mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2025-01-12 17:32:19 +00:00
Fill in some holes in ScalarEvolution's loop iteration condition
analysis. This allows indvars to emit a simpler loop trip count expression. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@76085 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
parent
358ab1dbb0
commit
850f791abc
@ -960,6 +960,22 @@ const SCEV *ScalarEvolution::getSignExtendExpr(const SCEV *Op,
|
||||
return getAddRecExpr(getSignExtendExpr(Start, Ty),
|
||||
getSignExtendExpr(Step, Ty),
|
||||
L);
|
||||
|
||||
// Similar to above, only this time treat the step value as unsigned.
|
||||
// This covers loops that count up with an unsigned step.
|
||||
const SCEV *UMul =
|
||||
getMulExpr(CastedMaxBECount,
|
||||
getTruncateOrZeroExtend(Step, Start->getType()));
|
||||
Add = getAddExpr(Start, UMul);
|
||||
OperandExtendedAdd =
|
||||
getAddExpr(getZeroExtendExpr(Start, WideTy),
|
||||
getMulExpr(getZeroExtendExpr(CastedMaxBECount, WideTy),
|
||||
getZeroExtendExpr(Step, WideTy)));
|
||||
if (getZeroExtendExpr(Add, WideTy) == OperandExtendedAdd)
|
||||
// Return the expression with the addrec on the outside.
|
||||
return getAddRecExpr(getSignExtendExpr(Start, Ty),
|
||||
getZeroExtendExpr(Step, Ty),
|
||||
L);
|
||||
}
|
||||
|
||||
// If the backedge is guarded by a comparison with the pre-inc value
|
||||
@ -4248,7 +4264,7 @@ bool ScalarEvolution::isKnownPredicate(ICmpInst::Predicate Pred,
|
||||
|
||||
switch (Pred) {
|
||||
default:
|
||||
assert(0 && "Unexpected ICmpInst::Predicate value!");
|
||||
llvm_unreachable("Unexpected ICmpInst::Predicate value!");
|
||||
break;
|
||||
case ICmpInst::ICMP_SGT:
|
||||
Pred = ICmpInst::ICMP_SLT;
|
||||
@ -4556,23 +4572,32 @@ ScalarEvolution::isNecessaryCondOperands(ICmpInst::Predicate Pred,
|
||||
const SCEV *FoundLHS,
|
||||
const SCEV *FoundRHS) {
|
||||
switch (Pred) {
|
||||
default: break;
|
||||
default: llvm_unreachable("Unexpected ICmpInst::Predicate value!");
|
||||
case ICmpInst::ICMP_EQ:
|
||||
case ICmpInst::ICMP_NE:
|
||||
if (HasSameValue(LHS, FoundLHS) && HasSameValue(RHS, FoundRHS))
|
||||
return true;
|
||||
break;
|
||||
case ICmpInst::ICMP_SLT:
|
||||
case ICmpInst::ICMP_SLE:
|
||||
if (isKnownPredicate(ICmpInst::ICMP_SLE, LHS, FoundLHS) &&
|
||||
isKnownPredicate(ICmpInst::ICMP_SGE, RHS, FoundRHS))
|
||||
return true;
|
||||
break;
|
||||
case ICmpInst::ICMP_SGT:
|
||||
case ICmpInst::ICMP_SGE:
|
||||
if (isKnownPredicate(ICmpInst::ICMP_SGE, LHS, FoundLHS) &&
|
||||
isKnownPredicate(ICmpInst::ICMP_SLE, RHS, FoundRHS))
|
||||
return true;
|
||||
break;
|
||||
case ICmpInst::ICMP_ULT:
|
||||
case ICmpInst::ICMP_ULE:
|
||||
if (isKnownPredicate(ICmpInst::ICMP_ULE, LHS, FoundLHS) &&
|
||||
isKnownPredicate(ICmpInst::ICMP_UGE, RHS, FoundRHS))
|
||||
return true;
|
||||
break;
|
||||
case ICmpInst::ICMP_UGT:
|
||||
case ICmpInst::ICMP_UGE:
|
||||
if (isKnownPredicate(ICmpInst::ICMP_UGE, LHS, FoundLHS) &&
|
||||
isKnownPredicate(ICmpInst::ICMP_ULE, RHS, FoundRHS))
|
||||
return true;
|
||||
|
38
test/Transforms/IndVarSimplify/lftr-promote.ll
Normal file
38
test/Transforms/IndVarSimplify/lftr-promote.ll
Normal file
@ -0,0 +1,38 @@
|
||||
; RUN: llvm-as < %s | opt -indvars | llvm-dis | grep add | count 1
|
||||
|
||||
; Indvars should be able to compute the exit value of this loop
|
||||
; without any additional arithmetic. The only add needed should
|
||||
; be the canonical IV increment.
|
||||
|
||||
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
|
||||
|
||||
define void @foo(double* %p, i32 %n) nounwind {
|
||||
entry:
|
||||
%0 = icmp sgt i32 %n, 0 ; <i1> [#uses=1]
|
||||
br i1 %0, label %bb.nph, label %return
|
||||
|
||||
bb.nph: ; preds = %entry
|
||||
br label %bb2
|
||||
|
||||
bb2: ; preds = %bb3, %bb.nph
|
||||
%i.01 = phi i32 [ %7, %bb3 ], [ 0, %bb.nph ] ; <i32> [#uses=3]
|
||||
%1 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
||||
%2 = getelementptr double* %p, i64 %1 ; <double*> [#uses=1]
|
||||
%3 = load double* %2, align 8 ; <double> [#uses=1]
|
||||
%4 = fmul double %3, 1.100000e+00 ; <double> [#uses=1]
|
||||
%5 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
||||
%6 = getelementptr double* %p, i64 %5 ; <double*> [#uses=1]
|
||||
store double %4, double* %6, align 8
|
||||
%7 = add i32 %i.01, 1 ; <i32> [#uses=2]
|
||||
br label %bb3
|
||||
|
||||
bb3: ; preds = %bb2
|
||||
%8 = icmp slt i32 %7, %n ; <i1> [#uses=1]
|
||||
br i1 %8, label %bb2, label %bb3.return_crit_edge
|
||||
|
||||
bb3.return_crit_edge: ; preds = %bb3
|
||||
br label %return
|
||||
|
||||
return: ; preds = %bb3.return_crit_edge, %entry
|
||||
ret void
|
||||
}
|
Loading…
x
Reference in New Issue
Block a user