- User Since
- Dec 14 2017, 6:53 AM (145 w, 43 m)
- addressing comment
- addressing comments
No problem, the tempo works for me. :) Thanks!
Tue, Sep 22
- Update the algorithm for finding the entry values
- Improve the tests
Mon, Sep 21
- addressing comments
- fixing tests
- adding RegState::Debug to the Reg from the dbg_value
@aprantl thanks for your comments.
@dstenb Please let me know if this looks OK to you now.
Fri, Sep 18
Please investigate the test failures.
($ make check-all)
LGTM, thanks. The motivation is D87903?
Thu, Sep 17
- Addressing comments
- Rename debugify passes into more general name (in order to support original debug info check) - VerifyDIPreserve
- Rename the options
- Fix the test cases
- Update the docs
Fri, Sep 11
will make right fix for this (coming soon)
Thu, Sep 10
- Test new expression
Wed, Sep 9
-Split into 2 parts
-Remove unused parts
@dstenb Thanks a lot for your comments! I will split this into two patches.
Mon, Sep 7
(test.c was listed within llvm/test/DebugInfo/X86/entry-values-for-isel-invalidated-nodes.ll)
Looks like LLDB has a problem with parsing "complex" expression with DW_OP_entry_value:
This is actually ISel phase, but we still consider SSA, so this should be the latest point we are able to use the findEntryValue() in the pipeline.
Tue, Sep 1
@dblaikie Thanks for your comment!
Mon, Aug 31
Wed, Aug 26
Aug 25 2020
Please close the revision as well.
Aug 24 2020
nit included, otherwise lgtm, thanks!
@Orlando Thanks for doing this!
Aug 7 2020
Aug 6 2020
Aug 5 2020
+1 for this question(s).
Aug 4 2020
According to the --show-section-sizes output, the .debug_loc has been increased?
From the , I've understood that this should affect debug build only (-g + -O0) ?
Aug 3 2020
Jul 31 2020
Jul 28 2020
@StephenTozer Thanks for addressing the comments! Can you please share with us if you have any measurements with the respect to build time increase (e.g. on clang build itself) with this patch? I think that this patch is the only one from the stack that could give us potential performance issues.
- additional formatting
Jul 27 2020
- refactor the comments
@craig.topper Thanks for addressing the comments, lgtm now.
Hi @vsk, is this in better shape now? :)
Jul 24 2020
@craig.topper Thanks for fixing this!
Jul 21 2020
Jul 20 2020
Any additional comment on this? :)
Jul 16 2020
Thanks for doing this!
Jul 15 2020
My initial impression went in that direction. The implementation of the callee side of the entry values feature (production of the DW_OP_entry_values) within GCC Value Tracking System was trivial as well. I think this new implementation of LLVM LDV will give us more freedom to use entry-values even for modified parameters, if we can express the modification in terms of its entry value (i.e. param = param + 2 ==> DW_OP_plus (DW_OP_entry_value (param), 2)).
- Improve the test
Jul 14 2020
- TODO: Consider another name for mode. 'original' might be too sharp.
- Move the def of DebugifyCustomPassManager into the Debugify.h
- Update the comment around ShouldEmitDebugEntryValues()
Jul 13 2020