netdev - Re: [RFC bpf-next 00/16] bpf: Speed up trampoline attach

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]

Message-ID: <20201025190147.GA2681365@krava>
Date:   Sun, 25 Oct 2020 20:01:47 +0100
From:   Jiri Olsa <jolsa@...hat.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Jiri Olsa <jolsa@...nel.org>, Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Andrii Nakryiko <andriin@...com>, netdev@...r.kernel.org,
        bpf@...r.kernel.org, Martin KaFai Lau <kafai@...com>,
        Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
        John Fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...omium.org>, Daniel Xu <dxu@...uu.xyz>,
        Jesper Brouer <jbrouer@...hat.com>,
        Toke Høiland-Jørgensen <toke@...hat.com>,
        Viktor Malik <vmalik@...hat.com>
Subject: Re: [RFC bpf-next 00/16] bpf: Speed up trampoline attach

On Fri, Oct 23, 2020 at 09:50:20AM -0400, Steven Rostedt wrote:

SNIP

> Is there something to keep an eBPF program from tracing a function with 6
> args? If the program saves only 5 args, but traces a function that has 6
> args, then the tracing program may end up using the register that the 6
> argument is in, and corrupting it.
> 
> I'm looking at bpf/trampoline.c, that has:
> 
> 	arch_prepare_bpf_trampoline(new_image, ...)
> 
> and that new_image is passed into:
> 
> 	register_ftrace_direct(ip, new_addr);
> 
> where new_addr == new_image.
> 
> And I don't see anywhere in the creating on that new_image that saves the
> 6th parameter.

  arch_prepare_bpf_trampoline
    ...
    save_regs(m, &prog, nr_args, stack_size);

> 
> The bpf program calls some helper functions which are allowed to clobber
> %r9 (where the 6th parameter is stored on x86_64). That means, when it
> returns to the function it traced, the 6th parameter is no longer correct.
> 
> At a minimum, direct callers must save all the parameters used by the
> function, not just what the eBPF code may use.
> 
> > 
> > > 
> > > The code in question is this:
> > > 
> > > int btf_distill_func_proto(struct bpf_verifier_log *log,
> > > 			   struct btf *btf,
> > > 			   const struct btf_type *func,
> > > 			   const char *tname,
> > > 			   struct btf_func_model *m)
> > > {
> > > 	const struct btf_param *args;
> > > 	const struct btf_type *t;
> > > 	u32 i, nargs;
> > > 	int ret;
> > > 
> > > 	if (!func) {
> > > 		/* BTF function prototype doesn't match the verifier types.
> > > 		 * Fall back to 5 u64 args.
> > > 		 */
> > > 		for (i = 0; i < 5; i++)
> > > 			m->arg_size[i] = 8;
> > > 		m->ret_size = 8;
> > > 		m->nr_args = 5;
> > > 		return 0;
> > > 	}

the fallback code in btf_distill_func_proto you're reffering to
is for case of tracing another ebpf program, when hooking to
kernel function, all args are used with no fallback to 5 args

I'm not sure what are the rules wrt args count when tracing
another ebpf program

jirka