[PATCH 00/20] arm64: FPSIMD/SVE/SME fixes + re-eanble SME

Mark Brown broonie at kernel.org
Wed May 7 04:26:30 PDT 2025


On Wed, May 07, 2025 at 10:56:54AM +0100, Mark Rutland wrote:

> If you have a specific objection, I can consider massaging the wording,
> but as it stands I don't see any problem with the commit message on
> patch 17. There clearly was an incorrect assumption during development,
> as indicated by the code, the comments, and the documentation. Providing
> that context makes review *much* easier.

I don't have a specific wording in the style you like right now, and
like I say I'm not sure it's too important since the actual chagne looks
good.  The main thing is just that the original thought was that we'd
not supply any register data unless in streaming mode which does due to
the use of one bit boil down to setting the FPSIMD value for the flag,
but the intent was more that the data format flag was rendered
irrelevant by the lack of data.  I was definitely aware that the two
flags were just a single bit - I distinctly remember seeing that and
thinking how unhelpful it was.  

Life would be a lot simpler and we probably wouldn't loose much if we
didn't support the FPSIMD format in the SVE regset but that ship is far
over the horizon.

> > I'm on holiday for a week and a bit more so probably
> > won't have time to review properly, other than a couple of the smaller
> > patches.

> No problem. If it's possible to throw any automated testing at this,
> that'd be much appreciated.

Yeah, might take a little bit since some other stuff blew up and my CI's
a bit miserable right now.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20250507/7e0aa2cc/attachment.sig>


More information about the linux-arm-kernel mailing list