e949b54a5b
Update LAA to use PSE::getSymbolicMaxBackedgeTakenCount which returns the minimum of the countable exits. When analyzing dependences and computing runtime checks, we need the smallest upper bound on the number of iterations. In terms of memory safety, it shouldn't matter if any uncomputable exits leave the loop, as long as we prove that there are no dependences given the minimum of the countable exits. The same should apply also for generating runtime checks. Note that this shifts the responsiblity of checking whether all exit counts are computable or handling early-exits to the users of LAA. Depends on https://github.com/llvm/llvm-project/pull/93498 PR: https://github.com/llvm/llvm-project/pull/93499
221 lines
6.7 KiB
LLVM
221 lines
6.7 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -passes='print<access-info>' -disable-output %s 2>&1 | FileCheck %s
|
|
|
|
define void @all_exits_dominate_latch_countable_exits_at_most_500_iterations(ptr %A, ptr %B) {
|
|
; CHECK-LABEL: 'all_exits_dominate_latch_countable_exits_at_most_500_iterations'
|
|
; CHECK-NEXT: loop.header:
|
|
; CHECK-NEXT: Memory dependences are safe with run-time checks
|
|
; CHECK-NEXT: Dependences:
|
|
; CHECK-NEXT: Run-time memory checks:
|
|
; CHECK-NEXT: Check 0:
|
|
; CHECK-NEXT: Comparing group ([[GRP1:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
; CHECK-NEXT: Against group ([[GRP2:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
; CHECK-NEXT: Grouped accesses:
|
|
; CHECK-NEXT: Group [[GRP1]]:
|
|
; CHECK-NEXT: (Low: %B High: (2000 + %B))
|
|
; CHECK-NEXT: Member: {%B,+,4}<nuw><%loop.header>
|
|
; CHECK-NEXT: Group [[GRP2]]:
|
|
; CHECK-NEXT: (Low: %A High: (2000 + %A))
|
|
; CHECK-NEXT: Member: {%A,+,4}<nuw><%loop.header>
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
|
|
; CHECK-NEXT: SCEV assumptions:
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Expressions re-written:
|
|
;
|
|
entry:
|
|
br label %loop.header
|
|
|
|
loop.header:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
|
|
%gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
%gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
%l = load i32, ptr %gep.A, align 4
|
|
store i32 0, ptr %gep.B, align 4
|
|
%cntable.c.1 = icmp ult i64 %iv, 1000
|
|
%iv.next = add nuw nsw i64 %iv, 1
|
|
br i1 %cntable.c.1, label %b2, label %e.1
|
|
|
|
b2:
|
|
%uncntable.c.0 = icmp eq i32 %l, 0
|
|
br i1 %uncntable.c.0, label %e.2, label %b3
|
|
|
|
b3:
|
|
%cntable.c.2 = icmp eq i64 %iv.next, 500
|
|
br i1 %cntable.c.2, label %cleanup4, label %latch
|
|
|
|
latch:
|
|
br label %loop.header
|
|
|
|
cleanup4:
|
|
ret void
|
|
|
|
e.1:
|
|
ret void
|
|
e.2:
|
|
ret void
|
|
|
|
}
|
|
|
|
|
|
|
|
define i32 @all_exits_dominate_latch_countable_exits_at_most_1000_iterations(ptr %A, ptr %B) {
|
|
; CHECK-LABEL: 'all_exits_dominate_latch_countable_exits_at_most_1000_iterations'
|
|
; CHECK-NEXT: loop.header:
|
|
; CHECK-NEXT: Memory dependences are safe with run-time checks
|
|
; CHECK-NEXT: Dependences:
|
|
; CHECK-NEXT: Run-time memory checks:
|
|
; CHECK-NEXT: Check 0:
|
|
; CHECK-NEXT: Comparing group ([[GRP3:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
; CHECK-NEXT: Against group ([[GRP4:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
; CHECK-NEXT: Grouped accesses:
|
|
; CHECK-NEXT: Group [[GRP3]]:
|
|
; CHECK-NEXT: (Low: %B High: (4004 + %B))
|
|
; CHECK-NEXT: Member: {%B,+,4}<nuw><%loop.header>
|
|
; CHECK-NEXT: Group [[GRP4]]:
|
|
; CHECK-NEXT: (Low: %A High: (4004 + %A))
|
|
; CHECK-NEXT: Member: {%A,+,4}<nuw><%loop.header>
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
|
|
; CHECK-NEXT: SCEV assumptions:
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Expressions re-written:
|
|
;
|
|
entry:
|
|
br label %loop.header
|
|
|
|
loop.header:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
|
|
%gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
%gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
%l = load i32, ptr %gep.A, align 4
|
|
store i32 0, ptr %gep.B, align 4
|
|
%cntable.c.1 = icmp ult i64 %iv, 1000
|
|
br i1 %cntable.c.1, label %b2, label %e.1
|
|
|
|
b2:
|
|
%uncntable.c.0 = icmp eq i32 %l, 0
|
|
br i1 %uncntable.c.0, label %e.2, label %b3
|
|
|
|
b3:
|
|
%iv.next = add nuw nsw i64 %iv, 1
|
|
%cntable.c.2 = icmp eq i64 %iv.next, 2000
|
|
br i1 %cntable.c.2, label %e.0, label %latch
|
|
|
|
latch:
|
|
br label %loop.header
|
|
|
|
e.0:
|
|
ret i32 0
|
|
|
|
e.1:
|
|
ret i32 1
|
|
|
|
e.2:
|
|
ret i32 2
|
|
}
|
|
|
|
|
|
define i32 @not_all_exits_dominate_latch(ptr %A, ptr %B) {
|
|
; CHECK-LABEL: 'not_all_exits_dominate_latch'
|
|
; CHECK-NEXT: loop.header:
|
|
; CHECK-NEXT: Report: could not determine number of loop iterations
|
|
; CHECK-NEXT: Dependences:
|
|
; CHECK-NEXT: Run-time memory checks:
|
|
; CHECK-NEXT: Grouped accesses:
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
|
|
; CHECK-NEXT: SCEV assumptions:
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Expressions re-written:
|
|
;
|
|
entry:
|
|
br label %loop.header
|
|
|
|
loop.header:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
|
|
%gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
%gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
%l = load i32, ptr %gep.A, align 4
|
|
store i32 0, ptr %gep.B, align 4
|
|
%cntable.c.1 = icmp ult i64 %iv, 1000
|
|
%iv.next = add nuw nsw i64 %iv, 1
|
|
br i1 %cntable.c.1, label %b2, label %latch
|
|
|
|
b2:
|
|
%uncntable.c.0 = icmp eq i32 %l, 0
|
|
br i1 %uncntable.c.0, label %e.2, label %b3
|
|
|
|
b3:
|
|
%cntable.c.2 = icmp eq i64 %iv.next, 2000
|
|
br i1 %cntable.c.2, label %e.0, label %latch
|
|
|
|
latch:
|
|
br label %loop.header
|
|
|
|
e.0:
|
|
ret i32 0
|
|
|
|
e.2:
|
|
ret i32 1
|
|
}
|
|
|
|
define i32 @b3_does_not_dominate_latch(ptr %A, ptr %B) {
|
|
; CHECK-LABEL: 'b3_does_not_dominate_latch'
|
|
; CHECK-NEXT: loop.header:
|
|
; CHECK-NEXT: Memory dependences are safe with run-time checks
|
|
; CHECK-NEXT: Dependences:
|
|
; CHECK-NEXT: Run-time memory checks:
|
|
; CHECK-NEXT: Check 0:
|
|
; CHECK-NEXT: Comparing group ([[GRP5:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
; CHECK-NEXT: Against group ([[GRP6:0x[0-9a-f]+]]):
|
|
; CHECK-NEXT: %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
; CHECK-NEXT: Grouped accesses:
|
|
; CHECK-NEXT: Group [[GRP5]]:
|
|
; CHECK-NEXT: (Low: %B High: (4004 + %B))
|
|
; CHECK-NEXT: Member: {%B,+,4}<nuw><%loop.header>
|
|
; CHECK-NEXT: Group [[GRP6]]:
|
|
; CHECK-NEXT: (Low: %A High: (4004 + %A))
|
|
; CHECK-NEXT: Member: {%A,+,4}<nuw><%loop.header>
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
|
|
; CHECK-NEXT: SCEV assumptions:
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Expressions re-written:
|
|
;
|
|
entry:
|
|
br label %loop.header
|
|
|
|
loop.header:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
|
|
%gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
|
|
%gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
|
|
%l = load i32, ptr %gep.A, align 4
|
|
store i32 0, ptr %gep.B, align 4
|
|
%cntable.c.1 = icmp ult i64 %iv, 1000
|
|
%iv.next = add nuw nsw i64 %iv, 1
|
|
br i1 %cntable.c.1, label %b2, label %e.1
|
|
|
|
b2:
|
|
%uncntable.c.0 = icmp eq i32 %l, 0
|
|
br i1 %uncntable.c.0, label %latch, label %b3
|
|
|
|
b3:
|
|
%cntable.c.2 = icmp eq i64 %iv.next, 500
|
|
br i1 %cntable.c.2, label %e.0, label %latch
|
|
|
|
latch:
|
|
br label %loop.header
|
|
|
|
e.0:
|
|
ret i32 0
|
|
|
|
e.1:
|
|
ret i32 1
|
|
}
|