On Fri, May 27, 2022 at 2:16 PM Jason A. Donenfeld <Jason@xxxxxxxxx> wrote:
>
> Whoops. I sent this to linux-crypto/Herbert assuming that it'd work
> with the "2 weeks pass and nobody complains --> queued up" flow of
> that list.
Well, it's more like "4 months since I complained about this thing",
so I thought it was overdue and applied it..
That said, I don't think CRYPTO_LIB_POLY1305_GENERIC can be set unless
CRYPTO_LIB_POLY1305 is set. And CRYPTO_ARCH_HAVE_LIB_POLY1305 looks
like a bogus thing to test for too - those two just decide if the
arch-specific one or the generic one should be used.
So I think v1 was actually the right one. Possibly
depends on CRYPTO_LIB_POLY1305 != n
due to the tristate?
Linus
