Re: [Cocci] [PATCH 1/1] kprobes: Deletion of an unnecessary check before the function call "module_put"

[Julia Lawall]

On Sun, 16 Nov 2014, SF Markus Elfring wrote:

> >> diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> >> index 3995f54..f1e7d45 100644
> >> --- a/kernel/kprobes.c
> >> +++ b/kernel/kprobes.c
> >> @@ -1527,8 +1527,7 @@ int register_kprobe(struct kprobe *p)
> >>  out:
> >>  	mutex_unlock(&kprobe_mutex);
> >>
> >> -	if (probed_mod)
> >> -		module_put(probed_mod);
> >> +	module_put(probed_mod);
> >
> > There is an out label, so please check whether the labels could not be
> > better positioned to avoid calling module_put when it is not needed.
>
> I do not see refactoring opportunities around jump labels in this use case
> for the implementation of the register_kprobe() function so far because
> the mutex_unlock() function must be called.
> Would you like to suggest any other source code fine-tuning?

OK.  I don't think that removing the if is a good choice in this case.
The code ret = check_kprobe_address_safe(p, &probed_mod); is unusual, in
that it can fail to do anything in two ways.  One is by setting ret, on
detecting an error, and the other is by returning 0 but still putting a
NULL value in probed_mod when there is nothing to do.  Thus, in the
successful execution of the rest of the function, a probed module might or
might not exist.  The if around the module_put is helpful to the reader to
understand that this possibility exists.

julia
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/