Re: [PATCH 1/4] arm64/sve: Remove bitrotted comment about syscall behaviour

From: Dave Martin <Dave.Martin@arm.com>
To: Mark Brown <broonie@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will@kernel.org>, Jonathan Corbet <corbet@lwn.net>,
	linux-arm-kernel@lists.infradead.org, linux-doc@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Edmund Grimley-Evans <edmund.grimley-evans@arm.com>
Subject: Re: [PATCH 1/4] arm64/sve: Remove bitrotted comment about syscall behaviour
Date: Wed, 24 Jan 2024 14:04:42 +0000	[thread overview]
Message-ID: <ZbEY+n11M9lQGsWA@e133380.arm.com> (raw)
In-Reply-To: <ede05e2d-0cde-4de1-b2db-d40df19d7075@sirena.org.uk>

On Tue, Jan 23, 2024 at 06:11:52PM +0000, Mark Brown wrote:
> On Tue, Jan 23, 2024 at 05:54:17PM +0000, Dave Martin wrote:
> 
> > I still feel that it is iffy practice for userspace to rely on the
> > extra bits being zeroed -- I think the architecture hides this
> > guarantee anyway whenever you go through a function call confirming to
> > the regular procedure call standard (including the syscall wrappers).
> > But there may not be a lot of point trying to put people off if we
> > can't force them not to rely on it.
> 
> I do tend to agree that the requirement to zero is excessively zealous
> and that the risk from relaxing it is minor (it's stricter than the
> function call ABI), I did leave a sysctl as a mechanism for restoring
> compatibility in the case where we did run into issues in my original
> series but I didn't expect to need it.  If you convince everyone else
> I'd be happy to relax things but I don't super care either way.

[...]

I don't feel that strongly about it.

Ideally we'd have gone for the fully relaxed approach from the start,
but it's hard to test whether "unspecified" registers aren't leaking
data from somewhere they shouldn't.

Given that the decision has been made anyway, the documentation should
not send mixed messages, so:

Reviewed-by: Dave Martin <Dave.Martin@arm.com>