[ARM] Remove dead mov's in preheader of tail predicated loops

[dmgreen]

Summary

With t2DoLoopDec we can be left with some extra MOV's in the preheaders of tail predicated loops. This removes them, in the same way we remove other dead variables.

[samtebbs]

llvm/lib/Target/ARM/ARMLowOverheadLoops.cpp

793

Is it guaranteed that these extra instructions are unused (and so it's ok to remove them without any extra RDA checks)?

[dmgreen]

llvm/lib/Target/ARM/ARMLowOverheadLoops.cpp

793

I am hoping that is what TryRemove will do. It checks for things like corrupting IT blocks too.

I'm not super confident about this code late-backend dead code removal not causing some problems, but hope it's at least as sane as the existing checks. I have not seen any problems so far at least.

[samparker]

llvm/lib/Target/ARM/ARMLowOverheadLoops.cpp

802

TryRemove is, relatively, robust :) I think you should just be able to pass ToRemove in place of ElementChain here, though I can't immediately remember why that isn't the case for the loop above.

[dmgreen]

Rebase and cleanup the ToRemove array

[dmgreen]

ping :)

[samparker]

cheers!

llvm/lib/Target/ARM/ARMLowOverheadLoops.cpp

if (TryRemove(Def, RDA, ElementChain, Ignore)) {
           LLVM_DEBUG(dbgs() << "ARM Loops: Unexpected instruction in element"
                      " count: " << *MI);
           return false;
         }
       }
       ToRemove.insert(ElementChain.begin(), ElementChain.end());
     }
   }
+  // If we converted the LoopStart to a t2DoLoopStartTP, we can also remove any
+  // extra instructions in the preheader, which often includes a now unused MOV.
+  if (Start->getOpcode() == ARM::t2DoLoopStartTP && Preheader &&
+      !Preheader->empty() &&
+      !RDA.hasLocalDefBefore(VCTP, VCTP->getOperand(1).getReg())) {
+    if (auto *Def = RDA.getUniqueReachingMIDef(
+            &Preheader->back(), VCTP->getOperand(1).getReg().asMCReg())) {
+      SmallPtrSet<MachineInstr*, 2> ElementChain;
+      SmallPtrSet<MachineInstr*, 2> Ignore;
+      Ignore.insert(VCTPs.begin(), VCTPs.end());
+      if (TryRemove(Def, RDA, ElementChain, Ignore)) {
+        ToRemove.insert(ElementChain.begin(), ElementChain.end());
+      }
+    }
+  }
   return true;
 }
 
 static bool isRegInClass(const MachineOperand &MO,
                          const TargetRegisterClass *Class) {
   return MO.isReg() && MO.getReg() && Class->contains(MO.getReg());
 }
 

llvm/test/CodeGen/Thumb2/LowOverheadLoops/exitcount.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp -verify-machineinstrs -tail-predication=enabled -o - %s | FileCheck %s
 %struct.SpeexPreprocessState_ = type { i32, i32, half*, half* }
 
 define void @foo(%struct.SpeexPreprocessState_* nocapture readonly %st, i16* %x) {
 ; CHECK-LABEL: foo:
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    .save {r4, lr}
 ; CHECK-NEXT:    push {r4, lr}
 ; CHECK-NEXT:    ldrd r12, r2, [r0]
 ; CHECK-NEXT:    ldrd r4, r3, [r0, #8]
 ; CHECK-NEXT:    rsb r12, r12, r2, lsl #1
-; CHECK-NEXT:    mov r2, r12
 ; CHECK-NEXT:    dlstp.16 lr, r12
 ; CHECK-NEXT:  .LBB0_1: @ %do.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vldrh.u16 q0, [r3], #16
 ; CHECK-NEXT:    vstrh.16 q0, [r4], #16
 ; CHECK-NEXT:    letp lr, .LBB0_1
 ; CHECK-NEXT:  @ %bb.2: @ %do.end
 ; CHECK-NEXT:    ldr r2, [r0]

llvm/test/CodeGen/Thumb2/LowOverheadLoops/mov-operand.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp -verify-machineinstrs -tail-predication=enabled -o - %s | FileCheck %s
 
 define arm_aapcs_vfpcc void @arm_var_f32_mve(float* %pSrc, i32 %blockSize, float* nocapture %pResult) {
 ; CHECK-LABEL: arm_var_f32_mve:
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    .save {r7, lr}
 ; CHECK-NEXT:    push {r7, lr}
 ; CHECK-NEXT:    vmov.i32 q0, #0x0
-; CHECK-NEXT:    mov r3, r1
 ; CHECK-NEXT:    mov r12, r0
 ; CHECK-NEXT:    dlstp.32 lr, r1
 ; CHECK-NEXT:  .LBB0_1: @ %do.body.i
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vldrw.u32 q1, [r12], #16
 ; CHECK-NEXT:    vadd.f32 q0, q0, q1
 ; CHECK-NEXT:    letp lr, .LBB0_1
 ; CHECK-NEXT:  @ %bb.2: @ %arm_mean_f32_mve.exit
 ; CHECK-NEXT:    vmov s4, r1
 ; CHECK-NEXT:    vadd.f32 s0, s3, s3
 ; CHECK-NEXT:    vcvt.f32.u32 s4, s4
-; CHECK-NEXT:    mov r3, r1
 ; CHECK-NEXT:    vdiv.f32 s0, s0, s4
 ; CHECK-NEXT:    vmov r12, s0
 ; CHECK-NEXT:    vmov.i32 q0, #0x0
 ; CHECK-NEXT:    dlstp.32 lr, r1
 ; CHECK-NEXT:  .LBB0_3: @ %do.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vldrw.u32 q1, [r0], #16
 ; CHECK-NEXT:    vsub.f32 q1, q1, r12

llvm/test/CodeGen/Thumb2/LowOverheadLoops/while-loops.ll

 ; CHECK-NEXT:    @ =>This Loop Header: Depth=1
 ; CHECK-NEXT:    @ Child Loop BB1_6 Depth 2
 ; CHECK-NEXT:    adds r7, r5, #3
 ; CHECK-NEXT:    cmp.w r12, r7, lsr #2
 ; CHECK-NEXT:    beq .LBB1_2
 ; CHECK-NEXT:  @ %bb.5: @ %do.body.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB1_4 Depth=1
 ; CHECK-NEXT:    bic r9, r7, #3
-; CHECK-NEXT:    mov r7, r5
 ; CHECK-NEXT:    mov r4, r3
 ; CHECK-NEXT:    add.w r8, r0, r9, lsl #2
 ; CHECK-NEXT:    dlstp.32 lr, r5
 ; CHECK-NEXT:  .LBB1_6: @ %do.body
 ; CHECK-NEXT:    @ Parent Loop BB1_4 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrw.u32 q0, [r0], #16
 ; CHECK-NEXT:    vaddva.s32 r4, q0

llvm/test/CodeGen/Thumb2/mve-postinc-dct.ll

 ; CHECK-NEXT:    movs r0, #1
 ; CHECK-NEXT:    lsl.w r9, r5, #2
 ; CHECK-NEXT:  .LBB0_2: @ %for.body
 ; CHECK-NEXT:    @ =>This Loop Header: Depth=1
 ; CHECK-NEXT:    @ Child Loop BB0_3 Depth 2
 ; CHECK-NEXT:    vmov.i32 q0, #0x0
 ; CHECK-NEXT:    mov r6, r1
 ; CHECK-NEXT:    mov r7, r3
-; CHECK-NEXT:    mov r4, r5
 ; CHECK-NEXT:    dlstp.32 lr, r5
 ; CHECK-NEXT:  .LBB0_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB0_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrw.u32 q1, [r6], #16
 ; CHECK-NEXT:    vldrw.u32 q2, [r7], #16
 ; CHECK-NEXT:    vfma.f32 q0, q2, q1
 ; CHECK-NEXT:    letp lr, .LBB0_3

 ; CHECK-NEXT:    @ =>This Loop Header: Depth=1
 ; CHECK-NEXT:    @ Child Loop BB1_3 Depth 2
 ; CHECK-NEXT:    ldr r5, [sp] @ 4-byte Reload
 ; CHECK-NEXT:    vmov.i32 q0, #0x0
 ; CHECK-NEXT:    add.w r10, r4, #1
 ; CHECK-NEXT:    mov r3, r11
 ; CHECK-NEXT:    mov r0, r7
 ; CHECK-NEXT:    vmov q1, q0
-; CHECK-NEXT:    mov r6, r12
 ; CHECK-NEXT:    dlstp.32 lr, r12
 ; CHECK-NEXT:  .LBB1_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB1_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrw.u32 q2, [r5], #16
 ; CHECK-NEXT:    vldrw.u32 q3, [r3], #16
 ; CHECK-NEXT:    vfma.f32 q1, q3, q2
 ; CHECK-NEXT:    vldrw.u32 q3, [r0], #16

 ; CHECK-NEXT:    adds r0, r5, #2
 ; CHECK-NEXT:    adds r2, r5, #1
 ; CHECK-NEXT:    str r0, [sp, #20] @ 4-byte Spill
 ; CHECK-NEXT:    mov r3, r9
 ; CHECK-NEXT:    mov r0, r12
 ; CHECK-NEXT:    mov r4, r10
 ; CHECK-NEXT:    vmov q2, q0
 ; CHECK-NEXT:    vmov q1, q0
-; CHECK-NEXT:    mov r8, r7
 ; CHECK-NEXT:    dlstp.32 lr, r7
 ; CHECK-NEXT:  .LBB2_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB2_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrw.u32 q3, [r6], #16
 ; CHECK-NEXT:    vldrw.u32 q4, [r3], #16
 ; CHECK-NEXT:    vfma.f32 q1, q4, q3
 ; CHECK-NEXT:    vldrw.u32 q4, [r0], #16

 ; CHECK-NEXT:    str r0, [sp, #28] @ 4-byte Spill
 ; CHECK-NEXT:    mov r3, r12
 ; CHECK-NEXT:    mov r0, r8
 ; CHECK-NEXT:    mov r5, r11
 ; CHECK-NEXT:    mov r4, r9
 ; CHECK-NEXT:    vmov q1, q0
 ; CHECK-NEXT:    vmov q2, q0
 ; CHECK-NEXT:    vmov q3, q0
-; CHECK-NEXT:    mov r10, r7
 ; CHECK-NEXT:    dlstp.32 lr, r7
 ; CHECK-NEXT:  .LBB3_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB3_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrw.u32 q4, [r1], #16
 ; CHECK-NEXT:    vldrw.u32 q5, [r0], #16
 ; CHECK-NEXT:    vfma.f32 q3, q5, q4
 ; CHECK-NEXT:    vldrw.u32 q5, [r3], #16

 ; CHECK-NEXT:    adds r1, r0, #3
 ; CHECK-NEXT:    add.w r11, r0, #1
 ; CHECK-NEXT:    str r1, [sp, #24] @ 4-byte Spill
 ; CHECK-NEXT:    mov r3, r8
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    vmov q3, q1
 ; CHECK-NEXT:    vmov q2, q1
 ; CHECK-NEXT:    vmov q4, q1
-; CHECK-NEXT:    mov r1, r7
 ; CHECK-NEXT:    dlstp.32 lr, r7
 ; CHECK-NEXT:  .LBB4_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB4_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    add.w r9, r3, r5
 ; CHECK-NEXT:    vldrw.u32 q5, [r4], #16
 ; CHECK-NEXT:    vldrw.u32 q6, [r3], #16
 ; CHECK-NEXT:    vfma.f32 q3, q6, q5

 ; CHECK-NEXT:    ldr r6, [sp, #4] @ 4-byte Reload
 ; CHECK-NEXT:    adds r4, r0, #1
 ; CHECK-NEXT:    mov r3, r8
 ; CHECK-NEXT:    vmov q3, q1
 ; CHECK-NEXT:    vmov q4, q1
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    vmov q5, q1
 ; CHECK-NEXT:    vmov q2, q1
-; CHECK-NEXT:    mov r9, r7
 ; CHECK-NEXT:    dlstp.32 lr, r7
 ; CHECK-NEXT:  .LBB5_3: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB5_2 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    add.w r12, r3, r5
 ; CHECK-NEXT:    vldrw.u32 q6, [r1], #16
 ; CHECK-NEXT:    vldrw.u32 q7, [r3], #16
 ; CHECK-NEXT:    vfma.f32 q4, q7, q6

llvm/test/CodeGen/Thumb2/mve-postinc-lsr.ll

 ; CHECK-NEXT:  @ %bb.6: @ %for.body24.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB5_5 Depth=1
 ; CHECK-NEXT:    ldr r2, [sp, #92]
 ; CHECK-NEXT:    ldr r0, [sp, #12] @ 4-byte Reload
 ; CHECK-NEXT:    mov r6, r12
 ; CHECK-NEXT:    ldr r1, [sp, #16] @ 4-byte Reload
 ; CHECK-NEXT:    mov r8, r12
 ; CHECK-NEXT:    mla r3, r11, r2, r0
-; CHECK-NEXT:    mov r9, r2
 ; CHECK-NEXT:    mov r10, r12
 ; CHECK-NEXT:    ldm.w sp, {r0, r5, r7} @ 12-byte Folded Reload
 ; CHECK-NEXT:    dlstp.16 lr, r2
 ; CHECK-NEXT:  .LBB5_7: @ %for.body24
 ; CHECK-NEXT:    @ Parent Loop BB5_5 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrb.s16 q0, [r0], #8
 ; CHECK-NEXT:    vadd.i16 q1, q0, r4

 ; CHECK-NEXT:  @ %bb.4: @ %for.body24.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB6_3 Depth=1
 ; CHECK-NEXT:    ldr r2, [sp, #92]
 ; CHECK-NEXT:    ldr r0, [sp, #12] @ 4-byte Reload
 ; CHECK-NEXT:    mov r6, r12
 ; CHECK-NEXT:    ldr r1, [sp, #16] @ 4-byte Reload
 ; CHECK-NEXT:    mov r8, r12
 ; CHECK-NEXT:    mla r3, r11, r2, r0
-; CHECK-NEXT:    mov r9, r2
 ; CHECK-NEXT:    mov r10, r12
 ; CHECK-NEXT:    ldm.w sp, {r0, r5, r7} @ 12-byte Folded Reload
 ; CHECK-NEXT:    dlstp.16 lr, r2
 ; CHECK-NEXT:  .LBB6_5: @ %for.body24
 ; CHECK-NEXT:    @ Parent Loop BB6_3 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    vldrb.s16 q0, [r0], #8
 ; CHECK-NEXT:    vadd.i16 q1, q0, r4

llvm/test/CodeGen/Thumb2/mve-pred-vctpvpsel.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp %s -verify-machineinstrs -o - | FileCheck %s
 
 define void @arm_min_helium_f32(float* %pSrc, i32 %blockSize, float* nocapture %pResult, i32* nocapture %pIndex) {
 ; CHECK-LABEL: arm_min_helium_f32:
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    .save {r4, r6, r7, lr}
 ; CHECK-NEXT:    push {r4, r6, r7, lr}
 ; CHECK-NEXT:    .vsave {d8, d9}
 ; CHECK-NEXT:    vpush {d8, d9}
 ; CHECK-NEXT:    movs r6, #0
 ; CHECK-NEXT:    vidup.u32 q2, r6, #1
 ; CHECK-NEXT:    adr r4, .LCPI0_0
 ; CHECK-NEXT:    vmov.i32 q0, #0x0
 ; CHECK-NEXT:    vldrw.u32 q1, [r4]
 ; CHECK-NEXT:    vmov.i32 q3, #0x4
-; CHECK-NEXT:    mov r12, r1
 ; CHECK-NEXT:    dlstp.32 lr, r1
 ; CHECK-NEXT:  .LBB0_1: @ %do.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vldrw.u32 q4, [r0], #16
 ; CHECK-NEXT:    vptt.f32 ge, q1, q4
 ; CHECK-NEXT:    vmovt q1, q4
 ; CHECK-NEXT:    vmovt q0, q2
 ; CHECK-NEXT:    vadd.i32 q2, q2, q3