Re: [PATCH] sparc64: Read qemu,hwcap-list from the OF cpu node on sun4u
From: Stian Halseth
Date: Wed Nov 18 2026 - 06:16:02 EST
Hi Eric,
On Tue, 2026-11-17 at 20:38 -0800, Eric Biggers wrote:
> The following series migrates the remaining SPARC crypto opcodes
> accelerated AES modes into lib/crypto/:
> https://lore.kernel.org/linux-crypto/20261118041033.50014-1-ebiggers@xxxxxxxxxx/
Thanks, that was quick! I'll send a v2 of this patch with your
Tested-by. It also gets a new first patch that addresses the Sashiko
finding.
I've also tested directly from your branch. With +crypto, the KUnit
tests and the full self-tests pass. To make sure the library really
uses the opcodes, I broke the AES round instruction in a local QEMU
build. The AES tests then failed with +crypto and passed without it.
I'll also try it on real hardware, a T4 and an M7, and reply on your
series.
> I think this is the only case where a Linux architecture has
> restrictions on when memcpy() can be called, though. It would be worth
> finding a way to remove that footgun
Agreed. Not a full proposition (yet), but perhaps U1memcpy and
U3memcpy could fall back to integer copies when the FPU is in use,
like NG4memcpy and M7memcpy do. I haven't properly looked into it,
but I might later.
/Stian
Attachment:
signature.asc
Description: This is a digitally signed message part