[PATCH] D31443: [LTO] Do not reorder global variables unnecessarily during merging

Mehdi Amini via llvm-commits llvm-commits at lists.llvm.org
Tue Apr 4 14:40:43 PDT 2017


I’m midly in favor of merging this (i.e. unless someone has a strong objection): the patch is harmless, and if it can help one of the LTO consumers (Qualcomm), why not? 
My view is that I’d rather have Qualcomm more involved upstream, and I’m inclined to see this as trying to be welcoming.

Best,

— 
Mehdi

> On Apr 4, 2017, at 2:37 PM, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
> 
> Someone that knows thin lto has to take a look at it.
> 
> I am more opposed to this now that it requires changes to a common
> utility function. I will side with Mehdi if he thinks this is OK for
> thinlto and will not lock us down in our ability to change this order
> at any time.
> 
> Cheers,
> Rafael
> 
> 
> On 4 April 2017 at 17:34, Tobias Edler von Koch via Phabricator
> <reviews at reviews.llvm.org> wrote:
>> tobiasvk added a comment.
>> 
>> Any further comments? I'd like to go ahead and merge this.
>> 
>> 
>> https://reviews.llvm.org/D31443
>> 
>> 
>> 



More information about the llvm-commits mailing list