[llvm-dev] Stop bugpoint from removing operand of metadata.

Mehdi Amini via llvm-dev llvm-dev at lists.llvm.org
Fri Aug 19 10:32:11 PDT 2016


Hi,


> On Aug 19, 2016, at 10:24 AM, Hongbin Zheng via llvm-dev <llvm-dev at lists.llvm.org> wrote:
> 
> Hi,
> 
> I am trying to use bugpoint to reduce testcase. And bugpoint try to reduce the number of operands for metadata, and this generate corrupted metadata:
> 
> *** Attempting to remove named metadata: Checking for crash with only these named metadata nodes: llvm.dbg.cu <http://llvm.dbg.cu/> llvm.module.flags llvm.ident: <crash>
> Checking for crash with only these named metadata nodes: llvm.module.flags llvm.ident: <crash>
> Checking for crash with only these named metadata nodes: llvm.ident:
> Checking for crash with only these named metadata nodes: llvm.module.flags: <crash>
> Checking for crash with only 3 named metadata operands: <crash>
> Checking for crash with only 2 named metadata operands: All DICompileUnits must be listed in llvm.dbg.cu <http://llvm.dbg.cu/>
> LLVM ERROR: Broken module found, compilation aborted!

I had the same issue, and have some local patches to help with this.
Have you tried the   -disable-namedmd-remove option?

Also, at this point where bug point crashes, it should already have dumped a module.

— 
Mehdi


> 
> 
> This crash bugpoint itself. Is there away to prevent bugpoint from doing this rather than changing the source code of bugpoint?
> 
> I tried to run a "bugpoint --help-hidden" and get the following options related to metadata:
>     -lowertypetests                                        - Lower type metadata
>   -disable-namedmd-remove                                  - Do not remove global named metadata
>   -enable-import-metadata                                  - Enable import metadata like 'thinlto_src_module'
>   -enable-noalias-to-md-conversion                         - Convert noalias attributes to metadata during inlining.
>   -lto-strip-invalid-debug-info                            - Strip invalid debug info metadata during LTO instead of aborting.
> 
> And I don't think these help.
> 
> Any suggestion?
> 
> Thanks
> Hongbin
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20160819/356b4636/attachment.html>


More information about the llvm-dev mailing list