[clang] [llvm] [ARM64EC] Implement `-mlong-double-{64,80,128}` options (PR #237057)
Martin Storsjö via cfe-commits
cfe-commits at lists.llvm.org
Mon Nov 2 00:10:25 PST 2026
================
@@ -2381,11 +2392,104 @@ def AArch64SystemLibrary : SystemRuntimeLibrary<isAArch64_ExceptArm64EC,
["compiler-rt", "libatomic", "llvm-client-runtime", "libm", "libc", "unwind",
"sme"]>;
+//===----------------------------------------------------------------------===//
+// ARM64EC helpers for the x87 `long double` format
+//===----------------------------------------------------------------------===//
+
+// These are not the exact x87 soft-float helpers of libgcc: The value stays
+// an 80-bit x87 image in a 16-byte slot, but every operation on it is carried
+// out in `double`. These are named with an `__arm64ec_` prefix so they will
+// not be confused with, or silently bound to, an exact implementation of the
+// same format.
----------------
mstorsjo wrote:
It would probably be good to put them up for review in parallel with the compiler code changes. (Back when I suggested holding off of submitting code changes for the runtimes, I hadn't noticed that the changes for compiler-rt were new functions that the compiler generates calls to.) But landing them should probably wait until there's there's agreement that the compiler changes look acceptable as well.
https://github.com/llvm/llvm-project/pull/237057
More information about the cfe-commits
mailing list