[PATCH] ARM & AArch64: merge the semantic checking of NEON intrinsics

Tim Northover t.p.northover at gmail.com
Tue Feb 11 11:31:58 PST 2014



================
Comment at: lib/Sema/SemaChecking.cpp:423
@@ -425,3 +422,3 @@
     QualType RHSTy = RHS.get()->getType();
-    QualType EltTy = getNeonEltType(NeonTypeFlags(TV), Context, true);
+    QualType EltTy = getNeonEltType(NeonTypeFlags(TV), Context, false);
     if (HasConstPtr)
----------------
Oh yes, this is wrong since it's always checking the non-AArch64 case (that final "false"). I'll put a more sensible check in before committing, though it should only lead to spurious warnings on polynomial types.

The real solution is to poke GCC maintainers until they agree that, yes, a signed polyN_t is clinically insane and should be scourged from the surface of the Earth (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56024).


http://llvm-reviews.chandlerc.com/D2738



More information about the cfe-commits mailing list