[PATCH] D26343: [RFC] Refactor TargetParserTests

[Renato Golin via Phabricator via llvm-commits]

rengolin added a comment.

Sorry for the delay, I'm wondering if the `&=` is the reason why you're getting the wrong expectation on the tests...



================
Comment at: unittests/Support/TargetParserTest.cpp:541
+                              ARMBuildAttrs::CPUArch::v8_A));
+  EXPECT_TRUE(testAArch64Arch("armv8.2-a", "generic", "v8.2a",
+                              ARMBuildAttrs::CPUArch::v8_A));
----------------
jojo wrote:
> rengolin wrote:
> > We now have v8M and v8R.
> AFAIK, both of them are 32bit. 
> And v8R has been supported in ARMTargetParser. 
> There is still no support in ARM backend for v8M, 
> Should  we add it to targetparser after it is actually
> supported in the backend?
Yes, because this tests the parser, not the target support.


================
Comment at: unittests/Support/TargetParserTest.cpp:40
+  unsigned FPUKind = ARM::getDefaultFPU(CPUName, ArchKind);
+  ret &= ARM::getFPUName(FPUKind).equals(ExpectedFPU);
+
----------------
Shouldn't this be `ret |=`?


================
Comment at: unittests/Support/TargetParserTest.cpp:582
+  unsigned FPUKind = AArch64::getDefaultFPU(CPUName, ArchKind);
+  ret &= AArch64::getFPUName(FPUKind).equals(ExpectedFPU);
+
----------------
Same here.


https://reviews.llvm.org/D26343