OK, enough time has passed, let's do this. ================ Comment at: include/clang/Basic/DiagnosticSemaKinds.td:6881 @@ +6880,3 @@ + "the newer semantics are provided here">, + InGroup<DiagGroup<"gcc-intrinsic-semantics-changed">>; + ---------------- I think you should pick a warning flag name that is specific to this particular change. We'd never want to put two different semantic changes in the same warning group, because we would want users to be able to turn off one of them without turning off the other.
================ Comment at: test/CodeGen/Atomics.c:168-183 @@ -149,1 +167,18 @@ + sc = __sync_nand_and_fetch (&sc, uc); // CHECK: atomicrmw nand + // CHECK: xor + uc = __sync_nand_and_fetch (&uc, uc); // CHECK: atomicrmw nand + // CHECK: xor + ss = __sync_nand_and_fetch (&ss, uc); // CHECK: atomicrmw nand + // CHECK: xor + us = __sync_nand_and_fetch (&us, uc); // CHECK: atomicrmw nand + // CHECK: xor + si = __sync_nand_and_fetch (&si, uc); // CHECK: atomicrmw nand + // CHECK: xor + ui = __sync_nand_and_fetch (&ui, uc); // CHECK: atomicrmw nand + // CHECK: xor + sll = __sync_nand_and_fetch (&sll, uc); // CHECK: atomicrmw nand + // CHECK: xor + ull = __sync_nand_and_fetch (&ull, uc); // CHECK: atomicrmw nand + // CHECK: xor + ---------------- Please also check for the `and` instruction here. http://reviews.llvm.org/D5429 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
