Skip to contentSkip to conversation

bpf: Return error for missed kprobe multi bpf program execution

PATCH Applied Jiri Olsa 5 participants 7 messages Jan 6, 2025 – Jan 8, 2025

Subsystems (6)
Kselftest BPF Tracing BPF selftests +2 more subsystems
lore
Share & tools
lore

Jiri Olsa

author reviewer
#0 lore raw

PATCH 1/2 bpf: Return error for missed kprobe multi bpf program execution

When kprobe multi bpf program can't be executed due to recursion check, we currently return 0 (success) to fprobe layer where it's ignored for standard kprobe multi probes.

For kprobe session the success return value will make fprobe layer to install return probe and try to execute it as well.

But the return session probe should not get executed, because the entry part did not run. FWIW the return probe bpf program most likely won't get executed, because its recursion check will likely fail as well, but we don't need to run it in the first place.. also we can make this clear and obvious.

It also affects missed counts for kprobe session program execution, which are now doubled (extra count for not executed return probe).

---
 kernel/trace/bpf_trace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/bpf_trace.c +1 -1
index 48db147c6c7d..1f3d4b72a3f2 100644
@@ -2797,7 +2797,7 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link,
 
 	if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
 		bpf_prog_inc_misses_counter(link->link.prog);
-		err = 0;
+		err = 1;
 		goto out;
 	}
 
Signature · 1 line
2.47.0
author reviewer
#1 lore raw

PATCH 2/2 selftests/bpf: Add kprobe session recursion check test

Adding kprobe.session probe to bpf_kfunc_common_test that misses bpf program execution due to recursion check and making sure it increases the program missed count properly.

---
 tools/testing/selftests/bpf/prog_tests/missed.c             | 1 +
 tools/testing/selftests/bpf/progs/missed_kprobe_recursion.c | 6 ++++++
 2 files changed, 7 insertions(+)
tools/testing/selftests/bpf/prog_tests/missed.c +1 -0
index 70d90c43537c..ed8857ae914a 100644
@@ -85,6 +85,7 @@ static void test_missed_kprobe_recursion(void)
 	ASSERT_GE(get_missed_count(bpf_program__fd(skel->progs.test3)), 1, "test3_recursion_misses");
 	ASSERT_GE(get_missed_count(bpf_program__fd(skel->progs.test4)), 1, "test4_recursion_misses");
 	ASSERT_GE(get_missed_count(bpf_program__fd(skel->progs.test5)), 1, "test5_recursion_misses");
+	ASSERT_EQ(get_missed_count(bpf_program__fd(skel->progs.test6)), 1, "test6_recursion_misses");
 
 cleanup:
 	missed_kprobe_recursion__destroy(skel);
tools/testing/selftests/bpf/progs/missed_kprobe_recursion.c +6 -0
index c4bf679a9876..29c18d869ec1 100644
@@ -46,3 +46,9 @@ int test5(struct pt_regs *ctx)
 {
 	return 0;
 }
+
+SEC("kprobe.session/bpf_kfunc_common_test")
+int test6(struct pt_regs *ctx)
+{
+	return 0;
+}
Signature · 1 line
2.47.0
maintainer
#2 lore raw
In reply to Jiri Olsa
Quoted 27 lines · Jiri Olsa

When kprobe multi bpf program can't be executed due to recursion check, we currently return 0 (success) to fprobe layer where it's ignored for standard kprobe multi probes.

For kprobe session the success return value will make fprobe layer to install return probe and try to execute it as well.

But the return session probe should not get executed, because the entry part did not run. FWIW the return probe bpf program most likely won't get executed, because its recursion check will likely fail as well, but we don't need to run it in the first place.. also we can make this clear and obvious.

It also affects missed counts for kprobe session program execution, which are now doubled (extra count for not executed return probe).

---
 kernel/trace/bpf_trace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/bpf_trace.c +0 -0
index 48db147c6c7d..1f3d4b72a3f2 100644

kernel/trace/bpf_trace.c kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link, · line 2800

Quoted 5 lines
kernel/trace/bpf_trace.c +1 -1
@@ -2797,7 +2797,7 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link,
 
        if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
                bpf_prog_inc_misses_counter(link->link.prog);
-               err = 0;
+               err = 1;

nit: Shall we return -EBUSY or some other error code?

Quoted 8 lines
kernel/trace/bpf_trace.c +0 -1
                goto out;
        }
 
 
--

2.47.0

#3 lore raw
Quoted 29 lines · Song Liu
Quoted 28 lines · Jiri Olsa

When kprobe multi bpf program can't be executed due to recursion check, we currently return 0 (success) to fprobe layer where it's ignored for standard kprobe multi probes.

For kprobe session the success return value will make fprobe layer to install return probe and try to execute it as well.

But the return session probe should not get executed, because the entry part did not run. FWIW the return probe bpf program most likely won't get executed, because its recursion check will likely fail as well, but we don't need to run it in the first place.. also we can make this clear and obvious.

It also affects missed counts for kprobe session program execution, which are now doubled (extra count for not executed return probe).

---
 kernel/trace/bpf_trace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/bpf_trace.c +0 -0
index 48db147c6c7d..1f3d4b72a3f2 100644
@@ -2797,7 +2797,7 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link,
 
Quoted 6 lines

kernel/trace/bpf_trace.c kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link, · line 2800

Quoted 4 lines
kernel/trace/bpf_trace.c +1 -1
        if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
                bpf_prog_inc_misses_counter(link->link.prog);
-               err = 0;
+               err = 1;

nit: Shall we return -EBUSY or some other error code?

it's processed in __fprobe_handler and it's treated as bool, so technically it does not matter.. but I'd rather keep the 0/1 return values in here, because it's what the session program is forced to return

thanks,
jirka

maintainer
#4 lore raw

On Wed, Jan 8, 2025 at 3:38 AM Jiri Olsa <[email protected]> wrote:
[...]

Quoted 14 lines
Quoted 7 lines
Quoted 7 lines
kernel/trace/bpf_trace.c +0 -0
index 48db147c6c7d..1f3d4b72a3f2 100644
@@ -2797,7 +2797,7 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link,
 
Quoted 6 lines

kernel/trace/bpf_trace.c kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link, · line 2800

Quoted 4 lines
kernel/trace/bpf_trace.c +1 -1
        if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
                bpf_prog_inc_misses_counter(link->link.prog);
-               err = 0;
+               err = 1;

nit: Shall we return -EBUSY or some other error code?

Quoted 3 lines

it's processed in __fprobe_handler and it's treated as bool, so technically it does not matter.. but I'd rather keep the 0/1 return values in here, because it's what the session program is forced to return

Got it. Thanks for the explanation.

Song

maintainer Acked-by on v1 1/2
#5 lore raw
In reply to Jiri Olsa
Quoted 4 lines · On Mon, 6 Jan 2025 18:50:47 +0100 Jiri Olsa <[email protected]> wrote:

When kprobe multi bpf program can't be executed due to recursion check, we currently return 0 (success) to fprobe layer where it's ignored for standard kprobe multi probes.

Quoted 8 lines

For kprobe session the success return value will make fprobe layer to install return probe and try to execute it as well.

But the return session probe should not get executed, because the entry part did not run. FWIW the return probe bpf program most likely won't get executed, because its recursion check will likely fail as well, but we don't need to run it in the first place.. also we can make this clear and obvious.

Yeah, that's right.

Quoted 5 lines

It also affects missed counts for kprobe session program execution, which are now doubled (extra count for not executed return probe).

Looks good to me.

Thanks!

Quoted 20 lines
---
 kernel/trace/bpf_trace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/bpf_trace.c +1 -1
index 48db147c6c7d..1f3d4b72a3f2 100644
@@ -2797,7 +2797,7 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link *link,
 
 	if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
 		bpf_prog_inc_misses_counter(link->link.prog);
-		err = 0;
+		err = 1;
 		goto out;
 	}
 
Signature · 1 line
2.47.0
Signature · 1 line
Masami Hiramatsu (Google) <[email protected]>
#6 lore raw
In reply to Jiri Olsa
This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Hello:

This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Quoted 8 lines · On Mon, 6 Jan 2025 18:50:47 +0100 you wrote:

When kprobe multi bpf program can't be executed due to recursion check, we currently return 0 (success) to fprobe layer where it's ignored for standard kprobe multi probes.

For kprobe session the success return value will make fprobe layer to install return probe and try to execute it as well.

[...]

Here is the summary with links:
- [bpf-next,1/2] bpf: Return error for missed kprobe multi bpf program execution

    https://git.kernel.org/bpf/bpf-next/c/2ebadb60cb36
  - [bpf-next,2/2] selftests/bpf: Add kprobe session recursion check test
    https://git.kernel.org/bpf/bpf-next/c/bfaac2a0b9e5

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html