Re: kprobes: propagate error from arm_kprobe_ftrace()

From: Jessica Yu
Date: Wed Jan 03 2018 - 16:00:08 EST

+++ Steven Rostedt [03/01/18 09:33 -0500]:
On Wed, 3 Jan 2018 02:40:47 +0100
Jessica Yu <jeyu@xxxxxxxxxx> wrote:

Improve error handling when arming ftrace-based kprobes. Specifically, if
we fail to arm a ftrace-based kprobe, register_kprobe()/enable_kprobe()
should report an error instead of success. Previously, this has lead to
confusing situations where register_kprobe() would return 0 indicating
success, but the kprobe would not be functional if ftrace registration
during the kprobe arming process had failed. We should therefore take any
errors returned by ftrace into account and propagate this error so that we
do not register/enable kprobes that cannot be armed. This can happen if,
for example, register_ftrace_function() finds an IPMODIFY conflict (since
kprobe_ftrace_ops has this flag set) and returns an error. Such a conflict
is possible since livepatches also set the IPMODIFY flag for their ftrace_ops.

arm_all_kprobes() keeps its current behavior and attempts to arm all
kprobes. It returns the last encountered error and gives a warning if
not all probes could be armed.

This patch is based on Petr Mladek's original patchset (patches 2 and 3)
back in 2015, which improved kprobes error handling, found here:

https://lkml.org/lkml/2015/2/26/452

However, further work on this had been paused since then and the patches
were not upstreamed.

Based-on-patches-by: Petr Mladek <pmladek@xxxxxxxx>
Signed-off-by: Jessica Yu <jeyu@xxxxxxxxxx>
---
kernel/kprobes.c | 94 +++++++++++++++++++++++++++++++++++++++++---------------
1 file changed, 69 insertions(+), 25 deletions(-)

diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index b4aab48ad258..ae6b6fe79de3 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -988,18 +988,32 @@ static int prepare_kprobe(struct kprobe *p)
}

/* Caller must lock kprobe_mutex */
-static void arm_kprobe_ftrace(struct kprobe *p)
+static int arm_kprobe_ftrace(struct kprobe *p)
{
- int ret;
+ int ret = 0;

ret = ftrace_set_filter_ip(&kprobe_ftrace_ops,
(unsigned long)p->addr, 0, 0);
- WARN(ret < 0, "Failed to arm kprobe-ftrace at %p (%d)\n", p->addr, ret);
- kprobe_ftrace_enabled++;
- if (kprobe_ftrace_enabled == 1) {
+ if (WARN(ret < 0, "Failed to arm kprobe-ftrace at %p (%d)\n", p->addr, ret))
+ return ret;

I wonder if we should change this from a WARN to a printk(). No reason
to do stack dumps here.

Yeah, I was trying to preserve the current behavior. I'll leave it up
to Masami.

+
+ if (kprobe_ftrace_enabled == 0) {
ret = register_ftrace_function(&kprobe_ftrace_ops);
- WARN(ret < 0, "Failed to init kprobe-ftrace (%d)\n", ret);
+ if (WARN(ret < 0, "Failed to init kprobe-ftrace (%d)\n", ret))
+ goto err_ftrace;
}
+
+ kprobe_ftrace_enabled++;
+ return ret;
+
+err_ftrace:
+ /*
+ * Note: Since kprobe_ftrace_ops has IPMODIFY set, and ftrace requires a
+ * non-empty filter_hash for IPMODIFY ops, we're safe from an accidental
+ * empty filter_hash which would undesirably trace all functions.
+ */
+ ftrace_set_filter_ip(&kprobe_ftrace_ops, (unsigned long)p->addr, 1, 0);
+ return ret;
}

/* Caller must lock kprobe_mutex */
@@ -1018,22 +1032,23 @@ static void disarm_kprobe_ftrace(struct kprobe *p)
}
#else /* !CONFIG_KPROBES_ON_FTRACE */
#define prepare_kprobe(p) arch_prepare_kprobe(p)
-#define arm_kprobe_ftrace(p) do {} while (0)
+#define arm_kprobe_ftrace(p) (0)

Hmm. Perhaps we should have arm_kprobe_ftrace() return a failure.

#define disarm_kprobe_ftrace(p) do {} while (0)
#endif

/* Arm a kprobe with text_mutex */
-static void arm_kprobe(struct kprobe *kp)
+static int arm_kprobe(struct kprobe *kp)
{
- if (unlikely(kprobe_ftrace(kp))) {
- arm_kprobe_ftrace(kp);
- return;
- }
+ if (unlikely(kprobe_ftrace(kp)))
+ return arm_kprobe_ftrace(kp);

If CONFIG_KPROBES_ON_FTRACE is not defined, this if should always be
false. But if for some reason in the future, it is not false, we just
had arm_kprobe_ftrace() return success when it really is a failure.

-ENODEV?

Good point, I will include this change in v4, unless there are
objections.

+
cpus_read_lock();
mutex_lock(&text_mutex);
__arm_kprobe(kp);
mutex_unlock(&text_mutex);
cpus_read_unlock();
+
+ return 0;
}

/* Disarm a kprobe with text_mutex */
@@ -1372,9 +1387,15 @@ static int register_aggr_kprobe(struct kprobe *orig_p, struct kprobe *p)

if (ret == 0 && kprobe_disabled(ap) && !kprobe_disabled(p)) {
ap->flags &= ~KPROBE_FLAG_DISABLED;
- if (!kprobes_all_disarmed)
+ if (!kprobes_all_disarmed) {
/* Arm the breakpoint again. */
- arm_kprobe(ap);
+ ret = arm_kprobe(ap);
+ if (ret) {
+ ap->flags |= KPROBE_FLAG_DISABLED;
+ list_del_rcu(&p->list);

Don't we need to hold the mutex to modify the list?

It is unfortunately unclear from this snippet, but we do hold the
kprobe_mutex here. It's held for most of the duration of
register_kprobe(), where register_aggr_kprobe() is called.

+ synchronize_sched();
+ }
+ }
}
return ret;
}
@@ -1594,8 +1615,14 @@ int register_kprobe(struct kprobe *p)
hlist_add_head_rcu(&p->hlist,
&kprobe_table[hash_ptr(p->addr, KPROBE_HASH_BITS)]);

- if (!kprobes_all_disarmed && !kprobe_disabled(p))
- arm_kprobe(p);
+ if (!kprobes_all_disarmed && !kprobe_disabled(p)) {
+ ret = arm_kprobe(p);
+ if (ret) {
+ hlist_del_rcu(&p->hlist);

Same here.

We do hold kprobe_mutex here as well (see above comment).

+ synchronize_sched();
+ goto out;
+ }
+ }

/* Try to optimize kprobe */
try_to_optimize_kprobe(p);
@@ -2137,7 +2164,9 @@ int enable_kprobe(struct kprobe *kp)

if (!kprobes_all_disarmed && kprobe_disabled(p)) {
p->flags &= ~KPROBE_FLAG_DISABLED;
- arm_kprobe(p);
+ ret = arm_kprobe(p);
+ if (ret)
+ p->flags |= KPROBE_FLAG_DISABLED;
}
out:
mutex_unlock(&kprobe_mutex);
@@ -2565,11 +2594,12 @@ static const struct file_operations debugfs_kprobe_ei_ops = {
.release = seq_release,
};

-static void arm_all_kprobes(void)
+static int arm_all_kprobes(void)
{
struct hlist_head *head;
struct kprobe *p;
- unsigned int i;
+ unsigned int i, errors = 0;
+ int err, ret = 0;

mutex_lock(&kprobe_mutex);

@@ -2586,16 +2616,26 @@ static void arm_all_kprobes(void)
/* Arming kprobes doesn't optimize kprobe itself */
for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
head = &kprobe_table[i];
- hlist_for_each_entry_rcu(p, head, hlist)
- if (!kprobe_disabled(p))
- arm_kprobe(p);
+ /* Arm all kprobes on a best-effort basis */
+ hlist_for_each_entry_rcu(p, head, hlist) {
+ if (!kprobe_disabled(p)) {
+ err = arm_kprobe(p);
+ if (err) {
+ errors++;
+ ret = err;
+ }
+ }
+ }
}

- printk(KERN_INFO "Kprobes globally enabled\n");
+ if (errors)
+ pr_warn("Kprobes globally enabled, but failed to arm %d probes\n", errors);

Perhaps we should have a count of all kprobes that were tried, and
write something like:

"Kprobes globally enabled, but failed to arm %d out of %d probes\n",
errors, total

Sure, ok.

Thank you for the review!

Jessica