Received-SPF: pass (google.com: domain of linux-kernel+bounces-134293-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Subject: Re: [syzbot] [bpf?] KASAN: stack-out-of-bounds Read in hash
To: syzbot <syzbot+9459b5d7fab774cf182f@syzkaller.appspotmail.com>,
 bpf@vger.kernel.org, syzkaller-bugs@googlegroups.com, netdev@vger.kernel.org
References: <000000000000b97fba06156dc57b@google.com>
From: Hou Tao <houtao@huaweicloud.com>
Cc: andrii@kernel.org, ast@kernel.org, daniel@iogearbox.net,
 eddyz87@gmail.com, haoluo@google.com, john.fastabend@gmail.com,
 kpsingh@kernel.org, jolsa@kernel.org, martin.lau@linux.dev, sdf@google.com,
 song@kernel.org, yonghong.song@linux.dev, linux-kernel@vger.kernel.org
Message-ID: <b764fde2-cbf3-6446-d437-45af0964b062@huaweicloud.com>
Date: Sun, 7 Apr 2024 17:09:28 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.6.0
Precedence: bulk
MIME-Version: 1.0
In-Reply-To: <000000000000b97fba06156dc57b@google.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Content-Language: en-US

Hi,

On 4/6/2024 9:44 PM, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:
>
> HEAD commit:    443574b03387 riscv, bpf: Fix kfunc parameters incompatibil..
> git tree:       bpf
> console+strace: https://syzkaller.appspot.com/x/log.txt?x=148ad855180000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=6fb1be60a193d440
> dashboard link: https://syzkaller.appspot.com/bug?extid=9459b5d7fab774cf182f
> compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=13d86795180000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=143eff76180000

According to the reproducer, it passes a big value_size (0xfffffe00)
when creating bloom filter map. The big value_size bypasses the check in
check_stack_access_within_bounds(). I think a proper fix needs to add
these following two checks:
(1) in check_stack_access_within_bounds()  add check for negative
access_size
(2) in bloom_map_alloc() limit the max value of bloom_map_alloc().

Will post a patch to fix the syzbot report. Will also check whether or
not there are similar problems for other bpf maps.