maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[ARM] Constant pool promotion - fix alignment calculation 

Global variables are GlobalValues, so they have explicit alignment. Querying
DataLayout for the alignment was incorrect.

Testcase added.

llvm-svn: 283423
This commit is contained in:
James Molloy 2016-10-06 07:56:00 +00:00
parent 78561c4917
commit 6215fad0e9
2 changed files with 11 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/Target/ARM/ARMISelLowering.cpp
View File

@ -3084,7 +3084,7 @@ static SDValue promoteToConstantPool(const GlobalValue *GV, SelectionDAG &DAG,
// that are strings for simplicity.
auto *CDAInit = dyn_cast<ConstantDataArray>(Init);
unsigned Size = DAG.getDataLayout().getTypeAllocSize(Init->getType());
unsigned Align = DAG.getDataLayout().getABITypeAlignment(Init->getType());
unsigned Align = GVar->getAlignment();
unsigned RequiredPadding = 4 - (Size % 4);
bool PaddingPossible =
RequiredPadding == 4 || (CDAInit && CDAInit->isString());

10
llvm/test/CodeGen/ARM/constantpool-promote.ll
View File

@ -15,6 +15,7 @@ target triple = "armv7--linux-gnueabihf"
@.arr2 = private unnamed_addr constant [2 x i16] [i16 7, i16 8], align 2
@.arr3 = private unnamed_addr constant [2 x i16*] [i16* null, i16* null], align 4
@.ptr = private unnamed_addr constant [2 x i16*] [i16* getelementptr inbounds ([2 x i16], [2 x i16]* @.arr2, i32 0, i32 0), i16* null], align 2
@.arr4 = private unnamed_addr constant [2 x i16] [i16 3, i16 4], align 16
; CHECK-LABEL: @test1
; CHECK: adr r0, [[x:.*]]
@ -125,6 +126,15 @@ entry:
ret void
}
; This shouldn't be promoted, as the global requires >4 byte alignment.
; CHECK-LABEL: @test9
; CHECK-NOT: adr
define void @test9() #0 {
tail call void @c(i16* getelementptr inbounds ([2 x i16], [2 x i16]* @.arr4, i32 0, i32 0)) #2
ret void
}
declare void @b(i8*) #1
declare void @c(i16*) #1
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture writeonly, i8* nocapture readonly, i32, i32, i1)