The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to stable@vger.kernel.org.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From b0399092ccebd9feef68d4ceb8d6219a8c0caa05 Mon Sep 17 00:00:00 2001
From: Muchun Song songmuchun@bytedance.com Date: Fri, 18 Sep 2020 21:20:21 -0700 Subject: [PATCH] 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: e8386a0cb22f ("kprobes: support probing module __exit function") Co-developed-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Muchun Song songmuchun@bytedance.com Signed-off-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Andrew Morton akpm@linux-foundation.org Acked-by: Masami Hiramatsu mhiramat@kernel.org Cc: "Naveen N . Rao" naveen.n.rao@linux.ibm.com Cc: Anil S Keshavamurthy anil.s.keshavamurthy@intel.com Cc: David S. Miller davem@davemloft.net Cc: Song Liu songliubraving@fb.com Cc: Steven Rostedt rostedt@goodmis.org Cc: stable@vger.kernel.org Link: https://lkml.kernel.org/r/20200822030055.32383-1-songmuchun@bytedance.com Signed-off-by: Linus Torvalds torvalds@linux-foundation.org
diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 287b263c9cb9..049da84e1952 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -2140,6 +2140,9 @@ static void kill_kprobe(struct kprobe *p)
lockdep_assert_held(&kprobe_mutex);
+ if (WARN_ON_ONCE(kprobe_gone(p))) + return; + p->flags |= KPROBE_FLAG_GONE; if (kprobe_aggrprobe(p)) { /* @@ -2419,7 +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))) { @@ -2436,6 +2442,7 @@ static int kprobes_module_callback(struct notifier_block *nb, */ kill_kprobe(p); } + } } if (val == MODULE_STATE_GOING) remove_module_kprobe_blacklist(mod);
On Mon, 21 Sep 2020 14:25:19 +0200 gregkh@linuxfoundation.org wrote:
The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to stable@vger.kernel.org.
thanks,
greg k-h
The below should apply and work for 5.4 - 4.4.
Masami, can you verify that I did this correctly. It appears to pass all the ftrace tests, at least all the ones that pass without the patch ;-)
-- Steve
From b0399092ccebd9feef68d4ceb8d6219a8c0caa05 Mon Sep 17 00:00:00 2001
From: Muchun Song songmuchun@bytedance.com Date: Fri, 18 Sep 2020 21:20:21 -0700 Subject: [PATCH] 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: e8386a0cb22f ("kprobes: support probing module __exit function") Co-developed-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Muchun Song songmuchun@bytedance.com Signed-off-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Andrew Morton akpm@linux-foundation.org Acked-by: Masami Hiramatsu mhiramat@kernel.org Cc: "Naveen N . Rao" naveen.n.rao@linux.ibm.com Cc: Anil S Keshavamurthy anil.s.keshavamurthy@intel.com Cc: David S. Miller davem@davemloft.net Cc: Song Liu songliubraving@fb.com Cc: Steven Rostedt rostedt@goodmis.org Cc: stable@vger.kernel.org Link: https://lkml.kernel.org/r/20200822030055.32383-1-songmuchun@bytedance.com Signed-off-by: Linus Torvalds torvalds@linux-foundation.org
Index: linux-test.git/kernel/kprobes.c =================================================================== --- linux-test.git.orig/kernel/kprobes.c +++ linux-test.git/kernel/kprobes.c @@ -2088,6 +2088,9 @@ static void kill_kprobe(struct kprobe *p { struct kprobe *kp;
+ if (WARN_ON_ONCE(kprobe_gone(p))) + return; + p->flags |= KPROBE_FLAG_GONE; if (kprobe_aggrprobe(p)) { /* @@ -2270,7 +2273,10 @@ static int kprobes_module_callback(struc mutex_lock(&kprobe_mutex); for (i = 0; i < KPROBE_TABLE_SIZE; i++) { head = &kprobe_table[i]; - hlist_for_each_entry_rcu(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))) { @@ -2287,6 +2293,7 @@ static int kprobes_module_callback(struc */ kill_kprobe(p); } + } } mutex_unlock(&kprobe_mutex); return NOTIFY_DONE;
On Thu, 24 Sep 2020 18:06:06 -0400 Steven Rostedt rostedt@goodmis.org wrote:
On Mon, 21 Sep 2020 14:25:19 +0200 gregkh@linuxfoundation.org wrote:
The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to stable@vger.kernel.org.
thanks,
greg k-h
The below should apply and work for 5.4 - 4.4.
Masami, can you verify that I did this correctly. It appears to pass all the ftrace tests, at least all the ones that pass without the patch ;-)
Thanks Steve! I think we should keep using hlist_for_each_entry_rcu() unless backporting commit 7e6a71d8e601 ("kprobes: Use non RCU traversal APIs on kprobe_tables if possible") (and this commit involves another improvement.)
From b2dafed7da5e4ee5e82fdfe309d69ca6e189ef0d Mon Sep 17 00:00:00 2001
From: Muchun Song songmuchun@bytedance.com Date: Fri, 18 Sep 2020 21:20:21 -0700 Subject: [PATCH] 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: e8386a0cb22f ("kprobes: support probing module __exit function") Co-developed-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Muchun Song songmuchun@bytedance.com Signed-off-by: Chengming Zhou zhouchengming@bytedance.com Signed-off-by: Andrew Morton akpm@linux-foundation.org Acked-by: Masami Hiramatsu mhiramat@kernel.org Cc: "Naveen N . Rao" naveen.n.rao@linux.ibm.com Cc: Anil S Keshavamurthy anil.s.keshavamurthy@intel.com Cc: David S. Miller davem@davemloft.net Cc: Song Liu songliubraving@fb.com Cc: Steven Rostedt rostedt@goodmis.org Cc: stable@vger.kernel.org Link: https://lkml.kernel.org/r/20200822030055.32383-1-songmuchun@bytedance.com Signed-off-by: Linus Torvalds torvalds@linux-foundation.org --- kernel/kprobes.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/kernel/kprobes.c b/kernel/kprobes.c index bbff4bccb885..5646f291eb70 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -2088,6 +2088,9 @@ static void kill_kprobe(struct kprobe *p) { struct kprobe *kp;
+ if (WARN_ON_ONCE(kprobe_gone(p))) + return; + p->flags |= KPROBE_FLAG_GONE; if (kprobe_aggrprobe(p)) { /* @@ -2270,7 +2273,10 @@ 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_rcu(p, head, hlist) + hlist_for_each_entry_rcu(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))) { @@ -2287,6 +2293,7 @@ static int kprobes_module_callback(struct notifier_block *nb, */ kill_kprobe(p); } + } } mutex_unlock(&kprobe_mutex); return NOTIFY_DONE;
linux-stable-mirror@lists.linaro.org