Re: aes: self-test failed

[Nick Hudson <nick.hudson%gmx.co.uk@localhost>]

On 18/03/2023 16:38, Taylor R Campbell wrote:
> > Date: Sat, 18 Mar 2023 08:16:07 +0900 (JST)
> > From: Mori Hiroki <yamori813%yahoo.co.jp@localhost>
> >
> > I have as flow error sometime at boot.
> > [...]
> > [   1.0300028] BearSSL aes_ct AES-192-CBC enc: self-test failed
>
> This sounds likely to be a hardware error.  We run the same tests of
> the same code on many different architectures including arm32, and we
> probably would have noticed by now if the BearSSL aes_ct code were
> broken.
>
> Another possibility is that there is a bug in restoring registers on
> interrupt or context switch in armv6.  But I think we can rule out
> this possibility because interrupts should still be blocked by this
> point.

I think interrupts are enabled here.

We saw a similar BearSSL self-test failed on risc-v until this fix

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/riscv/riscv/cpu_switch.S.diff?r1=1.1&r2=1.2&only_with_tag=MAIN

That said, I be surprised if there is a unnoticed bug in the armv6 
exception handlers, but you never know.

Nick