linux-kernel - Re: [PATCH v2 00/17] SHA-512 library functions

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20250617060523.GH8289@sol>
Date: Mon, 16 Jun 2025 23:05:23 -0700
From: Eric Biggers <ebiggers@...nel.org>
To: linux-crypto@...r.kernel.org
Cc: linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
	linux-mips@...r.kernel.org, linux-riscv@...ts.infradead.org,
	linux-s390@...r.kernel.org, sparclinux@...r.kernel.org,
	x86@...nel.org, Ard Biesheuvel <ardb@...nel.org>,
	"Jason A . Donenfeld" <Jason@...c4.com>,
	Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH v2 00/17] SHA-512 library functions

On Sun, Jun 15, 2025 at 06:40:02PM -0700, Eric Biggers wrote:
> - Tests are KUnit tests, and they are fairly thorough (more thorough
>   than crypto/testmgr.c) and also optionally include benchmarks.

An additional note on testing: I have scripts that build the kernel for all the
arches that have arch-specific code in lib/crc/ or lib/crypto/, launch them in
QEMU with various -cpu options, and gather the results of the tests and any
other issues like warns or panics.

I'll get it into a sharable form at some point.

As far as the coverage of the arch-specific code in this specific patchset goes,
I've verified that my testing strategy covers all sha512_blocks() code paths,
including fallbacks, on arm, arm64, s390, riscv, and x86.

The two incomplete ones are mips and sparc, where I cannot test their optimized
code paths in sha512_blocks() because QEMU does not support it.

Still, I don't expect any issues.  That code is ultimately doing the same thing
as it was before for SHA-512 block processing, just integrated in a simpler way.

FWIW, my policy going forward is that any new arch-specific code in lib/crc/ or
lib/crypto/ *MUST* come with QEMU support so that it can be tested.  It's only
migration of existing code (usually from arch/*/crypto/) like this where I may
tolerate not being able to test it; that code gets "grandfathered in"...

- Eric