[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: About i386 support



On Mon, 2024-05-20 at 13:39 +0200, Ansgar 🙀 wrote:
> Hi,
> 
> On Sun, 2024-05-19 at 10:30 -0500, rhys@neoquasar.org wrote:
> > I have an N270 system I can use to contribute, if someone is willing
> > to explain what I need to do to make it useful. 
> > 
> > From: Victor Gamper <victor@wenzeslaus.de>
> > Sent: Sunday, May 19, 2024 08:03
> > To: debian-devel@lists.debian.org
> > Subject: Re: About i386 support
> > 
> > I believe I could swap out the processor on my T60,
> > however, I'd both need to have that processor and
> > make sure that it is actually possible. It still would
> > not really make sense on a platform that only supports
> > 3G of physical RAM.
> > 
> > Anyways, if the only reason why i386 cd images are not
> > supported anymore is the lack of contributors,
> > I'd be willing to contribute in that area, if it's possible.
> 
> If you look at https://release.debian.org/testing/arch_qualify.html
> there is at least several things that can be done:
> 
> 1. Add CPU security mitigations to Linux kernel.

With few exceptions, the CPU security issues that have been discovered
in tha past few years are not thought to affect 32-bit-only CPUs. 
(Meltdown did affect some of them but was eventually mitigated on
i386.)  I have raised the possibility of making i386 kernel builds warn
very loudly or refuse to boot on 64-bit-capable hardware, but didn't
try implementing it yet.

[...]
> 3. Look at other arch-specific issues (porter); this can also include
> baseline violations and other issues for real i386 hardware.
[...]

There is a tension here between the interests of (a) users that want to
run proprietary i386 binaries on 64-bit CPUs, and (b) those who want to
keep using 32-bit CPUs.  If i386 is meant for group (a) then the
baseline should be raised to include the features that 64-bit CPUs
provide, but if it's also for group (b) then this mustn't happen.

This also showed up in the 64-bit time_t transition, where the
transition would be desirable for group (b) but would make i386 useless
for group (a).  There was some talk then about creating a new port like
"i386t64" for use by group (b), while i386 would serve group (a) only.
Did anything come of this?

Ben.

-- 
Ben Hutchings
I say we take off; nuke the site from orbit.
It's the only way to be sure.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: