Re: [PATCH bpf-next 2/2] bpftool: show flow_dissector attachment status

From: Jakub Kicinski <jakub.kicinski@netronome.com>
To: Stanislav Fomichev <sdf@google.com>
Cc: netdev@vger.kernel.org, bpf@vger.kernel.org, davem@davemloft.net,
	ast@kernel.org, daniel@iogearbox.net,
	quentin.monnet@netronome.com
Subject: Re: [PATCH bpf-next 2/2] bpftool: show flow_dissector attachment status
Date: Tue, 23 Apr 2019 16:32:44 -0700	[thread overview]
Message-ID: <20190423163244.0a665bb0@cakuba.netronome.com> (raw)
In-Reply-To: <20190423232200.101627-2-sdf@google.com>

On Tue, 23 Apr 2019 16:22:00 -0700, Stanislav Fomichev wrote:
> +static int query_flow_dissector(struct bpf_attach_info *attach_info)
> +{
> +	__u32 prog_ids[1] = {0};
> +	__u32 prog_cnt = ARRAY_SIZE(prog_ids);
> +	__u32 attach_flags;
> +	int fd;
> +	int err;

reverse christmas tree, please :)

Move the init into the code if necessary.

> +	fd = open("/proc/self/ns/net", O_RDONLY);
> +	if (fd < 0) {
> +		p_err("can't open /proc/self/ns/net: %d",
> +		      strerror(errno));
> +		return -1;
> +	}
> +	err = bpf_prog_query(fd, BPF_FLOW_DISSECTOR, 0,
> +			     &attach_flags, prog_ids, &prog_cnt);
> +	close(fd);
> +	if (err) {
> +		if (errno == EINVAL) {
> +			/* Older kernel's don't support querying
> +			 * flow dissector programs.
> +			 */
> +			return 0;
> +		}
> +		p_err("can't query prog: %s", strerror(errno));
> +		return -1;
> +	}
> +
> +	if (prog_cnt == 1)
> +		attach_info->flow_dissector_id = prog_ids[0];

So the count can only be 0 or 1?  Hm.

> +	return 0;
> +}