From 896c1e9b70dd2db5697c374e366ba04468e617e9 Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@gmail.com>
Date: Sun, 25 Jan 2015 02:21:16 +0000
Subject: [PATCH] [X86] Replace i32i8imm on SSE/AVX instructions with i32u8imm
 which will make the assembler bounds check them. It will also make them print
 as unsigned.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@227032 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/X86/X86InstrAVX512.td        | 12 ++++----
 lib/Target/X86/X86InstrInfo.td          |  8 +++++
 lib/Target/X86/X86InstrMMX.td           |  8 ++---
 lib/Target/X86/X86InstrSSE.td           | 40 ++++++++++++-------------
 utils/TableGen/X86RecognizableInstr.cpp |  3 ++
 5 files changed, 41 insertions(+), 30 deletions(-)

diff --git a/lib/Target/X86/X86InstrAVX512.td b/lib/Target/X86/X86InstrAVX512.td
index 57d0e273cdd..57b638e2ea4 100644
--- a/lib/Target/X86/X86InstrAVX512.td
+++ b/lib/Target/X86/X86InstrAVX512.td
@@ -4155,12 +4155,12 @@ multiclass avx512_cvtph2ps<RegisterClass destRC, RegisterClass srcRC,
 multiclass avx512_cvtps2ph<RegisterClass destRC, RegisterClass srcRC,
                              X86MemOperand x86memop> {
   def rr : AVX512AIi8<0x1D, MRMDestReg, (outs destRC:$dst),
-               (ins srcRC:$src1, i32i8imm:$src2),
+               (ins srcRC:$src1, i32u8imm:$src2),
                "vcvtps2ph\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                []>, EVEX;
   let hasSideEffects = 0, mayStore = 1 in
   def mr : AVX512AIi8<0x1D, MRMDestMem, (outs),
-               (ins x86memop:$dst, srcRC:$src1, i32i8imm:$src2),
+               (ins x86memop:$dst, srcRC:$src1, i32u8imm:$src2),
                "vcvtps2ph\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>, EVEX;
 }
 
@@ -4558,14 +4558,14 @@ let ExeDomain = d in {
   // Intrinsic operation, reg.
   // Vector intrinsic operation, reg
   def r : AVX512AIi8<opc, MRMSrcReg,
-                    (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins RC:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     []>, EVEX;
 
   // Vector intrinsic operation, mem
   def m : AVX512AIi8<opc, MRMSrcMem,
-                    (outs RC:$dst), (ins x86memop:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins x86memop:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     []>, EVEX;
@@ -4596,13 +4596,13 @@ multiclass avx512_rndscale_scalar<bits<8> opc, string OpcodeStr,
                      Operand x86memop, RegisterClass RC, Domain d> {
 let ExeDomain = d in {
   def r : AVX512AIi8<opc, MRMSrcReg,
-                    (outs RC:$dst), (ins RC:$src1, RC:$src2, i32i8imm:$src3),
+                    (outs RC:$dst), (ins RC:$src1, RC:$src2, i32u8imm:$src3),
                     !strconcat(OpcodeStr,
                     "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     []>, EVEX_4V;
 
   def m : AVX512AIi8<opc, MRMSrcMem,
-                    (outs RC:$dst), (ins RC:$src1, x86memop:$src2,  i32i8imm:$src3),
+                    (outs RC:$dst), (ins RC:$src1, x86memop:$src2,  i32u8imm:$src3),
                     !strconcat(OpcodeStr,
                     "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     []>, EVEX_4V;
diff --git a/lib/Target/X86/X86InstrInfo.td b/lib/Target/X86/X86InstrInfo.td
index aa27345b7be..8320100ef90 100644
--- a/lib/Target/X86/X86InstrInfo.td
+++ b/lib/Target/X86/X86InstrInfo.td
@@ -648,6 +648,14 @@ def u8imm : Operand<i8> {
   let OperandType = "OPERAND_IMMEDIATE";
 }
 
+// 32-bit immediate but only 8-bits are significant and they are unsigned.
+// Used by some SSE/AVX instructions that use intrinsics.
+def i32u8imm : Operand<i32> {
+  let PrintMethod = "printU8Imm";
+  let ParserMatchClass = ImmUnsignedi8AsmOperand;
+  let OperandType = "OPERAND_IMMEDIATE";
+}
+
 // 64-bits but only 32 bits are significant, and those bits are treated as being
 // pc relative.
 def i64i32imm_pcrel : Operand<i64> {
diff --git a/lib/Target/X86/X86InstrMMX.td b/lib/Target/X86/X86InstrMMX.td
index dca5a77875c..54821ee2404 100644
--- a/lib/Target/X86/X86InstrMMX.td
+++ b/lib/Target/X86/X86InstrMMX.td
@@ -125,7 +125,7 @@ let Constraints = "$src1 = $dst" in {
                                     (bitconvert (load_mmx addr:$src2))))],
                   itins.rm>, Sched<[WriteVecShiftLd, ReadAfterLd]>;
     def ri : MMXIi8<opc2, ImmForm, (outs VR64:$dst),
-                                   (ins VR64:$src1, i32i8imm:$src2),
+                                   (ins VR64:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
            [(set VR64:$dst, (IntId2 VR64:$src1, imm:$src2))], itins.ri>,
            Sched<[WriteVecShift]>;
@@ -559,7 +559,7 @@ let Constraints = "$src1 = $dst" in {
 
 // Extract / Insert
 def MMX_PEXTRWirri: MMXIi8<0xC5, MRMSrcReg,
-                       (outs GR32orGR64:$dst), (ins VR64:$src1, i32i8imm:$src2),
+                       (outs GR32orGR64:$dst), (ins VR64:$src1, i32u8imm:$src2),
                        "pextrw\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                        [(set GR32orGR64:$dst, (int_x86_mmx_pextr_w VR64:$src1,
                                                imm:$src2))],
@@ -567,7 +567,7 @@ def MMX_PEXTRWirri: MMXIi8<0xC5, MRMSrcReg,
 let Constraints = "$src1 = $dst" in {
   def MMX_PINSRWirri : MMXIi8<0xC4, MRMSrcReg,
                       (outs VR64:$dst),
-                      (ins VR64:$src1, GR32orGR64:$src2, i32i8imm:$src3),
+                      (ins VR64:$src1, GR32orGR64:$src2, i32u8imm:$src3),
                       "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
                       [(set VR64:$dst, (int_x86_mmx_pinsr_w VR64:$src1,
                                         GR32orGR64:$src2, imm:$src3))],
@@ -575,7 +575,7 @@ let Constraints = "$src1 = $dst" in {
 
   def MMX_PINSRWirmi : MMXIi8<0xC4, MRMSrcMem,
                      (outs VR64:$dst),
-                     (ins VR64:$src1, i16mem:$src2, i32i8imm:$src3),
+                     (ins VR64:$src1, i16mem:$src2, i32u8imm:$src3),
                      "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
                      [(set VR64:$dst, (int_x86_mmx_pinsr_w VR64:$src1,
                                          (i32 (anyext (loadi16 addr:$src2))),
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index d0d390a8f39..6a9dfa127b0 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -4323,13 +4323,13 @@ defm VPSRAD : PDI_binop_rmi<0xE2, 0x72, MRM4r, "vpsrad", X86vsra, X86vsrai,
 let ExeDomain = SSEPackedInt, SchedRW = [WriteVecShift] in {
   // 128-bit logical shifts.
   def VPSLLDQri : PDIi8<0x73, MRM7r,
-                    (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
+                    (outs VR128:$dst), (ins VR128:$src1, i32u8imm:$src2),
                     "vpslldq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                     [(set VR128:$dst,
                       (int_x86_sse2_psll_dq_bs VR128:$src1, imm:$src2))]>,
                     VEX_4V;
   def VPSRLDQri : PDIi8<0x73, MRM3r,
-                    (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
+                    (outs VR128:$dst), (ins VR128:$src1, i32u8imm:$src2),
                     "vpsrldq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                     [(set VR128:$dst,
                       (int_x86_sse2_psrl_dq_bs VR128:$src1, imm:$src2))]>,
@@ -4369,13 +4369,13 @@ defm VPSRADY : PDI_binop_rmi<0xE2, 0x72, MRM4r, "vpsrad", X86vsra, X86vsrai,
 let ExeDomain = SSEPackedInt, SchedRW = [WriteVecShift] in {
   // 256-bit logical shifts.
   def VPSLLDQYri : PDIi8<0x73, MRM7r,
-                    (outs VR256:$dst), (ins VR256:$src1, i32i8imm:$src2),
+                    (outs VR256:$dst), (ins VR256:$src1, i32u8imm:$src2),
                     "vpslldq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                     [(set VR256:$dst,
                       (int_x86_avx2_psll_dq_bs VR256:$src1, imm:$src2))]>,
                     VEX_4V, VEX_L;
   def VPSRLDQYri : PDIi8<0x73, MRM3r,
-                    (outs VR256:$dst), (ins VR256:$src1, i32i8imm:$src2),
+                    (outs VR256:$dst), (ins VR256:$src1, i32u8imm:$src2),
                     "vpsrldq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                     [(set VR256:$dst,
                       (int_x86_avx2_psrl_dq_bs VR256:$src1, imm:$src2))]>,
@@ -4415,13 +4415,13 @@ defm PSRAD : PDI_binop_rmi<0xE2, 0x72, MRM4r, "psrad", X86vsra, X86vsrai,
 let ExeDomain = SSEPackedInt, SchedRW = [WriteVecShift] in {
   // 128-bit logical shifts.
   def PSLLDQri : PDIi8<0x73, MRM7r,
-                       (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
+                       (outs VR128:$dst), (ins VR128:$src1, i32u8imm:$src2),
                        "pslldq\t{$src2, $dst|$dst, $src2}",
                        [(set VR128:$dst,
                          (int_x86_sse2_psll_dq_bs VR128:$src1, imm:$src2))],
                          IIC_SSE_INTSHDQ_P_RI>;
   def PSRLDQri : PDIi8<0x73, MRM3r,
-                       (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
+                       (outs VR128:$dst), (ins VR128:$src1, i32u8imm:$src2),
                        "psrldq\t{$src2, $dst|$dst, $src2}",
                        [(set VR128:$dst,
                          (int_x86_sse2_psrl_dq_bs VR128:$src1, imm:$src2))],
@@ -4805,7 +4805,7 @@ let ExeDomain = SSEPackedInt in {
 multiclass sse2_pinsrw<bit Is2Addr = 1> {
   def rri : Ii8<0xC4, MRMSrcReg,
        (outs VR128:$dst), (ins VR128:$src1,
-        GR32orGR64:$src2, i32i8imm:$src3),
+        GR32orGR64:$src2, u8imm:$src3),
        !if(Is2Addr,
            "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
            "vpinsrw\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
@@ -4814,7 +4814,7 @@ multiclass sse2_pinsrw<bit Is2Addr = 1> {
        IIC_SSE_PINSRW>, Sched<[WriteShuffle]>;
   def rmi : Ii8<0xC4, MRMSrcMem,
                        (outs VR128:$dst), (ins VR128:$src1,
-                        i16mem:$src2, i32i8imm:$src3),
+                        i16mem:$src2, u8imm:$src3),
        !if(Is2Addr,
            "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
            "vpinsrw\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
@@ -6607,7 +6607,7 @@ let ExeDomain = SSEPackedSingle in {
   // Intrinsic operation, reg.
   // Vector intrinsic operation, reg
   def PSr : SS4AIi8<opcps, MRMSrcReg,
-                    (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins RC:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     [(set RC:$dst, (V4F32Int RC:$src1, imm:$src2))],
@@ -6615,7 +6615,7 @@ let ExeDomain = SSEPackedSingle in {
 
   // Vector intrinsic operation, mem
   def PSm : SS4AIi8<opcps, MRMSrcMem,
-                    (outs RC:$dst), (ins x86memop:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins x86memop:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     [(set RC:$dst,
@@ -6626,7 +6626,7 @@ let ExeDomain = SSEPackedSingle in {
 let ExeDomain = SSEPackedDouble in {
   // Vector intrinsic operation, reg
   def PDr : SS4AIi8<opcpd, MRMSrcReg,
-                    (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins RC:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     [(set RC:$dst, (V2F64Int RC:$src1, imm:$src2))],
@@ -6634,7 +6634,7 @@ let ExeDomain = SSEPackedDouble in {
 
   // Vector intrinsic operation, mem
   def PDm : SS4AIi8<opcpd, MRMSrcMem,
-                    (outs RC:$dst), (ins x86memop:$src1, i32i8imm:$src2),
+                    (outs RC:$dst), (ins x86memop:$src1, i32u8imm:$src2),
                     !strconcat(OpcodeStr,
                     "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
                     [(set RC:$dst,
@@ -6651,7 +6651,7 @@ let ExeDomain = GenericDomain in {
   // Operation, reg.
   let hasSideEffects = 0 in
   def SSr : SS4AIi8<opcss, MRMSrcReg,
-      (outs FR32:$dst), (ins FR32:$src1, FR32:$src2, i32i8imm:$src3),
+      (outs FR32:$dst), (ins FR32:$src1, FR32:$src2, i32u8imm:$src3),
       !if(Is2Addr,
           !strconcat(OpcodeStr,
               "ss\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -6662,7 +6662,7 @@ let ExeDomain = GenericDomain in {
   // Intrinsic operation, reg.
   let isCodeGenOnly = 1 in
   def SSr_Int : SS4AIi8<opcss, MRMSrcReg,
-        (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
+        (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32u8imm:$src3),
         !if(Is2Addr,
             !strconcat(OpcodeStr,
                 "ss\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -6673,7 +6673,7 @@ let ExeDomain = GenericDomain in {
 
   // Intrinsic operation, mem.
   def SSm : SS4AIi8<opcss, MRMSrcMem,
-        (outs VR128:$dst), (ins VR128:$src1, ssmem:$src2, i32i8imm:$src3),
+        (outs VR128:$dst), (ins VR128:$src1, ssmem:$src2, i32u8imm:$src3),
         !if(Is2Addr,
             !strconcat(OpcodeStr,
                 "ss\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -6686,7 +6686,7 @@ let ExeDomain = GenericDomain in {
   // Operation, reg.
   let hasSideEffects = 0 in
   def SDr : SS4AIi8<opcsd, MRMSrcReg,
-        (outs FR64:$dst), (ins FR64:$src1, FR64:$src2, i32i8imm:$src3),
+        (outs FR64:$dst), (ins FR64:$src1, FR64:$src2, i32u8imm:$src3),
         !if(Is2Addr,
             !strconcat(OpcodeStr,
                 "sd\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -6697,7 +6697,7 @@ let ExeDomain = GenericDomain in {
   // Intrinsic operation, reg.
   let isCodeGenOnly = 1 in
   def SDr_Int : SS4AIi8<opcsd, MRMSrcReg,
-        (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
+        (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32u8imm:$src3),
         !if(Is2Addr,
             !strconcat(OpcodeStr,
                 "sd\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -6708,7 +6708,7 @@ let ExeDomain = GenericDomain in {
 
   // Intrinsic operation, mem.
   def SDm : SS4AIi8<opcsd, MRMSrcMem,
-        (outs VR128:$dst), (ins VR128:$src1, sdmem:$src2, i32i8imm:$src3),
+        (outs VR128:$dst), (ins VR128:$src1, sdmem:$src2, i32u8imm:$src3),
         !if(Is2Addr,
             !strconcat(OpcodeStr,
                 "sd\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -8482,14 +8482,14 @@ multiclass f16c_ph2ps<RegisterClass RC, X86MemOperand x86memop, Intrinsic Int> {
 
 multiclass f16c_ps2ph<RegisterClass RC, X86MemOperand x86memop, Intrinsic Int> {
   def rr : Ii8<0x1D, MRMDestReg, (outs VR128:$dst),
-               (ins RC:$src1, i32i8imm:$src2),
+               (ins RC:$src1, i32u8imm:$src2),
                "vcvtps2ph\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                [(set VR128:$dst, (Int RC:$src1, imm:$src2))]>,
                TAPD, VEX, Sched<[WriteCvtF2F]>;
   let hasSideEffects = 0, mayStore = 1,
       SchedRW = [WriteCvtF2FLd, WriteRMW] in
   def mr : Ii8<0x1D, MRMDestMem, (outs),
-               (ins x86memop:$dst, RC:$src1, i32i8imm:$src2),
+               (ins x86memop:$dst, RC:$src1, i32u8imm:$src2),
                "vcvtps2ph\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>,
                TAPD, VEX;
 }
diff --git a/utils/TableGen/X86RecognizableInstr.cpp b/utils/TableGen/X86RecognizableInstr.cpp
index 4f23db0908d..88d5cfbdbdf 100644
--- a/utils/TableGen/X86RecognizableInstr.cpp
+++ b/utils/TableGen/X86RecognizableInstr.cpp
@@ -933,6 +933,7 @@ OperandType RecognizableInstr::typeFromString(const std::string &s,
   TYPE("i8mem",               TYPE_M8)
   TYPE("i8imm",               TYPE_IMM8)
   TYPE("u8imm",               TYPE_UIMM8)
+  TYPE("i32u8imm",            TYPE_UIMM8)
   TYPE("GR8",                 TYPE_R8)
   TYPE("VR128",               TYPE_XMM128)
   TYPE("VR128X",              TYPE_XMM128)
@@ -1044,6 +1045,7 @@ RecognizableInstr::immediateEncodingFromString(const std::string &s,
   ENCODING("i64i8imm",        ENCODING_IB)
   ENCODING("i8imm",           ENCODING_IB)
   ENCODING("u8imm",           ENCODING_IB)
+  ENCODING("i32u8imm",        ENCODING_IB)
   // This is not a typo.  Instructions like BLENDVPD put
   // register IDs in 8-bit immediates nowadays.
   ENCODING("FR32",            ENCODING_IB)
@@ -1213,6 +1215,7 @@ RecognizableInstr::relocationEncodingFromString(const std::string &s,
   ENCODING("i64i8imm",        ENCODING_IB)
   ENCODING("i8imm",           ENCODING_IB)
   ENCODING("u8imm",           ENCODING_IB)
+  ENCODING("i32u8imm",        ENCODING_IB)
   ENCODING("i64i32imm_pcrel", ENCODING_ID)
   ENCODING("i16imm_pcrel",    ENCODING_IW)
   ENCODING("i32imm_pcrel",    ENCODING_ID)
-- 
2.34.1