[PATCH] D86185: [Cloning] Fix to cloning DISubprograms.
Amy Huang via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Thu Aug 20 17:51:09 PDT 2020
akhuang added inline comments.
================
Comment at: llvm/unittests/Transforms/Utils/CloningTest.cpp:740-744
+ !3 = distinct !DISubprogram(name: "my_operator", scope: !1, retainedNodes: !{!5})
+ !4 = distinct !DICompositeType(tag: DW_TAG_structure_type, scope: !3)
+ !5 = !DILocalVariable(name: "awaitables", scope: !3)
+ !6 = distinct !DISubprogram(name: "test", scope: !1, retainedNodes: !{!7})
+ !7 = !DILocalVariable(name: "w", scope: !6, type: !4)
----------------
dblaikie wrote:
> This debug info metadata looks like it has two cycles (my_operator -> awaitables -> my_operator and test -> w -> test). Are they both necessary? do they both test different cases? Might be worth a few more words in the test comment above about what they're exercising/what aspects of them are relevant/necessary to reproduce the issue?
I don't think either is necessary, actually. The important part for reproducing is just that there's a path from test -> my_operator. Maybe I can get rid of some more stuff; this is just what the metadata happened to look like in the repro.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D86185/new/
https://reviews.llvm.org/D86185
More information about the llvm-commits
mailing list