Files
clang-p2996/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
Nikita Popov 90ba33099c [InstCombine] Canonicalize constant GEPs to i8 source element type (#68882)
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.
2024-01-24 15:25:29 +01:00

97 lines
4.9 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes="default<O1>" -S < %s | FileCheck --check-prefixes=O1 %s
; RUN: opt -passes="default<O2>" -S < %s | FileCheck --check-prefixes=O23 %s
; RUN: opt -passes="default<O3>" -S < %s | FileCheck --check-prefixes=O23 %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
; We should retain the TBAA on the load here, not lose it.
define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
; O1-LABEL: @licm(
; O1-NEXT: entry:
; O1-NEXT: [[CMP1_NOT:%.*]] = icmp eq i64 [[NUMELEM:%.*]], 0
; O1-NEXT: br i1 [[CMP1_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY_LR_PH:%.*]]
; O1: for.body.lr.ph:
; O1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_START_I:%.*]], align 8, !tbaa [[TBAA3:![0-9]+]]
; O1-NEXT: br label [[FOR_BODY:%.*]]
; O1: for.body:
; O1-NEXT: [[K_02:%.*]] = phi i64 [ 0, [[FOR_BODY_LR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[K_02]]
; O1-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[TBAA8:![0-9]+]]
; O1-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
; O1-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
; O1-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]]
; O1: for.cond.cleanup:
; O1-NEXT: ret void
;
; O23-LABEL: @licm(
; O23-NEXT: entry:
; O23-NEXT: [[CMP1_NOT:%.*]] = icmp eq i64 [[NUMELEM:%.*]], 0
; O23-NEXT: br i1 [[CMP1_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY_LR_PH:%.*]]
; O23: for.body.lr.ph:
; O23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_START_I:%.*]], align 8, !tbaa [[TBAA3:![0-9]+]]
; O23-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[NUMELEM]], 4
; O23-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY_PREHEADER:%.*]], label [[VECTOR_PH:%.*]]
; O23: vector.ph:
; O23-NEXT: [[N_VEC:%.*]] = and i64 [[NUMELEM]], -4
; O23-NEXT: br label [[VECTOR_BODY:%.*]]
; O23: vector.body:
; O23-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; O23-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[INDEX]]
; O23-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; O23-NEXT: store <2 x double> <double 2.000000e+00, double 2.000000e+00>, ptr [[TMP1]], align 8, !tbaa [[TBAA8:![0-9]+]]
; O23-NEXT: store <2 x double> <double 2.000000e+00, double 2.000000e+00>, ptr [[TMP2]], align 8, !tbaa [[TBAA8]]
; O23-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; O23-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; O23-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; O23: middle.block:
; O23-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[NUMELEM]]
; O23-NEXT: br i1 [[CMP_N]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY_PREHEADER]]
; O23: for.body.preheader:
; O23-NEXT: [[K_02_PH:%.*]] = phi i64 [ 0, [[FOR_BODY_LR_PH]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
; O23-NEXT: br label [[FOR_BODY:%.*]]
; O23: for.body:
; O23-NEXT: [[K_02:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[K_02_PH]], [[FOR_BODY_PREHEADER]] ]
; O23-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[K_02]]
; O23-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[TBAA8]]
; O23-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
; O23-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
; O23-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; O23: for.cond.cleanup:
; O23-NEXT: ret void
;
entry:
br label %for.cond
for.cond: ; preds = %for.body, %entry
%k.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
%cmp = icmp ult i64 %k.0, %numElem
br i1 %cmp, label %for.body, label %for.cond.cleanup
for.body: ; preds = %for.cond
%0 = load ptr, ptr %_M_start.i, align 8, !tbaa !3
%add.ptr.i = getelementptr inbounds double, ptr %0, i64 %k.0
store double 2.000000e+00, ptr %add.ptr.i, align 8, !tbaa !8
%inc = add nuw i64 %k.0, 1
br label %for.cond
for.cond.cleanup: ; preds = %for.cond
ret void
}
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"uwtable", i32 1}
!2 = !{!"clang version 15.0.0 (https://github.com/llvm/llvm-project.git fc510998f7c287df2bc1304673e0cd8452d50b31)"}
!3 = !{!4, !5, i64 0}
!4 = !{!"_ZTSNSt12_Vector_baseIdSaIdEE17_Vector_impl_dataE", !5, i64 0, !5, i64 8, !5, i64 16}
!5 = !{!"any pointer", !6, i64 0}
!6 = !{!"omnipotent char", !7, i64 0}
!7 = !{!"Simple C++ TBAA"}
!8 = !{!9, !9, i64 0}
!9 = !{!"double", !6, i64 0}