defm RCP : sse1_fp_unop_s<0x53, "rcp", X86frcp, int_x86_sse_rcp_ss>,
sse1_fp_unop_p<0x53, "rcp", X86frcp, int_x86_sse_rcp_ps>;
+// There is no f64 version of the reciprocal approximation instructions.
+
//===----------------------------------------------------------------------===//
// SSE 1 & 2 - Non-temporal stores
//===----------------------------------------------------------------------===//
TB, Requires<[HasSSE2]>;
//===----------------------------------------------------------------------===//
-// SSE 1 & 2 - Misc Instructions
+// SSE 1 & 2 - Misc Instructions (No AVX form)
//===----------------------------------------------------------------------===//
// Prefetch intrinsic.
def SFENCE : I<0xAE, MRM_F8, (outs), (ins), "sfence", [(int_x86_sse_sfence)]>,
TB, Requires<[HasSSE1]>;
-// MXCSR register
-def LDMXCSR : PSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
- "ldmxcsr\t$src", [(int_x86_sse_ldmxcsr addr:$src)]>;
-def STMXCSR : PSI<0xAE, MRM3m, (outs), (ins i32mem:$dst),
- "stmxcsr\t$dst", [(int_x86_sse_stmxcsr addr:$dst)]>;
-
// Alias instructions that map zero vector to pxor / xorp* for sse.
// We set canFoldAsLoad because this can be converted to a constant-pool
// load of an all-zeros value if folding it would be beneficial.
def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))),
(f32 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss))>;
+//===----------------------------------------------------------------------===//
+// SSE 1 & 2 - Load/Store XCSR register
+//===----------------------------------------------------------------------===//
+
+let isAsmParserOnly = 1 in {
+ def VLDMXCSR : VPSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
+ "ldmxcsr\t$src", [(int_x86_sse_ldmxcsr addr:$src)]>, VEX;
+ def VSTMXCSR : VPSI<0xAE, MRM3m, (outs), (ins i32mem:$dst),
+ "stmxcsr\t$dst", [(int_x86_sse_stmxcsr addr:$dst)]>, VEX;
+}
+
+def LDMXCSR : PSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
+ "ldmxcsr\t$src", [(int_x86_sse_ldmxcsr addr:$src)]>;
+def STMXCSR : PSI<0xAE, MRM3m, (outs), (ins i32mem:$dst),
+ "stmxcsr\t$dst", [(int_x86_sse_stmxcsr addr:$dst)]>;
+
//===---------------------------------------------------------------------===//
// SSE2 Instructions
//===---------------------------------------------------------------------===//
-
-// There is no f64 version of the reciprocal approximation instructions.
-
//===---------------------------------------------------------------------===//
// SSE integer instructions
let ExeDomain = SSEPackedInt in {
unsigned NumOps = MI.getNumOperands();
unsigned CurOp = 0;
- if ((TSFlags & X86II::FormMask) == X86II::MRMDestMem)
- NumOps = CurOp = X86AddrNumOperands;
-
switch (TSFlags & X86II::FormMask) {
case X86II::MRMInitReg: assert(0 && "FIXME: Remove this!");
- case X86II::MRMSrcMem:
+ case X86II::MRM0m: case X86II::MRM1m:
+ case X86II::MRM2m: case X86II::MRM3m:
+ case X86II::MRM4m: case X86II::MRM5m:
+ case X86II::MRM6m: case X86II::MRM7m:
case X86II::MRMDestMem:
+ NumOps = CurOp = X86AddrNumOperands;
+ case X86II::MRMSrcMem:
case X86II::MRMSrcReg:
- if (MI.getOperand(CurOp).isReg() &&
+ if (MI.getNumOperands() > CurOp && MI.getOperand(CurOp).isReg() &&
X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(CurOp).getReg()))
VEX_R = 0x0;
// CHECK: encoding: [0xc5,0xf8,0x2b,0x08]
vmovntps %xmm1, (%eax)
+// CHECK: vldmxcsr (%eax)
+// CHECK: encoding: [0xc5,0xf8,0xae,0x10]
+ vldmxcsr (%eax)
+
+// CHECK: vstmxcsr (%eax)
+// CHECK: encoding: [0xc5,0xf8,0xae,0x18]
+ vstmxcsr (%eax)
+
+// CHECK: vldmxcsr 3735928559
+// CHECK: encoding: [0xc5,0xf8,0xae,0x15,0xef,0xbe,0xad,0xde]
+ vldmxcsr 0xdeadbeef
+
+// CHECK: vstmxcsr 3735928559
+// CHECK: encoding: [0xc5,0xf8,0xae,0x1d,0xef,0xbe,0xad,0xde]
+ vstmxcsr 0xdeadbeef
+