[PATCH 1/4] locking/atomic/x86: Silence intentional wrapping addition

[Peter Zijlstra]

On Wed, Apr 24, 2024 at 04:30:50PM -0700, Kees Cook wrote:

> > That is, anything that actively warns about signed overflow when build
> > with -fno-strict-overflow is a bug. If you want this warning you have to
> > explicitly mark things.
> 
> This is confusing UB with "overflow detection". We're doing the latter.

Well, all of this is confusing to me because it is not presented
coherently.

The traditional 'must not let signed overflow' is because of the UB
nonsense, which we fixed.

> > Signed overflow is not UB, is not a bug.
> > 
> > Now, it might be unexpected in some places, but fundamentally we run on
> > 2s complement and expect 2s complement. If you want more, mark it so.
> 
> Regular C never provided us with enough choice in types to be able to
> select the overflow resolution strategy. :( So we're stuck mixing
> expectations into our types.

Traditionally C has explicit wrapping for unsigned and UB on signed. We
fixed the UB, so now expect wrapping for everything.

You want to add overflow, so you should make that a special and preserve
semantics for existing code.

Also I would very strongly suggest you add an overflow qualifier to the
type system and please provide sane means of qualifier manipulation --
stripping qualifiers is painful :/

> Regardless, yes, someone intent on wrapping gets their expected 2s
> complement results, but in the cases were a few values started collecting
> in some dark corner of protocol handling, having a calculation wrap around
> is at best a behavioral bug and at worst a total system compromise.
> Wrapping is the uncommon case here, so we mark those.

Then feel free to sprinkle copious amounts of 'overflow' qualifiers in
the protocol handling code.