netdev - Re: [PATCH bpf-next 09/10] tools: bpftool: add simple perf event output reader

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:   Fri, 4 May 2018 23:53:34 +0200
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Jakub Kicinski <jakub.kicinski@...ronome.com>,
        alexei.starovoitov@...il.com
Cc:     oss-drivers@...ronome.com, netdev@...r.kernel.org
Subject: Re: [PATCH bpf-next 09/10] tools: bpftool: add simple perf event
 output reader

On 05/04/2018 03:37 AM, Jakub Kicinski wrote:
> Users of BPF sooner or later discover perf_event_output() helpers
> and BPF_MAP_TYPE_PERF_EVENT_ARRAY.  Dumping this array type is
> not possible, however, we can add simple reading of perf events.
> Create a new event_pipe subcommand for maps, this sub command
> will only work with BPF_MAP_TYPE_PERF_EVENT_ARRAY maps.
> 
> Parts of the code from samples/bpf/trace_output_user.c.
> 
> Signed-off-by: Jakub Kicinski <jakub.kicinski@...ronome.com>
> Reviewed-by: Quentin Monnet <quentin.monnet@...ronome.com>
[...]

One remark below:

[...]
> +static void
> +print_bpf_output(struct event_ring_info *ring, struct perf_event_sample *e)
> +{
> +	struct {
> +		struct perf_event_header header;
> +		__u64 id;
> +		__u64 lost;
> +	} *lost = (void *)e;
> +	struct timespec ts;
> +
> +	if (clock_gettime(CLOCK_MONOTONIC, &ts)) {
> +		perror("Can't read clock for timestamp");
> +		return;
> +	}
Instead of the timestamp above, probably better to pick it up via
PERF_SAMPLE_TIME which needs to be added to sample_type so it also
ends up in the RB. Given below you poll with 200 and you don't set
a wakeup event for perf RB (it's probably fine not to here, but it
can be done based on watermark or events), the clock_gettime() will
be off compared to when it was actually put into the RB.

> +	if (json_output) {
> +		jsonw_start_object(json_wtr);
> +		jsonw_name(json_wtr, "timestamp");
> +		jsonw_uint(json_wtr, ts.tv_sec * 1000000000ull + ts.tv_nsec);
> +		jsonw_name(json_wtr, "type");
> +		jsonw_uint(json_wtr, e->header.type);
> +		jsonw_name(json_wtr, "cpu");
> +		jsonw_uint(json_wtr, ring->cpu);
> +		jsonw_name(json_wtr, "index");
> +		jsonw_uint(json_wtr, ring->key);
> +		if (e->header.type == PERF_RECORD_SAMPLE) {
> +			jsonw_name(json_wtr, "data");
> +			print_data_json(e->data, e->size);
> +		} else if (e->header.type == PERF_RECORD_LOST) {
> +			jsonw_name(json_wtr, "lost");
> +			jsonw_start_object(json_wtr);
> +			jsonw_name(json_wtr, "id");
> +			jsonw_uint(json_wtr, lost->id);
> +			jsonw_name(json_wtr, "count");
> +			jsonw_uint(json_wtr, lost->lost);
> +			jsonw_end_object(json_wtr);
> +		}
> +		jsonw_end_object(json_wtr);
> +	} else {
> +		if (e->header.type == PERF_RECORD_SAMPLE) {
> +			printf("== @%ld.%ld CPU: %d index: %d =====\n",
> +			       (long)ts.tv_sec, ts.tv_nsec,
> +			       ring->cpu, ring->key);
> +			fprint_hex(stdout, e->data, e->size, " ");
> +			printf("\n");
> +		} else if (e->header.type == PERF_RECORD_LOST) {
> +			printf("lost %lld events\n", lost->lost);
> +		} else {
> +			printf("unknown event type=%d size=%d\n",
> +			       e->header.type, e->header.size);
> +		}