Re: [PATCH] kretprobe: avoid re-registration of the same kretprobe earlier

[Masami Hiramatsu]

Hi ShaoBo,

On Wed, 2 Dec 2020 09:23:35 +0800
"Wangshaobo (bobo)" <bobo.shaobowang@xxxxxxxxxx> wrote:

> Hi steve, Masami,
> 
> Thanks for your works, i will check code again and modify properly 
> according to steve's suggestion.
> 

Can you update your patch and resend it?

Thank you,

> -- ShaoBo
> 
> 在 2020/12/2 7:32, Masami Hiramatsu 写道:
> > On Mon, 30 Nov 2020 16:18:50 -0500
> > Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> >
> >> Masami,
> >>
> >> Can you review this patch, and also, should this go to -rc and stable?
> >>
> >> -- Steve
> > Thanks for ping me!
> >
> >> On Tue, 24 Nov 2020 19:57:19 +0800
> >> Wang ShaoBo <bobo.shaobowang@xxxxxxxxxx> wrote:
> >>
> >>> Our system encountered a re-init error when re-registering same kretprobe,
> >>> where the kretprobe_instance in rp->free_instances is illegally accessed
> >>> after re-init.
> > Ah, OK. Anyway if re-register happens on kretprobe, it must lose instances
> > on the list before checking re-register in register_kprobe().
> > So the idea looks good to me.
> >
> >
> >>> Implementation to avoid re-registration has been introduced for kprobe
> >>> before, but lags for register_kretprobe(). We must check if kprobe has
> >>> been re-registered before re-initializing kretprobe, otherwise it will
> >>> destroy the data struct of kretprobe registered, which can lead to memory
> >>> leak, system crash, also some unexpected behaviors.
> >>>
> >>> we use check_kprobe_rereg() to check if kprobe has been re-registered
> >>> before calling register_kretprobe(), for giving a warning message and
> >>> terminate registration process.
> >>>
> >>> Signed-off-by: Wang ShaoBo <bobo.shaobowang@xxxxxxxxxx>
> >>> Signed-off-by: Cheng Jian <cj.chengjian@xxxxxxxxxx>
> >>> ---
> >>>   kernel/kprobes.c | 8 ++++++++
> >>>   1 file changed, 8 insertions(+)
> >>>
> >>> diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> >>> index 41fdbb7953c6..7f54a70136f3 100644
> >>> --- a/kernel/kprobes.c
> >>> +++ b/kernel/kprobes.c
> >>> @@ -2117,6 +2117,14 @@ int register_kretprobe(struct kretprobe *rp)
> >>>   		}
> >>>   	}
> >>>   
> >>> +	/*
> >>> +	 * Return error if it's being re-registered,
> >>> +	 * also give a warning message to the developer.
> >>> +	 */
> >>> +	ret = check_kprobe_rereg(&rp->kp);
> >>> +	if (WARN_ON(ret))
> >>> +		return ret;
> > If you call this here, you must make sure kprobe_addr() is called on rp->kp.
> > But if kretprobe_blacklist_size == 0, kprobe_addr() is not called before
> > this check. So it should be in between kprobe_on_func_entry() and
> > kretprobe_blacklist_size check, like this
> >
> > 	if (!kprobe_on_func_entry(rp->kp.addr, rp->kp.symbol_name, rp->kp.offset))
> > 		return -EINVAL;
> >
> > 	addr = kprobe_addr(&rp->kp);
> > 	if (IS_ERR(addr))
> > 		return PTR_ERR(addr);
> > 	rp->kp.addr = addr;
> >
> > 	ret = check_kprobe_rereg(&rp->kp);
> > 	if (WARN_ON(ret))
> > 		return ret;
> >
> >          if (kretprobe_blacklist_size) {
> > 		for (i = 0; > > +	ret = check_kprobe_rereg(&rp->kp);
> >
> >
> > Thank you,
> >
> >


-- 
Masami Hiramatsu <mhiramat@xxxxxxxxxx>