Since they are just llvm debug intrinsics, would running StripDebugInfo prior to
yours help?
Zhang
> 在 2018年7月17日,17:42,Muhui Jiang via llvm-dev <llvm-dev at
lists.llvm.org> 写道:
>
> Hi
>
> I write a LLVM function pass to try to analysis all the functions inside
one binary.
>
> However, this pass has been run for 80 CPU hours.
>
> You can refer to
https://groups.google.com/forum/#!searchin/llvm-dev/slow$20pass%7Csort:date/llvm-dev/BNbqiSbrh70/mQFZkAj-BgAJ
>
> I look at the functions it analyzed. Then I found some very strange
function name. There are more than one hundred functions whose name is
output_xxx. I randomly select some of them and found that the IR are all the
same like below:
>
> Source Line:115|Column:11|File:insn-output.c|IR: call void
@llvm.dbg.value(metadata %struct.rtx_def** %operands, metadata !441244, metadata
!DIExpression()), !dbg !441246
> Source Line:116|Column:10|File:insn-output.c|IR: call void
@llvm.dbg.value(metadata %struct.rtx_def* %insn, metadata !441245, metadata
!DIExpression()), !dbg !441247
> Source Line:118|Column:9|File:insn-output.c|IR: %call = tail call fastcc
i8* @output_fp_compare(%struct.rtx_def* %insn, %struct.rtx_def** %operands, i32
0, i32 0) #5, !dbg !43814
> Source Line:118|Column:2|File:insn-output.c|IR: ret i8* %call, !dbg !43815
>
> I am not sure this is the reason that make my pass running such a long
time. The target bitcode is just less than 50Mb. I don't think it will cost
me so much time. Do anyone has any ideas? I
>
> Regards
> Muhui
> _______________________________________________
> 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/20180717/f5ea1f25/attachment.html>