[ARM] Add scheduling info for VFMS

[olista01]

Summary

The scalar VFMS instructions did not have scheduling information attached (but VFMA did), which was causing assertion failures with the Cortex-A57 scheduling model and -fp-contract=fast.

[javed.absar]

Thanks Oliver for adding the missing sched defs. I am surprised there was assertion failures for missing scheds as default behaviour of machine scheduler used to be 'to-silently-ignore'. Anyways, that's a separate issue. Still good to have complete defs.

lib/Target/ARM/ARMInstrVFP.td

def : Pat<(f32 (fma SPR:$Sn, SPR:$Sm, SPR:$Sdin)),
       Requires<[HasVFP4]>;
 
 def VFMSD : ADbI<0b11101, 0b10, 1, 0,
                  (outs DPR:$Dd), (ins DPR:$Ddin, DPR:$Dn, DPR:$Dm),
                  IIC_fpFMAC64, "vfms", ".f64\t$Dd, $Dn, $Dm",
                  [(set DPR:$Dd, (fadd_mlx (fneg (fmul_su DPR:$Dn,DPR:$Dm)),
                                           (f64 DPR:$Ddin)))]>,
               RegConstraint<"$Ddin = $Dd">,
-              Requires<[HasVFP4,HasDPVFP,UseFusedMAC]>;
+              Requires<[HasVFP4,HasDPVFP,UseFusedMAC]>,
+              Sched<[WriteFPMAC64, ReadFPMAC, ReadFPMUL, ReadFPMUL]>;
 
 def VFMSS : ASbIn<0b11101, 0b10, 1, 0,
                   (outs SPR:$Sd), (ins SPR:$Sdin, SPR:$Sn, SPR:$Sm),
                   IIC_fpFMAC32, "vfms", ".f32\t$Sd, $Sn, $Sm",
                   [(set SPR:$Sd, (fadd_mlx (fneg (fmul_su SPR:$Sn, SPR:$Sm)),
                                            SPR:$Sdin))]>,
               RegConstraint<"$Sdin = $Sd">,
-              Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]> {
+              Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>,
+              Sched<[WriteFPMAC32, ReadFPMAC, ReadFPMUL, ReadFPMUL]> {
   // Some single precision VFP instructions may be executed on both NEON and
   // VFP pipelines.
 }
 
 def VFMSH : AHbI<0b11101, 0b10, 1, 0,
                   (outs SPR:$Sd), (ins SPR:$Sdin, SPR:$Sn, SPR:$Sm),
                   IIC_fpFMAC16, "vfms", ".f16\t$Sd, $Sn, $Sm",
                   []>,
               RegConstraint<"$Sdin = $Sd">,
-              Requires<[HasFullFP16,UseFusedMAC]>;
+              Requires<[HasFullFP16,UseFusedMAC]>,
+              Sched<[WriteFPMAC32, ReadFPMAC, ReadFPMUL, ReadFPMUL]>;
 
 def : Pat<(fsub_mlx DPR:$dstin, (fmul_su DPR:$a, (f64 DPR:$b))),
           (VFMSD DPR:$dstin, DPR:$a, DPR:$b)>,
           Requires<[HasVFP4,HasDPVFP,UseFusedMAC]>;
 def : Pat<(fsub_mlx SPR:$dstin, (fmul_su SPR:$a, SPR:$b)),
           (VFMSS SPR:$dstin, SPR:$a, SPR:$b)>,
           Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>;
 

test/CodeGen/ARM/cortex-a57-misched-vfma.ll

 ; REQUIRES: asserts
-; RUN: llc < %s -mtriple=armv8r-eabi -mcpu=cortex-a57 -enable-misched -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null | FileCheck %s
+; RUN: llc < %s -mtriple=armv8r-eabi -mcpu=cortex-a57 -enable-misched -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK-DEFAULT
+; RUN: llc < %s -mtriple=armv8r-eabi -mcpu=cortex-a57 -enable-misched -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null -fp-contract=fast | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK-FAST
 ; Check latencies of vmul/vfma accumulate chains.
 
 define float @Test1(float %f1, float %f2, float %f3, float %f4, float %f5, float %f6) {
 ; CHECK:       ********** MI Scheduling **********
 ; CHECK:       Test1:BB#0
 
 ; CHECK:       VMULS
 ; > VMULS common latency = 5
 ; CHECK:       Latency            : 5
 ; CHECK:       Successors:
 ; CHECK:       data
 ; > VMULS read-advanced latency to VMLAS = 0
 ; CHECK-SAME:  Latency=0
 
-; CHECK:       VMLAS
+; CHECK-DEFAULT: VMLAS
+; CHECK-FAST:    VFMAS
 ; > VMLAS common latency = 9
 ; CHECK:       Latency            : 9
 ; CHECK:       Successors:
 ; CHECK:       data
 ; > VMLAS read-advanced latency to the next VMLAS = 4
 ; CHECK-SAME:  Latency=4
 
-; CHECK:       VMLAS
+; CHECK-DEFAULT: VMLAS
+; CHECK-FAST:    VFMAS
 ; CHECK:       Latency            : 9
 ; CHECK:       Successors:
 ; CHECK:       data
 ; > VMLAS not-optimized latency to VMOVRS = 9
 ; CHECK-SAME:  Latency=9
 
 ; f1 * f2 + f3 * f4 + f5 * f6  ==>  VMULS, VMLAS, VMLAS
   %mul1 = fmul float %f1, %f2

 ; CHECK:       VMULfd
 ; > VMULfd common latency = 5
 ; CHECK:       Latency            : 5
 ; CHECK:       Successors:
 ; CHECK:       data
 ; VMULfd read-advanced latency to VMLAfd = 0
 ; CHECK-SAME:  Latency=0
 
-; CHECK:       VMLAfd
+; CHECK-DEFAULT: VMLAfd
+; CHECK-FAST:    VFMAfd
 ; > VMLAfd common latency = 9
 ; CHECK:       Latency            : 9
 ; CHECK:       Successors:
 ; CHECK:       data
 ; > VMLAfd read-advanced latency to the next VMLAfd = 4
 ; CHECK-SAME:  Latency=4
 
-; CHECK:       VMLAfd
+; CHECK-DEFAULT: VMLAfd
+; CHECK-FAST:    VFMAfd
 ; CHECK:       Latency            : 9
 ; CHECK:       Successors:
 ; CHECK:       data
 ; > VMLAfd not-optimized latency to VMOVRRD = 9
 ; CHECK-SAME:  Latency=9
 
 ; f1 * f2 + f3 * f4 + f5 * f6  ==>  VMULS, VMLAS, VMLAS
   %mul1 = fmul <2 x float> %f1, %f2
   %mul2 = fmul <2 x float> %f3, %f4
   %mul3 = fmul <2 x float> %f5, %f6
   %add1 = fadd <2 x float> %mul1, %mul2
   %add2 = fadd <2 x float> %add1, %mul3
   ret <2 x float> %add2
 }
 
+define float @Test3(float %f1, float %f2, float %f3, float %f4, float %f5, float %f6) {
+; CHECK:       ********** MI Scheduling **********
+; CHECK:       Test3:BB#0
+
+; CHECK:       VMULS
+; > VMULS common latency = 5
+; CHECK:       Latency            : 5
+; CHECK:       Successors:
+; CHECK:       data
+; > VMULS read-advanced latency to VMLSS = 0
+; CHECK-SAME:  Latency=0
+
+; CHECK-DEFAULT: VMLSS
+; CHECK-FAST:    VFMSS
+; > VMLSS common latency = 9
+; CHECK:       Latency            : 9
+; CHECK:       Successors:
+; CHECK:       data
+; > VMLSS read-advanced latency to the next VMLSS = 4
+; CHECK-SAME:  Latency=4
+
+; CHECK-DEFAULT: VMLSS
+; CHECK-FAST:    VFMSS
+; CHECK:       Latency            : 9
+; CHECK:       Successors:
+; CHECK:       data
+; > VMLSS not-optimized latency to VMOVRS = 9
+; CHECK-SAME:  Latency=9
+
+; f1 * f2 + f3 * f4 + f5 * f6  ==>  VMULS, VMLSS, VMLSS
+  %mul1 = fmul float %f1, %f2
+  %mul2 = fmul float %f3, %f4
+  %mul3 = fmul float %f5, %f6
+  %sub1 = fsub float %mul1, %mul2
+  %sub2 = fsub float %sub1, %mul3
+  ret float %sub2
+}
+
+; ASIMD form
+define <2 x float> @Test4(<2 x float> %f1, <2 x float> %f2, <2 x float> %f3, <2 x float> %f4, <2 x float> %f5, <2 x float> %f6) {
+; CHECK:       ********** MI Scheduling **********
+; CHECK:       Test4:BB#0
+
+; CHECK:       VMULfd
+; > VMULfd common latency = 5
+; CHECK:       Latency            : 5
+; CHECK:       Successors:
+; CHECK:       data
+; VMULfd read-advanced latency to VMLSfd = 0
+; CHECK-SAME:  Latency=0
+
+; CHECK-DEFAULT: VMLSfd
+; CHECK-FAST:    VFMSfd
+; > VMLSfd common latency = 9
+; CHECK:       Latency            : 9
+; CHECK:       Successors:
+; CHECK:       data
+; > VMLSfd read-advanced latency to the next VMLSfd = 4
+; CHECK-SAME:  Latency=4
+
+; CHECK-DEFAULT: VMLSfd
+; CHECK-FAST:    VFMSfd
+; CHECK:       Latency            : 9
+; CHECK:       Successors:
+; CHECK:       data
+; > VMLSfd not-optimized latency to VMOVRRD = 9
+; CHECK-SAME:  Latency=9
+
+; f1 * f2 + f3 * f4 + f5 * f6  ==>  VMULS, VMLSS, VMLSS
+  %mul1 = fmul <2 x float> %f1, %f2
+  %mul2 = fmul <2 x float> %f3, %f4
+  %mul3 = fmul <2 x float> %f5, %f6
+  %sub1 = fsub <2 x float> %mul1, %mul2
+  %sub2 = fsub <2 x float> %sub1, %mul3
+  ret <2 x float> %sub2
+}