From: Tom Stellard <thomas.stellard@amd.com>
Date: Thu, 5 Sep 2013 18:37:57 +0000 (+0000)
Subject: R600: Fix incorrect LDS size calculation
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=470c451574609adcaab9b279cc74c6ff0f91b00f;p=oota-llvm.git

R600: Fix incorrect LDS size calculation

GlobalAdderss nodes that appeared in more than one basic block were
being counted twice.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@190078 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/R600/AMDGPUISelLowering.cpp b/lib/Target/R600/AMDGPUISelLowering.cpp
index 1237323ee89..d6b7cbed598 100644
--- a/lib/Target/R600/AMDGPUISelLowering.cpp
+++ b/lib/Target/R600/AMDGPUISelLowering.cpp
@@ -246,12 +246,18 @@ SDValue AMDGPUTargetLowering::LowerGlobalAddress(AMDGPUMachineFunction* MFI,
   assert(G->getOffset() == 0 &&
          "Do not know what to do with an non-zero offset");
 
-  unsigned Offset = MFI->LDSSize;
   const GlobalValue *GV = G->getGlobal();
-  uint64_t Size = TD->getTypeAllocSize(GV->getType()->getElementType());
 
-  // XXX: Account for alignment?
-  MFI->LDSSize += Size;
+  unsigned Offset;
+  if (MFI->LocalMemoryObjects.count(GV) == 0) {
+    uint64_t Size = TD->getTypeAllocSize(GV->getType()->getElementType());
+    Offset = MFI->LDSSize;
+    MFI->LocalMemoryObjects[GV] = Offset;
+    // XXX: Account for alignment?
+    MFI->LDSSize += Size;
+  } else {
+    Offset = MFI->LocalMemoryObjects[GV];
+  }
 
   return DAG.getConstant(Offset, getPointerTy(G->getAddressSpace()));
 }
diff --git a/lib/Target/R600/AMDGPUMachineFunction.h b/lib/Target/R600/AMDGPUMachineFunction.h
index 789b96a8fa0..fe80ce3f015 100644
--- a/lib/Target/R600/AMDGPUMachineFunction.h
+++ b/lib/Target/R600/AMDGPUMachineFunction.h
@@ -14,6 +14,7 @@
 #define AMDGPUMACHINEFUNCTION_H
 
 #include "llvm/CodeGen/MachineFunction.h"
+#include <map>
 
 namespace llvm {
 
@@ -21,6 +22,9 @@ class AMDGPUMachineFunction : public MachineFunctionInfo {
 public:
   AMDGPUMachineFunction(const MachineFunction &MF);
   unsigned ShaderType;
+  /// A map to keep track of local memory objects and their offsets within
+  /// the local memory space.
+  std::map<const GlobalValue *, unsigned> LocalMemoryObjects;
   /// Number of bytes in the LDS that are being used.
   unsigned LDSSize;
 };
diff --git a/test/CodeGen/R600/lds-size.ll b/test/CodeGen/R600/lds-size.ll
new file mode 100644
index 00000000000..2185180fd83
--- /dev/null
+++ b/test/CodeGen/R600/lds-size.ll
@@ -0,0 +1,26 @@
+; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s
+
+; This test makes sure we do not double count global values when they are
+; used in different basic blocks.
+
+; CHECK-LABEL: @test
+; CHECK: .long   166120
+; CHECK-NEXT: .long   1
+@lds = internal addrspace(3) unnamed_addr global i32 zeroinitializer, align 4
+
+define void @test(i32 addrspace(1)* %out, i32 %cond) {
+entry:
+  %0 = icmp eq i32 %cond, 0
+  br i1 %0, label %if, label %else
+
+if:
+  store i32 1, i32 addrspace(3)* @lds
+  br label %endif
+
+else:
+  store i32 2, i32 addrspace(3)* @lds
+  br label %endif
+
+endif:
+  ret void
+}