Re: [PATCH v3 1/2] minmax: sanity check constant bounds when clamping

From: Kees Cook
Date: Tue Oct 04 2022 - 10:40:14 EST

On October 4, 2022 6:41:48 AM PDT, "Jason A. Donenfeld" <Jason@xxxxxxxxx> wrote:
>On Mon, Sep 26, 2022 at 8:26 PM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>> On Mon, Sep 26, 2022 at 03:34:34PM +0200, Jason A. Donenfeld wrote:
>> > The clamp family of functions only makes sense if hi>=lo. If hi and lo
>> > are compile-time constants, then raise a build error. Doing so has
>> > already caught buggy code. This also introduces the infrastructure to
>> > improve the clamping function in subsequent commits.
>> >
>> > Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
>> > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
>> > Cc: Kees Cook <keescook@xxxxxxxxxxxx>
>> > Signed-off-by: Jason A. Donenfeld <Jason@xxxxxxxxx>
>> Reviewed-by: Kees Cook <keescook@xxxxxxxxxxxx>
>Wondering - did you ever queue this up for 6.1? I assume the plan is
>to hold off on 2/2 for the time being, but this 1/2 is good to have
>either way.

Since it produced at least one warning, there may be others in weird archs/configs, so I wanted it to bake in -next after the merge window for 6.1 closes. It's a good feature, but I didn't want to risk new build warnings so close to the merge. :)

Kees Cook