diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2022-11-23 21:23:47 -0500 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2022-12-02 15:20:57 -0500 |
| commit | f883e75497b610de81b3a6af106c976bb3c2bbb1 (patch) | |
| tree | c9feeddd6c11cb6437a732a644977ae068dd81c1 /clang/lib/Frontend/CompilerInvocation.cpp | |
| parent | 27387896cfe898e66f411ff26b8e34698a8f92fc (diff) | |
| download | llvm-f883e75497b610de81b3a6af106c976bb3c2bbb1.zip llvm-f883e75497b610de81b3a6af106c976bb3c2bbb1.tar.gz llvm-f883e75497b610de81b3a6af106c976bb3c2bbb1.tar.bz2 | |
GlobalIFunc: Make allowed constant expressions stricter
This was allowing getelementptr with offsets, which doesn't make
sense. My initial attempt to use stripPointerCasts broke a few tests
involving aliases; add a new targeted verifier test for aliases.
This also provides the fix from D138537 for free, and also adds
support for addrspacecast (D138538) for free. Merge the tests in from
those.
I'm not really sure why findBaseObject exists; it seems redundant with
stripPointerCasts* (I'm also not really sure why getelementptrs are
allowed off of functions).
Diffstat (limited to 'clang/lib/Frontend/CompilerInvocation.cpp')
0 files changed, 0 insertions, 0 deletions