[NVPTX] Properly handle bitcast ConstantExpr when checking for the alignment of function parameters

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194410 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 64 insertions, 13 deletions

diff --git a/lib/Target/NVPTX/NVPTXISelLowering.cpp b/lib/Target/NVPTX/NVPTXISelLowering.cpp
index 55caaad65b..98b9ce6e26 100644
--- a/lib/Target/NVPTX/NVPTXISelLowering.cpp
+++ b/lib/Target/NVPTX/NVPTXISelLowering.cpp
@@ -471,22 +471,47 @@ NVPTXTargetLowering::getArgumentAlignment(SDValue Callee,
                                           Type *Ty,
                                           unsigned Idx) const {
   const DataLayout *TD = getDataLayout();
-  unsigned align = 0;
-  GlobalAddressSDNode *Func = dyn_cast<GlobalAddressSDNode>(Callee.getNode());
+  unsigned Align = 0;
+  const Value *DirectCallee = CS->getCalledFunction();
+
+  if (!DirectCallee) {
+    // We don't have a direct function symbol, but that may be because of
+    // constant cast instructions in the call.
+    const Instruction *CalleeI = CS->getInstruction();
+    assert(CalleeI && "Call target is not a function or derived value?");
+
+    // With bitcast'd call targets, the instruction will be the call
+    if (isa<CallInst>(CalleeI)) {
+      // Check if we have call alignment metadata
+      if (llvm::getAlign(*cast<CallInst>(CalleeI), Idx, Align))
+        return Align;
+
+      const Value *CalleeV = cast<CallInst>(CalleeI)->getCalledValue();
+      // Ignore any bitcast instructions
+      while(isa<ConstantExpr>(CalleeV)) {
+        const ConstantExpr *CE = cast<ConstantExpr>(CalleeV);
+        if (!CE->isCast())
+          break;
+        // Look through the bitcast
+        CalleeV = cast<ConstantExpr>(CalleeV)->getOperand(0);
+      }
 
-  if (Func) { // direct call
-    assert(CS->getCalledFunction() &&
-           "direct call cannot find callee");
-    if (!llvm::getAlign(*(CS->getCalledFunction()), Idx, align))
-      align = TD->getABITypeAlignment(Ty);
-  }
-  else { // indirect call
-    const CallInst *CallI = dyn_cast<CallInst>(CS->getInstruction());
-    if (!llvm::getAlign(*CallI, Idx, align))
-      align = TD->getABITypeAlignment(Ty);
+      // We have now looked past all of the bitcasts.  Do we finally have a
+      // Function?
+      if (isa<Function>(CalleeV))
+        DirectCallee = CalleeV;
+    }
   }
 
-  return align;
+  // Check for function alignment information if we found that the
+  // ultimate target is a Function
+  if (DirectCallee)
+    if (llvm::getAlign(*cast<Function>(DirectCallee), Idx, Align))
+      return Align;
+
+  // Call is indirect or alignment information is not available, fall back to
+  // the ABI type alignment
+  return TD->getABITypeAlignment(Ty);
 }
 
 SDValue NVPTXTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
diff --git a/test/CodeGen/NVPTX/bug17709.ll b/test/CodeGen/NVPTX/bug17709.ll
new file mode 100644
index 0000000000..92f0fcb11e
--- /dev/null
+++ b/test/CodeGen/NVPTX/bug17709.ll
@@ -0,0 +1,26 @@
+; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s
+
+; ModuleID = '__kernelgen_main_module'
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16-v32:32:32-v64:64:64-v128:128:128-n16:32:64"
+target triple = "nvptx64-nvidia-cuda"
+
+define linker_private ptx_device { double, double } @__utils1_MOD_trace(%"struct.array2_complex(kind=8).43.5.57"* noalias %m) {
+entry:
+  ;unreachable
+  %t0 = insertvalue {double, double} undef, double 1.0, 0
+  %t1 = insertvalue {double, double} %t0, double 1.0, 1
+  ret { double, double } %t1
+}
+
+%struct.descriptor_dimension.0.52 = type { i64, i64, i64 }
+%"struct.array2_complex(kind=8).37.18.70" = type { i8*, i64, i64, [2 x %struct.descriptor_dimension.0.52] }
+%"struct.array2_complex(kind=8).43.5.57" = type { i8*, i64, i64, [2 x %struct.descriptor_dimension.0.52] }
+@replacementOfAlloca8 = private global %"struct.array2_complex(kind=8).37.18.70" zeroinitializer, align 4096
+
+; CHECK: .visible .entry __kernelgen_main
+define ptx_kernel void @__kernelgen_main(i32* nocapture %args, i32*) {
+entry:
+  %1 = tail call ptx_device { double, double } bitcast ({ double, double } (%"struct.array2_complex(kind=8).43.5.57"*)* @__utils1_MOD_trace to { double, double } (%"struct.array2_complex(kind=8).37.18.70"*)*)(%"struct.array2_complex(kind=8).37.18.70"* noalias @replacementOfAlloca8)
+  ret void
+}
+