D41113: [LLVMgold] Don't set resolutions for undefined symbols to 'Prevailing'

Teresa Johnson via llvm-commits llvm-commits at lists.llvm.org
Fri Dec 15 07:33:40 PST 2017


On Fri, Dec 15, 2017 at 7:22 AM, Evgeny Leviant <eleviant at accesssoftek.com>
wrote:

> This change causes following tests to fail:
>
>     LLVM :: LTO/Resolution/X86/common2.ll
>     LLVM :: LTO/Resolution/X86/dead-strip-fulllto.ll
>     LLVM :: ThinLTO/X86/deadstrip.ll
>

Multiple prevailing defs in these tests I suppose (looks like it in
deadstrip.ll)? That's wrong, so good to fix these. Thanks, Teresa


> I'll fix and commit.
> ________________________________________
> От: Teresa Johnson via Phabricator <reviews at reviews.llvm.org>
> Отправлено: 15 декабря 2017 г. 17:52
> Кому: Evgeny Leviant; rafael.espindola at gmail.com; tejohnson at google.com;
> joker.eph at gmail.com; peter at pcc.me.uk
> Копия: eraman at google.com; llvm-commits at lists.llvm.org;
> llvm.mail.list at gmail.com; George Rimar; Igor Kudrin
> Тема: [PATCH] D41113: [LLVMgold] Don't set resolutions for undefined
> symbols to 'Prevailing'
>
> tejohnson added inline comments.
>
>
> ================
> Comment at: llvm/trunk/lib/LTO/LTO.cpp:420
> +    if (Res.Prevailing) {
> +      assert((GlobalRes.IRName.empty() || !Sym.getIRName().empty()) &&
> +             "Overriding existing resolution with undefined asm symbol");
> ----------------
> I had suggested changing this to simply:
>
> assert(GlobalRes.IRName.empty() && "....
>
> since we should only have one prevailing def. Can you do that?
>
>
> Repository:
>   rL LLVM
>
> https://reviews.llvm.org/D41113
>
>
>
>


-- 
Teresa Johnson |  Software Engineer |  tejohnson at google.com |  408-460-2413
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20171215/594fff25/attachment.html>


More information about the llvm-commits mailing list