Re: [PATCH v6 9/9] vsprintf: Avoid confusion between invalid address and value

From: Petr Mladek
Date: Wed Feb 20 2019 - 04:24:30 EST


On Tue 2019-02-19 13:06:17, Andy Shevchenko wrote:
> On Tue, Feb 19, 2019 at 5:07 AM Sergey Senozhatsky
> <sergey.senozhatsky.work@xxxxxxxxx> wrote:
>
> > Suppose, in my driver I want to sprintf() IPv4 address. The longest
> > possible address is 3 * 4 (%d%d%d) + 3 bytes (dots) + terminating NULL.
> > E.g. 111.111.111.111
> >
> > forcing sprintf() to write "(invalid address)" to a 16-bytes buffer,
> > but the thing is - strlen("(invalid address)") > 16.

Good catch!

> It would print as many characters as buffer allows. In your case above
> the use of sprintf() is a bit fragile.

Heh, I do not see it used anywhere in the code. Only "%pi6" is used
in a real code and always a safe way.

> But yes, the error messages should not be longer than 8 / 16 bytes
> depending on 32- or 64-bit build we have.

Well, better be on the safe side. I'll move it at the beginning
of the patchset.

Best Regards,
Petr

PS: I am a bit busy with some other things. I'll send v7 later.
I think that it is a material for 5.2. This patchset would deserve
some testing in linux-next and we are already too close to
the 5.1 merge window.