1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 19:23:23 +01:00
llvm-mirror/test/CodeGen/AArch64/misched-stp.ll
Nirav Dave 3633380341 Elide stores which are overwritten without being observed.
Summary:
In SelectionDAG, when a store is immediately chained to another store
to the same address, elide the first store as it has no observable
effects. This is causes small improvements dealing with intrinsics
lowered to stores.

Test notes:

* Many testcases overwrite store addresses multiple times and needed
  minor changes, mainly making stores volatile to prevent the
  optimization from optimizing the test away.

* Many X86 test cases optimized out instructions associated with
  associated with va_start.

* Note that test_splat in CodeGen/AArch64/misched-stp.ll no longer has
  dependencies to check and can probably be removed and potentially
  replaced with another test.

Reviewers: rnk, john.brawn

Subscribers: aemerson, rengolin, qcolombet, jyknight, nemanjai, nhaehnle, javed.absar, llvm-commits

Differential Revision: https://reviews.llvm.org/D33206

llvm-svn: 303198
2017-05-16 19:43:56 +00:00

51 lines
1.9 KiB
LLVM

; REQUIRES: asserts
; RUN: llc < %s -mtriple=aarch64 -mcpu=cyclone -mattr=+use-aa -enable-misched -verify-misched -o - | FileCheck %s
; Tests to check that the scheduler dependencies derived from alias analysis are
; correct when we have loads that have been split up so that they can later be
; merged into STP.
; Now that overwritten stores are elided in SelectionDAG, dependencies
; are resolved and removed before MISCHED. Check that we have
; equivalent pair of stp calls as a baseline.
; CHECK-LABEL: test_splat
; CHECK: ldr [[REG:w[0-9]+]], [x2]
; CHECK-DAG: stp w0, [[REG]], [x2, #12]
; CHECK-DAG: stp [[REG]], w1, [x2, #4]
define void @test_splat(i32 %x, i32 %y, i32* %p) {
entry:
%val = load i32, i32* %p, align 4
%0 = getelementptr inbounds i32, i32* %p, i64 1
%1 = getelementptr inbounds i32, i32* %p, i64 2
%2 = getelementptr inbounds i32, i32* %p, i64 3
%vec0 = insertelement <4 x i32> undef, i32 %val, i32 0
%vec1 = insertelement <4 x i32> %vec0, i32 %val, i32 1
%vec2 = insertelement <4 x i32> %vec1, i32 %val, i32 2
%vec3 = insertelement <4 x i32> %vec2, i32 %val, i32 3
%3 = bitcast i32* %0 to <4 x i32>*
store <4 x i32> %vec3, <4 x i32>* %3, align 4
store i32 %x, i32* %2, align 4
store i32 %y, i32* %1, align 4
ret void
}
declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1)
%struct.tree_common = type { i8*, i8*, i32 }
; CHECK-LABEL: test_zero
; CHECK-DAG: stp x2, xzr, [x0, #8]
; CHECK-DAG: str w1, [x0, #16]
; CHECK-DAG: str xzr, [x0]
define void @test_zero(%struct.tree_common* %t, i32 %code, i8* %type) {
entry:
%0 = bitcast %struct.tree_common* %t to i8*
tail call void @llvm.memset.p0i8.i64(i8* %0, i8 0, i64 24, i32 8, i1 false)
%code1 = getelementptr inbounds %struct.tree_common, %struct.tree_common* %t, i64 0, i32 2
store i32 %code, i32* %code1, align 8
%type2 = getelementptr inbounds %struct.tree_common, %struct.tree_common* %t, i64 0, i32 1
store i8* %type, i8** %type2, align 8
ret void
}