Fix cc -ftrapping-math on macppc

Handle CALL_RM like CALL for 32-bit ELF.  If a function call has the
strictfp attribute, its opcode changes from CALL to CALL_RM.  If a
call uses the secure PLT, then it must getGlobalBaseReg() to set r30.

After I rebuilt xenocara/lib/pixman with this change, Xorg stopped
crashing on my macppc.  pixman uses cc -ftrapping-math which puts
strictfp on each function call.

https://github.com/llvm/llvm-project/pull/72758

ok jca@ tobhe@ deraadt@

diff --git a/gnu/llvm/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp b/gnu/llvm/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
index d80a33f..7704d1e 100644 (file)
--- a/gnu/llvm/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
+++ b/gnu/llvm/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
@@ -5293,7 +5293,8 @@ void PPCDAGToDAGISel::Select(SDNode *N) {
     // generate secure plt code for TLS symbols.
     getGlobalBaseReg();
   } break;
-  case PPCISD::CALL: {
+  case PPCISD::CALL:
+  case PPCISD::CALL_RM: {
     if (PPCLowering->getPointerTy(CurDAG->getDataLayout()) != MVT::i32 ||
         !TM.isPositionIndependent() || !Subtarget->isSecurePlt() ||
         !Subtarget->isTargetELF())