netdev - Re: [PATCH v2 bpf-next 2/6] bpf: propagate poke descriptors to subprograms

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: <29a3dcfc-9d85-c113-19d2-e33f80ce5430@iogearbox.net>
Date:   Wed, 22 Jul 2020 16:40:42 +0200
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Maciej Fijalkowski <maciej.fijalkowski@...el.com>, ast@...nel.org
Cc:     bpf@...r.kernel.org, netdev@...r.kernel.org, bjorn.topel@...el.com,
        magnus.karlsson@...el.com
Subject: Re: [PATCH v2 bpf-next 2/6] bpf: propagate poke descriptors to
 subprograms

On 7/21/20 1:53 PM, Maciej Fijalkowski wrote:
> Previously, there was no need for poke descriptors being present in
> subprogram's bpf_prog_aux struct since tailcalls were simply not allowed
> in them. Each subprog is JITed independently so in order to enable
> JITing such subprograms, simply copy poke descriptors from main program
> to subprogram's poke tab.
> 
> Add also subprog's aux struct to the BPF map poke_progs list by calling
> on it map_poke_track().
> 
> Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
> ---
>   kernel/bpf/verifier.c | 20 ++++++++++++++++++++
>   1 file changed, 20 insertions(+)
> 
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 3c1efc9d08fd..3428edf85220 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -9936,6 +9936,9 @@ static int jit_subprogs(struct bpf_verifier_env *env)
>   		goto out_undo_insn;
>   
>   	for (i = 0; i < env->subprog_cnt; i++) {
> +		struct bpf_map *map_ptr;
> +		int j;
> +
>   		subprog_start = subprog_end;
>   		subprog_end = env->subprog_info[i + 1].start;
>   
> @@ -9960,6 +9963,23 @@ static int jit_subprogs(struct bpf_verifier_env *env)
>   		func[i]->aux->btf = prog->aux->btf;
>   		func[i]->aux->func_info = prog->aux->func_info;
>   
> +		for (j = 0; j < prog->aux->size_poke_tab; j++) {
> +			int ret;
> +
> +			ret = bpf_jit_add_poke_descriptor(func[i],
> +							  &prog->aux->poke_tab[j]);
> +			if (ret < 0) {
> +				verbose(env, "adding tail call poke descriptor failed\n");
> +				goto out_free;
> +			}
> +			map_ptr = func[i]->aux->poke_tab[j].tail_call.map;
> +			ret = map_ptr->ops->map_poke_track(map_ptr, func[i]->aux);
> +			if (ret < 0) {
> +				verbose(env, "tracking tail call prog failed\n");
> +				goto out_free;
> +			}

Hmm, I don't think this is correct/complete. If some of these have been registered or
if later on the JIT'ing fails but the subprog is already exposed to the prog array then
it's /public/ at this point, so a later bpf_jit_free() in out_free will rip them mem
while doing live patching on prog updates leading to UAF.

> +		}
> +
>   		/* Use bpf_prog_F_tag to indicate functions in stack traces.
>   		 * Long term would need debug info to populate names
>   		 */
>