Re: [PATCH v6] kernel.h: Retain constant expression output for max()/min()

[Kees Cook]

On Fri, Mar 30, 2018 at 10:34 PM, Ingo Molnar <mingo@xxxxxxxxxx> wrote:
>
> * Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
>> On Mon, Mar 26, 2018 at 10:47 PM, Ingo Molnar <mingo@xxxxxxxxxx> wrote:
>> >
>> > * Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>> >
>> >> In the effort to remove all VLAs from the kernel[1], it is desirable to
>> >> build with -Wvla. However, this warning is overly pessimistic, in that
>> >> it is only happy with stack array sizes that are declared as constant
>> >> expressions, and not constant values. One case of this is the evaluation
>> >> of the max() macro which, due to its construction, ends up converting
>> >> constant expression arguments into a constant value result.
>> >>
>> >> All attempts to rewrite this macro with __builtin_constant_p() failed with
>> >> older compilers (e.g. gcc 4.4)[2]. However, Martin Uecker constructed[3] a
>> >> mind-shattering solution that works everywhere. Cthulhu fhtagn!
>> >>
>> >> This patch updates the min()/max() macros to evaluate to a constant
>> >> expression when called on constant expression arguments. This removes
>> >> several false-positive stack VLA warnings from an x86 allmodconfig
>> >> build when -Wvla is added:
>> >
>> > Cool!
>> >
>> > Acked-by: Ingo Molnar <mingo@xxxxxxxxxx>
>> >
>> > How many warnings are left in an allmodconfig build?
>>
>> For -Wvla? Out of the original 112 files with VLAs, 42 haven't had a
>> patch applied yet. Doing a linux-next allmodconfig build with the
>> max() patch and my latest ecc patch, we've gone from 316 warning
>> instances to 205. More than half of those are in
>> include/crypto/skcipher.h and include/crypto/hash.h.
>
> Great - once the number of warnings is zero, is the plan to enable the warning
> unconditionally?

That's the plan, yes. Like inbox-zero but for VLAs. ;)

-Kees

-- 
Kees Cook
Pixel Security