[PATCH] D22356: [ThinLTO] Perform conservative weak/linkonce resolution in distributed backend case

[Mehdi Amini via llvm-commits]

> On Jul 19, 2016, at 5:23 PM, Teresa Johnson <tejohnson at google.com> wrote:
> 
> 
> 
> On Tue, Jul 19, 2016 at 5:18 PM, Mehdi Amini <mehdi.amini at apple.com <mailto:mehdi.amini at apple.com>> wrote:
> 
>> On Jul 19, 2016, at 6:12 AM, Teresa Johnson <tejohnson at google.com <mailto:tejohnson at google.com>> wrote:
>> 
>> 
>> 
>> On Mon, Jul 18, 2016 at 10:40 PM, Xinliang David Li <davidxl at google.com <mailto:davidxl at google.com>> wrote:
>> Good example.  I think I am convinced :)
>> 
>> Yes, I can't think of another way to get this case to work without being far more conservative about importing, unfortunately. Ok, let me figure out what the best way is to transmit this info on to the final link through the build system. Probably through a file that can be directly used with the @ linker option in place of input files.
> 
> Do you have an example where it matters?
> The only case I can imagine is mixing ThinLTO files with non-ThinLTO files, and the same weak symbol defined in both.
> 
> Yes, this and the follow on fix both were for cases that came from real (internal) applications.
> IIRC, the weak symbols came from headers included in multiple places (all ThinLTO files).

I understand the issue with back-reference because the use of —start-lib/—end-lib, but I don’t understand why you got multiple weak out of the ThinLTO backends?
I.e. we should turn all but one to available externally, so no duplicate should be seen during the second link.

— 
Mehdi


> 
> Working on build system changes right now, I've confirmed with a recent binutils that we have enough info that the llvm side to emit these should be trivial.
> 
> 
> 
> — 
> Mehdi
> 
> 
> 
> -- 
> Teresa Johnson |	 Software Engineer |	 tejohnson at google.com <mailto:tejohnson at google.com> |	 408-460-2413

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160719/c3b4bbd1/attachment.html>