netdev - Re: [PATCH bpf-next 4/7] libbpf: sanitize BPF program code for bpf_probe_read_{kernel,user}[

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:   Wed, 19 Aug 2020 13:11:52 -0700
From:   Andrii Nakryiko <andrii.nakryiko@...il.com>
To:     Yonghong Song <yhs@...com>
Cc:     Andrii Nakryiko <andriin@...com>, bpf <bpf@...r.kernel.org>,
        Networking <netdev@...r.kernel.org>,
        Alexei Starovoitov <ast@...com>,
        Daniel Borkmann <daniel@...earbox.net>,
        Kernel Team <kernel-team@...com>
Subject: Re: [PATCH bpf-next 4/7] libbpf: sanitize BPF program code for bpf_probe_read_{kernel,user}[_str]

On Tue, Aug 18, 2020 at 6:42 PM Yonghong Song <yhs@...com> wrote:
>
>
>
> On 8/18/20 2:33 PM, Andrii Nakryiko wrote:
> > Add BPF program code sanitization pass, replacing calls to BPF
> > bpf_probe_read_{kernel,user}[_str]() helpers with bpf_probe_read[_str](), if
> > libbpf detects that kernel doesn't support new variants.
>
> I know this has been merged. The whole patch set looks good to me.
> A few nit or questions below.
>
> >
> > Signed-off-by: Andrii Nakryiko <andriin@...com>
> > ---
> >   tools/lib/bpf/libbpf.c | 80 ++++++++++++++++++++++++++++++++++++++++++
> >   1 file changed, 80 insertions(+)
> >
> > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > index ab0c3a409eea..bdc08f89a5c0 100644
> > --- a/tools/lib/bpf/libbpf.c
> > +++ b/tools/lib/bpf/libbpf.c
> > @@ -180,6 +180,8 @@ enum kern_feature_id {
> >       FEAT_ARRAY_MMAP,
> >       /* kernel support for expected_attach_type in BPF_PROG_LOAD */
> >       FEAT_EXP_ATTACH_TYPE,
> > +     /* bpf_probe_read_{kernel,user}[_str] helpers */
> > +     FEAT_PROBE_READ_KERN,
> >       __FEAT_CNT,
> >   };
> >
> > @@ -3591,6 +3593,27 @@ static int probe_kern_exp_attach_type(void)
> >       return probe_fd(bpf_load_program_xattr(&attr, NULL, 0));
> >   }
> >
> [...]
> >
> > +static bool insn_is_helper_call(struct bpf_insn *insn, enum bpf_func_id *func_id)
> > +{
> > +     __u8 class = BPF_CLASS(insn->code);
> > +
> > +     if ((class == BPF_JMP || class == BPF_JMP32) &&
>
> Do we support BPF_JMP32 + BPF_CALL ... as a helper call?
> I am not aware of this.

Verifier seems to support both. Check do_check in
kernel/bpf/verifier.c, around line 9000. So I decided to also support
it, even if Clang doesn't emit it (yet?).

>
> > +         BPF_OP(insn->code) == BPF_CALL &&
> > +         BPF_SRC(insn->code) == BPF_K &&
> > +         insn->src_reg == 0 && insn->dst_reg == 0) {
> > +                 if (func_id)
> > +                         *func_id = insn->imm;
>
> looks like func_id is always non-NULL. Unless this is to support future
> usage where func_id may be NULL, the above condition probably not needed.

Yeah, not sure why I assumed it might be optional, maybe the first
version of the code used to pass NULL in some other place. But I think
it's fine, this is a generic helper function that might be used later
as well. So I'd just keep it as is, it doesn't hurt.

>
> > +                 return true;
> > +     }
> > +     return false;
> > +}
> > +
> [...]