[X86] Add FPCW as a register and start using it as an implicit use on floating point instructions.

[craig.topper]

Summary

FPCW contains the rounding mode control which we manipulate to implement fp to integer conversion by changing the roudning mode, storing the value to the stack, and then changing the rounding mode back. Because we didn't model FPCW and its dependency chain, other instructions could be scheduled into the middle of the sequence.

This patch introduces the register and adds it as an implciit def of FLDCW and implicit use of the FP binary arithmetic instructions and store instructions. There are more instructions that need to be updated, but this is a good start. I believe this fixes at least the reduced test case from PR40529.

[dim]

For what it's worth, this builds and tests correctly for me, and also appears to solve the test cases from bug 40206.

lib/Target/X86/X86RegisterInfo.td

290

It looks like you recently renamed this to fpsr, so this must be rebased.

[craig.topper]

Rebase

[craig.topper]

Really rebase

[RKSimon]

Is it possible to get PR40206 + PR40529 test coverage?

[craig.topper]

Well I added the test for pr40529.ll already but I guess I didn't really reproduce the bug when doing so. Let me try again to see if I can find the right CPU name and triple to get it to really be wrong.

[craig.topper]

Rebase after fixing the test case for pr40529

[craig.topper]

test/CodeGen/X86/pr34080.ll

73

This looks like we were still miscompiling this test despite this bug being closed

[dim]

test/CodeGen/X86/pr34080.ll

73

Hm yes, in rL321424 @RKSimon added a pr34080-2.ll test case, but didn't touch this one, for some reason.

[craig.topper]

Add FPCW to more instructions.

[RKSimon]

LGTM but a PR40206 test case as well would be useful

test/CodeGen/X86/pr34080.ll

73

I think it shows the limits of the fix I put in - this approach seems a lot more thorough

[craig.topper]

In D57735#1388651, @RKSimon wrote:

LGTM but a PR40206 test case as well would be useful

Both PR40529 and PR40206 are caused by a multiply instruction being moved between the fldcw. They aren't really different. The case I put in for pr40529 is the same reduced test case that Dimitry put in both bugs.

lib/Target/X86/X86InstrFPStack.td

def _FpI32m80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, i32mem:$src2),
                            (OpNode (X86fild addr:$src2, i32), RFP80:$src1)))]>;
 def _FI16m  : FPI<0xDE, fp, (outs), (ins i16mem:$src),
                   !strconcat("fi", asmstring, "{s}\t$src")>;
 def _FI32m  : FPI<0xDA, fp, (outs), (ins i32mem:$src),
                   !strconcat("fi", asmstring, "{l}\t$src")>;
 } // mayLoad = 1, hasSideEffects = 1
 }
 
-let Defs = [FPSW] in {
+let Defs = [FPSW], Uses = [FPCW] in {
 // FPBinary_rr just defines pseudo-instructions, no need to set a scheduling
 // resources.
 let hasNoSchedulingInfo = 1 in {
 defm ADD : FPBinary_rr<fadd>;
 defm SUB : FPBinary_rr<fsub>;
 defm MUL : FPBinary_rr<fmul>;
 defm DIV : FPBinary_rr<fdiv>;
 }

 class FPrST0Inst<Format fp, string asm>
   : FPI<0xDC, fp, (outs), (ins RSTi:$op), asm>;
 class FPrST0PInst<Format fp, string asm>
   : FPI<0xDE, fp, (outs), (ins RSTi:$op), asm>;
 
 // NOTE: GAS and apparently all other AT&T style assemblers have a broken notion
 // of some of the 'reverse' forms of the fsub and fdiv instructions.  As such,
 // we have to put some 'r's in and take them out of weird places.
-let SchedRW = [WriteFAdd], Defs = [FPSW] in {
+let SchedRW = [WriteFAdd], Defs = [FPSW], Uses = [FPCW] in {
 def ADD_FST0r   : FPST0rInst <MRM0r, "fadd\t{$op, %st|st, $op}">;
 def ADD_FrST0   : FPrST0Inst <MRM0r, "fadd\t{%st, $op|$op, st}">;
 def ADD_FPrST0  : FPrST0PInst<MRM0r, "faddp\t{%st, $op|$op, st}">;
 def SUBR_FST0r  : FPST0rInst <MRM5r, "fsubr\t{$op, %st|st, $op}">;
 def SUB_FrST0   : FPrST0Inst <MRM5r, "fsub{r}\t{%st, $op|$op, st}">;
 def SUB_FPrST0  : FPrST0PInst<MRM5r, "fsub{r}p\t{%st, $op|$op, st}">;
 def SUB_FST0r   : FPST0rInst <MRM4r, "fsub\t{$op, %st|st, $op}">;
 def SUBR_FrST0  : FPrST0Inst <MRM4r, "fsub{|r}\t{%st, $op|$op, st}">;
 def SUBR_FPrST0 : FPrST0PInst<MRM4r, "fsub{|r}p\t{%st, $op|$op, st}">;
 } // SchedRW
-let SchedRW = [WriteFCom], Defs = [FPSW] in {
+let SchedRW = [WriteFCom], Defs = [FPSW], Uses = [FPCW] in {
 def COM_FST0r   : FPST0rInst <MRM2r, "fcom\t$op">;
 def COMP_FST0r  : FPST0rInst <MRM3r, "fcomp\t$op">;
 } // SchedRW
-let SchedRW = [WriteFMul], Defs = [FPSW] in {
+let SchedRW = [WriteFMul], Defs = [FPSW], Uses = [FPCW] in {
 def MUL_FST0r   : FPST0rInst <MRM1r, "fmul\t{$op, %st|st, $op}">;
 def MUL_FrST0   : FPrST0Inst <MRM1r, "fmul\t{%st, $op|$op, st}">;
 def MUL_FPrST0  : FPrST0PInst<MRM1r, "fmulp\t{%st, $op|$op, st}">;
 } // SchedRW
-let SchedRW = [WriteFDiv], Defs = [FPSW] in {
+let SchedRW = [WriteFDiv], Defs = [FPSW], Uses = [FPCW] in {
 def DIVR_FST0r  : FPST0rInst <MRM7r, "fdivr\t{$op, %st|st, $op}">;
 def DIV_FrST0   : FPrST0Inst <MRM7r, "fdiv{r}\t{%st, $op|$op, st}">;
 def DIV_FPrST0  : FPrST0PInst<MRM7r, "fdiv{r}p\t{%st, $op|$op, st}">;
 def DIV_FST0r   : FPST0rInst <MRM6r, "fdiv\t{$op, %st|st, $op}">;
 def DIVR_FrST0  : FPrST0Inst <MRM6r, "fdiv{|r}\t{%st, $op|$op, st}">;
 def DIVR_FPrST0 : FPrST0PInst<MRM6r, "fdiv{|r}p\t{%st, $op|$op, st}">;
 } // SchedRW
 
 // Unary operations.
 multiclass FPUnary<SDNode OpNode, Format fp, string asmstring> {
 def _Fp32  : FpIf32<(outs RFP32:$dst), (ins RFP32:$src), OneArgFPRW,
                  [(set RFP32:$dst, (OpNode RFP32:$src))]>;
 def _Fp64  : FpIf64<(outs RFP64:$dst), (ins RFP64:$src), OneArgFPRW,
                  [(set RFP64:$dst, (OpNode RFP64:$src))]>;
 def _Fp80  : FpI_<(outs RFP80:$dst), (ins RFP80:$src), OneArgFPRW,
                  [(set RFP80:$dst, (OpNode RFP80:$src))]>;
 def _F     : FPI<0xD9, fp, (outs), (ins), asmstring>;
 }
 
-let Defs = [FPSW] in {
+let Defs = [FPSW], Uses = [FPCW] in {
 
 let SchedRW = [WriteFSign] in {
 defm CHS : FPUnary<fneg, MRM_E0, "fchs">;
 defm ABS : FPUnary<fabs, MRM_E1, "fabs">;
 }
 
 let SchedRW = [WriteFSqrt80] in
 defm SQRT: FPUnary<fsqrt,MRM_FA, "fsqrt">;

 } // hasSideEffects
 
 def TST_F  : FPI<0xD9, MRM_E4, (outs), (ins), "ftst">;
 } // SchedRW
 } // Defs = [FPSW]
 
 // Versions of FP instructions that take a single memory operand.  Added for the
 //   disassembler; remove as they are included with patterns elsewhere.
-let SchedRW = [WriteFComLd], Defs = [FPSW] in {
+let SchedRW = [WriteFComLd], Defs = [FPSW], Uses = [FPCW] in {
 def FCOM32m  : FPI<0xD8, MRM2m, (outs), (ins f32mem:$src), "fcom{s}\t$src">;
 def FCOMP32m : FPI<0xD8, MRM3m, (outs), (ins f32mem:$src), "fcomp{s}\t$src">;
 
 def FCOM64m  : FPI<0xDC, MRM2m, (outs), (ins f64mem:$src), "fcom{l}\t$src">;
 def FCOMP64m : FPI<0xDC, MRM3m, (outs), (ins f64mem:$src), "fcomp{l}\t$src">;
 
 def FICOM16m : FPI<0xDE, MRM2m, (outs), (ins i16mem:$src), "ficom{s}\t$src">;
 def FICOMP16m: FPI<0xDE, MRM3m, (outs), (ins i16mem:$src), "ficomp{s}\t$src">;

 def ILD_Fp16m80: FpI_<(outs RFP80:$dst), (ins i16mem:$src), ZeroArgFP,
                   [(set RFP80:$dst, (X86fild addr:$src, i16))]>;
 def ILD_Fp32m80: FpI_<(outs RFP80:$dst), (ins i32mem:$src), ZeroArgFP,
                   [(set RFP80:$dst, (X86fild addr:$src, i32))]>;
 def ILD_Fp64m80: FpI_<(outs RFP80:$dst), (ins i64mem:$src), ZeroArgFP,
                   [(set RFP80:$dst, (X86fild addr:$src, i64))]>;
 } // SchedRW
 
-let SchedRW = [WriteStore] in {
+let SchedRW = [WriteStore], Uses = [FPCW] in {
 def ST_Fp32m   : FpIf32<(outs), (ins f32mem:$op, RFP32:$src), OneArgFP,
                   [(store RFP32:$src, addr:$op)]>;
 def ST_Fp64m32 : FpIf64<(outs), (ins f32mem:$op, RFP64:$src), OneArgFP,
                   [(truncstoref32 RFP64:$src, addr:$op)]>;
 def ST_Fp64m   : FpIf64<(outs), (ins f64mem:$op, RFP64:$src), OneArgFP,
                   [(store RFP64:$src, addr:$op)]>;
 def ST_Fp80m32 : FpI_<(outs), (ins f32mem:$op, RFP80:$src), OneArgFP,
                   [(truncstoref32 RFP80:$src, addr:$op)]>;

 def IST_Fp64m32  : FpIf32<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP, []>;
 def IST_Fp16m64  : FpIf64<(outs), (ins i16mem:$op, RFP64:$src), OneArgFP, []>;
 def IST_Fp32m64  : FpIf64<(outs), (ins i32mem:$op, RFP64:$src), OneArgFP, []>;
 def IST_Fp64m64  : FpIf64<(outs), (ins i64mem:$op, RFP64:$src), OneArgFP, []>;
 def IST_Fp16m80  : FpI_<(outs), (ins i16mem:$op, RFP80:$src), OneArgFP, []>;
 def IST_Fp32m80  : FpI_<(outs), (ins i32mem:$op, RFP80:$src), OneArgFP, []>;
 def IST_Fp64m80  : FpI_<(outs), (ins i64mem:$op, RFP80:$src), OneArgFP, []>;
 } // mayStore
-} // SchedRW
+} // SchedRW, Uses = [FPCW]
 
 let mayLoad = 1, SchedRW = [WriteLoad] in {
 def LD_F32m   : FPI<0xD9, MRM0m, (outs), (ins f32mem:$src), "fld{s}\t$src">;
 def LD_F64m   : FPI<0xDD, MRM0m, (outs), (ins f64mem:$src), "fld{l}\t$src">;
 def LD_F80m   : FPI<0xDB, MRM5m, (outs), (ins f80mem:$src), "fld{t}\t$src">;
 def ILD_F16m  : FPI<0xDF, MRM0m, (outs), (ins i16mem:$src), "fild{s}\t$src">;
 def ILD_F32m  : FPI<0xDB, MRM0m, (outs), (ins i32mem:$src), "fild{l}\t$src">;
 def ILD_F64m  : FPI<0xDF, MRM5m, (outs), (ins i64mem:$src), "fild{ll}\t$src">;
 }
-let mayStore = 1, SchedRW = [WriteStore] in {
+let mayStore = 1, SchedRW = [WriteStore], Uses = [FPCW] in {
 def ST_F32m   : FPI<0xD9, MRM2m, (outs), (ins f32mem:$dst), "fst{s}\t$dst">;
 def ST_F64m   : FPI<0xDD, MRM2m, (outs), (ins f64mem:$dst), "fst{l}\t$dst">;
 def ST_FP32m  : FPI<0xD9, MRM3m, (outs), (ins f32mem:$dst), "fstp{s}\t$dst">;
 def ST_FP64m  : FPI<0xDD, MRM3m, (outs), (ins f64mem:$dst), "fstp{l}\t$dst">;
 def ST_FP80m  : FPI<0xDB, MRM7m, (outs), (ins f80mem:$dst), "fstp{t}\t$dst">;
 def IST_F16m  : FPI<0xDF, MRM2m, (outs), (ins i16mem:$dst), "fist{s}\t$dst">;
 def IST_F32m  : FPI<0xDB, MRM2m, (outs), (ins i32mem:$dst), "fist{l}\t$dst">;
 def IST_FP16m : FPI<0xDF, MRM3m, (outs), (ins i16mem:$dst), "fistp{s}\t$dst">;
 def IST_FP32m : FPI<0xDB, MRM3m, (outs), (ins i32mem:$dst), "fistp{l}\t$dst">;
 def IST_FP64m : FPI<0xDF, MRM7m, (outs), (ins i64mem:$dst), "fistp{ll}\t$dst">;
 }
 
 // FISTTP requires SSE3 even though it's a FPStack op.
-let Predicates = [HasSSE3], SchedRW = [WriteStore] in {
+let Predicates = [HasSSE3], SchedRW = [WriteStore], Uses = [FPCW] in {
 def ISTT_Fp16m32 : FpI_<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP,
                     [(X86fp_to_i16mem RFP32:$src, addr:$op)]>;
 def ISTT_Fp32m32 : FpI_<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP,
                     [(X86fp_to_i32mem RFP32:$src, addr:$op)]>;
 def ISTT_Fp64m32 : FpI_<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP,
                     [(X86fp_to_i64mem RFP32:$src, addr:$op)]>;
 def ISTT_Fp16m64 : FpI_<(outs), (ins i16mem:$op, RFP64:$src), OneArgFP,
                     [(X86fp_to_i16mem RFP64:$src, addr:$op)]>;
 def ISTT_Fp32m64 : FpI_<(outs), (ins i32mem:$op, RFP64:$src), OneArgFP,
                     [(X86fp_to_i32mem RFP64:$src, addr:$op)]>;
 def ISTT_Fp64m64 : FpI_<(outs), (ins i64mem:$op, RFP64:$src), OneArgFP,
                     [(X86fp_to_i64mem RFP64:$src, addr:$op)]>;
 def ISTT_Fp16m80 : FpI_<(outs), (ins i16mem:$op, RFP80:$src), OneArgFP,
                     [(X86fp_to_i16mem RFP80:$src, addr:$op)]>;
 def ISTT_Fp32m80 : FpI_<(outs), (ins i32mem:$op, RFP80:$src), OneArgFP,
                     [(X86fp_to_i32mem RFP80:$src, addr:$op)]>;
 def ISTT_Fp64m80 : FpI_<(outs), (ins i64mem:$op, RFP80:$src), OneArgFP,
                     [(X86fp_to_i64mem RFP80:$src, addr:$op)]>;
 } // Predicates = [HasSSE3]
 
-let mayStore = 1, SchedRW = [WriteStore] in {
+let mayStore = 1, SchedRW = [WriteStore], Uses = [FPCW] in {
 def ISTT_FP16m : FPI<0xDF, MRM1m, (outs), (ins i16mem:$dst), "fisttp{s}\t$dst">;
 def ISTT_FP32m : FPI<0xDB, MRM1m, (outs), (ins i32mem:$dst), "fisttp{l}\t$dst">;
 def ISTT_FP64m : FPI<0xDD, MRM1m, (outs), (ins i64mem:$dst), "fisttp{ll}\t$dst">;
 }
 
 // FP Stack manipulation instructions.
 let SchedRW = [WriteMove] in {
 def LD_Frr   : FPI<0xD9, MRM0r, (outs), (ins RSTi:$op), "fld\t$op">;

 }
 
 let SchedRW = [WriteFLD0] in
 def LD_F0 : FPI<0xD9, MRM_EE, (outs), (ins), "fldz">;
 
 let SchedRW = [WriteFLD1] in
 def LD_F1 : FPI<0xD9, MRM_E8, (outs), (ins), "fld1">;
 
-let SchedRW = [WriteFLDC], Defs = [FPSW] in {
+let SchedRW = [WriteFLDC] in {
 def FLDL2T : I<0xD9, MRM_E9, (outs), (ins), "fldl2t", []>;
 def FLDL2E : I<0xD9, MRM_EA, (outs), (ins), "fldl2e", []>;
 def FLDPI : I<0xD9, MRM_EB, (outs), (ins), "fldpi", []>;
 def FLDLG2 : I<0xD9, MRM_EC, (outs), (ins), "fldlg2", []>;
 def FLDLN2 : I<0xD9, MRM_ED, (outs), (ins), "fldln2", []>;
 } // SchedRW
 
 // Floating point compares.
-let SchedRW = [WriteFCom] in {
+let SchedRW = [WriteFCom], Uses = [FPCW] in {
 def UCOM_Fpr32 : FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
                         [(set FPSW, (trunc (X86cmp RFP32:$lhs, RFP32:$rhs)))]>;
 def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
                         [(set FPSW, (trunc (X86cmp RFP64:$lhs, RFP64:$rhs)))]>;
 def UCOM_Fpr80 : FpI_  <(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
                         [(set FPSW, (trunc (X86cmp RFP80:$lhs, RFP80:$rhs)))]>;
 } // SchedRW
 } // Defs = [FPSW]
 
 let SchedRW = [WriteFCom] in {
 // CC = ST(0) cmp ST(i)
-let Defs = [EFLAGS, FPSW] in {
-let Predicates = [FPStackf32, HasCMov] in
-def UCOM_FpIr32: FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
-                  [(set EFLAGS, (X86cmp RFP32:$lhs, RFP32:$rhs))]>;
-let Predicates = [FPStackf64, HasCMov] in
-def UCOM_FpIr64: FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
-                  [(set EFLAGS, (X86cmp RFP64:$lhs, RFP64:$rhs))]>;
-let Predicates = [HasCMov] in
+let Defs = [EFLAGS, FPSW], Uses = [FPCW] in {
+def UCOM_FpIr32: FpI_<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
+                  [(set EFLAGS, (X86cmp RFP32:$lhs, RFP32:$rhs))]>,
+                  Requires<[FPStackf32, HasCMov]>;
+def UCOM_FpIr64: FpI_<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
+                  [(set EFLAGS, (X86cmp RFP64:$lhs, RFP64:$rhs))]>,
+                  Requires<[FPStackf64, HasCMov]>;
 def UCOM_FpIr80: FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
-                  [(set EFLAGS, (X86cmp RFP80:$lhs, RFP80:$rhs))]>;
+                  [(set EFLAGS, (X86cmp RFP80:$lhs, RFP80:$rhs))]>,
+                  Requires<[HasCMov]>;
 }
 
-let Defs = [FPSW], Uses = [ST0] in {
+let Defs = [FPSW], Uses = [ST0, FPCW] in {
 def UCOM_Fr    : FPI<0xDD, MRM4r,    // FPSW = cmp ST(0) with ST(i)
                     (outs), (ins RSTi:$reg), "fucom\t$reg">;
 def UCOM_FPr   : FPI<0xDD, MRM5r,    // FPSW = cmp ST(0) with ST(i), pop
                     (outs), (ins RSTi:$reg), "fucomp\t$reg">;
 def UCOM_FPPr  : FPI<0xDA, MRM_E9,       // cmp ST(0) with ST(1), pop, pop
                     (outs), (ins), "fucompp">;
 }
 
-let Defs = [EFLAGS, FPSW], Uses = [ST0] in {
+let Defs = [EFLAGS, FPSW], Uses = [ST0, FPCW] in {
 def UCOM_FIr   : FPI<0xDB, MRM5r,     // CC = cmp ST(0) with ST(i)
                     (outs), (ins RSTi:$reg), "fucomi\t{$reg, %st|st, $reg}">;
 def UCOM_FIPr  : FPI<0xDF, MRM5r,     // CC = cmp ST(0) with ST(i), pop
                     (outs), (ins RSTi:$reg), "fucompi\t{$reg, %st|st, $reg}">;
-}
 
-let Defs = [EFLAGS, FPSW] in {
 def COM_FIr : FPI<0xDB, MRM6r, (outs), (ins RSTi:$reg),
                   "fcomi\t{$reg, %st|st, $reg}">;
 def COM_FIPr : FPI<0xDF, MRM6r, (outs), (ins RSTi:$reg),
                    "fcompi\t{$reg, %st|st, $reg}">;
 }
 } // SchedRW
 
 // Floating point flag ops.
 let SchedRW = [WriteALU] in {
 let Defs = [AX], Uses = [FPSW] in
 def FNSTSW16r : I<0xDF, MRM_E0,                  // AX = fp flags
                   (outs), (ins), "fnstsw\t{%ax|ax}",
                   [(set AX, (X86fp_stsw FPSW))]>;
-let Defs = [FPSW] in
+let Defs = [FPSW], Uses = [FPCW] in
 def FNSTCW16m : I<0xD9, MRM7m,                   // [mem16] = X87 control world
                   (outs), (ins i16mem:$dst), "fnstcw\t$dst",
                   [(X86fp_cwd_get16 addr:$dst)]>;
 } // SchedRW
-let Defs = [FPSW], mayLoad = 1 in
+let Defs = [FPSW,FPCW], mayLoad = 1 in
 def FLDCW16m  : I<0xD9, MRM5m,                   // X87 control world = [mem16]
                   (outs), (ins i16mem:$dst), "fldcw\t$dst", []>,
                 Sched<[WriteLoad]>;
 
 // FPU control instructions
 let SchedRW = [WriteMicrocoded] in {
 let Defs = [FPSW] in {
 def FNINIT : I<0xDB, MRM_E3, (outs), (ins), "fninit", []>;

lib/Target/X86/X86RegisterInfo.cpp

 const uint32_t *X86RegisterInfo::getDarwinTLSCallPreservedMask() const {
   return CSR_64_TLS_Darwin_RegMask;
 }
 
 BitVector X86RegisterInfo::getReservedRegs(const MachineFunction &MF) const {
   BitVector Reserved(getNumRegs());
   const X86FrameLowering *TFI = getFrameLowering(MF);
 
+  // Set the floating point control register as reserved.
+  Reserved.set(X86::FPCW);
+
   // Set the stack-pointer register and its aliases as reserved.
   for (MCSubRegIterator I(X86::RSP, this, /*IncludeSelf=*/true); I.isValid();
        ++I)
     Reserved.set(*I);
 
   // Set the Shadow Stack Pointer as reserved.
   Reserved.set(X86::SSP);
 

lib/Target/X86/X86RegisterInfo.td

 def ST2 : X86Reg<"st(2)", 2>, DwarfRegNum<[35, 14, 13]>;
 def ST3 : X86Reg<"st(3)", 3>, DwarfRegNum<[36, 15, 14]>;
 def ST4 : X86Reg<"st(4)", 4>, DwarfRegNum<[37, 16, 15]>;
 def ST5 : X86Reg<"st(5)", 5>, DwarfRegNum<[38, 17, 16]>;
 def ST6 : X86Reg<"st(6)", 6>, DwarfRegNum<[39, 18, 17]>;
 def ST7 : X86Reg<"st(7)", 7>, DwarfRegNum<[40, 19, 18]>;
 
 // Floating-point status word
 def FPSW : X86Reg<"fpsr", 0>;
 
+// Floating-point control word
+def FPCW : X86Reg<"fpcr", 0>;
+
 // Status flags register.
 //
 // Note that some flags that are commonly thought of as part of the status
 // flags register are modeled separately. Typically this is due to instructions
 // reading and updating those flags independently of all the others. We don't
 // want to create false dependencies between these instructions and so we use
 // a separate register to model them.
 def EFLAGS : X86Reg<"flags", 0>;

test/CodeGen/MIR/X86/memory-operands.mir

 fixedStack:
   - { id: 0, offset: 0, size: 10, alignment: 16, isImmutable: true, isAliased: false }
 body: |
   bb.0.entry:
     $rsp = frame-setup SUB64ri8 $rsp, 24, implicit-def dead $eflags
     CFI_INSTRUCTION def_cfa_offset 32
     LD_F80m $rsp, 1, $noreg, 32, $noreg, implicit-def dead $fpsw
   ; CHECK: name: stack_psv
-  ; CHECK: ST_FP80m $rsp, 1, $noreg, 0, $noreg, implicit-def dead $fpsw :: (store 10 into stack, align 16)
-    ST_FP80m $rsp, 1, _, 0, _, implicit-def dead $fpsw :: (store 10 into stack, align 16)
+  ; CHECK: ST_FP80m $rsp, 1, $noreg, 0, $noreg, implicit-def dead $fpsw, implicit $fpcw :: (store 10 into stack, align 16)
+    ST_FP80m $rsp, 1, _, 0, _, implicit-def dead $fpsw, implicit $fpcw :: (store 10 into stack, align 16)
     CALL64pcrel32 &cosl, csr_64, implicit $rsp, implicit-def $rsp, implicit-def $fp0
     $rsp = ADD64ri8 $rsp, 24, implicit-def dead $eflags
     RETQ
 ...
 ---
 name:            got_psv
 tracksRegLiveness: true
 body: |

test/CodeGen/X86/ipra-reg-usage.ll

 ; RUN: llc -enable-ipra -print-regusage -o /dev/null 2>&1 < %s | FileCheck %s
 
 target triple = "x86_64-unknown-unknown"
 declare void @bar1()
 define preserve_allcc void @foo()#0 {
-; CHECK: foo Clobbered Registers: $cs $df $ds $eflags $eip $eiz $es $fpsw $fs $gs $hip $ip $rip $riz $ss $ssp $bnd0 $bnd1 $bnd2 $bnd3 $cr0 $cr1 $cr2 $cr3 $cr4 $cr5 $cr6 $cr7 $cr8 $cr9 $cr10 $cr11 $cr12 $cr13 $cr14 $cr15 $dr0 $dr1 $dr2 $dr3 $dr4 $dr5 $dr6 $dr7 $dr8 $dr9 $dr10 $dr11 $dr12 $dr13 $dr14 $dr15 $fp0 $fp1 $fp2 $fp3 $fp4 $fp5 $fp6 $fp7 $k0 $k1 $k2 $k3 $k4 $k5 $k6 $k7 $mm0 $mm1 $mm2 $mm3 $mm4 $mm5 $mm6 $mm7 $r11 $st0 $st1 $st2 $st3 $st4 $st5 $st6 $st7 $xmm16 $xmm17 $xmm18 $xmm19 $xmm20 $xmm21 $xmm22 $xmm23 $xmm24 $xmm25 $xmm26 $xmm27 $xmm28 $xmm29 $xmm30 $xmm31 $ymm0 $ymm1 $ymm2 $ymm3 $ymm4 $ymm5 $ymm6 $ymm7 $ymm8 $ymm9 $ymm10 $ymm11 $ymm12 $ymm13 $ymm14 $ymm15 $ymm16 $ymm17 $ymm18 $ymm19 $ymm20 $ymm21 $ymm22 $ymm23 $ymm24 $ymm25 $ymm26 $ymm27 $ymm28 $ymm29 $ymm30 $ymm31 $zmm0 $zmm1 $zmm2 $zmm3 $zmm4 $zmm5 $zmm6 $zmm7 $zmm8 $zmm9 $zmm10 $zmm11 $zmm12 $zmm13 $zmm14 $zmm15 $zmm16 $zmm17 $zmm18 $zmm19 $zmm20 $zmm21 $zmm22 $zmm23 $zmm24 $zmm25 $zmm26 $zmm27 $zmm28 $zmm29 $zmm30 $zmm31 $r11b $r11bh $r11d $r11w $r11wh
+; CHECK: foo Clobbered Registers: $cs $df $ds $eflags $eip $eiz $es $fpcw $fpsw $fs $gs $hip $ip $rip $riz $ss $ssp $bnd0 $bnd1 $bnd2 $bnd3 $cr0 $cr1 $cr2 $cr3 $cr4 $cr5 $cr6 $cr7 $cr8 $cr9 $cr10 $cr11 $cr12 $cr13 $cr14 $cr15 $dr0 $dr1 $dr2 $dr3 $dr4 $dr5 $dr6 $dr7 $dr8 $dr9 $dr10 $dr11 $dr12 $dr13 $dr14 $dr15 $fp0 $fp1 $fp2 $fp3 $fp4 $fp5 $fp6 $fp7 $k0 $k1 $k2 $k3 $k4 $k5 $k6 $k7 $mm0 $mm1 $mm2 $mm3 $mm4 $mm5 $mm6 $mm7 $r11 $st0 $st1 $st2 $st3 $st4 $st5 $st6 $st7 $xmm16 $xmm17 $xmm18 $xmm19 $xmm20 $xmm21 $xmm22 $xmm23 $xmm24 $xmm25 $xmm26 $xmm27 $xmm28 $xmm29 $xmm30 $xmm31 $ymm0 $ymm1 $ymm2 $ymm3 $ymm4 $ymm5 $ymm6 $ymm7 $ymm8 $ymm9 $ymm10 $ymm11 $ymm12 $ymm13 $ymm14 $ymm15 $ymm16 $ymm17 $ymm18 $ymm19 $ymm20 $ymm21 $ymm22 $ymm23 $ymm24 $ymm25 $ymm26 $ymm27 $ymm28 $ymm29 $ymm30 $ymm31 $zmm0 $zmm1 $zmm2 $zmm3 $zmm4 $zmm5 $zmm6 $zmm7 $zmm8 $zmm9 $zmm10 $zmm11 $zmm12 $zmm13 $zmm14 $zmm15 $zmm16 $zmm17 $zmm18 $zmm19 $zmm20 $zmm21 $zmm22 $zmm23 $zmm24 $zmm25 $zmm26 $zmm27 $zmm28 $zmm29 $zmm30 $zmm31 $r11b $r11bh $r11d $r11w $r11wh
   call void @bar1()
   call void @bar2()
   ret void
 }
 declare void @bar2()
 attributes #0 = {nounwind}

test/CodeGen/X86/pr34080.ll

 ; SSE2-SCHEDULE-NEXT:    fldt 16(%rbp)
 ; SSE2-SCHEDULE-NEXT:    movw %ax, -4(%rbp)
 ; SSE2-SCHEDULE-NEXT:    fistl -8(%rbp)
 ; SSE2-SCHEDULE-NEXT:    fldcw -4(%rbp)
 ; SSE2-SCHEDULE-NEXT:    cvtsi2sdl -8(%rbp), %xmm0
 ; SSE2-SCHEDULE-NEXT:    movsd %xmm0, -64(%rbp)
 ; SSE2-SCHEDULE-NEXT:    movsd %xmm0, -32(%rbp)
 ; SSE2-SCHEDULE-NEXT:    fsubl -32(%rbp)
-; SSE2-SCHEDULE-NEXT:    fnstcw -2(%rbp)
 ; SSE2-SCHEDULE-NEXT:    flds {{.*}}(%rip)
+; SSE2-SCHEDULE-NEXT:    fnstcw -2(%rbp)
+; SSE2-SCHEDULE-NEXT:    fmul %st, %st(1)
 ; SSE2-SCHEDULE-NEXT:    movzwl -2(%rbp), %eax
 ; SSE2-SCHEDULE-NEXT:    movw $3199, -2(%rbp) ## imm = 0xC7F
 ; SSE2-SCHEDULE-NEXT:    fldcw -2(%rbp)
-; SSE2-SCHEDULE-NEXT:    fmul %st, %st(1)
 ; SSE2-SCHEDULE-NEXT:    movw %ax, -2(%rbp)
 ; SSE2-SCHEDULE-NEXT:    fxch %st(1)
 ; SSE2-SCHEDULE-NEXT:    fistl -12(%rbp)
 ; SSE2-SCHEDULE-NEXT:    fldcw -2(%rbp)
 ; SSE2-SCHEDULE-NEXT:    xorps %xmm0, %xmm0
 ; SSE2-SCHEDULE-NEXT:    cvtsi2sdl -12(%rbp), %xmm0
 ; SSE2-SCHEDULE-NEXT:    movsd %xmm0, -56(%rbp)
 ; SSE2-SCHEDULE-NEXT:    movsd %xmm0, -24(%rbp)

test/CodeGen/X86/pr40529.ll

 ; CHECK-NEXT:    fldcw -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fldt {{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    movw %ax, -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fistl -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fldcw -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    movl -{{[0-9]+}}(%rsp), %eax
 ; CHECK-NEXT:    movl %eax, -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fisubl -{{[0-9]+}}(%rsp)
-; CHECK-NEXT:    fnstcw -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    flds {{.*}}(%rip)
+; CHECK-NEXT:    fnstcw -{{[0-9]+}}(%rsp)
+; CHECK-NEXT:    fmul %st, %st(1)
 ; CHECK-NEXT:    movzwl -{{[0-9]+}}(%rsp), %eax
 ; CHECK-NEXT:    movw $3199, -{{[0-9]+}}(%rsp) # imm = 0xC7F
 ; CHECK-NEXT:    fldcw -{{[0-9]+}}(%rsp)
-; CHECK-NEXT:    fmul %st, %st(1)
 ; CHECK-NEXT:    movw %ax, -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fxch %st(1)
 ; CHECK-NEXT:    fistl -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fldcw -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    movl -{{[0-9]+}}(%rsp), %eax
 ; CHECK-NEXT:    movl %eax, -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fisubl -{{[0-9]+}}(%rsp)
 ; CHECK-NEXT:    fmulp %st, %st(1)