Page MenuHomePhabricator

[llvm-readobj] Dump GNU_PROPERTY_X86_ISA_1_{NEEDED,USED} notes in .note.gnu.property
ClosedPublic

Authored by MaskRay on Feb 13 2019, 6:05 AM.

Details

Summary

And rename pr_data to PrData. This is a simple variable that does
not have strong needs to stick with the name used in System V
Application Binary Interface Linux Extensions.

Diff Detail

Repository
rL LLVM

Event Timeline

MaskRay created this revision.Feb 13 2019, 6:05 AM
Herald added a project: Restricted Project. · View Herald TranscriptFeb 13 2019, 6:05 AM

You do not need to do cosmetic changes in the same patch where you do functional changes generally.
i.e. renaming of the pr_data -> PrData is LGTM and it can be committed as NFC change that does not need a review.

Regarding the rest of the patch, seems it is not fully covered with a test case? I see no output for AVX512BW for example?

There is no output for AVX512BW, as I'm not sure enumerating every bit would be useful. Every bit can be tested individually, but that is tedious and makes the test hard to read (there would just be a long line duplicating the ISA names in the source code). I think that checking the 2 most significant bits and the 2 least significant bits should be good enough.

There is no output for AVX512BW, as I'm not sure enumerating every bit would be useful. Every bit can be tested individually, but that is tedious and makes the test hard to read (there would just be a long line duplicating the ISA names in the source code). I think that checking the 2 most significant bits and the 2 least significant bits should be good enough.

I am not suggesting to test each bit individually but suggesting to prepare an input that would test all of them. I.e. a case where all bits are set.
I think that makes sense because we want to test printing of all of the bits actually.
That is how we test them in LLD and in tests for tools like llvm-objdump.

Not sure this is the best example, but we have lld/test/ELF/dt_flags.s in LLD and its idea is to test we set all the dynamic entries flags for
the command line passed. The same is here - just show we can dump the flags correctly.

MaskRay updated this revision to Diff 186647.Feb 13 2019, 7:02 AM

Print all bits as grimar suggested

This looks fine. But I still suggest renaming the variable in the different patch ;)

MaskRay updated this revision to Diff 186672.Feb 13 2019, 7:59 AM

Committed pr_data -> PrData in a separate patch

This revision is now accepted and ready to land.Feb 13 2019, 9:53 AM
This revision was automatically updated to reflect the committed changes.