Re: x86/kprobes: kretprobe fails to triggered if kprobe at function entry is not optimized (trigger by int3 breakpoint)

[Masami Hiramatsu <mhiramat@kernel.org>]

Hi Peter,

On Tue, 25 Aug 2020 14:09:11 +0200
peterz@infradead.org wrote:

> 
> @@ -1934,50 +1884,28 @@ unsigned long __weak arch_deref_entry_point(void *entry)
>  static int pre_handler_kretprobe(struct kprobe *p, struct pt_regs *regs)
>  {
>  	struct kretprobe *rp = container_of(p, struct kretprobe, kp);
> -	unsigned long hash, flags = 0;
>  	struct kretprobe_instance *ri;
> +	struct llist_node *llist;
>  
> -	/*
> -	 * To avoid deadlocks, prohibit return probing in NMI contexts,
> -	 * just skip the probe and increase the (inexact) 'nmissed'
> -	 * statistical counter, so that the user is informed that
> -	 * something happened:
> -	 */
> -	if (unlikely(in_nmi())) {
> +	llist = llist_del_first(&rp->free_instances);
> +	if (!llist) {
>  		rp->nmissed++;
>  		return 0;
>  	}

Would we need a lock around llist_del_first(&rp->free_instance) here?

linux/llist.h said,

 * Cases where locking is not needed:
 * If there are multiple producers and multiple consumers, llist_add can be
 * used in producers and llist_del_all can be used in consumers simultaneously
 * without locking. Also a single consumer can use llist_del_first while
                               ^^^^^^^^^^^^^^^^^^^^^^^
 * multiple producers simultaneously use llist_add, without any locking.
 *
 * Cases where locking is needed:
 * If we have multiple consumers with llist_del_first used in one consumer, and
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 * llist_del_first or llist_del_all used in other consumers, then a lock is
 * needed.

pre_handler_kretprobe() can be invoked simultaneously on the different CPUs
if those are calling the same probed function.


Thank you,

-- 
Masami Hiramatsu <mhiramat@kernel.org>