Re: [PATCH] fix calls to request_module()

[Al Viro]

On Thu, Dec 11, 2008 at 01:23:36PM +0900, Nguyen Anh Quynh wrote:

> So I checked again by fixing the code that should be compiled
> (sound/core/sound.c), and can confirm that without the patch we got
> warning like below:
> 
> sound/core/sound.c: In function 'snd_request_other':
> sound/core/sound.c:91: warning: format not a string literal and no
> format arguments

Ah, but that's different.  Take a look at that warning and think _why_
it is given and what is it about.  Getting an untrusted string as
format argument is a real security hole, but it has nothing to do
with a pile of cases in your patch.

FWIW, even in this case (where the warning is a false positive, BTW -
the string passed there can have one of two values and both are safe)
I would simply do
	switch(...) {
	case .... : request_module("snd-seq"); break;
	case .... : request_module("snd-timer"); break;
	}
and that's it.

Slapping "%s" here actually obfuscates the code, without making it safer.
--
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/