.c \
@@ -531,6 +532,7 @@ xdp_features.skel.h-deps := xdp_features.bpf.o
tracing_multi.skel.h-deps := tracing_multi_attach.bpf.o tracing_multi_check.bpf.o
tracing_multi_module.skel.h-deps := tracing_multi_attach_module.bpf.o tracing_multi_check.bpf.o
tracing_multi_intersect.skel.h-deps := tracing_multi_intersect_attach.bpf.o tracing_multi_check.bpf.o
+tracing_multi_session.skel.h-deps := tracing_multi_session_attach.bpf.o tracing_multi_check.bpf.o
LINKED_BPF_OBJS := $(foreach skel,$(LINKED_SKELS),$($(skel)-deps))
LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.c,$(LINKED_BPF_OBJS))
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index 0f066063cb82..05683b8d0680 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -7,6 +7,7 @@
#include "tracing_multi.skel.h"
#include "tracing_multi_module.skel.h"
#include "tracing_multi_intersect.skel.h"
+#include "tracing_multi_session.skel.h"
#include "trace_helpers.h"
static __u64 bpf_fentry_test_cookies[] = {
@@ -455,6 +456,48 @@ static void test_intersect(void)
tracing_multi_intersect__destroy(skel);
}
+static void test_session(void)
+{
+ LIBBPF_OPTS(bpf_test_run_opts, topts);
+ struct tracing_multi_session *skel;
+ int err, prog_fd;
+
+ skel = tracing_multi_session__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi_session__open_and_load"))
+ return;
+
+ skel->bss->pid = getpid();
+
+ err = tracing_multi_session__attach(skel);
+ if (!ASSERT_OK(err, "tracing_multi_session__attach"))
+ goto cleanup;
+
+ /* execute kernel session */
+ prog_fd = bpf_program__fd(skel->progs.test_session_1);
+ err = bpf_prog_test_run_opts(prog_fd, &topts);
+ ASSERT_OK(err, "test_run");
+
+ /* 10 for test_session_1, 1 for test_fsession_s */
+ ASSERT_EQ(skel->bss->test_result_fentry, 11, "test_result_fentry");
+ /* extra count (+1 for each fexit execution) for test_result_fexit cookie check/inc */
+ ASSERT_EQ(skel->bss->test_result_fexit, 22, "test_result_fexit");
+
+ skel->bss->test_result_fentry = 0;
+ skel->bss->test_result_fexit = 0;
+
+ /* execute bpf_testmo.ko session */
+ ASSERT_OK(trigger_module_test_read(1), "trigger_read");
+
+ /* 5 for test_session_2 */
+ ASSERT_EQ(skel->bss->test_result_fentry, 5, "test_result_fentry");
+ /* extra count (+1 for each fexit execution) for test_result_fexit cookie */
+ ASSERT_EQ(skel->bss->test_result_fexit, 10, "test_result_fexit");
+
+
+cleanup:
+ tracing_multi_session__destroy(skel);
+}
+
void test_tracing_multi_test(void)
{
#ifndef __x86_64__
@@ -478,4 +521,6 @@ void test_tracing_multi_test(void)
test_intersect();
if (test__start_subtest("cookies"))
test_link_api_ids(true);
+ if (test__start_subtest("session"))
+ test_session();
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_session_attach.c b/tools/testing/selftests/bpf/progs/tracing_multi_session_attach.c
new file mode 100644
index 000000000000..7c9a46016ccd
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_session_attach.c
@@ -0,0 +1,65 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+__hidden extern int tracing_multi_arg_check(__u64 *ctx, __u64 *test_result, bool is_return);
+
+__u64 test_result_fentry = 0;
+__u64 test_result_fexit = 0;
+
+SEC("fsession.multi/bpf_fentry_test*")
+int BPF_PROG(test_session_1)
+{
+ volatile __u64 *cookie = bpf_session_cookie(ctx);
+
+ if (bpf_session_is_return(ctx)) {
+ if (tracing_multi_arg_check(ctx, &test_result_fexit, true))
+ return 0;
+ /* extra count for test_result_fexit cookie */
+ test_result_fexit += *cookie == 0xbeafbeafbeafbeaf;
+ } else {
+ if (tracing_multi_arg_check(ctx, &test_result_fentry, false))
+ return 0;
+ *cookie = 0xbeafbeafbeafbeaf;
+ }
+ return 0;
+}
+
+SEC("fsession.multi.s/bpf_fentry_test1")
+int BPF_PROG(test_fsession_s)
+{
+ volatile __u64 *cookie = bpf_session_cookie(ctx);
+
+ if (bpf_session_is_return(ctx)) {
+ if (tracing_multi_arg_check(ctx, &test_result_fexit, true))
+ return 0;
+ /* extra count for test_result_fexit cookie */
+ test_result_fexit += *cookie == 0xbeafbeafbeafbeaf;
+ } else {
+ if (tracing_multi_arg_check(ctx, &test_result_fentry, false))
+ return 0;
+ *cookie = 0xbeafbeafbeafbeaf;
+ }
+ return 0;
+}
+
+SEC("fsession.multi/bpf_testmod:bpf_testmod_fentry_test*")
+int BPF_PROG(test_session_2)
+{
+ volatile __u64 *cookie = bpf_session_cookie(ctx);
+
+ if (bpf_session_is_return(ctx)) {
+ if (tracing_multi_arg_check(ctx, &test_result_fexit, true))
+ return 0;
+ /* extra count for test_result_fexit cookie */
+ test_result_fexit += *cookie == 0xbeafbeafbeafbeaf;
+ } else {
+ if (tracing_multi_arg_check(ctx, &test_result_fentry, false))
+ return 0;
+ *cookie = 0xbeafbeafbeafbeaf;
+ }
+ return 0;
+}
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 26/29] selftests/bpf: Add tracing multi attach fails test
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for attach fails on tracing multi link.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/tracing_multi.c | 96 +++++++++++++++++++
.../selftests/bpf/progs/tracing_multi_fail.c | 18 ++++
2 files changed, 114 insertions(+)
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_fail.c
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index 05683b8d0680..7e1bb071ce2a 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -8,6 +8,7 @@
#include "tracing_multi_module.skel.h"
#include "tracing_multi_intersect.skel.h"
#include "tracing_multi_session.skel.h"
+#include "tracing_multi_fail.skel.h"
#include "trace_helpers.h"
static __u64 bpf_fentry_test_cookies[] = {
@@ -498,6 +499,99 @@ static void test_session(void)
tracing_multi_session__destroy(skel);
}
+static void test_attach_api_fails(void)
+{
+ LIBBPF_OPTS(bpf_tracing_multi_opts, opts);
+ static const char * const func[] = {
+ "bpf_fentry_test2",
+ };
+ struct tracing_multi_fail *skel = NULL;
+ __u32 ids[2] = {}, *ids2 = NULL;
+ __u64 cookies[2];
+
+ skel = tracing_multi_fail__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "tracing_multi_fail__open_and_load"))
+ return;
+
+ /* fail#1 (libbpf) pattern and opts NULL */
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ NULL, NULL);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -EINVAL, "fail_1"))
+ goto cleanup;
+
+ /* fail#2 (libbpf) pattern and ids */
+ LIBBPF_OPTS_RESET(opts,
+ .ids = ids,
+ .cnt = 2,
+ );
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ "bpf_fentry_test*", &opts);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -EINVAL, "fail_2"))
+ goto cleanup;
+
+ /* fail#3 (libbpf) pattern and cookies */
+ LIBBPF_OPTS_RESET(opts,
+ .ids = NULL,
+ .cnt = 2,
+ .cookies = cookies,
+ );
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ "bpf_fentry_test*", &opts);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -EINVAL, "fail_3"))
+ goto cleanup;
+
+ /* fail#4 (libbpf) bogus pattern */
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ "bpf_not_really_a_function*", NULL);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -EINVAL, "fail_4"))
+ goto cleanup;
+
+ /* fail#5 (kernel) abnormal cnt */
+ LIBBPF_OPTS_RESET(opts,
+ .ids = ids,
+ .cnt = INT_MAX,
+ );
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ NULL, &opts);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -E2BIG, "fail_5"))
+ goto cleanup;
+
+ /* fail#6 (kernel) attach sleepable program to not-allowed function */
+ ids2 = get_ids(func, 1, NULL);
+ if (!ASSERT_OK_PTR(ids2, "get_ids"))
+ goto cleanup;
+
+ LIBBPF_OPTS_RESET(opts,
+ .ids = ids2,
+ .cnt = 1,
+ );
+
+ skel->links.test_fentry_s = bpf_program__attach_tracing_multi(skel->progs.test_fentry_s,
+ NULL, &opts);
+ if (!ASSERT_EQ(libbpf_get_error(skel->links.test_fentry_s), -EINVAL, "fail_6"))
+ goto cleanup;
+
+ /* fail#7 (kernel) attach with duplicate id */
+ ids[0] = ids2[0];
+ ids[1] = ids2[0];
+
+ LIBBPF_OPTS_RESET(opts,
+ .ids = ids,
+ .cnt = 2,
+ );
+
+ skel->links.test_fentry = bpf_program__attach_tracing_multi(skel->progs.test_fentry,
+ NULL, &opts);
+ ASSERT_EQ(libbpf_get_error(skel->links.test_fentry), -EINVAL, "fail_7");
+
+cleanup:
+ tracing_multi_fail__destroy(skel);
+ free(ids2);
+}
+
void test_tracing_multi_test(void)
{
#ifndef __x86_64__
@@ -523,4 +617,6 @@ void test_tracing_multi_test(void)
test_link_api_ids(true);
if (test__start_subtest("session"))
test_session();
+ if (test__start_subtest("attach_api_fails"))
+ test_attach_api_fails();
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_fail.c b/tools/testing/selftests/bpf/progs/tracing_multi_fail.c
new file mode 100644
index 000000000000..7f0375f4213d
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_fail.c
@@ -0,0 +1,18 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+SEC("fentry.multi")
+int BPF_PROG(test_fentry)
+{
+ return 0;
+}
+
+SEC("fentry.multi.s")
+int BPF_PROG(test_fentry_s)
+{
+ return 0;
+}
--
2.54.0
^ permalink raw reply related
* [PATCHv7 bpf-next 27/29] selftests/bpf: Add tracing multi verifier fails test
From: Jiri Olsa @ 2026-06-03 11:05 UTC (permalink / raw)
To: Alexei Starovoitov, Daniel Borkmann, Andrii Nakryiko
Cc: bpf, linux-trace-kernel, Martin KaFai Lau, Eduard Zingerman,
Song Liu, Yonghong Song, Menglong Dong, Steven Rostedt
In-Reply-To: <20260603110554.29590-1-jolsa@kernel.org>
Adding tests for verifier fails on tracing multi programs.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
.../selftests/bpf/prog_tests/tracing_multi.c | 2 ++
.../bpf/progs/tracing_multi_verifier.c | 31 +++++++++++++++++++
2 files changed, 33 insertions(+)
create mode 100644 tools/testing/selftests/bpf/progs/tracing_multi_verifier.c
diff --git a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
index 7e1bb071ce2a..9e026f2b254d 100644
--- a/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
+++ b/tools/testing/selftests/bpf/prog_tests/tracing_multi.c
@@ -9,6 +9,7 @@
#include "tracing_multi_intersect.skel.h"
#include "tracing_multi_session.skel.h"
#include "tracing_multi_fail.skel.h"
+#include "tracing_multi_verifier.skel.h"
#include "trace_helpers.h"
static __u64 bpf_fentry_test_cookies[] = {
@@ -619,4 +620,5 @@ void test_tracing_multi_test(void)
test_session();
if (test__start_subtest("attach_api_fails"))
test_attach_api_fails();
+ RUN_TESTS(tracing_multi_verifier);
}
diff --git a/tools/testing/selftests/bpf/progs/tracing_multi_verifier.c b/tools/testing/selftests/bpf/progs/tracing_multi_verifier.c
new file mode 100644
index 000000000000..7b6ed41bf452
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/tracing_multi_verifier.c
@@ -0,0 +1,31 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+#include "bpf_misc.h"
+
+char _license[] SEC("license") = "GPL";
+
+SEC("fentry.multi/bpf_fentry_test1")
+__failure
+__msg("func 'bpf_multi_func' doesn't have 1-th argument")
+int BPF_PROG(fentry_direct_access, int a)
+{
+ return a;
+}
+
+SEC("fexit.multi/bpf_fentry_test3")
+__failure
+__msg("invalid bpf_context access off=24 size=8")
+int BPF_PROG(fexit_direct_access, char a, int b, __u64 c, int ret)
+{
+ return ret;
+}
+
+SEC("fsession.multi/bpf_fentry_test4")
+__failure
+__msg("invalid bpf_context access off=16 size=8")
+int BPF_PROG(fsession_direct_access, void *a, char b, int c, __u64 d, int ret)
+{
+ return c;
+}
--
2.54.0
^ permalink raw reply related
* Re: [PATCHv7 bpf-next 13/29] bpf: Add support for tracing multi link
From: bot+bpf-ci @ 2026-06-03 11:52 UTC (permalink / raw)
To: jolsa, ast, daniel, andrii
Cc: bpf, linux-trace-kernel, kafai, eddyz87, songliubraving, yhs,
menglong8.dong, rostedt, ast, andrii, daniel, martin.lau, eddyz87,
yonghong.song, clm, ihor.solodrai
In-Reply-To: <20260603110554.29590-14-jolsa@kernel.org>
[-- Attachment #1: Type: text/plain, Size: 1792 bytes --]
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index d853f97bd154..9e3cb547651e 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -42,6 +42,7 @@
>
> #define MAX_UPROBE_MULTI_CNT (1U << 20)
> #define MAX_KPROBE_MULTI_CNT (1U << 20)
> +#define MAX_TRACING_MULTI_CNT (1U << 20)
[ ... ]
> @@ -3641,3 +3642,132 @@ __bpf_kfunc int bpf_copy_from_user_task_str_dynptr(const struct bpf_dynptr *dptr
[ ... ]
> +static int check_dup_ids(u32 *ids, u32 cnt)
> +{
> + int err = 0;
> +
> + /*
> + * Sort ids array (together with cookies array if defined)
> + * and check it for duplicates. The ids and cookies arrays
> + * are left sorted.
> + */
> + sort_r_nonatomic(ids, cnt, sizeof(ids[0]), ids_cmp_r, ids_swap_r, NULL);
The sort here can run over up to MAX_TRACING_MULTI_CNT (1 << 20) elements
without ever yielding the CPU. Can this hold the CPU long enough to trip
RCU stalls or the soft lockup watchdog when cnt is large?
This was raised on an earlier posting of the series, with the suggestion
to add a cond_resched() somewhere in the path. The reply noted that in
practice the count is expected to be closer to 100k rather than the full
1M limit, and that it was unclear where the cond_resched() should go, so
the code was left as is. Since MAX_TRACING_MULTI_CNT is still 1 << 20,
would it be worth revisiting whether a reschedule point is needed here?
> +
> + for (int i = 1; i < cnt; i++) {
> + if (ids[i] == ids[i - 1]) {
> + err = -EINVAL;
> + break;
> + }
> + }
> + return err;
> +}
---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/26881310426
^ permalink raw reply
* Re: [PATCHv7 bpf-next 10/29] bpf: Add multi tracing attach types
From: bot+bpf-ci @ 2026-06-03 11:52 UTC (permalink / raw)
To: jolsa, ast, daniel, andrii
Cc: bpf, linux-trace-kernel, kafai, eddyz87, songliubraving, yhs,
menglong8.dong, rostedt, ast, andrii, daniel, martin.lau, eddyz87,
yonghong.song, clm, ihor.solodrai
In-Reply-To: <20260603110554.29590-11-jolsa@kernel.org>
[-- Attachment #1: Type: text/plain, Size: 4122 bytes --]
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index 4764b4aa7081..bc5a57a9faee 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -2094,6 +2094,11 @@ static inline void bpf_prog_put_recursion_context(struct bpf_prog *prog)
> #endif
> }
>
> +static inline bool is_tracing_multi(enum bpf_attach_type type)
> +{
> + return type == BPF_TRACE_FENTRY_MULTI || type == BPF_TRACE_FEXIT_MULTI;
> +}
> +
> #if defined(CONFIG_BPF_JIT) && defined(CONFIG_BPF_SYSCALL)
> /* This macro helps developer to register a struct_ops type and generate
> * type information correctly. Developers should use this macro to register
[ ... ]
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index aec171ccb6ef..0cde0e2f3d4b 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -1155,6 +1155,8 @@ enum bpf_attach_type {
> BPF_TRACE_KPROBE_SESSION,
> BPF_TRACE_UPROBE_SESSION,
> BPF_TRACE_FSESSION,
> + BPF_TRACE_FENTRY_MULTI,
> + BPF_TRACE_FEXIT_MULTI,
> __MAX_BPF_ATTACH_TYPE
> };
>
[ ... ]
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 8ed484cb1a8a..521ec2a9051e 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
[ ... ]
> @@ -18699,6 +18701,11 @@ static int check_attach_modify_return(unsigned long addr, const char *func_name)
>
> #endif /* CONFIG_FUNCTION_ERROR_INJECTION */
>
> +static bool is_tracing_multi_id(const struct bpf_prog *prog, u32 btf_id)
> +{
> + return is_tracing_multi(prog->expected_attach_type) && bpf_multi_func_btf_id[0] == btf_id;
> +}
> +
> int bpf_check_attach_target(struct bpf_verifier_log *log,
> const struct bpf_prog *prog,
> const struct bpf_prog *tgt_prog,
> @@ -18821,6 +18828,8 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
> prog_extension &&
> (tgt_prog->expected_attach_type == BPF_TRACE_FENTRY ||
> tgt_prog->expected_attach_type == BPF_TRACE_FEXIT ||
> + tgt_prog->expected_attach_type == BPF_TRACE_FENTRY_MULTI ||
> + tgt_prog->expected_attach_type == BPF_TRACE_FEXIT_MULTI ||
> tgt_prog->expected_attach_type == BPF_TRACE_FSESSION)) {
> /* Program extensions can extend all program types
> * except fentry/fexit. The reason is the following.
[ ... ]
> @@ -18956,7 +18967,18 @@ int bpf_check_attach_target(struct bpf_verifier_log *log,
> if (ret < 0)
> return ret;
>
> - if (tgt_prog) {
> + /*
> + * *.multi programs don't need an address during program
> + * verification, we just take the module ref if needed.
> + */
> + if (is_tracing_multi_id(prog, btf_id)) {
> + if (btf_is_module(btf)) {
> + mod = btf_try_get_module(btf);
> + if (!mod)
> + return -ENOENT;
> + }
> + addr = 0;
> + } else if (tgt_prog) {
> if (subprog == 0)
> addr = (long) tgt_prog->bpf_func;
> else
Should BPF_TRACE_FENTRY_MULTI and BPF_TRACE_FEXIT_MULTI be added to the
validation list in btf_validate_prog_ctx_type() in kernel/bpf/btf.c?
This concern was raised in v6 by bot+bpf-ci@kernel.org (message-id
23ab5817f619274a252b35119e760817c383e3c20f9ce9facccbf4b4fec9dd58
at mail.kernel.org, dated Wed, 27 May 2026):
"The multi types are not included in btf_validate_prog_ctx_type()
in kernel/bpf/btf.c. That function validates which attach types
allow u64* as their ctx parameter... Since BPF_TRACE_FENTRY_MULTI
and BPF_TRACE_FEXIT_MULTI also use u64* context, do they need to
be added to that validation list?"
The author acknowledged this in the v6 discussion but the change has
not been included in v7. Since these multi types use the same u64*
context layout as BPF_TRACE_FENTRY and BPF_TRACE_FEXIT, should they
be added alongside those types in btf_validate_prog_ctx_type()?
Lore reference:
https://lore.kernel.org/bpf/23ab5817f619274a252b35119e760817c383e3c20f9ce9facccbf4b4fec9dd58@mail.kernel.org/
---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/26881310426
^ permalink raw reply
* Re: [PATCHv7 bpf-next 12/29] bpf: Add bpf_trampoline_multi_attach/detach functions
From: bot+bpf-ci @ 2026-06-03 11:52 UTC (permalink / raw)
To: jolsa, ast, daniel, andrii
Cc: bpf, linux-trace-kernel, kafai, eddyz87, songliubraving, yhs,
menglong8.dong, rostedt, ast, andrii, daniel, martin.lau, eddyz87,
yonghong.song, clm, ihor.solodrai
In-Reply-To: <20260603110554.29590-13-jolsa@kernel.org>
[-- Attachment #1: Type: text/plain, Size: 4119 bytes --]
> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> index ae7e4fdfe2a3..957e5d7f9554 100644
> --- a/kernel/bpf/trampoline.c
> +++ b/kernel/bpf/trampoline.c
[ ... ]
> @@ -1447,6 +1447,277 @@ int __weak arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
> return -ENOTSUPP;
> }
>
> +#if defined(CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS) && \
> + defined(CONFIG_HAVE_SINGLE_FTRACE_DIRECT_OPS) && \
> + defined(CONFIG_BPF_SYSCALL)
[ ... ]
> +int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
> + struct bpf_tracing_multi_link *link)
> +{
> + struct bpf_tracing_multi_data *data = &link->data;
> + struct bpf_attach_target_info tgt_info = {};
> + struct btf *btf = prog->aux->attach_btf;
> + struct bpf_tracing_multi_node *mnode;
> + struct bpf_trampoline *tr;
> + int i, err, rollback_cnt;
> + u64 key;
> +
> + for_each_mnode(mnode, link) {
> + rollback_cnt = i;
> +
> + err = bpf_check_attach_btf_id_multi(btf, prog, ids[i], &tgt_info);
> + if (err)
> + goto rollback_put;
> +
> + key = bpf_trampoline_compute_key(NULL, btf, ids[i]);
> +
> + tr = bpf_trampoline_get(key, &tgt_info);
> + if (!tr) {
> + err = -ENOMEM;
> + goto rollback_put;
> + }
> +
> + mnode->trampoline = tr;
> + mnode->node.link = &link->link;
> +
> + cond_resched();
> + }
> +
> + err = init_tracing_multi_data(data);
> + if (err) {
> + rollback_cnt = link->nodes_cnt;
> + goto rollback_put;
> + }
> +
> + trampoline_lock_all();
> +
> + for_each_mnode(mnode, link) {
> + bpf_trampoline_multi_attach_init(mnode->trampoline);
> +
> + data->entry = &mnode->entry;
> + err = __bpf_trampoline_link_prog(&mnode->node, mnode->trampoline, NULL,
> + &trampoline_multi_ops, data);
> + if (err) {
> + rollback_cnt = i;
> + goto rollback_unlink;
> + }
> + }
[ ... ]
> +int bpf_trampoline_multi_detach(struct bpf_prog *prog, struct bpf_tracing_multi_link *link)
> +{
> + struct bpf_tracing_multi_data *data = &link->data;
> + struct bpf_tracing_multi_node *mnode;
> + int i;
> +
> + trampoline_lock_all();
> +
> + for_each_mnode(mnode, link) {
> + data->entry = &mnode->entry;
> + bpf_trampoline_multi_attach_init(mnode->trampoline);
> + WARN_ON_ONCE(__bpf_trampoline_unlink_prog(&mnode->node, mnode->trampoline,
> + NULL, &trampoline_multi_ops, data));
> + }
> +
> + if (ftrace_hash_count(data->unreg))
> + WARN_ON_ONCE(update_ftrace_direct_del(&direct_ops, data->unreg));
> + if (ftrace_hash_count(data->modify))
> + WARN_ON_ONCE(update_ftrace_direct_mod(&direct_ops, data->modify, true));
> +
> + for_each_mnode(mnode, link)
> + bpf_trampoline_multi_attach_free(mnode->trampoline);
> +
> + trampoline_unlock_all();
> +
> + for_each_mnode(mnode, link)
> + bpf_trampoline_put(mnode->trampoline);
> +
> + clear_tracing_multi_data(data);
> + return 0;
> +}
In v4, bot+bpf-ci@kernel.org raised a concern about the detach function
potentially failing due to memory allocation. The concern was that if
alloc_ftrace_hash() failed, the function would return -ENOMEM without
detaching any trampolines or releasing refcounts. Since the caller
(bpf_tracing_multi_link_release) uses WARN_ON_ONCE to drop the error,
the link struct would be freed while trampolines remained attached with
elevated refcounts, permanently leaking them and their ftrace hooks.
The reviewer suggested pre-allocating hashes during attach and storing
them in the link struct so that detach cannot fail.
In v7 (the current code), the bpf_tracing_multi_data structure is
embedded in the link structure and init_tracing_multi_data is called
during bpf_trampoline_multi_attach. The detach function now uses the
pre-allocated data from link->data rather than allocating new hashes,
which appears to resolve the original concern.
Reference: https://lore.kernel.org/bpf/711e8b90d27722e47cccfc30a1ccfe243ea7a696322997c15fda1a2147dba6a4@mail.kernel.org/
---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/26881310426
^ permalink raw reply
* Re: [PATCH v6 5/7] locking: Add contended_release tracepoint to qspinlock
From: Peter Zijlstra @ 2026-06-03 12:08 UTC (permalink / raw)
To: Dmitry Ilvokhin
Cc: Ingo Molnar, Will Deacon, Boqun Feng, Waiman Long,
Thomas Bogendoerfer, Juergen Gross, Ajay Kaher, Alexey Makhalov,
Broadcom internal kernel review list, Thomas Gleixner,
Borislav Petkov, Dave Hansen, x86, H. Peter Anvin, Arnd Bergmann,
Dennis Zhou, Tejun Heo, Christoph Lameter, Steven Rostedt,
Masami Hiramatsu, Mathieu Desnoyers, linux-kernel, linux-mips,
virtualization, linux-arch, linux-mm, linux-trace-kernel,
kernel-team, Paul E. McKenney
In-Reply-To: <agXBb0ga_6HJrrnm@shell.ilvokhin.com>
On Thu, May 14, 2026 at 12:34:55PM +0000, Dmitry Ilvokhin wrote:
> Baseline, in best case scenario of least number of executed
> instructions.
>
> 3e0: endbr64 ; 4 bytes (always executed)
> 3e4: movb $0x0,(%rdi) ; 3 bytes (unlock,
> ; always executed)
> 3e7: decl %gs:__preempt_count ; 7 bytes (always executed)
> 3ee: je 3f5 ; 2 bytes (always executed)
> 3f0: jmp __x86_return_thunk ; 5 bytes (executed if above
> ; je is not taken)
> ; rest is not executed
> 3f5: call __SCT__preempt_schedule ; 5 bytes
> 3fa: jmp __x86_return_thunk ; 5 bytes
>
> Tracepoint (again same case of least number of executed instructions).
>
> bc0: endbr64 ; 4 bytes (always executed)
> bc4: xchg %ax,%ax ; 2 bytes (always executed, this is an
> ; only addition on the execution path).
> bc6: movb $0x0,(%rdi) ; 3 bytes (unlock, always executed)
> bc9: decl %gs:__preempt_count ; 7 bytes (always executed)
> bd0: je bde ; 2 bytes (always executed)
> bd2: jmp __x86_return_thunk ; 5 bytes (executed if above
> ; je is not taken)
> ; rest is not executed
> bd7: call queued_spin_release_traced ; 5 bytes
> bdc: jmp bc9 ; 2 bytes
> bde: call __SCT__preempt_schedule ; 5 bytes
> be3: jmp __x86_return_thunk ; 5 bytes
>
So I've been playing with this a bit, and it is all really sad.
Now, since pretty much everybody+dog will have PARAVIRT_SPINLOCK=y, the
'best' solution would be changing that paravirt call with a
static_call(), that actually shrinks the code by 1 byte.
And then this tracepoint nonsense can simply use a different unlock
function, just like paravirt.
0000 00000000000001d0 <_raw_spin_unlock>:
0000 1d0: f3 0f 1e fa endbr64
0004 1d4: ff 15 00 00 00 00 call *0x0(%rip) # 1da <_raw_spin_unlock+0xa> 1d6: R_X86_64_PC32 pv_ops_lock+0x4
000a 1da: 65 ff 0d 00 00 00 00 decl %gs:0x0(%rip) # 1e1 <_raw_spin_unlock+0x11> 1dd: R_X86_64_PC32 __preempt_count-0x4
0011 1e1: 74 06 je 1e9 <_raw_spin_unlock+0x19>
0013 1e3: 2e e9 00 00 00 00 cs jmp 1e9 <_raw_spin_unlock+0x19> 1e5: R_X86_64_PLT32 __x86_return_thunk-0x4
0019 1e9: e8 00 00 00 00 call 1ee <_raw_spin_unlock+0x1e> 1ea: R_X86_64_PLT32 __SCT__preempt_schedule-0x4
001e 1ee: 2e e9 00 00 00 00 cs jmp 1f4 <_raw_spin_unlock+0x24> 1f0: R_X86_64_PLT32 __x86_return_thunk-0x4
0000 00000000000001d0 <_raw_spin_unlock>:
0000 1d0: f3 0f 1e fa endbr64
0004 1d4: e8 00 00 00 00 call 1d9 <_raw_spin_unlock+0x9> 1d5: R_X86_64_PLT32 __SCT__queued_spin_unlock-0x4
0009 1d9: 65 ff 0d 00 00 00 00 decl %gs:0x0(%rip) # 1e0 <_raw_spin_unlock+0x10> 1dc: R_X86_64_PC32 __preempt_count-0x4
0010 1e0: 74 06 je 1e8 <_raw_spin_unlock+0x18>
0012 1e2: 2e e9 00 00 00 00 cs jmp 1e8 <_raw_spin_unlock+0x18> 1e4: R_X86_64_PLT32 __x86_return_thunk-0x4
0018 1e8: e8 00 00 00 00 call 1ed <_raw_spin_unlock+0x1d> 1e9: R_X86_64_PLT32 __SCT__preempt_schedule-0x4
001d 1ed: 2e e9 00 00 00 00 cs jmp 1f3 <_raw_spin_unlock+0x23> 1ef: R_X86_64_PLT32 __x86_return_thunk-0x4
Something a little like so, which is completely untested, except to
build kernel/locking/spinlock.o (with clang-23).
Also, I think someone should go do some performance runs with
ARCH_INLINE_SPIN_* set for x86 just like for s390.
Of course, this is only x86 done, and it doesn't nicely tie in with
tracepoints, but it does give the sanest asm.
---
diff --git a/arch/x86/hyperv/hv_spinlock.c b/arch/x86/hyperv/hv_spinlock.c
index 210b494e4de0..6b4bdea18218 100644
--- a/arch/x86/hyperv/hv_spinlock.c
+++ b/arch/x86/hyperv/hv_spinlock.c
@@ -78,8 +78,8 @@ void __init hv_init_spinlocks(void)
pr_info("PV spinlocks enabled\n");
__pv_init_lock_hash();
- pv_ops_lock.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath;
- pv_ops_lock.queued_spin_unlock = PV_CALLEE_SAVE(__pv_queued_spin_unlock);
+ static_call_update(queued_spin_lock_slowpath, __pv_queued_spin_lock_slowpath);
+ static_call_update(queued_spin_unlock, __raw_callee_save___pv_queued_spin_unlock);
pv_ops_lock.wait = hv_qlock_wait;
pv_ops_lock.kick = hv_qlock_kick;
pv_ops_lock.vcpu_is_preempted = PV_CALLEE_SAVE(hv_vcpu_is_preempted);
diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
index 1d506e5d6f46..52e7ce10df57 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -225,7 +225,6 @@
#define X86_FEATURE_EPT_AD ( 8*32+17) /* "ept_ad" Intel Extended Page Table access-dirty bit */
#define X86_FEATURE_VMCALL ( 8*32+18) /* Hypervisor supports the VMCALL instruction */
#define X86_FEATURE_VMW_VMMCALL ( 8*32+19) /* VMware prefers VMMCALL hypercall instruction */
-#define X86_FEATURE_PVUNLOCK ( 8*32+20) /* PV unlock function */
#define X86_FEATURE_VCPUPREEMPT ( 8*32+21) /* PV vcpu_is_preempted function */
#define X86_FEATURE_TDX_GUEST ( 8*32+22) /* "tdx_guest" Intel Trust Domain Extensions Guest */
diff --git a/arch/x86/include/asm/paravirt-spinlock.h b/arch/x86/include/asm/paravirt-spinlock.h
index 7beffcb08ed6..553910f92906 100644
--- a/arch/x86/include/asm/paravirt-spinlock.h
+++ b/arch/x86/include/asm/paravirt-spinlock.h
@@ -3,6 +3,7 @@
#define _ASM_X86_PARAVIRT_SPINLOCK_H
#include <asm/paravirt_types.h>
+#include <linux/static_call_types.h>
#ifdef CONFIG_SMP
#include <asm/spinlock_types.h>
@@ -11,9 +12,6 @@
struct qspinlock;
struct pv_lock_ops {
- void (*queued_spin_lock_slowpath)(struct qspinlock *lock, u32 val);
- struct paravirt_callee_save queued_spin_unlock;
-
void (*wait)(u8 *ptr, u8 val);
void (*kick)(int cpu);
@@ -26,20 +24,23 @@ extern struct pv_lock_ops pv_ops_lock;
extern void native_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val);
extern void __pv_init_lock_hash(void);
extern void __pv_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val);
+extern void __raw_callee_save___native_queued_spin_unlock(struct qspinlock *lock);
extern void __raw_callee_save___pv_queued_spin_unlock(struct qspinlock *lock);
extern bool nopvspin;
+DECLARE_STATIC_CALL(queued_spin_lock_slowpath, native_queued_spin_lock_slowpath);
+DECLARE_STATIC_CALL(queued_spin_unlock, __raw_callee_save___native_queued_spin_unlock);
+
static __always_inline void pv_queued_spin_lock_slowpath(struct qspinlock *lock,
u32 val)
{
- PVOP_VCALL2(pv_ops_lock, queued_spin_lock_slowpath, lock, val);
+ static_call(queued_spin_lock_slowpath)(lock, val);
}
static __always_inline void pv_queued_spin_unlock(struct qspinlock *lock)
{
- PVOP_ALT_VCALLEE1(pv_ops_lock, queued_spin_unlock, lock,
- "movb $0, (%%" _ASM_ARG1 ")",
- ALT_NOT(X86_FEATURE_PVUNLOCK));
+ __STATIC_CALL_MOD_ADDRESSABLE(queued_spin_unlock);
+ asm volatile ("call " STATIC_CALL_TRAMP_STR(queued_spin_unlock) : ASM_CALL_CONSTRAINT);
}
static __always_inline bool pv_vcpu_is_preempted(long cpu)
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 29226d112029..5908d9fd94bb 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -1135,9 +1135,8 @@ void __init kvm_spinlock_init(void)
pr_info("PV spinlocks enabled\n");
__pv_init_lock_hash();
- pv_ops_lock.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath;
- pv_ops_lock.queued_spin_unlock =
- PV_CALLEE_SAVE(__pv_queued_spin_unlock);
+ static_call_update(queued_spin_lock_slowpath, __pv_queued_spin_lock_slowpath);
+ static_call_update(queued_spin_unlock, __raw_callee_save___pv_queued_spin_unlock);
pv_ops_lock.wait = kvm_wait;
pv_ops_lock.kick = kvm_kick_cpu;
diff --git a/arch/x86/kernel/paravirt-spinlocks.c b/arch/x86/kernel/paravirt-spinlocks.c
index 95452444868f..28407304f123 100644
--- a/arch/x86/kernel/paravirt-spinlocks.c
+++ b/arch/x86/kernel/paravirt-spinlocks.c
@@ -25,9 +25,12 @@ __visible void __native_queued_spin_unlock(struct qspinlock *lock)
}
PV_CALLEE_SAVE_REGS_THUNK(__native_queued_spin_unlock);
+DEFINE_STATIC_CALL(queued_spin_lock_slowpath, native_queued_spin_lock_slowpath);
+DEFINE_STATIC_CALL(queued_spin_unlock, __raw_callee_save___native_queued_spin_unlock);
+
bool pv_is_native_spin_unlock(void)
{
- return pv_ops_lock.queued_spin_unlock.func ==
+ return static_call_query(queued_spin_unlock) ==
__raw_callee_save___native_queued_spin_unlock;
}
@@ -45,16 +48,11 @@ bool pv_is_native_vcpu_is_preempted(void)
void __init paravirt_set_cap(void)
{
- if (!pv_is_native_spin_unlock())
- setup_force_cpu_cap(X86_FEATURE_PVUNLOCK);
-
if (!pv_is_native_vcpu_is_preempted())
setup_force_cpu_cap(X86_FEATURE_VCPUPREEMPT);
}
struct pv_lock_ops pv_ops_lock = {
- .queued_spin_lock_slowpath = native_queued_spin_lock_slowpath,
- .queued_spin_unlock = PV_CALLEE_SAVE(__native_queued_spin_unlock),
.wait = paravirt_nop,
.kick = paravirt_nop,
.vcpu_is_preempted = PV_CALLEE_SAVE(__native_vcpu_is_preempted),
diff --git a/arch/x86/kernel/static_call.c b/arch/x86/kernel/static_call.c
index 61592e41a6b1..1268b7dc93b1 100644
--- a/arch/x86/kernel/static_call.c
+++ b/arch/x86/kernel/static_call.c
@@ -3,6 +3,7 @@
#include <linux/memory.h>
#include <linux/bug.h>
#include <asm/text-patching.h>
+#include <asm/paravirt-spinlock.h>
enum insn_type {
CALL = 0, /* site call */
@@ -31,6 +32,15 @@ static const u8 retinsn[] = { RET_INSN_OPCODE, 0xcc, 0xcc, 0xcc, 0xcc };
*/
static const u8 warninsn[] = { 0x67, 0x48, 0x0f, 0xb9, 0x3a };
+/*
+ * ds ds movb $0, (_ASM_ARG1)
+ */
+#ifdef CONFIG_64BIT
+static const u8 unlockinsn[] = { 0x3e, 0x3e, 0xc6, 0x07, 0x00 };
+#else
+static const u8 unlockinsn[] = { 0x3e, 0x3e, 0xc6, 0x00, 0x00 };
+#endif
+
static u8 __is_Jcc(u8 *insn) /* Jcc.d32 */
{
u8 ret = 0;
@@ -78,6 +88,10 @@ static void __ref __static_call_transform(void *insn, enum insn_type type,
emulate = code;
code = &warninsn;
}
+ if (func == &__raw_callee_save___native_queued_spin_unlock) {
+ emulate = code;
+ code = &unlockinsn;
+ }
break;
case NOP:
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index 83ac24ead289..f718e535ea7c 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -134,9 +134,8 @@ void __init xen_init_spinlocks(void)
printk(KERN_DEBUG "xen: PV spinlocks enabled\n");
__pv_init_lock_hash();
- pv_ops_lock.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath;
- pv_ops_lock.queued_spin_unlock =
- PV_CALLEE_SAVE(__pv_queued_spin_unlock);
+ static_call_update(queued_spin_lock_slowpath, __pv_queued_spin_lock_slowpath);
+ static_call_update(queued_spin_unlock, __raw_callee_save___pv_queued_spin_unlock);
pv_ops_lock.wait = xen_qlock_wait;
pv_ops_lock.kick = xen_qlock_kick;
pv_ops_lock.vcpu_is_preempted = PV_CALLEE_SAVE(xen_vcpu_stolen);
diff --git a/tools/arch/x86/include/asm/cpufeatures.h b/tools/arch/x86/include/asm/cpufeatures.h
index 86d17b195e79..61541f042f74 100644
--- a/tools/arch/x86/include/asm/cpufeatures.h
+++ b/tools/arch/x86/include/asm/cpufeatures.h
@@ -225,7 +225,6 @@
#define X86_FEATURE_EPT_AD ( 8*32+17) /* "ept_ad" Intel Extended Page Table access-dirty bit */
#define X86_FEATURE_VMCALL ( 8*32+18) /* Hypervisor supports the VMCALL instruction */
#define X86_FEATURE_VMW_VMMCALL ( 8*32+19) /* VMware prefers VMMCALL hypercall instruction */
-#define X86_FEATURE_PVUNLOCK ( 8*32+20) /* PV unlock function */
#define X86_FEATURE_VCPUPREEMPT ( 8*32+21) /* PV vcpu_is_preempted function */
#define X86_FEATURE_TDX_GUEST ( 8*32+22) /* "tdx_guest" Intel Trust Domain Extensions Guest */
^ permalink raw reply related
* Re: [PATCH mm-unstable v18 11/14] mm/khugepaged: Introduce mTHP collapse support
From: Nico Pache @ 2026-06-03 12:16 UTC (permalink / raw)
To: David Hildenbrand (Arm)
Cc: linux-doc, linux-kernel, linux-mm, linux-trace-kernel, aarcange,
akpm, anshuman.khandual, apopple, baohua, baolin.wang, byungchul,
catalin.marinas, cl, corbet, dave.hansen, dev.jain, gourry,
hannes, hughd, jack, jackmanb, jannh, jglisse, joshua.hahnjy, kas,
lance.yang, liam, ljs, mathieu.desnoyers, matthew.brost, mhiramat,
mhocko, peterx, pfalcato, rakie.kim, raquini, rdunlap,
richard.weiyang, rientjes, rostedt, rppt, ryan.roberts, shivankg,
sunnanyong, surenb, thomas.hellstrom, tiwai, vbabka, vishal.moola,
wangkefeng.wang, will, willy, yang, ying.huang, ziy, zokeefe,
Usama Arif, usamaarif642
In-Reply-To: <ef5c470f-3c32-418d-9566-17ec01b691bb@kernel.org>
On Wed, Jun 3, 2026 at 4:01 AM David Hildenbrand (Arm) <david@kernel.org> wrote:
>
>
> > next_order:
> > - if ((BIT(order) - 1) & enabled_orders) {
> > - const u8 next_order = order - 1;
> > - const u16 mid_offset = offset + (nr_ptes / 2);
> > -
> > - collapse_mthp_stack_push(cc, &stack_size, mid_offset,
> > - next_order);
> > - collapse_mthp_stack_push(cc, &stack_size, offset,
> > - next_order);
> > + if (order > KHUGEPAGED_MIN_MTHP_ORDER &&
> > + (BIT(order) - 1) & enabled_orders) {
>
> Why not a test_bit() ?
The test bit is at the top of the loop. This adds a exit if the lower
orders are all disabled or we hit the last order.
>
>
> But, wouldn't you want to skip orders that are not enabled and try with the next
> smaller one in any case before you advance the offset?
We are currently iterating through each order (not skipping them).
There may be optimizations to avoid iterating through every order
(like your changes suggest), but currently, every collapse, whether it
succeeds or fails at the bottom order, must also iterate the offset.
lmk if that makes sense!
-- Nico
>
> --
> Cheers,
>
> David
>
^ permalink raw reply
* Re: [PATCH mm-unstable v18 11/14] mm/khugepaged: Introduce mTHP collapse support
From: David Hildenbrand (Arm) @ 2026-06-03 12:27 UTC (permalink / raw)
To: Nico Pache
Cc: linux-doc, linux-kernel, linux-mm, linux-trace-kernel, aarcange,
akpm, anshuman.khandual, apopple, baohua, baolin.wang, byungchul,
catalin.marinas, cl, corbet, dave.hansen, dev.jain, gourry,
hannes, hughd, jack, jackmanb, jannh, jglisse, joshua.hahnjy, kas,
lance.yang, liam, ljs, mathieu.desnoyers, matthew.brost, mhiramat,
mhocko, peterx, pfalcato, rakie.kim, raquini, rdunlap,
richard.weiyang, rientjes, rostedt, rppt, ryan.roberts, shivankg,
sunnanyong, surenb, thomas.hellstrom, tiwai, vbabka, vishal.moola,
wangkefeng.wang, will, willy, yang, ying.huang, ziy, zokeefe,
Usama Arif, usamaarif642
In-Reply-To: <CAA1CXcD_Bz=Si+vP0Sg_FhAHVjbeAJ4VdvtWGM-jVYiRXFHAnw@mail.gmail.com>
On 6/3/26 14:16, Nico Pache wrote:
> On Wed, Jun 3, 2026 at 4:01 AM David Hildenbrand (Arm) <david@kernel.org> wrote:
>>
>>
>>> next_order:
>>> - if ((BIT(order) - 1) & enabled_orders) {
>>> - const u8 next_order = order - 1;
>>> - const u16 mid_offset = offset + (nr_ptes / 2);
>>> -
>>> - collapse_mthp_stack_push(cc, &stack_size, mid_offset,
>>> - next_order);
>>> - collapse_mthp_stack_push(cc, &stack_size, offset,
>>> - next_order);
>>> + if (order > KHUGEPAGED_MIN_MTHP_ORDER &&
>>> + (BIT(order) - 1) & enabled_orders) {
>>
>> Why not a test_bit() ?
>
> The test bit is at the top of the loop. This adds a exit if the lower
> orders are all disabled or we hit the last order.
Ah, now I understand what you want to do here.
I guess you should add a () around the & and maybe add a comment.
And likely using GENMASK is clearer?
/*
* Continue with the next smaller order if there is still
* any smaller order enabled.
*/
if (order > KHUGEPAGED_MIN_MTHP_ORDER &&
(enabled_orders & GENMASK(order - 1, 0))) {
...
}
>
>>
>>
>> But, wouldn't you want to skip orders that are not enabled and try with the next
>> smaller one in any case before you advance the offset?
>
> We are currently iterating through each order (not skipping them).
> There may be optimizations to avoid iterating through every order
> (like your changes suggest), but currently, every collapse, whether it
> succeeds or fails at the bottom order, must also iterate the offset.
Right, we currently miss opportunities to just skip, that we can optimize later.
--
Cheers,
David
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox