Re: [PATCH v2] iptables: open eBPF programs in read only mode

From: Jakub Kicinski <kuba@kernel.org>
To: "Maciej Żenczykowski" <zenczykowski@gmail.com>
Cc: "Maciej Żenczykowski" <maze@google.com>,
	"Pablo Neira Ayuso" <pablo@netfilter.org>,
	"Florian Westphal" <fw@strlen.de>,
	"Linux Network Development Mailing List" <netdev@vger.kernel.org>,
	"Netfilter Development Mailing List"
	<netfilter-devel@vger.kernel.org>,
	"Chenbo Feng" <fengc@google.com>,
	"Alexei Starovoitov" <ast@kernel.org>,
	"Willem de Bruijn" <willemb@google.com>
Subject: Re: [PATCH v2] iptables: open eBPF programs in read only mode
Date: Thu, 26 Mar 2020 11:30:48 -0700	[thread overview]
Message-ID: <20200326113048.250e7098@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com> (raw)
In-Reply-To: <20200326142803.239183-1-zenczykowski@gmail.com>

On Thu, 26 Mar 2020 07:28:03 -0700 Maciej Żenczykowski wrote:
> diff --git a/extensions/libxt_bpf.c b/extensions/libxt_bpf.c
> index 92958247..44cdd5cb 100644
> --- a/extensions/libxt_bpf.c
> +++ b/extensions/libxt_bpf.c
> @@ -61,11 +61,22 @@ static const struct xt_option_entry bpf_opts_v1[] = {
>  	XTOPT_TABLEEND,
>  };
>  
> -static int bpf_obj_get(const char *filepath)
> +static int bpf_obj_get_readonly(const char *filepath)
>  {
>  #if defined HAVE_LINUX_BPF_H && defined __NR_bpf && defined BPF_FS_MAGIC
>  	union bpf_attr attr;
> +	// file_flags && BPF_F_RDONLY requires Linux 4.15+ uapi kernel headers
> +#ifdef BPF_F_RDONLY

FWIW the BPF subsystem is about to break uAPI backward-compat and
replace the defines with enums. See commit 1aae4bdd7879 ("bpf: Switch
BPF UAPI #define constants used from BPF program side to enums").

> +	int fd;
>  
> +	memset(&attr, 0, sizeof(attr));
> +	attr.pathname = (__u64) filepath;
> +	attr.file_flags = BPF_F_RDONLY;
> +	fd = syscall(__NR_bpf, BPF_OBJ_GET, &attr, sizeof(attr));
> +	if (fd >= 0) return fd;
> +
> +	// on any error fallback to default R/W access for pre-4.15-rc1 kernels
> +#endif