90ba33099c
This patch canonicalizes getelementptr instructions with constant indices to use the `i8` source element type. This makes it easier for optimizations to recognize that two GEPs are identical, because they don't need to see past many different ways to express the same offset. This is a first step towards https://discourse.llvm.org/t/rfc-replacing-getelementptr-with-ptradd/68699. This is limited to constant GEPs only for now, as they have a clear canonical form, while we're not yet sure how exactly to deal with variable indices. The test llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll gives two representative examples of the kind of optimization improvement we expect from this change. In the first test SimplifyCFG can now realize that all switch branches are actually the same. In the second test it can convert it into simple arithmetic. These are representative of common optimization failures we see in Rust. Fixes https://github.com/llvm/llvm-project/issues/69841.
125 lines
4.2 KiB
LLVM
125 lines
4.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt -O1 -S < %s | FileCheck %s
|
|
; RUN: opt -passes='default<O1>' -S < %s | FileCheck %s
|
|
|
|
; Don't simplify unconditional branches from empty blocks in simplifyCFG
|
|
; until late in the pipeline because it can destroy canonical loop structure.
|
|
|
|
define i1 @PR33605(i32 %a, i32 %b, ptr %c) {
|
|
; CHECK-LABEL: @PR33605(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[OR:%.*]] = or i32 [[B:%.*]], [[A:%.*]]
|
|
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 4
|
|
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp ne i32 [[OR]], [[TMP0]]
|
|
; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
|
|
; CHECK: if.then:
|
|
; CHECK-NEXT: store i32 [[OR]], ptr [[ARRAYIDX]], align 4
|
|
; CHECK-NEXT: tail call void @foo()
|
|
; CHECK-NEXT: br label [[IF_END]]
|
|
; CHECK: if.end:
|
|
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[C]], align 4
|
|
; CHECK-NEXT: [[CMP_1:%.*]] = icmp eq i32 [[OR]], [[TMP1]]
|
|
; CHECK-NEXT: br i1 [[CMP_1]], label [[IF_END_1:%.*]], label [[IF_THEN_1:%.*]]
|
|
; CHECK: if.then.1:
|
|
; CHECK-NEXT: store i32 [[OR]], ptr [[C]], align 4
|
|
; CHECK-NEXT: tail call void @foo()
|
|
; CHECK-NEXT: br label [[IF_END_1]]
|
|
; CHECK: if.end.1:
|
|
; CHECK-NEXT: [[CHANGED_1_OFF0_1:%.*]] = phi i1 [ true, [[IF_THEN_1]] ], [ [[CMP]], [[IF_END]] ]
|
|
; CHECK-NEXT: ret i1 [[CHANGED_1_OFF0_1]]
|
|
;
|
|
entry:
|
|
br label %for.cond
|
|
|
|
for.cond:
|
|
%i.0 = phi i32 [ 2, %entry ], [ %dec, %if.end ]
|
|
%changed.0.off0 = phi i1 [ false, %entry ], [ %changed.1.off0, %if.end ]
|
|
%dec = add nsw i32 %i.0, -1
|
|
%tobool = icmp eq i32 %i.0, 0
|
|
br i1 %tobool, label %for.cond.cleanup, label %for.body
|
|
|
|
for.cond.cleanup:
|
|
%changed.0.off0.lcssa = phi i1 [ %changed.0.off0, %for.cond ]
|
|
ret i1 %changed.0.off0.lcssa
|
|
|
|
for.body:
|
|
%or = or i32 %a, %b
|
|
%idxprom = sext i32 %dec to i64
|
|
%arrayidx = getelementptr inbounds i32, ptr %c, i64 %idxprom
|
|
%0 = load i32, ptr %arrayidx, align 4
|
|
%cmp = icmp eq i32 %or, %0
|
|
br i1 %cmp, label %if.end, label %if.then
|
|
|
|
if.then:
|
|
store i32 %or, ptr %arrayidx, align 4
|
|
call void @foo()
|
|
br label %if.end
|
|
|
|
if.end:
|
|
%changed.1.off0 = phi i1 [ true, %if.then ], [ %changed.0.off0, %for.body ]
|
|
br label %for.cond
|
|
}
|
|
|
|
declare void @foo()
|
|
|
|
; PR34603 - https://bugs.llvm.org/show_bug.cgi?id=34603
|
|
; We should have a select of doubles, not a select of double pointers.
|
|
; SimplifyCFG should not flatten this before early-cse has a chance to eliminate redundant ops.
|
|
|
|
define double @max_of_loads(ptr %x, ptr %y, i64 %i) {
|
|
; CHECK-LABEL: @max_of_loads(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[XI_PTR:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[I:%.*]]
|
|
; CHECK-NEXT: [[YI_PTR:%.*]] = getelementptr double, ptr [[Y:%.*]], i64 [[I]]
|
|
; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XI_PTR]], align 8
|
|
; CHECK-NEXT: [[YI:%.*]] = load double, ptr [[YI_PTR]], align 8
|
|
; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt double [[XI]], [[YI]]
|
|
; CHECK-NEXT: [[XI_YI:%.*]] = select i1 [[CMP]], double [[XI]], double [[YI]]
|
|
; CHECK-NEXT: ret double [[XI_YI]]
|
|
;
|
|
entry:
|
|
%xi_ptr = getelementptr double, ptr %x, i64 %i
|
|
%yi_ptr = getelementptr double, ptr %y, i64 %i
|
|
%xi = load double, ptr %xi_ptr
|
|
%yi = load double, ptr %yi_ptr
|
|
%cmp = fcmp ogt double %xi, %yi
|
|
br i1 %cmp, label %if, label %else
|
|
|
|
if:
|
|
%xi_ptr_again = getelementptr double, ptr %x, i64 %i
|
|
%xi_again = load double, ptr %xi_ptr_again
|
|
br label %end
|
|
|
|
else:
|
|
%yi_ptr_again = getelementptr double, ptr %y, i64 %i
|
|
%yi_again = load double, ptr %yi_ptr_again
|
|
br label %end
|
|
|
|
end:
|
|
%max = phi double [ %xi_again, %if ], [ %yi_again, %else ]
|
|
ret double %max
|
|
}
|
|
|
|
define i1 @PR58313(i1 %lhs, i1 %rhs) {
|
|
; CHECK-LABEL: @PR58313(
|
|
; CHECK-NEXT: andandend:
|
|
; CHECK-NEXT: [[SPEC_SELECT:%.*]] = xor i1 [[LHS:%.*]], [[RHS:%.*]]
|
|
; CHECK-NEXT: ret i1 [[SPEC_SELECT]]
|
|
;
|
|
andandend:
|
|
%0 = xor i1 %rhs, true
|
|
%andandval = select i1 %lhs, i1 %0, i1 false
|
|
br i1 %andandval, label %ororend, label %oror
|
|
|
|
oror:
|
|
%1 = xor i1 %lhs, true
|
|
%andandval3 = select i1 %1, i1 %rhs, i1 false
|
|
br label %ororend
|
|
|
|
ororend:
|
|
%ororval = phi i1 [ true, %andandend ], [ %andandval3, %oror ]
|
|
ret i1 %ororval
|
|
}
|
|
|