caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
1650f1b3d7f97ca95eb930984e74bdfd91b02b4e
clang-p2996/llvm/test/CodeGen/AArch64/nested-iv-regalloc.mir
David Green 303a7835ff [GreedyRA] Improve RA for nested loop induction variables (#72093) 
Imagine a loop of the form:
```
  preheader:
    %r = def
  header:
    bcc latch, inner
  inner1:
    ..
  inner2:
    b latch
  latch:
    %r = subs %r
    bcc header
```

It can be possible for code to spend a decent amount of time in the
header<->latch loop, not going into the inner part of the loop as much.
The greedy register allocator can prefer to spill _around_ %r though,
adding spills around the subs in the loop, which can be very detrimental
for performance. (The case I am looking at is actually a very deeply
nested set of loops that repeat the header<->latch pattern at multiple
different levels).

The greedy RA will apply a preference to spill to the IV, as it is live
through the header block. This patch attempts to add a heuristic to
prevent that in this case for variables that look like IVs, in a similar
regard to the extra spill weight that gets added to variables that look
like IVs, that are expensive to spill. That will mean spills are more
likely to be pushed into the inner blocks, where they are less likely to
be executed and not as expensive as spills around the IV.

This gives a 8% speedup in the exchange benchmark from spec2017 when
compiled with flang-new, whilst importantly stabilising the scores to be
less chaotic to other changes. Running ctmark showed no difference in
the compile time. I've tried to run a range of benchmarking for
performance, most of which were relatively flat not showing many large
differences. One matrix multiply case improved 21.3% due to removing a
cascading chains of spills, and some other knock-on effects happen which
usually cause small differences in the scores.
2023-11-18 09:55:19 +00:00

413 lines
21 KiB
YAML
Raw Blame History

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple aarch64 --run-pass=greedy,virtregrewriter -verify-machineinstrs %s -o - | FileCheck %s
# We should ideally not spill around any of the SUBSWri in the loop exit blocks (if.end and if.end27).
--- |
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64"
@g = dso_local local_unnamed_addr global [9 x [9 x i32]] zeroinitializer, align 4
define void @test(ptr nocapture noundef readonly %p, i32 noundef %m, ptr noundef %q) {
entry:
%0 = load i32, ptr %p, align 4
%spec.select = tail call i32 @llvm.smax.i32(i32 %0, i32 1)
%arrayidx2 = getelementptr inbounds i32, ptr %p, i64 1
%1 = load i32, ptr %arrayidx2, align 4
%cond8 = tail call i32 @llvm.smax.i32(i32 %1, i32 1)
br label %do.body
do.body: ; preds = %if.end27, %entry
%indvars.iv49 = phi i64 [ %indvars.iv.next50, %if.end27 ], [ 0, %entry ]
%n0.0 = phi i32 [ %dec30, %if.end27 ], [ %spec.select, %entry ]
%n1.0 = phi i32 [ %n1.2, %if.end27 ], [ %cond8, %entry ]
%arrayidx9 = getelementptr inbounds [9 x [9 x i32]], ptr @g, i64 0, i64 %indvars.iv49
%2 = load i32, ptr %arrayidx9, align 4
%cmp11 = icmp sgt i32 %2, 0
br i1 %cmp11, label %do.body12.preheader, label %if.end27
do.body12.preheader: ; preds = %do.body
br label %do.body12
do.body12: ; preds = %do.body12.preheader, %if.end
%indvars.iv = phi i64 [ %indvars.iv.next, %if.end ], [ 0, %do.body12.preheader ]
%n1.1 = phi i32 [ %dec, %if.end ], [ %n1.0, %do.body12.preheader ]
%arrayidx14 = getelementptr inbounds [9 x [9 x i32]], ptr @g, i64 0, i64 %indvars.iv
%3 = load i32, ptr %arrayidx14, align 4
%cmp16 = icmp sgt i32 %3, 0
br i1 %cmp16, label %if.then17, label %if.end
if.then17: ; preds = %do.body12
%arrayidx19 = getelementptr inbounds i32, ptr %q, i64 %indvars.iv
%4 = load volatile i32, ptr %arrayidx19, align 4
br label %for.body
for.cond.cleanup: ; preds = %for.body
tail call void asm sideeffect "nop;nop", "~{x0},~{x2},~{x3},~{x4},~{x5},~{x6},~{x7},~{x8},~{x9},~{x10},~{x11},~{x12},~{x13},~{x14},~{x15},~{x16},~{x17},~{x18},~{x19},~{x20},~{x21},~{x22},~{x23},~{x24},~{x25},~{x26},~{x27},~{x28},~{fp},~{lr}"() #3
%sunkaddr = mul i64 %indvars.iv, 4
%sunkaddr1 = getelementptr inbounds i8, ptr %q, i64 %sunkaddr
store volatile i32 %add, ptr %sunkaddr1, align 4
br label %if.end
for.body: ; preds = %for.body, %if.then17
%lsr.iv = phi i32 [ %lsr.iv.next, %for.body ], [ 100, %if.then17 ]
%s.046 = phi i32 [ %4, %if.then17 ], [ %add, %for.body ]
%call = tail call i32 @callee() #3
%add = add nsw i32 %call, %s.046
%lsr.iv.next = add nsw i32 %lsr.iv, -1
%exitcond.not = icmp eq i32 %lsr.iv.next, 0
br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
if.end: ; preds = %for.cond.cleanup, %do.body12
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%dec = add nsw i32 %n1.1, -1
%cmp24.not = icmp eq i32 %dec, 0
br i1 %cmp24.not, label %do.end, label %do.body12
do.end: ; preds = %if.end
%arrayidx26 = getelementptr inbounds i32, ptr %q, i64 %indvars.iv49
store volatile i32 0, ptr %arrayidx26, align 4
br label %if.end27
if.end27: ; preds = %do.end, %do.body
%n1.2 = phi i32 [ 0, %do.end ], [ %n1.0, %do.body ]
%indvars.iv.next50 = add nuw nsw i64 %indvars.iv49, 1
%dec30 = add nsw i32 %n0.0, -1
%cmp31.not = icmp eq i32 %dec30, 0
br i1 %cmp31.not, label %do.end32, label %do.body
do.end32: ; preds = %if.end27
ret void
}
declare i32 @callee(...)
declare i32 @llvm.smax.i32(i32, i32) #2
...
---
name: test
alignment: 4
exposesReturnsTwice: false
legalized: false
regBankSelected: false
selected: false
failedISel: false
tracksRegLiveness: true
hasWinCFI: false
callsEHReturn: false
callsUnwindInit: false
hasEHCatchret: false
hasEHScopes: false
hasEHFunclets: false
isOutlined: false
debugInstrRef: false
failsVerification: false
tracksDebugUserValues: false
registers:
- { id: 0, class: gpr32, preferred-register: '' }
- { id: 1, class: gpr32, preferred-register: '' }
- { id: 2, class: gpr64common, preferred-register: '' }
- { id: 3, class: gpr32sp, preferred-register: '' }
- { id: 4, class: gpr32all, preferred-register: '' }
- { id: 5, class: gpr64common, preferred-register: '' }
- { id: 6, class: gpr32sp, preferred-register: '' }
- { id: 7, class: gpr32, preferred-register: '' }
- { id: 8, class: gpr32sp, preferred-register: '' }
- { id: 9, class: gpr32, preferred-register: '' }
- { id: 10, class: gpr32, preferred-register: '' }
- { id: 11, class: gpr32, preferred-register: '' }
- { id: 12, class: gpr64sp, preferred-register: '' }
- { id: 13, class: gpr32, preferred-register: '' }
- { id: 14, class: gpr32all, preferred-register: '' }
- { id: 15, class: gpr64sp, preferred-register: '' }
- { id: 16, class: gpr32, preferred-register: '' }
- { id: 17, class: gpr64common, preferred-register: '' }
- { id: 18, class: gpr32, preferred-register: '' }
- { id: 19, class: gpr64common, preferred-register: '' }
- { id: 20, class: gpr64all, preferred-register: '' }
- { id: 21, class: gpr32common, preferred-register: '' }
- { id: 22, class: gpr32, preferred-register: '' }
- { id: 23, class: gpr32, preferred-register: '' }
- { id: 24, class: gpr32common, preferred-register: '' }
- { id: 25, class: gpr32, preferred-register: '' }
- { id: 26, class: gpr32, preferred-register: '' }
- { id: 27, class: gpr64all, preferred-register: '' }
- { id: 28, class: gpr32, preferred-register: '' }
- { id: 29, class: gpr64, preferred-register: '' }
- { id: 30, class: gpr64, preferred-register: '' }
- { id: 31, class: gpr64common, preferred-register: '' }
- { id: 32, class: gpr32common, preferred-register: '' }
- { id: 33, class: gpr32, preferred-register: '' }
- { id: 34, class: gpr64all, preferred-register: '' }
- { id: 35, class: gpr64all, preferred-register: '' }
- { id: 36, class: gpr32, preferred-register: '' }
- { id: 37, class: gpr64, preferred-register: '' }
- { id: 38, class: gpr64, preferred-register: '' }
- { id: 39, class: gpr64common, preferred-register: '' }
- { id: 40, class: gpr32common, preferred-register: '' }
- { id: 41, class: gpr32, preferred-register: '' }
- { id: 42, class: gpr32, preferred-register: '' }
- { id: 43, class: gpr32, preferred-register: '' }
- { id: 44, class: gpr32, preferred-register: '' }
- { id: 45, class: gpr32, preferred-register: '' }
- { id: 46, class: gpr32, preferred-register: '' }
- { id: 47, class: gpr32, preferred-register: '' }
- { id: 48, class: gpr64sp, preferred-register: '' }
- { id: 49, class: gpr32, preferred-register: '' }
- { id: 50, class: gpr32all, preferred-register: '' }
- { id: 51, class: gpr32, preferred-register: '' }
- { id: 52, class: gpr64sp, preferred-register: '' }
- { id: 53, class: gpr32, preferred-register: '' }
- { id: 54, class: gpr64common, preferred-register: '' }
- { id: 55, class: gpr32common, preferred-register: '' }
- { id: 56, class: gpr32common, preferred-register: '' }
- { id: 57, class: gpr64common, preferred-register: '' }
- { id: 58, class: gpr32common, preferred-register: '' }
- { id: 59, class: gpr32common, preferred-register: '' }
- { id: 60, class: gpr32, preferred-register: '' }
- { id: 61, class: gpr32all, preferred-register: '' }
liveins:
- { reg: '$x0', virtual-reg: '%17' }
- { reg: '$x2', virtual-reg: '%19' }
frameInfo:
isFrameAddressTaken: false
isReturnAddressTaken: false
hasStackMap: false
hasPatchPoint: false
stackSize: 0
offsetAdjustment: 0
maxAlignment: 1
adjustsStack: true
hasCalls: true
stackProtector: ''
functionContext: ''
maxCallFrameSize: 0
cvBytesOfCalleeSavedRegisters: 0
hasOpaqueSPAdjustment: false
hasVAStart: false
hasMustTailInVarArgFunc: false
hasTailCall: false
localFrameSize: 0
savePoint: ''
restorePoint: ''
fixedStack: []
stack: []
entry_values: []
callSites: []
debugValueSubstitutions: []
constants: []
machineFunctionInfo: {}
body: |
; CHECK-LABEL: name: test
; CHECK: bb.0.entry:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: liveins: $x0, $x2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $x10 = COPY $xzr
; CHECK-NEXT: renamable $w8 = LDRWui renamable $x0, 0 :: (load (s32) from %ir.p)
; CHECK-NEXT: renamable $w9 = LDRWui killed renamable $x0, 1 :: (load (s32) from %ir.arrayidx2)
; CHECK-NEXT: dead $wzr = SUBSWri renamable $w8, 1, 0, implicit-def $nzcv
; CHECK-NEXT: renamable $w11 = CSINCWr killed renamable $w8, $wzr, 12, implicit $nzcv
; CHECK-NEXT: renamable $x8 = COPY killed renamable $x10
; CHECK-NEXT: dead $wzr = SUBSWri renamable $w9, 1, 0, implicit-def $nzcv
; CHECK-NEXT: renamable $w10 = CSINCWr killed renamable $w9, $wzr, 12, implicit $nzcv
; CHECK-NEXT: STRXui renamable $x2, %stack.0, 0 :: (store (s64) into %stack.0)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1.do.body:
; CHECK-NEXT: successors: %bb.3(0x50000000), %bb.2(0x30000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2, $x8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: STRXui renamable $x8, %stack.1, 0 :: (store (s64) into %stack.1)
; CHECK-NEXT: renamable $w9 = MOVi32imm 36, implicit-def $x9
; CHECK-NEXT: renamable $x8 = MADDXrrr killed renamable $x8, killed renamable $x9, $xzr
; CHECK-NEXT: renamable $x9 = MOVaddr target-flags(aarch64-page) @g, target-flags(aarch64-pageoff, aarch64-nc) @g
; CHECK-NEXT: renamable $w8 = LDRWroX killed renamable $x9, killed renamable $x8, 0, 0 :: (load (s32) from %ir.arrayidx9)
; CHECK-NEXT: dead $wzr = SUBSWri killed renamable $w8, 1, 0, implicit-def $nzcv
; CHECK-NEXT: Bcc 10, %bb.3, implicit $nzcv
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.10(0x80000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $x8 = LDRXui %stack.1, 0 :: (load (s64) from %stack.1)
; CHECK-NEXT: B %bb.10
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3.do.body12.preheader:
; CHECK-NEXT: successors: %bb.4(0x80000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $x12 = COPY $xzr
; CHECK-NEXT: STRWui renamable $w11, %stack.2, 0 :: (store (s32) into %stack.2)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4.do.body12:
; CHECK-NEXT: successors: %bb.5(0x50000000), %bb.8(0x30000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2, $x12
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $w8 = MOVi32imm 36, implicit-def $x8
; CHECK-NEXT: renamable $x8 = MADDXrrr renamable $x12, killed renamable $x8, $xzr
; CHECK-NEXT: renamable $x9 = MOVaddr target-flags(aarch64-page) @g, target-flags(aarch64-pageoff, aarch64-nc) @g
; CHECK-NEXT: renamable $w8 = LDRWroX killed renamable $x9, killed renamable $x8, 0, 0 :: (load (s32) from %ir.arrayidx14)
; CHECK-NEXT: dead $wzr = SUBSWri killed renamable $w8, 1, 0, implicit-def $nzcv
; CHECK-NEXT: Bcc 11, %bb.8, implicit $nzcv
; CHECK-NEXT: B %bb.5
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5.if.then17:
; CHECK-NEXT: successors: %bb.7(0x80000000)
; CHECK-NEXT: liveins: $w10, $x2, $x12
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: STRWui killed renamable $w10, %stack.3, 0 :: (store (s32) into %stack.3)
; CHECK-NEXT: STRXui renamable $x12, %stack.4, 0 :: (store (s64) into %stack.4)
; CHECK-NEXT: renamable $w20 = LDRWroX killed renamable $x2, killed renamable $x12, 0, 1 :: (volatile load (s32) from %ir.arrayidx19)
; CHECK-NEXT: renamable $w19 = MOVi32imm 100
; CHECK-NEXT: B %bb.7
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.6.for.cond.cleanup:
; CHECK-NEXT: successors: %bb.8(0x80000000)
; CHECK-NEXT: liveins: $w20
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $w1 = COPY killed renamable $w20
; CHECK-NEXT: INLINEASM &"nop;nop", 1 /* sideeffect attdialect */, 12 /* clobber */, implicit-def dead early-clobber $x0, 12 /* clobber */, implicit-def dead early-clobber $x2, 12 /* clobber */, implicit-def dead early-clobber $x3, 12 /* clobber */, implicit-def dead early-clobber $x4, 12 /* clobber */, implicit-def dead early-clobber $x5, 12 /* clobber */, implicit-def dead early-clobber $x6, 12 /* clobber */, implicit-def dead early-clobber $x7, 12 /* clobber */, implicit-def dead early-clobber $x8, 12 /* clobber */, implicit-def dead early-clobber $x9, 12 /* clobber */, implicit-def dead early-clobber $x10, 12 /* clobber */, implicit-def dead early-clobber $x11, 12 /* clobber */, implicit-def dead early-clobber $x12, 12 /* clobber */, implicit-def dead early-clobber $x13, 12 /* clobber */, implicit-def dead early-clobber $x14, 12 /* clobber */, implicit-def dead early-clobber $x15, 12 /* clobber */, implicit-def dead early-clobber $x16, 12 /* clobber */, implicit-def dead early-clobber $x17, 12 /* clobber */, implicit-def dead early-clobber $x18, 12 /* clobber */, implicit-def dead early-clobber $x19, 12 /* clobber */, implicit-def dead early-clobber $x20, 12 /* clobber */, implicit-def dead early-clobber $x21, 12 /* clobber */, implicit-def dead early-clobber $x22, 12 /* clobber */, implicit-def dead early-clobber $x23, 12 /* clobber */, implicit-def dead early-clobber $x24, 12 /* clobber */, implicit-def dead early-clobber $x25, 12 /* clobber */, implicit-def dead early-clobber $x26, 12 /* clobber */, implicit-def dead early-clobber $x27, 12 /* clobber */, implicit-def dead early-clobber $x28, 12 /* clobber */, implicit-def dead early-clobber $fp, 12 /* clobber */, implicit-def dead early-clobber $lr
; CHECK-NEXT: renamable $x2 = LDRXui %stack.0, 0 :: (load (s64) from %stack.0)
; CHECK-NEXT: renamable $x12 = LDRXui %stack.4, 0 :: (load (s64) from %stack.4)
; CHECK-NEXT: STRWroX killed renamable $w1, renamable $x2, renamable $x12, 0, 1 :: (volatile store (s32) into %ir.sunkaddr1)
; CHECK-NEXT: renamable $w11 = LDRWui %stack.2, 0 :: (load (s32) from %stack.2)
; CHECK-NEXT: renamable $w10 = LDRWui %stack.3, 0 :: (load (s32) from %stack.3)
; CHECK-NEXT: B %bb.8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.7.for.body:
; CHECK-NEXT: successors: %bb.6(0x04000000), %bb.7(0x7c000000)
; CHECK-NEXT: liveins: $w19, $w20
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: ADJCALLSTACKDOWN 0, 0, implicit-def dead $sp, implicit $sp
; CHECK-NEXT: BL @callee, csr_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit-def $sp, implicit-def $w0
; CHECK-NEXT: ADJCALLSTACKUP 0, 0, implicit-def dead $sp, implicit $sp
; CHECK-NEXT: renamable $w19 = nsw SUBSWri killed renamable $w19, 1, 0, implicit-def $nzcv
; CHECK-NEXT: renamable $w20 = nsw ADDWrr killed renamable $w0, killed renamable $w20
; CHECK-NEXT: Bcc 0, %bb.6, implicit $nzcv
; CHECK-NEXT: B %bb.7
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.8.if.end:
; CHECK-NEXT: successors: %bb.9(0x04000000), %bb.4(0x7c000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2, $x12
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $w10 = nsw SUBSWri killed renamable $w10, 1, 0, implicit-def $nzcv
; CHECK-NEXT: renamable $x12 = nuw nsw ADDXri killed renamable $x12, 1, 0
; CHECK-NEXT: Bcc 1, %bb.4, implicit $nzcv
; CHECK-NEXT: B %bb.9
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.9.do.end:
; CHECK-NEXT: successors: %bb.10(0x80000000)
; CHECK-NEXT: liveins: $w11, $x2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $w10 = COPY $wzr
; CHECK-NEXT: renamable $x8 = LDRXui %stack.1, 0 :: (load (s64) from %stack.1)
; CHECK-NEXT: STRWroX $wzr, renamable $x2, renamable $x8, 0, 1 :: (volatile store (s32) into %ir.arrayidx26)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.10.if.end27:
; CHECK-NEXT: successors: %bb.11(0x04000000), %bb.1(0x7c000000)
; CHECK-NEXT: liveins: $w10, $w11, $x2, $x8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: renamable $w11 = nsw SUBSWri killed renamable $w11, 1, 0, implicit-def $nzcv
; CHECK-NEXT: renamable $x8 = nuw nsw ADDXri killed renamable $x8, 1, 0
; CHECK-NEXT: Bcc 1, %bb.1, implicit $nzcv
; CHECK-NEXT: B %bb.11
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.11.do.end32:
; CHECK-NEXT: RET_ReallyLR
bb.0.entry:
successors: %bb.1(0x80000000)
liveins: $x0, $x2
%19:gpr64common = COPY $x2
%17:gpr64common = COPY $x0
%54:gpr64common = COPY $xzr
%21:gpr32common = LDRWui %17, 0 :: (load (s32) from %ir.p)
%24:gpr32common = LDRWui %17, 1 :: (load (s32) from %ir.arrayidx2)
dead $wzr = SUBSWri %21, 1, 0, implicit-def $nzcv
%55:gpr32common = CSINCWr %21, $wzr, 12, implicit $nzcv
dead $wzr = SUBSWri %24, 1, 0, implicit-def $nzcv
%56:gpr32common = CSINCWr %24, $wzr, 12, implicit $nzcv
undef %29.sub_32:gpr64 = MOVi32imm 36
%31:gpr64common = MOVaddr target-flags(aarch64-page) @g, target-flags(aarch64-pageoff, aarch64-nc) @g
bb.1.do.body:
successors: %bb.2(0x50000000), %bb.11(0x30000000)
%30:gpr64 = MADDXrrr %54, %29, $xzr
%32:gpr32common = LDRWroX %31, %30, 0, 0 :: (load (s32) from %ir.arrayidx9)
dead $wzr = SUBSWri %32, 1, 0, implicit-def $nzcv
Bcc 10, %bb.2, implicit $nzcv
bb.11:
successors: %bb.9(0x80000000)
B %bb.9
bb.2.do.body12.preheader:
successors: %bb.3(0x80000000)
%57:gpr64common = COPY $xzr
bb.3.do.body12:
successors: %bb.4(0x50000000), %bb.7(0x30000000)
%38:gpr64 = MADDXrrr %57, %29, $xzr
%40:gpr32common = LDRWroX %31, %38, 0, 0 :: (load (s32) from %ir.arrayidx14)
dead $wzr = SUBSWri %40, 1, 0, implicit-def $nzcv
Bcc 11, %bb.7, implicit $nzcv
B %bb.4
bb.4.if.then17:
successors: %bb.6(0x80000000)
%60:gpr32 = LDRWroX %19, %57, 0, 1 :: (volatile load (s32) from %ir.arrayidx19)
%59:gpr32common = MOVi32imm 100
B %bb.6
bb.5.for.cond.cleanup:
successors: %bb.7(0x80000000)
INLINEASM &"nop;nop", 1 /* sideeffect attdialect */, 12 /* clobber */, implicit-def dead early-clobber $x0, 12 /* clobber */, implicit-def dead early-clobber $x2, 12 /* clobber */, implicit-def dead early-clobber $x3, 12 /* clobber */, implicit-def dead early-clobber $x4, 12 /* clobber */, implicit-def dead early-clobber $x5, 12 /* clobber */, implicit-def dead early-clobber $x6, 12 /* clobber */, implicit-def dead early-clobber $x7, 12 /* clobber */, implicit-def dead early-clobber $x8, 12 /* clobber */, implicit-def dead early-clobber $x9, 12 /* clobber */, implicit-def dead early-clobber $x10, 12 /* clobber */, implicit-def dead early-clobber $x11, 12 /* clobber */, implicit-def dead early-clobber $x12, 12 /* clobber */, implicit-def dead early-clobber $x13, 12 /* clobber */, implicit-def dead early-clobber $x14, 12 /* clobber */, implicit-def dead early-clobber $x15, 12 /* clobber */, implicit-def dead early-clobber $x16, 12 /* clobber */, implicit-def dead early-clobber $x17, 12 /* clobber */, implicit-def dead early-clobber $x18, 12 /* clobber */, implicit-def dead early-clobber $x19, 12 /* clobber */, implicit-def dead early-clobber $x20, 12 /* clobber */, implicit-def dead early-clobber $x21, 12 /* clobber */, implicit-def dead early-clobber $x22, 12 /* clobber */, implicit-def dead early-clobber $x23, 12 /* clobber */, implicit-def dead early-clobber $x24, 12 /* clobber */, implicit-def dead early-clobber $x25, 12 /* clobber */, implicit-def dead early-clobber $x26, 12 /* clobber */, implicit-def dead early-clobber $x27, 12 /* clobber */, implicit-def dead early-clobber $x28, 12 /* clobber */, implicit-def dead early-clobber $fp, 12 /* clobber */, implicit-def dead early-clobber $lr
STRWroX %60, %19, %57, 0, 1 :: (volatile store (s32) into %ir.sunkaddr1)
B %bb.7
bb.6.for.body:
successors: %bb.5(0x04000000), %bb.6(0x7c000000)
ADJCALLSTACKDOWN 0, 0, implicit-def dead $sp, implicit $sp
BL @callee, csr_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit-def $sp, implicit-def $w0
ADJCALLSTACKUP 0, 0, implicit-def dead $sp, implicit $sp
%45:gpr32 = COPY $w0
%59:gpr32common = nsw SUBSWri %59, 1, 0, implicit-def $nzcv
%60:gpr32 = nsw ADDWrr %45, %60
Bcc 0, %bb.5, implicit $nzcv
B %bb.6
bb.7.if.end:
successors: %bb.8(0x04000000), %bb.3(0x7c000000)
%56:gpr32common = nsw SUBSWri %56, 1, 0, implicit-def $nzcv
%57:gpr64common = nuw nsw ADDXri %57, 1, 0
Bcc 1, %bb.3, implicit $nzcv
B %bb.8
bb.8.do.end:
successors: %bb.9(0x80000000)
%56:gpr32common = COPY $wzr
STRWroX $wzr, %19, %54, 0, 1 :: (volatile store (s32) into %ir.arrayidx26)
bb.9.if.end27:
successors: %bb.10(0x04000000), %bb.1(0x7c000000)
%55:gpr32common = nsw SUBSWri %55, 1, 0, implicit-def $nzcv
%54:gpr64common = nuw nsw ADDXri %54, 1, 0
Bcc 1, %bb.1, implicit $nzcv
B %bb.10
bb.10.do.end32:
RET_ReallyLR
...
Reference in New Issue View Git Blame Copy Permalink