[PATCH] D133452: [ORC] Fix auto-claim of weak defs in ObjectLinkingLayer

Jonas Hahnfeld via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Oct 27 10:40:10 PDT 2022


Hahnfeld added a comment.

In D133452#3887989 <https://reviews.llvm.org/D133452#3887989>, @sgraenitz wrote:

> Hi Jonas, thats interesting results! I don't have the time right now to dig into the details, but..
>
> In D133452#3877904 <https://reviews.llvm.org/D133452#3877904>, @Hahnfeld wrote:
>
>>   JIT session error: In catch.ll-jitted-objectbuffer, encountered duplicate section ".group" while building debug object
>>
>> The first error about duplicate sections is there even without the change, but I can't remember it from September. Is that something recent?
>
> There was a discussion on Discord these days about it: https://discord.com/channels/636084430946959380/687692371038830597/1033427215448293396
> The error originates from JITLink's `DebugObjectManagerPlugin` and it should be fixed with: https://github.com/llvm/llvm-project/commit/b26f45e5a49ae363164e7dbbf57eadd9e78d612c
> Does it work for you? Otherwise, can you disable the plugin temporarily and try again?

Yes, the error is gone now. Thanks for letting me know!
Unfortunately, my patch above still doesn't work on x86 with the `EHFrame` support. I'll need to debug...

>>   JIT session error: Unexpected definitions in module catch.ll-jitted-objectbuffer: [ DW.ref.__gxx_personality_v0 ]
>>   ./bin/lli: Failed to materialize symbols: { (main, { __clang_call_terminate, main }) }
>
> Just a guess: Maybe the `EHFrame` plugin adds the extra dependency here. Do you need exception handling and/or backtraces? Maybe check and disable the plugin temporarily as well?

Yes, it obviously does. Yes, eventually I need and want exception handling (after all, I'm compiling with exceptions and want to throw and catch them at some point), but right now this isn't implemented for RISC-V. So there is nothing that I could disable to fix the problem, but I pointed out that disabling the plugin on x86 leads to similar problems there, so it's a general problem (as opposed to specific to RISC-V).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D133452/new/

https://reviews.llvm.org/D133452



More information about the llvm-commits mailing list