linux-kernel - Re: [PATCH -next 1/2] perf stat: Increase perf_attr

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] [day] [month] [year] [list]

Message-ID: <ZvTgHKl4eZvpyVml@google.com>
Date: Wed, 25 Sep 2024 21:16:28 -0700
From: Namhyung Kim <namhyung@...nel.org>
To: Tengda Wu <wutengda@...weicloud.com>
Cc: Peter Zijlstra <peterz@...radead.org>, song@...nel.org,
	Ingo Molnar <mingo@...hat.com>,
	Arnaldo Carvalho de Melo <acme@...nel.org>,
	Mark Rutland <mark.rutland@....com>,
	Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
	Jiri Olsa <jolsa@...nel.org>, Ian Rogers <irogers@...gle.com>,
	Adrian Hunter <adrian.hunter@...el.com>, kan.liang@...ux.intel.com,
	linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
	bpf@...r.kernel.org
Subject: Re: [PATCH -next 1/2] perf stat: Increase perf_attr_map entries

On Wed, Sep 25, 2024 at 01:55:22PM +0000, Tengda Wu wrote:
> bperf restricts the size of perf_attr_map's entries to 16, which
> cannot hold all events in many scenarios. A typical example is
> when the user specifies `-a -ddd` ([0]). And in other cases such as
> top-down analysis, which often requires a set of more than 16 PMUs
> to be collected simultaneously.
> 
> Fix this by increase perf_attr_map entries to 100, and an event
> number check has been introduced when bperf__load() to ensure that
> users receive a more friendly prompt when the event limit is reached.
> 
>   [0] https://lore.kernel.org/all/20230104064402.1551516-3-namhyung@kernel.org/

Apparently this patch was never applied.  I don't know how much you need
but having too many events at the same time won't be very useful because
multiplexing could reduce the accuracy.

Thanks,
Namhyung

> 
> Fixes: 7fac83aaf2ee ("perf stat: Introduce 'bperf' to share hardware PMCs with BPF")
> Signed-off-by: Tengda Wu <wutengda@...weicloud.com>
> ---
>  tools/perf/util/bpf_counter.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/perf/util/bpf_counter.c b/tools/perf/util/bpf_counter.c
> index 7a8af60e0f51..3346129c20cf 100644
> --- a/tools/perf/util/bpf_counter.c
> +++ b/tools/perf/util/bpf_counter.c
> @@ -28,7 +28,7 @@
>  #include "bpf_skel/bperf_leader.skel.h"
>  #include "bpf_skel/bperf_follower.skel.h"
>  
> -#define ATTR_MAP_SIZE 16
> +#define ATTR_MAP_SIZE 100
>  
>  static inline void *u64_to_ptr(__u64 ptr)
>  {
> @@ -451,6 +451,12 @@ static int bperf__load(struct evsel *evsel, struct target *target)
>  	enum bperf_filter_type filter_type;
>  	__u32 filter_entry_cnt, i;
>  
> +	if (evsel->evlist->core.nr_entries > ATTR_MAP_SIZE) {
> +		pr_err("Too many events, please limit to %d or less\n",
> +			ATTR_MAP_SIZE);
> +		return -1;
> +	}
> +
>  	if (bperf_check_target(evsel, target, &filter_type, &filter_entry_cnt))
>  		return -1;
>  
> -- 
> 2.34.1
>