https://llvm.org/bugs/show_bug.cgi?id=25320
Bug ID: 25320
Summary: [X86] [AVX2] incorrect code generated for insert
element
Product: new-bugs
Version: trunk
Hardware: PC
OS: All
Status: NEW
Severity: normal
Priority: P
Component: new bugs
Assignee: unassignedb...@nondot.org
Reporter: koffieya...@gmail.com
CC: llvm-bugs@lists.llvm.org
Classification: Unclassified
This is a follow up to PR24935, but since the is related to insert element, I'm
opening a new issue.
The following:
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
define void @__sample_test(<2 x i32>* %in, <8 x i32>* %out) {
body:
%0 = load <2 x i32>, <2 x i32>* %in, align 8
%1 = extractelement <2 x i32> %0, i64 0
%2 = insertelement <8 x i32> <i32 undef, i32 0, i32 0, i32 0, i32 0, i32 0,
i32 0, i32 0>, i32 %1, i64 0
store <8 x i32> %2, <8 x i32>* %out, align 32
ret void
}
Generates wrong code when AVX2 is enabled (a non-zero value occurs both in the
0th and in the 2nd position of the output vector due to the use of vpmovzxdq):
$llc test.ll -o - -mattr=+avx2
.text
.file "test.ll"
.globl __sample_test
.align 16, 0x90
.type __sample_test,@function
__sample_test: # @__sample_test
.cfi_startproc
# BB#0: # %body
vpmovzxdq (%rdi), %xmm0 # xmm0 = mem[0],zero,mem[1],zero
vmovdqa %xmm0, %xmm0
vmovdqa %ymm0, (%rsi)
vzeroupper
retq
.Lfunc_end0:
.size __sample_test, .Lfunc_end0-__sample_test
.cfi_endproc
.section ".note.GNU-stack","",@progbits
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
