Add in64BitMode/in32BitMode to the MMX/SSE2/AVX maskmovq/dq instructions. This way the asm parser will pick the right one based on the mode. Instruction selection already did the right thing based on the pointer size.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192266 91177308-0d34-0410-b5e6-96231b3b80d8

5 files changed, 19 insertions, 11 deletions

diff --git a/lib/Target/X86/X86InstrFormats.td b/lib/Target/X86/X86InstrFormats.td
index 8b38ee6065..0fd9011338 100644
--- a/lib/Target/X86/X86InstrFormats.td
+++ b/lib/Target/X86/X86InstrFormats.td
@@ -772,6 +772,7 @@ class VRS2I<bits<8> o, Format F, dag outs, dag ins, string asm,
 //
 
 // MMXI   - MMX instructions with TB prefix.
+// MMXI32 - MMX instructions with TB prefix valid only in 32 bit mode.
 // MMXI64 - MMX instructions with TB prefix valid only in 64 bit mode.
 // MMX2I  - MMX / SSE2 instructions with TB and OpSize prefixes.
 // MMXIi8 - MMX instructions with ImmT == Imm8 and TB prefix.
@@ -781,6 +782,9 @@ class VRS2I<bits<8> o, Format F, dag outs, dag ins, string asm,
 class MMXI<bits<8> o, Format F, dag outs, dag ins, string asm, 
            list<dag> pattern, InstrItinClass itin = NoItinerary>
       : I<o, F, outs, ins, asm, pattern, itin>, TB, Requires<[HasMMX]>;
+class MMXI32<bits<8> o, Format F, dag outs, dag ins, string asm, 
+             list<dag> pattern, InstrItinClass itin = NoItinerary>
+      : I<o, F, outs, ins, asm, pattern, itin>, TB, Requires<[HasMMX,In32BitMode]>;
 class MMXI64<bits<8> o, Format F, dag outs, dag ins, string asm, 
              list<dag> pattern, InstrItinClass itin = NoItinerary>
       : I<o, F, outs, ins, asm, pattern, itin>, TB, Requires<[HasMMX,In64BitMode]>;
diff --git a/lib/Target/X86/X86InstrMMX.td b/lib/Target/X86/X86InstrMMX.td
index 16fb4d7959..fac6b35ca5 100644
--- a/lib/Target/X86/X86InstrMMX.td
+++ b/lib/Target/X86/X86InstrMMX.td
@@ -595,10 +595,10 @@ def : Pat<(x86mmx (MMX_X86movdq2q (loadv2i64 addr:$src))),
 // Misc.
 let SchedRW = [WriteShuffle] in {
 let Uses = [EDI] in
-def MMX_MASKMOVQ : MMXI<0xF7, MRMSrcReg, (outs), (ins VR64:$src, VR64:$mask),
-                        "maskmovq\t{$mask, $src|$src, $mask}",
-                        [(int_x86_mmx_maskmovq VR64:$src, VR64:$mask, EDI)],
-                        IIC_MMX_MASKMOV>;
+def MMX_MASKMOVQ : MMXI32<0xF7, MRMSrcReg, (outs), (ins VR64:$src, VR64:$mask),
+                          "maskmovq\t{$mask, $src|$src, $mask}",
+                          [(int_x86_mmx_maskmovq VR64:$src, VR64:$mask, EDI)],
+                          IIC_MMX_MASKMOV>;
 let Uses = [RDI] in
 def MMX_MASKMOVQ64: MMXI64<0xF7, MRMSrcReg, (outs), (ins VR64:$src, VR64:$mask),
                            "maskmovq\t{$mask, $src|$src, $mask}",
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 0c05bd2475..4c1b95366e 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -4336,25 +4336,25 @@ def PMOVMSKBrr : PDI<0xD7, MRMSrcReg, (outs GR32:$dst), (ins VR128:$src),
 
 let ExeDomain = SSEPackedInt, SchedRW = [WriteStore] in {
 
-let Uses = [EDI] in
+let Uses = [EDI], Predicates = [HasAVX,In32BitMode] in
 def VMASKMOVDQU : VPDI<0xF7, MRMSrcReg, (outs),
            (ins VR128:$src, VR128:$mask),
            "maskmovdqu\t{$mask, $src|$src, $mask}",
            [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, EDI)],
            IIC_SSE_MASKMOV>, VEX;
-let Uses = [RDI] in
+let Uses = [RDI], Predicates = [HasAVX,In64BitMode] in
 def VMASKMOVDQU64 : VPDI<0xF7, MRMSrcReg, (outs),
            (ins VR128:$src, VR128:$mask),
            "maskmovdqu\t{$mask, $src|$src, $mask}",
            [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, RDI)],
            IIC_SSE_MASKMOV>, VEX;
 
-let Uses = [EDI] in
+let Uses = [EDI], Predicates = [UseSSE2,In32BitMode] in
 def MASKMOVDQU : PDI<0xF7, MRMSrcReg, (outs), (ins VR128:$src, VR128:$mask),
            "maskmovdqu\t{$mask, $src|$src, $mask}",
            [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, EDI)],
            IIC_SSE_MASKMOV>;
-let Uses = [RDI] in
+let Uses = [RDI], Predicates = [UseSSE2,In64BitMode] in
 def MASKMOVDQU64 : PDI<0xF7, MRMSrcReg, (outs), (ins VR128:$src, VR128:$mask),
            "maskmovdqu\t{$mask, $src|$src, $mask}",
            [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, RDI)],
diff --git a/test/CodeGen/X86/maskmovdqu.ll b/test/CodeGen/X86/maskmovdqu.ll
index 7796f0e9a1..0b3334d19f 100644
--- a/test/CodeGen/X86/maskmovdqu.ll
+++ b/test/CodeGen/X86/maskmovdqu.ll
@@ -1,5 +1,7 @@
-; RUN: llc < %s -march=x86    -mattr=+sse2 | grep -i EDI
-; RUN: llc < %s -march=x86-64 -mattr=+sse2 | grep -i RDI
+; RUN: llc < %s -march=x86    -mattr=+sse2,-avx | grep -i EDI
+; RUN: llc < %s -march=x86-64 -mattr=+sse2,-avx | grep -i RDI
+; RUN: llc < %s -march=x86    -mattr=+avx | grep -i EDI
+; RUN: llc < %s -march=x86-64 -mattr=+avx | grep -i RDI
 ; rdar://6573467
 
 define void @test(<16 x i8> %a, <16 x i8> %b, i32 %dummy, i8* %c) nounwind {
diff --git a/test/CodeGen/X86/mmx-builtins.ll b/test/CodeGen/X86/mmx-builtins.ll
index f5b3f765fe..aabdd53b09 100644
--- a/test/CodeGen/X86/mmx-builtins.ll
+++ b/test/CodeGen/X86/mmx-builtins.ll
@@ -1,5 +1,7 @@
-; RUN: llc < %s -march=x86 -mattr=+mmx,+ssse3 | FileCheck %s
+; RUN: llc < %s -march=x86 -mattr=+mmx,+ssse3,-avx | FileCheck %s
 ; RUN: llc < %s -march=x86 -mattr=+avx | FileCheck %s
+; RUN: llc < %s -march=x86-64 -mattr=+mmx,+ssse3,-avx | FileCheck %s
+; RUN: llc < %s -march=x86-64 -mattr=+avx | FileCheck %s
 
 declare x86_mmx @llvm.x86.ssse3.phadd.w(x86_mmx, x86_mmx) nounwind readnone