def RetCC_X86Common : CallingConv<[
// Scalar values are returned in AX first, then DX.
CCIfType<[i8] , CCAssignToReg<[AL]>>,
- CCIfType<[i16], CCAssignToReg<[AX]>>,
+ CCIfType<[i16], CCAssignToReg<[AX, DX]>>,
CCIfType<[i32], CCAssignToReg<[EAX, EDX]>>,
CCIfType<[i64], CCAssignToReg<[RAX, RDX]>>,
- // Vector types are returned in XMM0 and XMM1, when they fit. If the target
- // doesn't have XMM registers, it won't have vector types.
+ // Vector types are returned in XMM0 and XMM1, when they fit. XMMM2 and XMM3
+ // can only be used by ABI non-compliant code. If the target doesn't have XMM
+ // registers, it won't have vector types.
CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
- CCAssignToReg<[XMM0,XMM1]>>,
+ CCAssignToReg<[XMM0,XMM1,XMM2,XMM3]>>,
// MMX vector types are always returned in MM0. If the target doesn't have
// MM0, it doesn't support these vector types.
- CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToReg<[MM0]>>,
+ CCIfType<[v8i8, v4i16, v2i32, v1i64, v2f32], CCAssignToReg<[MM0]>>,
// Long double types are always returned in ST0 (even with SSE).
- CCIfType<[f80], CCAssignToReg<[ST0]>>
+ CCIfType<[f80], CCAssignToReg<[ST0, ST1]>>
]>;
// X86-32 C return-value convention.
def RetCC_X86_32_C : CallingConv<[
- // The X86-32 calling convention returns FP values in ST0, otherwise it is the
- // same as the common X86 calling conv.
- CCIfType<[f32], CCAssignToReg<[ST0]>>,
- CCIfType<[f64], CCAssignToReg<[ST0]>>,
+ // The X86-32 calling convention returns FP values in ST0, unless marked
+ // with "inreg" (used here to distinguish one kind of reg from another,
+ // weirdly; this is really the sse-regparm calling convention) in which
+ // case they use XMM0, otherwise it is the same as the common X86 calling
+ // conv.
+ CCIfInReg<CCIfSubtarget<"hasSSE2()",
+ CCIfType<[f32, f64], CCAssignToReg<[XMM0,XMM1,XMM2]>>>>,
+ CCIfType<[f32,f64], CCAssignToReg<[ST0, ST1]>>,
CCDelegateTo<RetCC_X86Common>
]>;
// X86-64 C return-value convention.
def RetCC_X86_64_C : CallingConv<[
// The X86-64 calling convention always returns FP values in XMM0.
- CCIfType<[f32], CCAssignToReg<[XMM0]>>,
- CCIfType<[f64], CCAssignToReg<[XMM0]>>,
+ CCIfType<[f32], CCAssignToReg<[XMM0, XMM1]>>,
+ CCIfType<[f64], CCAssignToReg<[XMM0, XMM1]>>,
+
+ // MMX vector types are always returned in XMM0.
+ CCIfType<[v8i8, v4i16, v2i32, v1i64, v2f32], CCAssignToReg<[XMM0, XMM1]>>,
CCDelegateTo<RetCC_X86Common>
]>;
+// X86-Win64 C return-value convention.
+def RetCC_X86_Win64_C : CallingConv<[
+ // The X86-Win64 calling convention always returns __m64 values in RAX.
+ CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToReg<[RAX]>>,
+
+ // And FP in XMM0 only.
+ CCIfType<[f32], CCAssignToReg<[XMM0]>>,
+ CCIfType<[f64], CCAssignToReg<[XMM0]>>,
+
+ // Otherwise, everything is the same as 'normal' X86-64 C CC.
+ CCDelegateTo<RetCC_X86_64_C>
+]>;
// This is the root return-value convention for the X86-32 backend.
// This is the root return-value convention for the X86-64 backend.
def RetCC_X86_64 : CallingConv<[
- // Always just the same as C calling conv for X86-64.
+ // Mingw64 and native Win64 use Win64 CC
+ CCIfSubtarget<"isTargetWin64()", CCDelegateTo<RetCC_X86_Win64_C>>,
+
+ // Otherwise, drop to normal X86-64 CC
CCDelegateTo<RetCC_X86_64_C>
]>;
//===----------------------------------------------------------------------===//
def CC_X86_64_C : CallingConv<[
+ // Handles byval parameters.
+ CCIfByVal<CCPassByVal<8, 8>>,
+
// Promote i8/i16 arguments to i32.
CCIfType<[i8, i16], CCPromoteToType<i32>>,
-
- CCIfStruct<CCStructAssign<[RDI, RSI, RDX, RCX, R8, R9 ]>>,
+
+ // The 'nest' parameter, if any, is passed in R10.
+ CCIfNest<CCAssignToReg<[R10]>>,
// The first 6 integer arguments are passed in integer registers.
CCIfType<[i32], CCAssignToReg<[EDI, ESI, EDX, ECX, R8D, R9D]>>,
// The first 8 FP/Vector arguments are passed in XMM registers.
CCIfType<[f32, f64, v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
- CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>,
+ CCIfSubtarget<"hasSSE1()",
+ CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>>,
+
+ // The first 8 MMX (except for v1i64) vector arguments are passed in XMM
+ // registers on Darwin.
+ CCIfType<[v8i8, v4i16, v2i32, v2f32],
+ CCIfSubtarget<"isTargetDarwin()",
+ CCIfSubtarget<"hasSSE2()",
+ CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>>>,
+
+ // The first 8 v1i64 vector arguments are passed in GPRs on Darwin.
+ CCIfType<[v1i64],
+ CCIfSubtarget<"isTargetDarwin()",
+ CCAssignToReg<[RDI, RSI, RDX, RCX, R8]>>>,
+
+ // Integer/FP values get stored in stack slots that are 8 bytes in size and
+ // 8-byte aligned if there are no more registers to hold them.
+ CCIfType<[i32, i64, f32, f64], CCAssignToStack<8, 8>>,
+
+ // Long doubles get stack slots whose size and alignment depends on the
+ // subtarget.
+ CCIfType<[f80], CCAssignToStack<0, 0>>,
- // The first 8 MMX vector arguments are passed in GPRs.
- CCIfType<[v8i8, v4i16, v2i32, v1i64],
- CCAssignToReg<[RDI, RSI, RDX, RCX, R8 , R9 ]>>,
+ // Vectors get 16-byte stack slots that are 16-byte aligned.
+ CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64], CCAssignToStack<16, 16>>,
+
+ // __m64 vectors get 8-byte stack slots that are 8-byte aligned.
+ CCIfType<[v8i8, v4i16, v2i32, v1i64, v2f32], CCAssignToStack<8, 8>>
+]>;
+
+// Calling convention used on Win64
+def CC_X86_Win64_C : CallingConv<[
+ // FIXME: Handle byval stuff.
+ // FIXME: Handle varargs.
+
+ // Promote i8/i16 arguments to i32.
+ CCIfType<[i8, i16], CCPromoteToType<i32>>,
// The 'nest' parameter, if any, is passed in R10.
CCIfNest<CCAssignToReg<[R10]>>,
+ // The first 4 integer arguments are passed in integer registers.
+ CCIfType<[i32], CCAssignToRegWithShadow<[ECX , EDX , R8D , R9D ],
+ [XMM0, XMM1, XMM2, XMM3]>>,
+ CCIfType<[i64], CCAssignToRegWithShadow<[RCX , RDX , R8 , R9 ],
+ [XMM0, XMM1, XMM2, XMM3]>>,
+
+ // The first 4 FP/Vector arguments are passed in XMM registers.
+ CCIfType<[f32, f64, v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
+ CCAssignToRegWithShadow<[XMM0, XMM1, XMM2, XMM3],
+ [RCX , RDX , R8 , R9 ]>>,
+
+ // The first 4 MMX vector arguments are passed in GPRs.
+ CCIfType<[v8i8, v4i16, v2i32, v1i64, v2f32],
+ CCAssignToRegWithShadow<[RCX , RDX , R8 , R9 ],
+ [XMM0, XMM1, XMM2, XMM3]>>,
+
// Integer/FP values get stored in stack slots that are 8 bytes in size and
- // 8-byte aligned if there are no more registers to hold them.
- CCIfType<[i32, i64, f32, f64], CCAssignToStack<8, 8>>,
-
+ // 16-byte aligned if there are no more registers to hold them.
+ CCIfType<[i32, i64, f32, f64], CCAssignToStack<8, 16>>,
+
// Long doubles get stack slots whose size and alignment depends on the
// subtarget.
CCIfType<[f80], CCAssignToStack<0, 0>>,
// Vectors get 16-byte stack slots that are 16-byte aligned.
CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64], CCAssignToStack<16, 16>>,
- // __m64 vectors get 8-byte stack slots that are 8-byte aligned.
- CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToStack<8, 8>>
+ // __m64 vectors get 8-byte stack slots that are 16-byte aligned.
+ CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToStack<8, 16>>
]>;
// Tail call convention (fast): One register is reserved for target address,
// namely R9
def CC_X86_64_TailCall : CallingConv<[
+ // Handles byval parameters.
+ CCIfByVal<CCPassByVal<8, 8>>,
+
// Promote i8/i16 arguments to i32.
CCIfType<[i8, i16], CCPromoteToType<i32>>,
-
- CCIfStruct<CCStructAssign<[RDI, RSI, RDX, RCX, R8]>>,
+
+ // The 'nest' parameter, if any, is passed in R10.
+ CCIfNest<CCAssignToReg<[R10]>>,
// The first 6 integer arguments are passed in integer registers.
CCIfType<[i32], CCAssignToReg<[EDI, ESI, EDX, ECX, R8D]>>,
// The first 8 FP/Vector arguments are passed in XMM registers.
CCIfType<[f32, f64, v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
- CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>,
-
- // The first 8 MMX vector arguments are passed in GPRs.
- CCIfType<[v8i8, v4i16, v2i32, v1i64],
- CCAssignToReg<[RDI, RSI, RDX, RCX, R8]>>,
-
- // The 'nest' parameter, if any, is passed in R10.
- CCIfNest<CCAssignToReg<[R10]>>,
-
+ CCIfSubtarget<"hasSSE1()",
+ CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>>,
+
+ // The first 8 MMX (except for v1i64) vector arguments are passed in XMM
+ // registers on Darwin.
+ CCIfType<[v8i8, v4i16, v2i32, v2f32],
+ CCIfSubtarget<"isTargetDarwin()",
+ CCAssignToReg<[XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7]>>>,
+
+ // The first 8 v1i64 vector arguments are passed in GPRs on Darwin.
+ CCIfType<[v1i64],
+ CCIfSubtarget<"isTargetDarwin()",
+ CCAssignToReg<[RDI, RSI, RDX, RCX, R8]>>>,
+
// Integer/FP values get stored in stack slots that are 8 bytes in size and
// 8-byte aligned if there are no more registers to hold them.
CCIfType<[i32, i64, f32, f64], CCAssignToStack<8, 8>>,
/// values are spilled on the stack, and the first 4 vector values go in XMM
/// regs.
def CC_X86_32_Common : CallingConv<[
+ // Handles byval parameters.
+ CCIfByVal<CCPassByVal<4, 4>>,
+
+ // The first 3 float or double arguments, if marked 'inreg' and if the call
+ // is not a vararg call and if SSE2 is available, are passed in SSE registers.
+ CCIfNotVarArg<CCIfInReg<CCIfType<[f32,f64],
+ CCIfSubtarget<"hasSSE2()",
+ CCAssignToReg<[XMM0,XMM1,XMM2]>>>>>,
+
+ // The first 3 __m64 (except for v1i64) vector arguments are passed in mmx
+ // registers if the call is not a vararg call.
+ CCIfNotVarArg<CCIfType<[v8i8, v4i16, v2i32, v2f32],
+ CCAssignToReg<[MM0, MM1, MM2]>>>,
+
// Integer/Float values get stored in stack slots that are 4 bytes in
// size and 4-byte aligned.
CCIfType<[i32, f32], CCAssignToStack<4, 4>>,
// Long doubles get slots whose size depends on the subtarget.
CCIfType<[f80], CCAssignToStack<0, 4>>,
- // The first 4 vector arguments are passed in XMM registers.
- CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
- CCAssignToReg<[XMM0, XMM1, XMM2, XMM3]>>,
+ // The first 4 SSE vector arguments are passed in XMM registers.
+ CCIfNotVarArg<CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],
+ CCAssignToReg<[XMM0, XMM1, XMM2, XMM3]>>>,
- // Other vectors get 16-byte stack slots that are 16-byte aligned.
+ // Other SSE vectors get 16-byte stack slots that are 16-byte aligned.
CCIfType<[v16i8, v8i16, v4i32, v2i64, v4f32, v2f64], CCAssignToStack<16, 16>>,
- // __m64 vectors get 8-byte stack slots that are 8-byte aligned. They are
+ // __m64 vectors get 8-byte stack slots that are 4-byte aligned. They are
// passed in the parameter area.
- CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToStack<8, 8>>
-]>;
+ CCIfType<[v8i8, v4i16, v2i32, v1i64], CCAssignToStack<8, 4>>]>;
def CC_X86_32_C : CallingConv<[
// Promote i8/i16 arguments to i32.
CCDelegateTo<CC_X86_32_Common>
]>;
-/// Same as C calling convention except for non-free ECX which is used for storing
-/// a potential pointer to the tail called function.
-def CC_X86_32_TailCall : CallingConv<[
+def CC_X86_32_FastCall : CallingConv<[
// Promote i8/i16 arguments to i32.
CCIfType<[i8, i16], CCPromoteToType<i32>>,
- // Nested function trampolines are currently not supported by fastcc.
-
- // The first 3 integer arguments, if marked 'inreg' and if the call is not
- // a vararg call, are passed in integer registers.
- CCIfNotVarArg<CCIfInReg<CCIfType<[i32], CCAssignToReg<[EAX, EDX]>>>>,
+ // The 'nest' parameter, if any, is passed in EAX.
+ CCIfNest<CCAssignToReg<[EAX]>>,
+
+ // The first 2 integer arguments are passed in ECX/EDX
+ CCIfType<[i32], CCAssignToReg<[ECX, EDX]>>,
// Otherwise, same as everything else.
CCDelegateTo<CC_X86_32_Common>
]>;
-def CC_X86_32_FastCall : CallingConv<[
+def CC_X86_32_FastCC : CallingConv<[
+ // Handles byval parameters. Note that we can't rely on the delegation
+ // to CC_X86_32_Common for this because that happens after code that
+ // puts arguments in registers.
+ CCIfByVal<CCPassByVal<4, 4>>,
+
// Promote i8/i16 arguments to i32.
CCIfType<[i8, i16], CCPromoteToType<i32>>,
// The first 2 integer arguments are passed in ECX/EDX
CCIfType<[i32], CCAssignToReg<[ECX, EDX]>>,
+ // The first 3 float or double arguments, if the call is not a vararg
+ // call and if SSE2 is available, are passed in SSE registers.
+ CCIfNotVarArg<CCIfType<[f32,f64],
+ CCIfSubtarget<"hasSSE2()",
+ CCAssignToReg<[XMM0,XMM1,XMM2]>>>>,
+
+ // Doubles get 8-byte slots that are 8-byte aligned.
+ CCIfType<[f64], CCAssignToStack<8, 8>>,
+
// Otherwise, same as everything else.
CCDelegateTo<CC_X86_32_Common>
]>;