Index: llvm/trunk/lib/Target/ARM/ARMTargetTransformInfo.cpp
===================================================================
--- llvm/trunk/lib/Target/ARM/ARMTargetTransformInfo.cpp
+++ llvm/trunk/lib/Target/ARM/ARMTargetTransformInfo.cpp
@@ -167,6 +167,27 @@
   if (!SrcTy.isSimple() || !DstTy.isSimple())
     return BaseT::getCastInstrCost(Opcode, Dst, Src);
 
+  // The extend of a load is free
+  if (I && isa<LoadInst>(I->getOperand(0))) {
+    static const TypeConversionCostTblEntry LoadConversionTbl[] = {
+        {ISD::SIGN_EXTEND, MVT::i32, MVT::i16, 0},
+        {ISD::ZERO_EXTEND, MVT::i32, MVT::i16, 0},
+        {ISD::SIGN_EXTEND, MVT::i32, MVT::i8, 0},
+        {ISD::ZERO_EXTEND, MVT::i32, MVT::i8, 0},
+        {ISD::SIGN_EXTEND, MVT::i16, MVT::i8, 0},
+        {ISD::ZERO_EXTEND, MVT::i16, MVT::i8, 0},
+        {ISD::SIGN_EXTEND, MVT::i64, MVT::i32, 1},
+        {ISD::ZERO_EXTEND, MVT::i64, MVT::i32, 1},
+        {ISD::SIGN_EXTEND, MVT::i64, MVT::i16, 1},
+        {ISD::ZERO_EXTEND, MVT::i64, MVT::i16, 1},
+        {ISD::SIGN_EXTEND, MVT::i64, MVT::i8, 1},
+        {ISD::ZERO_EXTEND, MVT::i64, MVT::i8, 1},
+    };
+    if (const auto *Entry = ConvertCostTableLookup(
+            LoadConversionTbl, ISD, DstTy.getSimpleVT(), SrcTy.getSimpleVT()))
+      return Entry->Cost;
+  }
+
   // Some arithmetic, load and store operations have specific instructions
   // to cast up/down their types automatically at no extra cost.
   // TODO: Get these tables to know at least what the related operations are.
Index: llvm/trunk/test/Analysis/CostModel/ARM/cast.ll
===================================================================
--- llvm/trunk/test/Analysis/CostModel/ARM/cast.ll
+++ llvm/trunk/test/Analysis/CostModel/ARM/cast.ll
@@ -814,15 +814,15 @@
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv2i16 = load <2 x i16>, <2 x i16>* undef
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv4i16 = load <4 x i16>, <4 x i16>* undef
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv2i32 = load <2 x i32>, <2 x i32>* undef
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r0 = sext i8 %loadi8 to i16
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r1 = zext i8 %loadi8 to i16
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r2 = sext i8 %loadi8 to i32
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r3 = zext i8 %loadi8 to i32
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r0 = sext i8 %loadi8 to i16
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r1 = zext i8 %loadi8 to i16
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r2 = sext i8 %loadi8 to i32
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r3 = zext i8 %loadi8 to i32
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r4 = sext i8 %loadi8 to i64
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r5 = zext i8 %loadi8 to i64
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r6 = sext i16 %loadi16 to i32
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r7 = zext i16 %loadi16 to i32
-; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r8 = sext i16 %loadi16 to i64
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r6 = sext i16 %loadi16 to i32
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r7 = zext i16 %loadi16 to i32
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r8 = sext i16 %loadi16 to i64
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r9 = zext i16 %loadi16 to i64
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r10 = sext i32 %loadi32 to i64
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r11 = zext i32 %loadi32 to i64
@@ -850,15 +850,15 @@
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv2i16 = load <2 x i16>, <2 x i16>* undef
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv4i16 = load <4 x i16>, <4 x i16>* undef
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %loadv2i32 = load <2 x i32>, <2 x i32>* undef
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r0 = sext i8 %loadi8 to i16
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r1 = zext i8 %loadi8 to i16
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r2 = sext i8 %loadi8 to i32
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r3 = zext i8 %loadi8 to i32
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r0 = sext i8 %loadi8 to i16
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r1 = zext i8 %loadi8 to i16
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r2 = sext i8 %loadi8 to i32
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r3 = zext i8 %loadi8 to i32
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r4 = sext i8 %loadi8 to i64
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r5 = zext i8 %loadi8 to i64
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r6 = sext i16 %loadi16 to i32
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r7 = zext i16 %loadi16 to i32
-; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r8 = sext i16 %loadi16 to i64
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r6 = sext i16 %loadi16 to i32
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %r7 = zext i16 %loadi16 to i32
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r8 = sext i16 %loadi16 to i64
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r9 = zext i16 %loadi16 to i64
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r10 = sext i32 %loadi32 to i64
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r11 = zext i32 %loadi32 to i64