Re: WARNING at arch/x86/kernel/alternative.c:707 text_poke+0x25d/0x270

From: Thomas Gleixner
Date: Tue May 16 2017 - 17:43:12 EST


On Wed, 17 May 2017, Masami Hiramatsu wrote:
> On Tue, 16 May 2017 09:48:02 -0400
> Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> > It appears that the kprobe_optimizer work thread call happened after
> > the init pages were freed, causing alternative.c to give the above
> > warning because the text that is being unoptimized happens to no longer
> > exist.
>
> Ah, I see. I need to check that case. Actually for the module
> init text area, kill_kprobe() correctly kicks kill_optimized_kprobe()
> so it should safe. But above case is on the init-text in kernel
> itself. I guess module_notifier may not be called for that area...

Find below the patch I'm using for now .

Thanks,

tglx

8<--------------------

--- a/include/linux/kprobes.h
+++ b/include/linux/kprobes.h
@@ -349,6 +349,9 @@ extern int proc_kprobes_optimization_han
int write, void __user *buffer,
size_t *length, loff_t *ppos);
#endif
+extern void wait_for_kprobe_optimizer(void);
+#else
+static inline void wait_for_kprobe_optimizer(void) { }
#endif /* CONFIG_OPTPROBES */
#ifdef CONFIG_KPROBES_ON_FTRACE
extern void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip,
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -595,7 +595,7 @@ static void kprobe_optimizer(struct work
}

/* Wait for completing optimization and unoptimization */
-static void wait_for_kprobe_optimizer(void)
+void wait_for_kprobe_optimizer(void)
{
mutex_lock(&kprobe_mutex);

--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -1535,6 +1535,11 @@ static __init int kprobe_trace_self_test

end:
release_all_trace_kprobes();
+ /*
+ * Wait for the optimizer work to finish. Otherwise it might fiddle
+ * with probes in already freed __init text.
+ */
+ wait_for_kprobe_optimizer();
if (warn)
pr_cont("NG: Some tests are failed. Please check them.\n");
else