[PATCH] D36789: [ARM] v8.3-a complex number support

[Sjoerd Meijer via Phabricator via llvm-commits]

SjoerdMeijer added inline comments.


================
Comment at: test/MC/ARM/neon-complex.s:12
+// RUN: not llvm-mc -triple thumb-none-linux-gnu -mattr=+v8.2a,+neon,+fullfp16 -show-encoding < %s 2>&1 | FileCheck %s --check-prefix=V82A
+// RUN: not llvm-mc -triple arm-none-linux-gnu -mattr=+v8.2a,+neon,+fullfp16 -show-encoding < %s 2>&1 | FileCheck %s --check-prefix=V82A
+
----------------
I think we also want to have negative tests for -neon


================
Comment at: test/MC/ARM/neon-complex.s:21
+// NO-FP16-STDERR: :[[@LINE-3]]:{{[0-9]*}}: error: instruction requires: full half-float
+// V82A: :[[@LINE-4]]:{{[0-9]*}}: error: instruction requires:
+  vcmla.f16 q0, q1, q2, #0
----------------
I think we want to have the full error message here, i.e. which extension is missing.


https://reviews.llvm.org/D36789