netdev - Re: [syzbot] WARNING in bpf_verifier

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 9 Sep 2022 14:15:40 -0700
From:   Peilin Ye <yepeilin.cs@...il.com>
To:     sdf@...gle.com
Cc:     Benjamin Tissoires <benjamin.tissoires@...hat.com>,
        Andrii Nakryiko <andrii@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>, bpf <bpf@...r.kernel.org>,
        Daniel Borkmann <daniel@...earbox.net>, davem@...emloft.net,
        haoluo@...gle.com, hawk@...nel.org,
        John Fastabend <john.fastabend@...il.com>, jolsa@...nel.org,
        KP Singh <kpsingh@...nel.org>, kuba@...nel.org,
        lkml <linux-kernel@...r.kernel.org>, llvm@...ts.linux.dev,
        martin.lau@...ux.dev, nathan@...nel.org,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Networking <netdev@...r.kernel.org>, Song Liu <song@...nel.org>,
        syzkaller-bugs@...glegroups.com, Tom Rix <trix@...hat.com>,
        Yonghong Song <yhs@...com>,
        Peilin Ye <peilin.ye@...edance.com>,
        Peilin Ye <yepeilin.cs@...il.com>
Subject: Re: [syzbot] WARNING in bpf_verifier_vlog

Hi all,

On Fri, Sep 09, 2022 at 12:54:06PM -0700, sdf@...gle.com wrote:
> On 09/09, Benjamin Tissoires wrote:
> Yeah, good point. I've run the repro. I think the issue is that
> syzkaller is able to pass btf with a super long random name which
> then hits BPF_VERIFIER_TMP_LOG_SIZE while printing the verifier
> log line. Seems like a non-issue to me, but maybe we need to
> add some extra validation..

In btf_func_proto_check_meta():

	if (t->name_off) {
		btf_verifier_log_type(env, t, "Invalid name");
		return -EINVAL;
	}

In the verifier log, maybe we should just say that BTF_KIND_FUNC_PROTO "must
not have a name" [1], instead of printing out the user-provided
(potentially very long) name and say it's "Invalid" ?

Similarly, for name-too-long errors, should we truncate the name to
KSYM_NAME_LEN bytes (see __btf_name_valid()) in the log ?

[1] commit 2667a2626f4d ("bpf: btf: Add BTF_KIND_FUNC and BTF_KIND_FUNC_PROTO")

Thanks,
Peilin Ye