Skip to contentSkip to conversation

tracing/kprobes: Check whether get_kretprobe() returns NULL in kretprobe_dispatcher()

PATCH In review Masami Hiramatsu (Google) 6 participants 15 messages May 27, 2022 – Jun 17, 2022

Subsystems (3)
lore
Share & tools
lore

Masami Hiramatsu (Google)

author maintainer
#0 lore raw

PATCH 1/1 tracing/kprobes: Check whether get_kretprobe() returns NULL in kretprobe_dispatcher()

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")
---
 kernel/trace/trace_kprobe.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
kernel/trace/trace_kprobe.c +10 -1
index 93507330462c..a245ea673715 100644
@@ -1718,8 +1718,17 @@ static int
 kretprobe_dispatcher(struct kretprobe_instance *ri, struct pt_regs *regs)
 {
 	struct kretprobe *rp = get_kretprobe(ri);
-	struct trace_kprobe *tk = container_of(rp, struct trace_kprobe, rp);
+	struct trace_kprobe *tk;
+
+	/*
+	 * There is a small chance that get_kretprobe(ri) returns NULL when
+	 * the kretprobe is unregister on another CPU between kretprobe's
+	 * trampoline_handler and this function.
+	 */
+	if (unlikely(!rp))
+		return 0;
 
+	tk = container_of(rp, struct trace_kprobe, rp);
 	raw_cpu_inc(*tk->nhit);
 
 	if (trace_probe_test_flag(&tk->tp, TP_FLAG_TRACE))
Acked-by on v1 1/1
#1 lore raw
Quoted 20 lines · Masami Hiramatsu (Google)

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Quoted 4 lines
Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")

[...]

maintainer
#2 lore raw
In reply to Masami Hiramatsu (Google)
Quoted 16 lines · Masami Hiramatsu (Google)

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Quoted 4 lines
Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")

Steven, I presume you'll pick this fix up?

Thanks,
Daniel

maintainer
#3 lore raw
Quoted 17 lines · Daniel Borkmann
Quoted 16 lines · Masami Hiramatsu (Google)

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Quoted 6 lines
Quoted 4 lines
Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")

Steven, I presume you'll pick this fix up?

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

-- Steve

Quoted 3 lines

Thanks,
Daniel

Signature · 1 line
Sent from my Android device with K-9 Mail. Please excuse my brevity and top posting.
maintainer
#4 lore raw
Quoted 18 lines · Steven Rostedt
Quoted 17 lines · Daniel Borkmann
Quoted 16 lines · Masami Hiramatsu (Google)

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")

Steven, I presume you'll pick this fix up?

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

Ok, thanks. If I don't hear back I presume you'll pick it up then.

maintainer
#5 lore raw
Quoted 7 lines · On Wed, 8 Jun 2022 14:38:39 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

Steven, I presume you'll pick this fix up?

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

Ok, thanks. If I don't hear back I presume you'll pick it up then.

Yeah, I'm way behind due to the conference. And I'll be on PTO from tomorrow and back on Tuesday. And registration for Linux Plumbers is supposed to open today (but of course there's issues with that!), thus, I'm really have too much on my plate today :-p

-- Steve

#6 lore raw
Quoted 8 lines · Steven Rostedt
Quoted 4 lines · On Wed, 8 Jun 2022 14:38:39 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 4 lines
Quoted 3 lines
Quoted 1 line

Steven, I presume you'll pick this fix up?

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

Ok, thanks. If I don't hear back I presume you'll pick it up then.

Yeah, I'm way behind due to the conference. And I'll be on PTO from tomorrow and back on Tuesday. And registration for Linux Plumbers is supposed to open today (but of course there's issues with that!), thus, I'm really have too much on my plate today :-p

Steven,

you missed Davem's presentation at KR about importance of delegation. You need watch it:
https://youtu.be/ELPENQrtUas?t=9085

maintainer
#8 lore raw
In reply to Steven Rostedt
Quoted 7 lines · Steven Rostedt
Quoted 4 lines · On Wed, 8 Jun 2022 14:38:39 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 4 lines
Quoted 3 lines
Quoted 1 line

Steven, I presume you'll pick this fix up?

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

Ok, thanks. If I don't hear back I presume you'll pick it up then.

Yeah, I'm way behind due to the conference. And I'll be on PTO from tomorrow and back on Tuesday. And registration for Linux Plumbers is supposed to open today (but of course there's issues with that!), thus, I'm really have too much on my plate today :-p

Steven, we still have this in our patchwork for tracking so it doesn't fall off the radar. The patch is 3 weeks old by now. Has this been picked up yet, or do you want to Ack and we ship the fix via bpf tree? Just asking as I didn't see any further updates ever since.

Thanks,
Daniel

maintainer
#9 lore raw
Quoted 17 lines · Daniel Borkmann
Quoted 7 lines · Steven Rostedt
Quoted 4 lines · On Wed, 8 Jun 2022 14:38:39 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 4 lines
Quoted 3 lines
Quoted 1 line

Steven, I presume you'll pick this fix up?

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

I'm currently at Embedded/Kernel Recipes, but yeah, I'll take a look at it. (Just need to finish my slides first ;-)

Ok, thanks. If I don't hear back I presume you'll pick it up then.

Yeah, I'm way behind due to the conference. And I'll be on PTO from tomorrow and back on Tuesday. And registration for Linux Plumbers is supposed to open today (but of course there's issues with that!), thus, I'm really have too much on my plate today :-p

Quoted 4 lines

Steven, we still have this in our patchwork for tracking so it doesn't fall off the radar. The patch is 3 weeks old by now. Has this been picked up yet, or do you want to Ack and we ship the fix via bpf tree? Just asking as I didn't see any further updates ever since.

Same goes for these two fixes waiting for Ack:

https://lore.kernel.org/bpf/165461825202.280167.12903689442217921817.stgit@devnote2/

Thanks,
Daniel

maintainer
#10 lore raw
In reply to Daniel Borkmann
Quoted 4 lines · On Fri, 17 Jun 2022 10:26:40 +0200 Daniel Borkmann <[email protected]> wrote:

Steven, we still have this in our patchwork for tracking so it doesn't fall off the radar. The patch is 3 weeks old by now. Has this been picked up yet, or do you want to Ack and we ship the fix via bpf tree? Just asking as I didn't see any further updates ever since.

Sorry, between traveling for conferences and PTO I fell behind. I'll pull this into my urgent queue and start running my tests on it.

-- Steve

maintainer
#11 lore raw
Quoted 10 lines · Steven Rostedt
Quoted 4 lines · On Fri, 17 Jun 2022 10:26:40 +0200 Daniel Borkmann <[email protected]> wrote:

Steven, we still have this in our patchwork for tracking so it doesn't fall off the radar. The patch is 3 weeks old by now. Has this been picked up yet, or do you want to Ack and we ship the fix via bpf tree? Just asking as I didn't see any further updates ever since.

Sorry, between traveling for conferences and PTO I fell behind. I'll pull this into my urgent queue and start running my tests on it.

Okay, if you pick these fixes up today then I'll toss them from our bpf patchwork.

maintainer
#12 lore raw
Quoted 4 lines · On Fri, 17 Jun 2022 18:09:44 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 2 lines

Sorry, between traveling for conferences and PTO I fell behind. I'll pull this into my urgent queue and start running my tests on it.

Okay, if you pick these fixes up today then I'll toss them from our bpf patchwork.

This patch I'll take, but please keep the fprobe/rethook one, as that's specific to your tree.

-- Steve

maintainer
#13 lore raw
Quoted 10 lines · Steven Rostedt
Quoted 4 lines · On Fri, 17 Jun 2022 18:09:44 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 2 lines

Sorry, between traveling for conferences and PTO I fell behind. I'll pull this into my urgent queue and start running my tests on it.

Okay, if you pick these fixes up today then I'll toss them from our bpf patchwork.

This patch I'll take, but please keep the fprobe/rethook one, as that's specific to your tree.

Ok, sounds good to me. I just applied them with the comment suggestion added.

Acked-by on v1 1/1
#14 lore raw
In reply to Masami Hiramatsu (Google)
Quoted 16 lines · Masami Hiramatsu (Google)

From: Masami Hiramatsu (Google) <[email protected]>

There is a small chance that get_kretprobe(ri) returns NULL in kretprobe_dispatcher() when another CPU unregisters the kretprobe right after __kretprobe_trampoline_handler().

To avoid this issue, kretprobe_dispatcher() checks the get_kretprobe() return value again. And if it is NULL, it returns soon because that kretprobe is under unregistering process.

This issue has been introduced when the kretprobe is decoupled from the struct kretprobe_instance by commit d741bf41d7c7 ("kprobes: Remove kretprobe hash"). Before that commit, the struct kretprob_instance::rp directly points the kretprobe and it is never be NULL.

Quoted 4 lines
Fixes d741bf41d7c7 ("kprobes: Remove kretprobe hash")

jirka

Quoted 28 lines
---
 kernel/trace/trace_kprobe.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
kernel/trace/trace_kprobe.c +10 -1
index 93507330462c..a245ea673715 100644
@@ -1718,8 +1718,17 @@ static int
 kretprobe_dispatcher(struct kretprobe_instance *ri, struct pt_regs *regs)
 {
 	struct kretprobe *rp = get_kretprobe(ri);
-	struct trace_kprobe *tk = container_of(rp, struct trace_kprobe, rp);
+	struct trace_kprobe *tk;
+
+	/*
+	 * There is a small chance that get_kretprobe(ri) returns NULL when
+	 * the kretprobe is unregister on another CPU between kretprobe's
+	 * trampoline_handler and this function.
+	 */
+	if (unlikely(!rp))
+		return 0;
 
+	tk = container_of(rp, struct trace_kprobe, rp);
 	raw_cpu_inc(*tk->nhit);
 
 	if (trace_probe_test_flag(&tk->tp, TP_FLAG_TRACE))