[llvm-dev] [lld] r271569 - Start adding tlsdesc support for aarch64.

[Rafael Espíndola via llvm-dev]

> Being said that, for this particular instance, I wouldn't submit right away
> but send it to review and explain why I think better. I'm totally fine if
> someone who knows more than me write a better patch than mine as a result of
> code review for my patch, but I would be surprised if an alternative is just
> submitted. I'm not sure if this needs to be reverted, but at least, could
> you send it review next time in a similar situation, Rafael?


I honestly don't see what is the issue, so I don't think I be able to
judge. The rule of the thumb for ahead of time commit is how confident
someone is. I spend quite a few hours figuring out how tls works on
aarch64 and as a result wrote a patch that used the same Expr names as
the original patch. Really, diff both diffs and notice it is quite
different. Also notice that the commit message includes information
about aarch64 that I simply haven't seen written anywhere, including
the original patch.

Once more is implemented I will document the details I found on how
dlopen works.

Cheers,
Rafael