llvm dev - Apr 2018 - [lld] Any chance to get review for a couple of patches in a couple of years?

Hi Rui, Rafael,

You are reviewers of two MIPS related patches for LLD linker:

- Handle cross-mode (regular <-> microMIPS) jumps
  https://reviews.llvm.org/D40147

- Multi-GOT implementation
  https://reviews.llvm.org/D31528

Both patches implement essential part of MIPS architecture. microMIPS
is something like ARM Thumb. Multi-GOT support required to link any
rather complex application for MIPS targets. For example, without
multi-GOT it is not possible to link Clang for MIPS. This
functionality is not experimental, rarely used, or obsoleted features.

Unfortunately both these patches could not be reviewed for a very long
time. The last comment for the microMIPS was received in February. As
to the Multi-GOT patch - it was submitted more than one year ago and
have not got any technical comments from you since that time. By
"technical comments" I mean some notes, suggestions, objections on
design and / or implementation details of the patch.

I would like to know do you plan and, if yes, when do you plan to
review the patches? At least any rough estimation. I think, guys who
work on MIPS FreeBSD port would like to get that estimation too.

Regards,
Simon Atanasyan

Hi,

Yes I'll do that. The situation has changed a lot since then and lld got
support for other architectures, in particular aarch64. But if I
remember correctly, there was a discussion whether we should bring this ABI
oddness to the ABI standard body (if exists) or not. I wonder if there was
some progress on that.

On Mon, Apr 23, 2018 at 10:28 AM Simon Atanasyan <simon at atanasyan.com>
wrote:
> Hi Rui, Rafael,
>
> You are reviewers of two MIPS related patches for LLD linker:
>
> - Handle cross-mode (regular <-> microMIPS) jumps
>   https://reviews.llvm.org/D40147
>
> - Multi-GOT implementation
>   https://reviews.llvm.org/D31528
>
> Both patches implement essential part of MIPS architecture. microMIPS
> is something like ARM Thumb. Multi-GOT support required to link any
> rather complex application for MIPS targets. For example, without
> multi-GOT it is not possible to link Clang for MIPS. This
> functionality is not experimental, rarely used, or obsoleted features.
>
> Unfortunately both these patches could not be reviewed for a very long
> time. The last comment for the microMIPS was received in February. As
> to the Multi-GOT patch - it was submitted more than one year ago and
> have not got any technical comments from you since that time. By
> "technical comments" I mean some notes, suggestions, objections
on
> design and / or implementation details of the patch.
>
> I would like to know do you plan and, if yes, when do you plan to
> review the patches? At least any rough estimation. I think, guys who
> work on MIPS FreeBSD port would like to get that estimation too.
>
> Regards,
> Simon Atanasyan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180424/18330b8e/attachment.html>

On Tue, Apr 24, 2018 at 8:21 AM, Rui Ueyama <ruiu at google.com>
wrote:
> if I remember correctly, there was a discussion whether we should bring
> this ABI oddness to the ABI standard body (if exists) or not. I wonder
> if there was some progress on that.
As far as I know there is no plan about new ABI. Anyway I do not think that
any new MIPS ABIs changes approaches to GOT handling. So things like multi-GOT
remain MIPS-specific feature because it provides benefits to MIPS architectures.

-- 
Simon Atanasyan