DMARC-Filter: OpenDMARC Filter v1.4.2 delorie.com 4AKJZmMa3752374
DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 1DF103858039
Date: Wed, 20 Nov 2024 11:34:36 -0800 (PST)
To: cygwin AT cygwin DOT com
Subject: include tag in uname -s for ARM64 hosts
Message-ID: <833327a2-3a62-ed26-e4ba-5302bcdb8000@jdrake.com>
MIME-Version: 1.0
From: Jeremy Drake via Cygwin <cygwin AT cygwin DOT com>
Reply-To: Jeremy Drake <cygwin AT jdrake DOT com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Sender: "Cygwin" <cygwin-bounces~archive-cygwin=delorie DOT com AT cygwin DOT com>

I mentioned this briefly on cygwin-patches during the review of the fix
for a deadlock when running under emulation on ARM64 hosts, but I wanted
to discuss it more thoroughly here now that that is merged.

It seems like it would be useful, perhaps for stats tracking purposes like
MSYS2 does [1], to indicate when Cygwin is running under emulation in
uname.  When i686 Cygwin was supported, there was a `-WOW64` tag appended
to the sysname field (uname -s) when it was running under WOW64, so it
seems to make sense to me to do something similar (say, appending
`-ARM64`).

There was some concern that appending -ARM64 might cause confusion with a
future native ARM64 version of Cygwin [2], but technically it would not
because the architecture of Cygwin is reported in the "machine" field of
uname (uname -m), while my proposal is to append the host architecture to
the sysname field (uname -s), only when it doesn't match the architecture
of Cygwin iteself.

I'm prototyping this based on Cygwin 3.3.6, because still supporting i686
results in a more complete test matrix (I don't want to come up with
something that isn't extensible if Microsoft ever adds something new to
the matrix):

Cygwin	Host	Suffix
i686	i686	(N/A)
i686	x86_64	-WOW64-x64 (or omit arch to match existing?)
i686	ARM64	-WOW64-ARM64
x86_64	x86_64	(N/A)
x86_64	ARM64	-ARM64 (x86_64-on-ARM64 is not considered WOW64 by MS)
Theoretical:
ARM64	ARM64	(N/A)
Not worth worrying about:
ARM	ARM64	-WOW64-ARM (?)
I'm mostly just curious how the algorithm would act there, there should be
*zero* interest in actually porting Cygwin to 32-bit ARM, MS just dropped
support for 32-bit ARM binaries on ARM64 in Windows 11 24H2.

Thoughts?

[1]: https://github.com/msys2/MSYS2-packages/discussions/2832
[2]: https://github.com/msys2/msys2-runtime/issues/238#issuecomment-2481438646

-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple