[anyregcc] Fix callee-save mask for anyregcc

Use separate callee-save masks for XMM and YMM registers for anyregcc on X86 and
select the proper mask depending on the target cpu we compile for.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@198985 91177308-0d34-0410-b5e6-96231b3b80d8

4 files changed, 141 insertions, 25 deletions

diff --git a/docs/BitCodeFormat.rst b/docs/BitCodeFormat.rst
index d9d1df0b8e..89531dcfea 100644
--- a/docs/BitCodeFormat.rst
+++ b/docs/BitCodeFormat.rst
@@ -729,6 +729,8 @@ function. The operand fields are:
   * ``ccc``: code 0
   * ``fastcc``: code 8
   * ``coldcc``: code 9
+  * ``webkit_jscc``: code 12
+  * ``anyregcc``: code 13
   * ``x86_stdcallcc``: code 64
   * ``x86_fastcallcc``: code 65
   * ``arm_apcscc``: code 66
diff --git a/lib/Target/X86/X86CallingConv.td b/lib/Target/X86/X86CallingConv.td
index 6185c1cd38..e733420e09 100644
--- a/lib/Target/X86/X86CallingConv.td
+++ b/lib/Target/X86/X86CallingConv.td
@@ -620,14 +620,15 @@ def CSR_64EHRet : CalleeSavedRegs<(add RAX, RDX, CSR_64)>;
 def CSR_Win64 : CalleeSavedRegs<(add RBX, RBP, RDI, RSI, R12, R13, R14, R15,
                                      (sequence "XMM%u", 6, 15))>;
 
-def CSR_MostRegs_64 : CalleeSavedRegs<(add RBX, RCX, RDX, RSI, RDI, R8, R9, R10,
+def CSR_64_MostRegs : CalleeSavedRegs<(add RBX, RCX, RDX, RSI, RDI, R8, R9, R10,
                                            R11, R12, R13, R14, R15, RBP,
                                            (sequence "XMM%u", 0, 15))>;
 
-def CSR_AllRegs_64 : CalleeSavedRegs<(add CSR_MostRegs_64, RAX, RSP,
-                                      (sequence "XMM%u", 16, 31),
-                                      (sequence "YMM%u", 0, 31),
-                                      (sequence "ZMM%u", 0, 31))>;
+def CSR_64_AllRegs     : CalleeSavedRegs<(add CSR_64_MostRegs, RAX, RSP,
+                                              (sequence "XMM%u", 16, 31))>;
+def CSR_64_AllRegs_AVX : CalleeSavedRegs<(sub (add CSR_64_MostRegs, RAX, RSP,
+                                                   (sequence "YMM%u", 0, 31)),
+                                              (sequence "XMM%u", 0, 15))>;
 
 // Standard C + YMM6-15
 def CSR_Win64_Intel_OCL_BI_AVX : CalleeSavedRegs<(add RBX, RBP, RDI, RSI, R12,
diff --git a/lib/Target/X86/X86RegisterInfo.cpp b/lib/Target/X86/X86RegisterInfo.cpp
index bdaf4ef58f..220dc433fe 100644
--- a/lib/Target/X86/X86RegisterInfo.cpp
+++ b/lib/Target/X86/X86RegisterInfo.cpp
@@ -234,17 +234,18 @@ X86RegisterInfo::getRegPressureLimit(const TargetRegisterClass *RC,
 
 const uint16_t *
 X86RegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
+  bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
+  bool HasAVX512 = TM.getSubtarget<X86Subtarget>().hasAVX512();
+
   switch (MF->getFunction()->getCallingConv()) {
   case CallingConv::GHC:
   case CallingConv::HiPE:
     return CSR_NoRegs_SaveList;
-
   case CallingConv::AnyReg:
-    return CSR_AllRegs_64_SaveList;
-
+    if (HasAVX)
+      return CSR_64_AllRegs_AVX_SaveList;
+    return CSR_64_AllRegs_SaveList;
   case CallingConv::Intel_OCL_BI: {
-    bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
-    bool HasAVX512 = TM.getSubtarget<X86Subtarget>().hasAVX512();
     if (HasAVX512 && IsWin64)
       return CSR_Win64_Intel_OCL_BI_AVX512_SaveList;
     if (HasAVX512 && Is64Bit)
@@ -257,12 +258,10 @@ X86RegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
       return CSR_64_Intel_OCL_BI_SaveList;
     break;
   }
-
   case CallingConv::Cold:
     if (Is64Bit)
-      return CSR_MostRegs_64_SaveList;
+      return CSR_64_MostRegs_SaveList;
     break;
-
   default:
     break;
   }
@@ -285,7 +284,15 @@ X86RegisterInfo::getCallPreservedMask(CallingConv::ID CC) const {
   bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
   bool HasAVX512 = TM.getSubtarget<X86Subtarget>().hasAVX512();
 
-  if (CC == CallingConv::Intel_OCL_BI) {
+  switch (CC) {
+  case CallingConv::GHC:
+  case CallingConv::HiPE:
+    return CSR_NoRegs_RegMask;
+  case CallingConv::AnyReg:
+    if (HasAVX)
+      return CSR_64_AllRegs_AVX_RegMask;
+    return CSR_64_AllRegs_RegMask;
+  case CallingConv::Intel_OCL_BI: {
     if (IsWin64 && HasAVX512)
       return CSR_Win64_Intel_OCL_BI_AVX512_RegMask;
     if (Is64Bit && HasAVX512)
@@ -297,17 +304,20 @@ X86RegisterInfo::getCallPreservedMask(CallingConv::ID CC) const {
     if (!HasAVX && !IsWin64 && Is64Bit)
       return CSR_64_Intel_OCL_BI_RegMask;
   }
-  if (CC == CallingConv::GHC || CC == CallingConv::HiPE)
-    return CSR_NoRegs_RegMask;
-  if (CC == CallingConv::AnyReg)
-    return CSR_AllRegs_64_RegMask;
-  if (!Is64Bit)
-    return CSR_32_RegMask;
-  if (CC == CallingConv::Cold)
-    return CSR_MostRegs_64_RegMask;
-  if (IsWin64)
-    return CSR_Win64_RegMask;
-  return CSR_64_RegMask;
+  case CallingConv::Cold:
+    if (Is64Bit)
+      return CSR_64_MostRegs_RegMask;
+    break;
+  default:
+    break;
+  }
+
+  if (Is64Bit) {
+    if (IsWin64)
+      return CSR_Win64_RegMask;
+    return CSR_64_RegMask;
+  }
+  return CSR_32_RegMask;
 }
 
 const uint32_t*
diff --git a/test/CodeGen/X86/anyregcc.ll b/test/CodeGen/X86/anyregcc.ll
index 0677023174..a00b9c433f 100644
--- a/test/CodeGen/X86/anyregcc.ll
+++ b/test/CodeGen/X86/anyregcc.ll
@@ -1,4 +1,7 @@
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -disable-fp-elim | FileCheck %s
+; RUN: llc < %s -march=x86-64 -mcpu=corei7     | FileCheck --check-prefix=SSE %s
+; RUN: llc < %s -march=x86-64 -mcpu=corei7-avx | FileCheck --check-prefix=AVX %s
+
 
 ; Stackmap Header: no constants - 6 callsites
 ; CHECK-LABEL: .section	__LLVM_STACKMAPS,__llvm_stackmaps
@@ -336,5 +339,105 @@ entry:
   ret i64 %result
 }
 
+; Make sure all regs are spilled
+define anyregcc void @anyregcc1() {
+entry:
+;SSE-LABEL: anyregcc1
+;SSE:      pushq %rax
+;SSE:      pushq %rbp
+;SSE:      pushq %r15
+;SSE:      pushq %r14
+;SSE:      pushq %r13
+;SSE:      pushq %r12
+;SSE:      pushq %r11
+;SSE:      pushq %r10
+;SSE:      pushq %r9
+;SSE:      pushq %r8
+;SSE:      pushq %rdi
+;SSE:      pushq %rsi
+;SSE:      pushq %rdx
+;SSE:      pushq %rcx
+;SSE:      pushq %rbx
+;SSE:      movaps %xmm15
+;SSE-NEXT: movaps %xmm14
+;SSE-NEXT: movaps %xmm13
+;SSE-NEXT: movaps %xmm12
+;SSE-NEXT: movaps %xmm11
+;SSE-NEXT: movaps %xmm10
+;SSE-NEXT: movaps %xmm9
+;SSE-NEXT: movaps %xmm8
+;SSE-NEXT: movaps %xmm7
+;SSE-NEXT: movaps %xmm6
+;SSE-NEXT: movaps %xmm5
+;SSE-NEXT: movaps %xmm4
+;SSE-NEXT: movaps %xmm3
+;SSE-NEXT: movaps %xmm2
+;SSE-NEXT: movaps %xmm1
+;SSE-NEXT: movaps %xmm0
+;AVX-LABEL:anyregcc1
+;AVX:      pushq %rax
+;AVX:      pushq %rbp
+;AVX:      pushq %r15
+;AVX:      pushq %r14
+;AVX:      pushq %r13
+;AVX:      pushq %r12
+;AVX:      pushq %r11
+;AVX:      pushq %r10
+;AVX:      pushq %r9
+;AVX:      pushq %r8
+;AVX:      pushq %rdi
+;AVX:      pushq %rsi
+;AVX:      pushq %rdx
+;AVX:      pushq %rcx
+;AVX:      pushq %rbx
+;AVX:      vmovups %ymm15
+;AVX-NEXT: vmovups %ymm14
+;AVX-NEXT: vmovups %ymm13
+;AVX-NEXT: vmovups %ymm12
+;AVX-NEXT: vmovups %ymm11
+;AVX-NEXT: vmovups %ymm10
+;AVX-NEXT: vmovups %ymm9
+;AVX-NEXT: vmovups %ymm8
+;AVX-NEXT: vmovups %ymm7
+;AVX-NEXT: vmovups %ymm6
+;AVX-NEXT: vmovups %ymm5
+;AVX-NEXT: vmovups %ymm4
+;AVX-NEXT: vmovups %ymm3
+;AVX-NEXT: vmovups %ymm2
+;AVX-NEXT: vmovups %ymm1
+;AVX-NEXT: vmovups %ymm0
+  call void asm sideeffect "", "~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15},~{rbp},~{xmm0},~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15}"()
+  ret void
+}
+
+; Make sure we don't spill any XMMs/YMMs
+declare anyregcc void @foo()
+define void @anyregcc2() {
+entry:
+;SSE-LABEL: anyregcc2
+;SSE-NOT: movaps %xmm
+;AVX-LABEL: anyregcc2
+;AVX-NOT: vmovups %ymm
+  %a0 = call <2 x double> asm sideeffect "", "={xmm0}"() nounwind
+  %a1 = call <2 x double> asm sideeffect "", "={xmm1}"() nounwind
+  %a2 = call <2 x double> asm sideeffect "", "={xmm2}"() nounwind
+  %a3 = call <2 x double> asm sideeffect "", "={xmm3}"() nounwind
+  %a4 = call <2 x double> asm sideeffect "", "={xmm4}"() nounwind
+  %a5 = call <2 x double> asm sideeffect "", "={xmm5}"() nounwind
+  %a6 = call <2 x double> asm sideeffect "", "={xmm6}"() nounwind
+  %a7 = call <2 x double> asm sideeffect "", "={xmm7}"() nounwind
+  %a8 = call <2 x double> asm sideeffect "", "={xmm8}"() nounwind
+  %a9 = call <2 x double> asm sideeffect "", "={xmm9}"() nounwind
+  %a10 = call <2 x double> asm sideeffect "", "={xmm10}"() nounwind
+  %a11 = call <2 x double> asm sideeffect "", "={xmm11}"() nounwind
+  %a12 = call <2 x double> asm sideeffect "", "={xmm12}"() nounwind
+  %a13 = call <2 x double> asm sideeffect "", "={xmm13}"() nounwind
+  %a14 = call <2 x double> asm sideeffect "", "={xmm14}"() nounwind
+  %a15 = call <2 x double> asm sideeffect "", "={xmm15}"() nounwind
+  call anyregcc void @foo()
+  call void asm sideeffect "", "{xmm0},{xmm1},{xmm2},{xmm3},{xmm4},{xmm5},{xmm6},{xmm7},{xmm8},{xmm9},{xmm10},{xmm11},{xmm12},{xmm13},{xmm14},{xmm15}"(<2 x double> %a0, <2 x double> %a1, <2 x double> %a2, <2 x double> %a3, <2 x double> %a4, <2 x double> %a5, <2 x double> %a6, <2 x double> %a7, <2 x double> %a8, <2 x double> %a9, <2 x double> %a10, <2 x double> %a11, <2 x double> %a12, <2 x double> %a13, <2 x double> %a14, <2 x double> %a15)
+  ret void
+}
+
 declare void @llvm.experimental.patchpoint.void(i64, i32, i8*, i32, ...)
 declare i64 @llvm.experimental.patchpoint.i64(i64, i32, i8*, i32, ...)