[Openembedded-architecture] Let's drop x86-x32 (was Re: [OE-core] [PATCH 05/51] rpm: update 4.17.0 -> 4.17.1)


Richard Purdie
 

On Wed, 2022-07-20 at 10:49 +0200, Alexander Kanavin wrote:
Note: this update fails on x32 with

configure: error: unrecognized GNU build triplet linux-gnux32
This time I want to put my foot down and suggest that we just drop the
whole x32 variant from the autobuilder (I had already sent a patch for
this previously). In all likelihood it was never used to ship anything
to customers, and was only devised to look better in benchmarks
against competition.
That amounts to dropping x32 support because as soon as we remove these
tests, it will bitrot.

There is still some value in the project being able to support
different architectures and different type sizes so I do still lean
towards keeping this alive at a minimal level.

Cheers,

Richard


Alexander Kanavin
 

On Wed, 20 Jul 2022 at 11:23, Richard Purdie
<richard.purdie@...> wrote:
That amounts to dropping x32 support because as soon as we remove these
tests, it will bitrot.

There is still some value in the project being able to support
different architectures and different type sizes so I do still lean
towards keeping this alive at a minimal level.
But then why not replace x32 with riscv32, which as well has 32 bit
pointers but 64 bit integers and thus trips over the same type size
issues?

Alex


Ross Burton
 

On 20 Jul 2022, at 10:28, Alexander Kanavin via lists.openembedded.org <alex.kanavin=gmail.com@...> wrote:

On Wed, 20 Jul 2022 at 11:23, Richard Purdie
<richard.purdie@...> wrote:
That amounts to dropping x32 support because as soon as we remove these
tests, it will bitrot.

There is still some value in the project being able to support
different architectures and different type sizes so I do still lean
towards keeping this alive at a minimal level.
But then why not replace x32 with riscv32, which as well has 32 bit
pointers but 64 bit integers and thus trips over the same type size
issues?
Does the RISC-V ecosystem care about riscv32?

The problem with Intel x32 is that very few people care, so we end up fixing upstream software. If RISC-V cares then we won’t be alone.

Also, Intel should get to have an opinion on this. If they actually care about x32 then they can help fix the issues, if they don’t then we can easily switch to a platform that has support.

Ross


Alexander Kanavin
 

On Wed, 20 Jul 2022 at 11:50, Ross Burton <Ross.Burton@...> wrote:
Also, Intel should get to have an opinion on this. If they actually care about x32 then they can help fix the issues, if they don’t then we can easily switch to a platform that has support.
Ok, let's ask Intel, specifically Anuj :-)

Anuj, does Intel still care about x32, and would anyone notice if
yocto drops x32 support from master branch?

Alex


Alexander Kanavin
 

On Wed, 20 Jul 2022 at 12:00, Marko Lindqvist <cazfi74@...> wrote:
How much difference is there between x32 and riscv32 in upstreams? As
they would trip on the same issues, one would assume that if the issue
is fixed for one, it gets fixed for the other too.
But might be that relevant upstreams need to have much of the code
duplicated (fixing one copy does not fix the other)
The fixing is often wrapped in target-specific conditionals, and so
fixing one does not address the other, until the conditional is
expanded with additional or statements, or checks are done from first
principles (e.g. actual type sizes).

Alex


Anuj Mittal
 

On Wed, 2022-07-20 at 11:58 +0200, Alexander Kanavin wrote:
On Wed, 20 Jul 2022 at 11:50, Ross Burton <Ross.Burton@...>
wrote:
Also, Intel should get to have an opinion on this.  If they
actually care about x32 then they can help fix the issues, if they
don’t then we can easily switch to a platform that has support.
Ok, let's ask Intel, specifically Anuj :-)

Anuj, does Intel still care about x32, and would anyone notice if
yocto drops x32 support from master branch?
I don't know if there are any Yocto users of it who might notice.

Instead of dropping the testing completely, may be we should switch to
building/testing just the core-image-minimal image on autobuilder and
keep at least some minimal support for now.

Thanks,

Anuj


Alexander Kanavin
 

On Wed, 20 Jul 2022 at 17:41, Mittal, Anuj <anuj.mittal@...> wrote:
I don't know if there are any Yocto users of it who might notice.

Instead of dropping the testing completely, may be we should switch to
building/testing just the core-image-minimal image on autobuilder and
keep at least some minimal support for now.
There's actually already a patch for that :)
https://lists.yoctoproject.org/g/yocto/message/56880

Alex


Alexander Kanavin
 

On Wed, 20 Jul 2022 at 17:46, Alexander Kanavin via
lists.yoctoproject.org <alex.kanavin=gmail.com@...>
wrote:

On Wed, 20 Jul 2022 at 17:41, Mittal, Anuj <anuj.mittal@...> wrote:
I don't know if there are any Yocto users of it who might notice.

Instead of dropping the testing completely, may be we should switch to
building/testing just the core-image-minimal image on autobuilder and
keep at least some minimal support for now.
There's actually already a patch for that :)
https://lists.yoctoproject.org/g/yocto/message/56880
Which I just resent. Now we have confirmation from Intel that nobody
minds reducing the scope of the AB test.

Alex


Anuj Mittal
 

On Wed, 2022-07-27 at 16:56 +0200, Alexander Kanavin wrote:
On Wed, 20 Jul 2022 at 17:46, Alexander Kanavin via
lists.yoctoproject.org
<alex.kanavin=gmail.com@...>
wrote:

On Wed, 20 Jul 2022 at 17:41, Mittal, Anuj <anuj.mittal@...>
wrote:
I don't know if there are any Yocto users of it who might notice.

Instead of dropping the testing completely, may be we should
switch to
building/testing just the core-image-minimal image on autobuilder
and
keep at least some minimal support for now.
There's actually already a patch for that :)
https://lists.yoctoproject.org/g/yocto/message/56880
Which I just resent. Now we have confirmation from Intel that nobody
minds reducing the scope of the AB test.
It was just a personal opinion saying that I won't mind. If it's
needed, then someone would hopefully fix the issues.

Thanks,

Anuj