[PATCH] D108872: [IR] Refactor GlobalIFunc to inherit from GlobalObject, Remove GlobalIndirectSymbol

[Itay Bookstein via Phabricator via llvm-commits]

ibookstein added a comment.

Ugh. should clang accept situations where the resolver of an ifunc is also an ifunc? The semantics are confusing, but neither compiler diagnoses anything:

  itay ~/tmp/gis> cat attr-ifunc.c
  int qux() { return 44; }
  int (*foo_resolver())() { return qux; }
  int foo() __attribute__((ifunc("foo_resolver")));
  
  int (*(*bar_resolver_resolver())())() { return foo_resolver; }
  int (*bar_resolver())() __attribute__((ifunc("bar_resolver_resolver")));
  int bar() __attribute__((ifunc("bar_resolver")));
  
  int main() { return foo() + bar(); }
  itay ~/tmp/gis> gcc attr-ifunc.c -o attr-ifunc -g ; ./attr-ifunc
  itay ~/tmp/gis [SIGTTIN]> clang-14 attr-ifunc.c -o attr-ifunc -g ; ./attr-ifunc
  itay ~/tmp/gis [108]>


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D108872/new/

https://reviews.llvm.org/D108872