Re: [PATCH 11/13] crypto: x86/sha256 - implement library instead of shash

[Eric Biggers]

On Sun, Apr 27, 2025 at 08:18:56AM +0800, Herbert Xu wrote:
> On Sat, Apr 26, 2025 at 11:03:26AM -0700, Eric Biggers wrote:
> >
> > The SHA-256 library functions currently work in any context, and this patch
> > series preserves that behavior.  Changing that would be a separate change.
> 
> I've already removed the SIMD fallback path and your patch is
> adding it back.

While you've been pushing out a lot of random broken changes to shash recently,
the SHA-256 library functions weren't SIMD-optimized until this patchset.

> > But also as I've explained before, for the library API the performance benefit
> > of removing the crypto_simd_usable() doesn't seem to be worth the footgun that
> > would be introduced.  Your position is, effectively, that if someone calls one
> > of the sha256*() functions from a hardirq, we should sometimes corrupt a random
> > task's FPU registers.  That's a really bad bug that is very difficult to
> > root-cause.  My position is that we should make it just work as expected.
> 
> kernel_fpu_begin already does a WARN_ON when called in hardirq
> context and it can't safely use the FPU, there is no silent
> corruption.

Only when CONFIG_X86_DEBUG_FPU is enabled, which people don't enable in
production.  And even if that is enabled, it's just a WARN, so the registers
still get used and corrupted anyway.

- Eric