Revision 271670 (which was comitted on behalf of me by @SjoerdMeijer) had a mistake where RAS extensions where assigned to 8.1 rather than 8.2.
Details
Diff Detail
Event Timeline
What does this affect? It seems like it ought to be possible to write a test for it. (try to assemble something with v8.1a and check that RAS isn't accepted?)
Tim.
Indeed this does not have any effect. I could even remove the enumerator from the macro and nothing changes.
AArch64 has a similar AArch64::AEK_RAS in armv8.2-a and it is handled in AArch64::getExtensionFeatures (in lib/Support/TargetParser.cpp) but (unless I'm missing something very obvious) it is not called from anywhere.
Conversely, ARM::getExtensionFeatures is called from tools/clang/lib/Basic/Targets.cpp but does not handle ARM::AEK_RAS. Handling it like AArch64 does not have any effect either.
Closing since @richard.barton.arm opened D23500 for the same issue but on top of the new AArch64 TargetParser
These macros are available for setting compiler-provided preprocessing macros like ACLE. We are not using any of the ARMv8.2-A ones because the ACLE has not caught up sufficiently with the architecture to have defined any.