[PATCH] D157110: llvm-cov: __cxx_global_var_init functions disrupt coverage

[Oleksii Odynochenko via Phabricator via llvm-commits]

tenta4 added a comment.

Thank you.
Yes, I agree that this is a wrong place for filtering.
And, furthermore it's wrong approach to filter out something weird just because we don't understand the reasons of its appearance.

I'll be able to proceed investigations on next week
Now I'll just try to answer your questions:
1 - This is the same pointers:

  p f.getFilename().Data == si.filename.Data
  $6 = true

2 - Is something going wrong in GCOVFile::readGCNO..? - Didn't investigate so deeper yet
3 - Is the data in the .gcno file itself corrupt? - Not sure how to check it
4 - Is the .gcno a newer version of the format that adds new information in a way that we don't handle it? - I almost sure they are compatible. I use clang-18 to generate gcno and latest llvm-project to analyze it (currently its version also 18)


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157110/new/

https://reviews.llvm.org/D157110