[llvm] r199244 - Reapply "LTO: add API to set strategy for -internalize"

[Duncan P. N. Exon Smith]

On Apr 2, 2014, at 14:55, Nick Kledzik <kledzik at apple.com> wrote:

> On Apr 2, 2014, at 1:22 PM, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
>>>> 
>>>> Have you found a case where the extra calls are actually noticeable?
>>> 
>>> Probably not.  Since my approach directly modelled what the linker wants, and
>>> certainly wasn't going to be slower, I never profiled.  (Sorry I missed that
>>> that conversation; I arrived too late.)
>>> 
>>> Nick, did you do any profiling of this?
> 
> Rafael,
> 
> Duncan and I came up with this API addition during discussions of why ld -r had some failure cases with bitcode.  We eventually mapped out that the linker has a couple of modes in which changes how symbols of different visibility are handled.  That model of modes led to this API. 
> 
> ld64 does not use the new API yet.
> 
> But, given that this can be implemented with extreme usage of lto_codegen_add_must_preserve_symbol(), we should just do that (that is, don’t do a premature optimization).  
> 
> Also, given that lld is coming and may do things differently.  No need to invest in a short lived new API.
> 
> 
> Duncan: I think I can do this all in the linker without a new API.
> 
> -Nick

Okay, great.  Coupled with Eric's reassurance that C API churn within a
release cycle is okay, I've pulled it out in r205471.  It always feels
good to delete code!