ARM Cortex-M FDPIC ABI

ARM Cortex-M FDPIC ABI

[Alexandru Onea]

[-- Attachment #1: Type: text/plain, Size: 322 bytes --]

Dear musl developers,

I am reading in the roadmap [1] that musl 1.1.26 introduces the ARM FDPIC
ABI, which is great news. Is there a development branch / snapshot
including this feature, which I can use for experimental work?

[1] https://wiki.musl-libc.org/roadmap.html#musl-1.1.26

Thanks in advance,
Alexandru N. Onea

[-- Attachment #2: Type: text/html, Size: 505 bytes --]

Re: ARM Cortex-M FDPIC ABI

[Rich Felker]

On Wed, Sep 11, 2019 at 05:08:57PM +0300, Alexandru Onea wrote:
> Dear musl developers,
> 
> I am reading in the roadmap [1] that musl 1.1.26 introduces the ARM FDPIC
> ABI, which is great news. Is there a development branch / snapshot
> including this feature, which I can use for experimental work?
> 
> [1] https://wiki.musl-libc.org/roadmap.html#musl-1.1.26

Roadmap is intent for future development, not secret branches that are
already finished. :-)

(Well, the switchover to time64 is kinda such a branch, but the
patches have been posted on the list.)

Anyway, now that the FDPIC ABI is upstream in GCC, it can definitely
happen. To make it easy to go forward, I need to get a version of GCC
that ARM FDPIC patches will apply to supported in musl-cross-make.
After that it shouldn't be hard to make it work.

Rich

Re: ARM Cortex-M FDPIC ABI

[Rich Felker]

On Wed, Sep 11, 2019 at 10:19:44AM -0400, Rich Felker wrote:
> On Wed, Sep 11, 2019 at 05:08:57PM +0300, Alexandru Onea wrote:
> > Dear musl developers,
> > 
> > I am reading in the roadmap [1] that musl 1.1.26 introduces the ARM FDPIC
> > ABI, which is great news. Is there a development branch / snapshot
> > including this feature, which I can use for experimental work?
> > 
> > [1] https://wiki.musl-libc.org/roadmap.html#musl-1.1.26
> 
> Roadmap is intent for future development, not secret branches that are
> already finished. :-)
> 
> (Well, the switchover to time64 is kinda such a branch, but the
> patches have been posted on the list.)

And to clarify *why* time64 switchover is just patches posted to the
list, rather than a public git branch, it's because this is a matter
of defining future ABI, and I really don't want to be giving ppl the
impression that using them in their current state is something
reasonable to do. Doing so would almost surely end up with binaries
that mismatch with the final ABI, probably in subtle ways that go
unnoticed until something catastrophic happens.

Rich