[lld] r250808 - [ELF2] - Lazy relocation support for x86_64.

[Rui Ueyama via llvm-commits]

On Tue, Oct 20, 2015 at 12:00 PM, Rafael Espíndola <
rafael.espindola at gmail.com> wrote:

> > I understand and agree with that. My first reaction to this lazy
> resolution
> > patch is "this is what we usually do, but I want to spend time to see if
> we
> > really need this feature on today's computer in the first place".
>
> Interesting, what have you found?
>

Well, I didn't do perf testing or something like that. I'd be interesting
to see how many microseconds are needed on average to resolve one dynamic
symbol.


> I would expect that it makes a difference when using lots of shared
> libraries and the old sysv hash.
>
> I am not sure if that would still be the case with .gnu.hash and
> specially if one day direct binding is implemented (-Bdirect)
>
> In other words, I think it is useful for now for at least some users
> and I am happy to have it.
>
> Cheers,
> Rafael
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20151020/088b6b76/attachment.html>