| 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
| ||
|
Date: Fri, 14 Aug 2020 08:53:18 -0300
From: Arnaldo Carvalho de Melo <acme@...nel.org>
To: Changbin Du <changbin.du@...il.com>
Cc: Jiri Olsa <jolsa@...hat.com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Namhyung Kim <namhyung@...nel.org>,
Steven Rostedt <rostedt@...dmis.org>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v8 04/18] perf ftrace: add option '-m/--buffer-size' to
set per-cpu buffer size
Em Sat, Aug 08, 2020 at 10:31:27AM +0800, Changbin Du escreveu:
> This adds an option '-m/--buffer-size' to allow us set the size of per-cpu
> tracing buffer.
>
> Signed-off-by: Changbin Du <changbin.du@...il.com>
So this is a little bit confusing, i.e. if I don't specify a suffix (B,
K, M, G) I get:
# perf ftrace -m 2048
Usage: perf ftrace [<options>] [<command>]
or: perf ftrace [<options>] -- <command> [<options>]
-m, --buffer-size <size>
size of per cpu buffer
#
Which is not very helpful, and you also forgot to add an entry to 'man
perf-ftrace' (tools/perf/Documentation/perf-ftrace.txt), so only by
looking at the patch I was able to figure out that a suffix is
required.
Now look how 'perf trace' works:
# perf trace -m 2048 -e open* --max-events 1
0.772 sh/343169 openat(dfd: CWD, filename: /etc/ld.so.cache, flags: RDONLY|CLOEXEC) = 3
#
I.e. it doesn't require a unit suffix, which is more natural.
# perf trace -h -m
-m, --mmap-pages <pages>
number of mmap data pages
#
It indicates that the argument is in units of 'pages', while you don't
state that in:
$ perf ftrace -h -m
-m, --buffer-size <size>
size of per cpu buffer
$
So please improve the option description and add its longer explanation
in the man page, also I suggest you interpret:
# perf ftrace -m 2048
as asking for 2048K since the file as _kb in its name, so people used to
ftrace will get what they ask for, i.e. N kb. Please add this to the one
line description in 'perf ftrace -h', namely that the default unit os
'K'.
All this can be made on top of this series, so I'm applying it now to
make progress.
- Arnaldo
> ---
> v2: support units as a suffix.
> ---
> tools/perf/Documentation/perf-ftrace.txt | 5 +++
> tools/perf/builtin-ftrace.c | 55 ++++++++++++++++++++++++
> 2 files changed, 60 insertions(+)
>
> diff --git a/tools/perf/Documentation/perf-ftrace.txt b/tools/perf/Documentation/perf-ftrace.txt
> index 4f5628445a63..7a5d915f60b0 100644
> --- a/tools/perf/Documentation/perf-ftrace.txt
> +++ b/tools/perf/Documentation/perf-ftrace.txt
> @@ -53,6 +53,11 @@ OPTIONS
> Ranges of CPUs are specified with -: 0-2.
> Default is to trace on all online CPUs.
>
> +-m::
> +--buffer-size::
> + Set the size of per-cpu tracing buffer, <size> is expected to
> + be a number with appended unit character - B/K/M/G.
> +
> -T::
> --trace-funcs=::
> Select function tracer and set function filter on the given
> diff --git a/tools/perf/builtin-ftrace.c b/tools/perf/builtin-ftrace.c
> index 4b3fcee5725a..a3a4f4be9dde 100644
> --- a/tools/perf/builtin-ftrace.c
> +++ b/tools/perf/builtin-ftrace.c
> @@ -26,6 +26,7 @@
> #include "thread_map.h"
> #include "util/cap.h"
> #include "util/config.h"
> +#include "util/units.h"
>
> #define DEFAULT_TRACER "function_graph"
>
> @@ -39,6 +40,7 @@ struct perf_ftrace {
> struct list_head graph_funcs;
> struct list_head nograph_funcs;
> int graph_depth;
> + unsigned long percpu_buffer_size;
> };
>
> struct filter_entry {
> @@ -324,6 +326,21 @@ static int set_tracing_depth(struct perf_ftrace *ftrace)
> return 0;
> }
>
> +static int set_tracing_percpu_buffer_size(struct perf_ftrace *ftrace)
> +{
> + int ret;
> +
> + if (ftrace->percpu_buffer_size == 0)
> + return 0;
> +
> + ret = write_tracing_file_int("buffer_size_kb",
> + ftrace->percpu_buffer_size / 1024);
> + if (ret < 0)
> + return ret;
> +
> + return 0;
> +}
> +
> static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
> {
> char *trace_file;
> @@ -388,6 +405,11 @@ static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
> goto out_reset;
> }
>
> + if (set_tracing_percpu_buffer_size(ftrace) < 0) {
> + pr_err("failed to set tracing per-cpu buffer size\n");
> + goto out_reset;
> + }
> +
> if (write_tracing_file("current_tracer", ftrace->tracer) < 0) {
> pr_err("failed to set current_tracer to %s\n", ftrace->tracer);
> goto out_reset;
> @@ -506,6 +528,37 @@ static void delete_filter_func(struct list_head *head)
> }
> }
>
> +static int parse_buffer_size(const struct option *opt,
> + const char *str, int unset)
> +{
> + unsigned long *s = (unsigned long *)opt->value;
> + static struct parse_tag tags_size[] = {
> + { .tag = 'B', .mult = 1 },
> + { .tag = 'K', .mult = 1 << 10 },
> + { .tag = 'M', .mult = 1 << 20 },
> + { .tag = 'G', .mult = 1 << 30 },
> + { .tag = 0 },
> + };
> + unsigned long val;
> +
> + if (unset) {
> + *s = 0;
> + return 0;
> + }
> +
> + val = parse_tag_value(str, tags_size);
> + if (val != (unsigned long) -1) {
> + if (val < 1024) {
> + pr_err("buffer size too small, must larger than 1KB.");
> + return -1;
> + }
> + *s = val;
> + return 0;
> + }
> +
> + return -1;
> +}
> +
> static void select_tracer(struct perf_ftrace *ftrace)
> {
> bool graph = !list_empty(&ftrace->graph_funcs) ||
> @@ -560,6 +613,8 @@ int cmd_ftrace(int argc, const char **argv)
> "Set nograph filter on given functions", parse_filter_func),
> OPT_INTEGER('D', "graph-depth", &ftrace.graph_depth,
> "Max depth for function graph tracer"),
> + OPT_CALLBACK('m', "buffer-size", &ftrace.percpu_buffer_size, "size",
> + "size of per cpu buffer", parse_buffer_size),
> OPT_END()
> };
>
> --
> 2.25.1
>
--
- Arnaldo
Powered by blists - more mailing lists