Skip to contentSkip to conversation

kprobes: Fix kill kprobe which has been marked as gone

PATCH Superseded Muchun Song 2 participants 4 messages Aug 20, 2020 – Aug 22, 2020 kprobes

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kprobes: Fix kill kprobe which has been marked as gone

If a kprobe is marked as gone, we should not kill it again. Otherwise, we can disarm the kprobe more than once. In that case, the statistics of kprobe_ftrace_enabled can unbalance which can lead to that kprobe do not work.

Fixes 0cb2f1372baa ("kprobes: Fix NULL pointer dereference at kprobe_ftrace_handler")
---
 kernel/kprobes.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
kernel/kprobes.c +5 -1
index d36e2b017588..7bac3ea44ff4 100644
@@ -2422,7 +2422,10 @@ static int kprobes_module_callback(struct notifier_block *nb,
 	mutex_lock(&kprobe_mutex);
 	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
 		head = &kprobe_table[i];
-		hlist_for_each_entry(p, head, hlist)
+		hlist_for_each_entry(p, head, hlist) {
+			if (kprobe_gone(p))
+				continue;
+
 			if (within_module_init((unsigned long)p->addr, mod) ||
 			    (checkcore &&
 			     within_module_core((unsigned long)p->addr, mod))) {
@@ -2439,6 +2442,7 @@ static int kprobes_module_callback(struct notifier_block *nb,
 				 */
 				kill_kprobe(p);
 			}
+		}
 	}
 	if (val == MODULE_STATE_GOING)
 		remove_module_kprobe_blacklist(mod);
Signature · 1 line
2.11.0
maintainer Acked-by on v1 1/1
#1 lore raw

Hi Muchun,

Quoted 5 lines · On Thu, 20 Aug 2020 11:19:33 +0800 Muchun Song <[email protected]> wrote:

If a kprobe is marked as gone, we should not kill it again. Otherwise, we can disarm the kprobe more than once. In that case, the statistics of kprobe_ftrace_enabled can unbalance which can lead to that kprobe do not work.

Good catch! Hmm, I think we also need an assertion in kill_kprobe() so that p is already gone. Anyway, this looks good to me.

Thanks!

Quoted 35 lines
Fixes 0cb2f1372baa ("kprobes: Fix NULL pointer dereference at kprobe_ftrace_handler")
---
 kernel/kprobes.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
kernel/kprobes.c +5 -1
index d36e2b017588..7bac3ea44ff4 100644
@@ -2422,7 +2422,10 @@ static int kprobes_module_callback(struct notifier_block *nb,
 	mutex_lock(&kprobe_mutex);
 	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
 		head = &kprobe_table[i];
-		hlist_for_each_entry(p, head, hlist)
+		hlist_for_each_entry(p, head, hlist) {
+			if (kprobe_gone(p))
+				continue;
+
 			if (within_module_init((unsigned long)p->addr, mod) ||
 			    (checkcore &&
 			     within_module_core((unsigned long)p->addr, mod))) {
@@ -2439,6 +2442,7 @@ static int kprobes_module_callback(struct notifier_block *nb,
 				 */
 				kill_kprobe(p);
 			}
+		}
 	}
 	if (val == MODULE_STATE_GOING)
 		remove_module_kprobe_blacklist(mod);
Signature · 1 line
2.11.0
Signature · 1 line
Masami Hiramatsu <[email protected]>
maintainer
#2 lore raw
Quoted 11 lines · On Fri, 21 Aug 2020 21:28:43 +0900 Masami Hiramatsu <[email protected]> wrote:

Hi Muchun,

Quoted 5 lines · On Thu, 20 Aug 2020 11:19:33 +0800 Muchun Song <[email protected]> wrote:

If a kprobe is marked as gone, we should not kill it again. Otherwise, we can disarm the kprobe more than once. In that case, the statistics of kprobe_ftrace_enabled can unbalance which can lead to that kprobe do not work.

Quoted 8 lines

Good catch! Hmm, I think we also need an assertion in kill_kprobe() so that p is already gone. Anyway, this looks good to me.

Thanks!

Quoted 1 line
Fixes 0cb2f1372baa ("kprobes: Fix NULL pointer dereference at kprobe_ftrace_handler")

BTW, this fixes older bug than this commit.

Fixes e8386a0cb22f ("kprobes: support probing module __exit function")

Thank you,

Quoted 38 lines
Quoted 34 lines
---
 kernel/kprobes.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
kernel/kprobes.c +5 -1
index d36e2b017588..7bac3ea44ff4 100644
@@ -2422,7 +2422,10 @@ static int kprobes_module_callback(struct notifier_block *nb,
 	mutex_lock(&kprobe_mutex);
 	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
 		head = &kprobe_table[i];
-		hlist_for_each_entry(p, head, hlist)
+		hlist_for_each_entry(p, head, hlist) {
+			if (kprobe_gone(p))
+				continue;
+
 			if (within_module_init((unsigned long)p->addr, mod) ||
 			    (checkcore &&
 			     within_module_core((unsigned long)p->addr, mod))) {
@@ -2439,6 +2442,7 @@ static int kprobes_module_callback(struct notifier_block *nb,
 				 */
 				kill_kprobe(p);
 			}
+		}
 	}
 	if (val == MODULE_STATE_GOING)
 		remove_module_kprobe_blacklist(mod);
Signature · 1 line
2.11.0
Signature · 1 line
Masami Hiramatsu <[email protected]>
Signature · 1 line
Masami Hiramatsu <[email protected]>
#3 lore raw
Quoted 20 lines · Masami Hiramatsu
Quoted 16 lines · On Fri, 21 Aug 2020 21:28:43 +0900 Masami Hiramatsu <[email protected]> wrote:

Hi Muchun,

Quoted 5 lines · On Thu, 20 Aug 2020 11:19:33 +0800 Muchun Song <[email protected]> wrote:

If a kprobe is marked as gone, we should not kill it again. Otherwise, we can disarm the kprobe more than once. In that case, the statistics of kprobe_ftrace_enabled can unbalance which can lead to that kprobe do not work.

Good catch! Hmm, I think we also need an assertion in kill_kprobe() so that p is already gone. Anyway, this looks good to me.

Quoted 7 lines
Quoted 3 lines

Thanks!

Quoted 1 line
Fixes 0cb2f1372baa ("kprobes: Fix NULL pointer dereference at kprobe_ftrace_handler")

BTW, this fixes older bug than this commit.

Fixes e8386a0cb22f ("kprobes: support probing module __exit function")

Yeah, it's right. Thanks.

Quoted 45 lines

Thank you,

Quoted 30 lines
Quoted 30 lines
---
 kernel/kprobes.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
kernel/kprobes.c +5 -1
index d36e2b017588..7bac3ea44ff4 100644
@@ -2422,7 +2422,10 @@ static int kprobes_module_callback(struct notifier_block *nb,
    mutex_lock(&kprobe_mutex);
    for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
            head = &kprobe_table[i];
-           hlist_for_each_entry(p, head, hlist)
+           hlist_for_each_entry(p, head, hlist) {
+                   if (kprobe_gone(p))
+                           continue;
+
                    if (within_module_init((unsigned long)p->addr, mod) ||
                        (checkcore &&
                         within_module_core((unsigned long)p->addr, mod))) {
@@ -2439,6 +2442,7 @@ static int kprobes_module_callback(struct notifier_block *nb,
                             */
                            kill_kprobe(p);
                    }
+           }
    }
    if (val == MODULE_STATE_GOING)
Quoted 8 lines
Quoted 4 lines
kernel/kprobes.c +0 -1
            remove_module_kprobe_blacklist(mod);
--

2.11.0

--
Masami Hiramatsu <[email protected]>

--
Masami Hiramatsu <[email protected]>

Signature · 2 lines
Yours,
Muchun