[llvm-bugs] [Bug 31005] inline asm global registers compatibility to gcc+icc

via llvm-bugs llvm-bugs at lists.llvm.org
Wed Dec 14 01:53:22 PST 2016


https://llvm.org/bugs/show_bug.cgi?id=31005

Renato Golin <renato.golin at linaro.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |WONTFIX

--- Comment #8 from Renato Golin <renato.golin at linaro.org> ---
Closing this as wontfix because that's our community's current position.

If anyone wants to change this position they'll have to:

1. Create a downstream prototype where registers can be allocated and
deallocated by the user on a function and module scope without destroying the
register allocators (plural) job.

2. Send an RFC to the list, with the design decisions taken to make that work,
and work with the community to adapt your downstream implementation to work
that way.

3. After addressing all the community's feedback, send a list of small and
incremental patches, implementing the functionality with enough tests to cover
each step, and making sure none of them break the build on their own, ie.
isolated and working changes.

4. Work with the community to push the patches upstream, address reviews, etc.

As it stands, the community doesn't need, nor want, arbitrary global named
registers, so don't expect people to implement that for you. We have
successfully convinced users to change their code, and most of the time to
better code, proving that this feature is not needed at all.

If it is needed for you, and you can show that to the community, you'll have to
go through the proper upstreaming process described above.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20161214/c664daa3/attachment.html>


More information about the llvm-bugs mailing list