[AArch64] Split Ampere1Write_Arith into rr/ri and rs/rx InstRWs. (#66384)

The ampere1 scheduling model uses IsCheapLSL predicates for ADDXri and
ADDWrr instructions, which only have 3 operands. In attempting to check
that the third is a shift, the predicate can attempt to access an out of
bounds operand, hitting an assert. This splits the rr/ri instructions
(which can never have shifts) from the rs/rx instructions to ensure they
both work correctly. Ampere1Write_1cyc_1AB was chosen for the rr/ir
instructions to match the cheap case.

This also sets CompleteModel = 0 for the ampere1 scheduling model, as at
runtime under debug it will attempt to check that as well as all
instructions having scheduling info, there is information for each
output operand.

DefIdx 1 exceeds machine model writes for
  renamable $w9, renamable $w8 = LDPWi renamable $x8, 0
(Try with MCSchedModel.CompleteModel set to false)incomplete machine
model

Author: davemgreen

llvm/lib/Target/AArch64/AArch64SchedAmpere1.td

@@ -22,7 +22,7 @@ def Ampere1Model : SchedMachineModel {
   let LoadLatency           =   4;  // Optimistic load latency
   let MispredictPenalty     =  10;  // Branch mispredict penalty
   let LoopMicroOpBufferSize =  32;  // Instruction queue size
-  let CompleteModel = 1;
+  let CompleteModel = 0;
 
   list<Predicate> UnsupportedFeatures = !listconcat(SVEUnsupported.F,
                                                     SMEUnsupported.F,
@@ -936,9 +936,13 @@ def : InstRW<[Ampere1Write_4cyc_1Z], (instregex "^FMOV[WX][HSD]r")>;
 def : InstRW<[Ampere1Write_1cyc_1A],
         (instregex "ADC(W|X)r", "SBC(W|X)r")>;
 def : InstRW<[Ampere1Write_Arith],
-        (instregex "(ADD|AND|BIC|EON|EOR|ORN|ORR|SUB)(W|X)r")>;
+        (instregex "(ADD|AND|BIC|EON|EOR|ORN|ORR|SUB)(W|X)r[sx]")>;
+def : InstRW<[Ampere1Write_1cyc_1AB],
+        (instregex "(ADD|AND|BIC|EON|EOR|ORN|ORR|SUB)(W|X)r[ri]")>;
 def : InstRW<[Ampere1Write_ArithFlagsetting],
-        (instregex "(ADD|AND|BIC|SUB)S(W|X)r")>;
+        (instregex "(ADD|AND|BIC|SUB)S(W|X)r[sx]")>;
+def : InstRW<[Ampere1Write_1cyc_1A],
+        (instregex "(ADD|AND|BIC|SUB)S(W|X)r[ri]")>;
 def : InstRW<[Ampere1Write_1cyc_1A],
         (instregex "(ADC|SBC)S(W|X)r")>;
 def : InstRW<[Ampere1Write_1cyc_1A], (instrs RMIF)>;

llvm/test/CodeGen/AArch64/ampere1-sched-add.mir

@@ -0,0 +1,55 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 3
+# RUN: llc -run-pass=machine-scheduler %s -o - | FileCheck %s
+
+--- |
+  target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+  target triple = "aarch64"
+
+  define i32 @test_add(ptr %0) #0 {
+    %2 = ptrtoint ptr %0 to i64
+    %3 = and i64 %2, -64
+    %4 = inttoptr i64 %3 to ptr
+    %5 = load i32, ptr %4, align 64
+    %6 = getelementptr inbounds i32, ptr %4, i64 1
+    %7 = load i32, ptr %6, align 4
+    %8 = add nsw i32 %7, %5
+    ret i32 %8
+  }
+
+  attributes #0 = { "target-cpu"="ampere1" }
+
+...
+---
+name:            test_add
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: gpr64, preferred-register: '' }
+  - { id: 1, class: gpr64sp, preferred-register: '' }
+  - { id: 2, class: gpr32, preferred-register: '' }
+  - { id: 3, class: gpr32, preferred-register: '' }
+  - { id: 4, class: gpr32, preferred-register: '' }
+liveins:
+  - { reg: '$x0', virtual-reg: '%0' }
+body:             |
+  bb.0 (%ir-block.1):
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_add
+    ; CHECK: liveins: $x0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr64 = COPY $x0
+    ; CHECK-NEXT: [[ANDXri:%[0-9]+]]:gpr64sp = ANDXri [[COPY]], 7865
+    ; CHECK-NEXT: [[LDRWui:%[0-9]+]]:gpr32 = LDRWui [[ANDXri]], 0 :: (load (s32) from %ir.4, align 64)
+    ; CHECK-NEXT: [[LDRWui1:%[0-9]+]]:gpr32 = LDRWui [[ANDXri]], 1 :: (load (s32) from %ir.6)
+    ; CHECK-NEXT: [[ADDWrr:%[0-9]+]]:gpr32 = nsw ADDWrr [[LDRWui1]], [[LDRWui]]
+    ; CHECK-NEXT: $w0 = COPY [[ADDWrr]]
+    ; CHECK-NEXT: RET_ReallyLR implicit $w0
+    %0:gpr64 = COPY $x0
+    %1:gpr64sp = ANDXri %0, 7865
+    %2:gpr32 = LDRWui %1, 0 :: (load (s32) from %ir.4, align 64)
+    %3:gpr32 = LDRWui %1, 1 :: (load (s32) from %ir.6)
+    %4:gpr32 = nsw ADDWrr %3, %2
+    $w0 = COPY %4
+    RET_ReallyLR implicit $w0
+
+...