[PATCH] D33880: COFF: Introduce LD shim around LINK

[Martin Storsjö via Phabricator via llvm-commits]

mstorsjo added a comment.

In https://reviews.llvm.org/D33880#831661, @martell wrote:

> In https://reviews.llvm.org/D33880#831570, @mstorsjo wrote:
>
> > If you're still too busy and don't mind, I could also take over trying to get this in a mergeable form.
>
>
> I can be on this to ensure it gets merged, there were 2 reasons why I left this slide for awhile.
>  I was having an issue with llvm-dlltool where if I used `new` to create memory for the strings it would work but if I done it the correct way as rui instructed it was not working correctly.
>  We now know that was because we were writing the strings beforehand the entry thanks to you :) So the 2 bugs because of the location of the the allocation seemed to nullify each other to some degree.
>  Which left me a little confused. So I left the lld support slide until llvm-dlltool was merged and also at the time I had started a new job.


Sure - it's great to have at least one of the components merged finally :-)

> I can still be somewhat useful here because I understand what is going on.
>  Where I probably won't be of any use is on the unwind exceptions, so might be an idea to commandeer that differential?
>  I'm sure this is something you know a lot more about and I can probably learn a lot from the patch that gets merged for it.

Nope, I know much less about that; I'll probably stay away from your libc++/libunwind patches until I've got everything necessary for building C programs upstreamed. The things left are this and the mingw ctor section handling (even though it's not used for C projects, it causes linker errors if not handled in one way or another).


Repository:
  rL LLVM

https://reviews.llvm.org/D33880