Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Thanks. Seems to fix the test for me FWIW, but
- why wasn't this caught by anyone and all of the bots?
- what is the actual underlying problem, if any?
Something is fishy here. Now the pre-merge tests fail... this should accept strictly more... I have to look into this, best guess: some non-determinism
For what it's worth, I have been affected by this and was drafting a note to D82822 today to ask about the status.
- what is the actual underlying problem, if any?
Root cause:
We sometimes, depending on the invocation and potentially other factors, use a full path and sometimes not:
So wen I run this with clang locally I get this:
@98 = private unnamed_addr constant [98 x i8] c";/data/src/llvm-project/clang/test/OpenMP/irbuilder_nested_parallel_for.c;parallel_for_2;294;16;;\00", align 1 @100 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([98 x i8], [98 x i8]* @98, i32 0, i32 0) }, align 8 @101 = private unnamed_addr constant [92 x i8] c";src/llvm-project/clang/test/OpenMP/irbuilder_nested_parallel_for.c;parallel_for_2;294;16;;\00", align 1 @103 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([92 x i8], [92 x i8]* @101, i32 0, i32 0) }, align 8
Note the 6 missing characters /data/ in the second string. Because they are missing, we generate two globals for this. When I run the same test with the update script I get one global as they agree on the location.
Looking for the location that strips the path now.
still lg, i guess
clang/test/OpenMP/irbuilder_nested_parallel_for.c | ||
---|---|---|
258 ↗ | (On Diff #285551) | Please precommit test regeneration first. |