Merging r261387:

[oota-llvm.git] / lib / Target / X86 / X86InstrCompiler.td

diff --git a/lib/Target/X86/X86InstrCompiler.td b/lib/Target/X86/X86InstrCompiler.td
index cc2703eb829fc9a6c540a7c7e2544e2600c55bc9..c709c8aca9faf1df156461d61c198ac6cac50ba3 100644 (file)
--- a/lib/Target/X86/X86InstrCompiler.td
+++ b/lib/Target/X86/X86InstrCompiler.td
@@ -262,13 +262,28 @@ def : Pat<(i64 0), (SUBREG_TO_REG (i64 0), (MOV32r0), sub_32bit)> {
   let AddedComplexity = 20;
 }
 
+let Predicates = [OptForSize, NotSlowIncDec, Not64BitMode],
+    AddedComplexity = 1 in {
+  // Pseudo instructions for materializing 1 and -1 using XOR+INC/DEC,
+  // which only require 3 bytes compared to MOV32ri which requires 5.
+  let Defs = [EFLAGS], isReMaterializable = 1, isPseudo = 1 in {
+    def MOV32r1 : I<0, Pseudo, (outs GR32:$dst), (ins), "",
+                        [(set GR32:$dst, 1)]>;
+    def MOV32r_1 : I<0, Pseudo, (outs GR32:$dst), (ins), "",
+                        [(set GR32:$dst, -1)]>;
+  }
+
+  // MOV16ri is 4 bytes, so the instructions above are smaller.
+  def : Pat<(i16 1), (EXTRACT_SUBREG (MOV32r1), sub_16bit)>;
+  def : Pat<(i16 -1), (EXTRACT_SUBREG (MOV32r_1), sub_16bit)>;
+}
+
 // Materialize i64 constant where top 32-bits are zero. This could theoretically
 // use MOV32ri with a SUBREG_TO_REG to represent the zero-extension, however
 // that would make it more difficult to rematerialize.
-let AddedComplexity = 1, isReMaterializable = 1, isAsCheapAsAMove = 1,
-    isCodeGenOnly = 1, hasSideEffects = 0 in
-def MOV32ri64 : Ii32<0xb8, AddRegFrm, (outs GR32:$dst), (ins i64i32imm:$src),
-                     "", [], IIC_ALU_NONMEM>, Sched<[WriteALU]>;
+let isReMaterializable = 1, isAsCheapAsAMove = 1,
+    isPseudo = 1, hasSideEffects = 0 in
+def MOV32ri64 : I<0, Pseudo, (outs GR32:$dst), (ins i64i32imm:$src), "", []>;
 
 // This 64-bit pseudo-move can be used for both a 64-bit constant that is
 // actually the zero-extension of a 32-bit constant and for labels in the
@@ -421,7 +436,7 @@ let Defs = [EAX, ECX, EDX, FP0, FP1, FP2, FP3, FP4, FP5, FP6, FP7,
             MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7,
             XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7,
             XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15, EFLAGS],
-    Uses = [ESP] in {
+    usesCustomInserter = 1, Uses = [ESP] in {
 def TLS_addr32 : I<0, Pseudo, (outs), (ins i32mem:$sym),
                   "# TLS_addr32",
                   [(X86tlsaddr tls32addr:$sym)]>,
@@ -441,7 +456,7 @@ let Defs = [RAX, RCX, RDX, RSI, RDI, R8, R9, R10, R11,
             MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7,
             XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7,
             XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15, EFLAGS],
-    Uses = [RSP] in {
+    usesCustomInserter = 1, Uses = [RSP] in {
 def TLS_addr64 : I<0, Pseudo, (outs), (ins i64mem:$sym),
                    "# TLS_addr64",
                   [(X86tlsaddr tls64addr:$sym)]>,
@@ -512,6 +527,7 @@ let usesCustomInserter = 1, Uses = [EFLAGS] in {
 
   defm _FR32   : CMOVrr_PSEUDO<FR32, f32>;
   defm _FR64   : CMOVrr_PSEUDO<FR64, f64>;
+  defm _FR128  : CMOVrr_PSEUDO<FR128, f128>;
   defm _V4F32  : CMOVrr_PSEUDO<VR128, v4f32>;
   defm _V2F64  : CMOVrr_PSEUDO<VR128, v2f64>;
   defm _V2I64  : CMOVrr_PSEUDO<VR128, v2i64>;
@@ -538,8 +554,8 @@ let usesCustomInserter = 1, Uses = [EFLAGS] in {
 // TODO: Get this to fold the constant into the instruction.
 let isCodeGenOnly = 1, Defs = [EFLAGS] in
 def OR32mrLocked  : I<0x09, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$zero),
-                      "or{l}\t{$zero, $dst|$dst, $zero}",
-                      [], IIC_ALU_MEM>, Requires<[Not64BitMode]>, LOCK,
+                      "or{l}\t{$zero, $dst|$dst, $zero}", [],
+                      IIC_ALU_MEM>, Requires<[Not64BitMode]>, OpSize32, LOCK,
                     Sched<[WriteALULd, WriteRMW]>;
 
 let hasSideEffects = 1 in