[PATCH] D39688: [Nios2] final infrastructure addition to provide compilation of simple return from a function.

[Andrei via Phabricator via llvm-commits]

AndreiGrischenko added inline comments.


================
Comment at: lib/Target/Nios2/MCTargetDesc/Nios2MCTargetDesc.cpp:49
   if (CPU.empty() || CPU == "generic") {
-    if (TT.getArch() == Triple::nios2) {
-      if (CPU.empty() || CPU == "nios2r2") {
-        ArchFS = "+nios2r2";
-      } else {
-        if (CPU == "nios2r1") {
-          ArchFS = "+nios2r1";
-        }
-      }
-    }
+    CPU == "nios2r1";
   }
----------------
echristo wrote:
> craig.topper wrote:
> > That's a comparison not an assignment.
> Which also means an untested code path - relatedly, there are no tests here.
omg! Very stupid typo.


================
Comment at: lib/Target/Nios2/MCTargetDesc/Nios2MCTargetDesc.cpp:49
   if (CPU.empty() || CPU == "generic") {
-    if (TT.getArch() == Triple::nios2) {
-      if (CPU.empty() || CPU == "nios2r2") {
-        ArchFS = "+nios2r2";
-      } else {
-        if (CPU == "nios2r1") {
-          ArchFS = "+nios2r1";
-        }
-      }
-    }
+    CPU == "nios2r1";
   }
----------------
AndreiGrischenko wrote:
> echristo wrote:
> > craig.topper wrote:
> > > That's a comparison not an assignment.
> > Which also means an untested code path - relatedly, there are no tests here.
> omg! Very stupid typo.
Not exactly. This patch is aimed to start printing "ret" into assembly output, so I've added appropriate test case for that. Further I plan to add more test cases when adding exact things, not just infrastructure. 
However I will add one more test checking that Nios2 processors are accepted.


https://reviews.llvm.org/D39688