Re: [PATCH v5 00/21] kprobes: Unify kretprobe trampoline handlers and make kretprobe lockless

[Masami Hiramatsu]

On Thu, 3 Sep 2020 10:39:54 +0900
Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:

> OK, I've confirmed that the lockdep warns on kretprobe from INT3
> with your fix. Of course make it lockless then warning is gone.
> But even without the lockless patch, this warning can be false-positive
> because we prohibit nested kprobe call, right?
> 
> If the kprobe user handler uses a spinlock, the spinlock is used
> only in that handler (and in the context between kprobe_busy_begin/end),
> it will be safe since the spinlock is not nested.
> But if the spinlock is shared with other context, it will be dangerous
> because it can be interrupted by NMI (including INT3). This also applied
> to the function which is called from kprobe user handlers, thus user
> has to take care of it.

Sorry, for noticing this point, I Cc'd to systemtap. Is systemtap taking
care of spinlock too?

Thank you,

-- 
Masami Hiramatsu <mhiramat@xxxxxxxxxx>