c65b4d64d4
Alignment of an alloca in IR can be lower than the preferred alignment on purpose, but this override essentially treats the preferred alignment as the minimum alignment. The patch changes this behavior to always use the specified alignment. If alignment is not set explicitly in LLVM IR, it is set to DL.getPrefTypeAlign(Ty) in computeAllocaDefaultAlign. Tests are changed as well: explicit alignment is increased to match the preferred alignment if it changes output, or omitted when it is hard to determine the right value (e.g. for pointers, some structs, or weird types). Differential Revision: https://reviews.llvm.org/D135462
30 lines
1.0 KiB
LLVM
30 lines
1.0 KiB
LLVM
; RUN: llc < %s -O0 -wasm-disable-explicit-locals -wasm-keep-registers | FileCheck %s
|
|
|
|
; Regression test for bug 40172. The problem was that FastISel assumed
|
|
; that CmpInst results did not need to be zero extended because
|
|
; WebAssembly's compare instructions always return 0 or 1. But in this
|
|
; test case FastISel falls back to DAG ISel, which combines away the
|
|
; comparison, invalidating FastISel's assumption.
|
|
|
|
target triple = "wasm32-unknown-unknown"
|
|
|
|
; CHECK: i32.sub $[[BASE:[0-9]+]]=,
|
|
; CHECK: local.copy $[[ARG:[0-9]+]]=, $0{{$}}
|
|
; CHECK: i32.const $[[A0:[0-9]+]]=, 1{{$}}
|
|
; CHECK: i32.and $[[A1:[0-9]+]]=, $[[ARG]], $[[A0]]{{$}}
|
|
; CHECK: i32.store8 8($[[BASE]]), $[[A1]]{{$}}
|
|
|
|
define void @test(i8 %byte) {
|
|
%t = alloca { i8, i8 }, align 8
|
|
%x4 = and i8 %byte, 1
|
|
%x5 = icmp eq i8 %x4, 1
|
|
%x6 = and i8 %byte, 2
|
|
%x7 = icmp eq i8 %x6, 2
|
|
%x9 = zext i1 %x5 to i8
|
|
store i8 %x9, ptr %t, align 1
|
|
%x10 = getelementptr inbounds { i8, i8 }, ptr %t, i32 0, i32 1
|
|
%x11 = zext i1 %x7 to i8
|
|
store i8 %x11, ptr %x10, align 1
|
|
ret void
|
|
}
|