Skip to contentSkip to conversation

perf kvm stat record: print kvm specific --help output

RFC PATCH New Alexander Yarygin 3 participants 4 messages Oct 6, 2014 – Oct 7, 2014 perf events

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 perf kvm stat record: print kvm specific --help output

Hello,

The 'perf kvm stat record' tool is an alias for 'perf record' tool. When 'perf kvm stat record' is running with '--help' option, it prints help output for 'perf record':

$ sudo perf kvm stat record --help
 usage: perf record [<options>] [<command>]
    or: perf record [<options>] -- <command> [<options>]

-e, --event <event> event selector. use 'perf list' to list available events
--filter <filter>
event filter
-p, --pid <pid> record events on existing process id
-t, --tid <tid> record events on existing thread id
-r, --realtime <n> collect data with this RT SCHED_FIFO priority
-D, --no-delay collect data without buffering
-R, --raw-samples collect raw sample records from all opened counters
-a, --all-cpus system-wide collection from all CPUs
-C, --cpu <cpu> list of cpus to monitor
-c, --count <n> event period to sample
-o, --output <file> output file name
.. snip ..

Some of these options can be used with 'kvm stat record', however others can break recording (i.e. --no-samples), or don't change anything kvm related (but they can add more details to the output file that are ignored by 'kvm stat report' and visible with 'perf report')

Plus the other 'perf kvm stat' tools have their own --help.

The patch is an proof-of-concept attempt to fix that inconsistency by printing useful for 'perf kvm stat record' options in addition to generic help. Like as follows:

$ sudo perf kvm stat record --help

usage: perf kvm stat record [<options>]

    -p, --pid <pid>      record events on existing process id
    -t, --tid <tid>      record events on existing thread id
    -r, --realtime <n>   collect data with this RT SCHED_FIFO priority
        --no-buffering   collect data without buffering
    -a, --all-cpus       system-wide collection from all CPUs
    -C, --cpu <cpu>      list of cpus to monitor
    -c, --count <n>      event period to sample
    -o, --output <file>  output file name
    -i, --no-inherit     child tasks do not inherit counters

-m, --mmap-pages <pages>
number of mmap data pages
-v, --verbose be more verbose (show counter open errors, etc)
-q, --quiet don't print any message
-D, --delay <n> ms to wait before starting measurement after program start
-u, --uid <user> user to profile
--per-thread use per-thread mmaps

Otherwise, it is the alias of 'perf record:'

usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]

-e, --event <event> event selector. use 'perf list' to list available events
--filter <filter>
event filter
-p, --pid <pid> record events on existing process id
.. here is the rest of the 'perf record' options ..

Problems:
1) I could drop some useful options.
2) Running 'perf kvm stat record' with invalid command leads to help message from 'perf record':

perf kvm stat record --i-want-help-from-record
Error: unknown option `i-want-help-from-record'

usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>] .. snip ..

Any ideas?

Thanks.

Alexander Yarygin (1):
perf kvm stat record: print kvm specific --help output

 tools/perf/builtin-kvm.c | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
Signature · 1 line
1.9.1
#1 lore raw

PATCH 1/1 perf kvm stat record: print kvm specific --help output

The 'perf kvm stat record' tool is an alias of 'perf record' with predefined kvm related options. All options that passed to 'perf kvm stat record' are processed by the 'perf record' tool. So, 'perf kvm stat record --help' prints help of usage for the 'perf record' command. There are a few options useful for 'perf kvm stat record', the rest either break kvm related output or don't change it.

Let's print safe for 'perf kvm stat record' options in addition to general 'perf record' --help output.

Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Christian Borntraeger <[email protected]>
Cc David Ahern <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Peter Zijlstra <[email protected]>
---
 tools/perf/builtin-kvm.c | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
tools/perf/builtin-kvm.c +38 -0
index 43367eb..0c169d7 100644
@@ -1138,6 +1138,40 @@ exit:
 		_p;			\
 	})
 
+static void print_record_help(void)
+{
+	fprintf(stderr, "\n usage: perf kvm stat record [<options>]\n\n");
+
+	fprintf(stderr, "    -p, --pid <pid>      record events on existing process id\n");
+	fprintf(stderr, "    -t, --tid <tid>      record events on existing thread id\n");
+	fprintf(stderr, "    -r, --realtime <n>   collect data with this RT SCHED_FIFO priority\n");
+	fprintf(stderr, "        --no-buffering   collect data without buffering\n");
+	fprintf(stderr, "    -a, --all-cpus       system-wide collection from all CPUs\n");
+	fprintf(stderr, "    -C, --cpu <cpu>      list of cpus to monitor\n");
+	fprintf(stderr, "    -c, --count <n>      event period to sample\n");
+	fprintf(stderr, "    -o, --output <file>  output file name\n");
+	fprintf(stderr, "    -i, --no-inherit     child tasks do not inherit counters\n");
+	fprintf(stderr, "    -m, --mmap-pages <pages>\n");
+	fprintf(stderr, "                         number of mmap data pages\n");
+	fprintf(stderr, "    -v, --verbose        be more verbose (show counter open errors, etc)\n");
+	fprintf(stderr, "    -q, --quiet          don't print any message\n");
+	fprintf(stderr, "    -D, --delay <n>      ms to wait before starting measurement after program start\n");
+	fprintf(stderr, "    -u, --uid <user>     user to profile\n");
+	fprintf(stderr, "        --per-thread     use per-thread mmaps\n");
+
+	fprintf(stderr, "\nOtherwise, it is the alias of 'perf record:'\n");
+}
+
+static bool is_help_option(int argc, const char **argv)
+{
+	int i;
+
+	for (i = 0; i < argc; i++)
+		if (!strcmp(argv[i], "--help") || !strcmp(argv[i], "-h"))
+			return true;
+	return false;
+}
+
 static int
 kvm_events_record(struct perf_kvm_stat *kvm, int argc, const char **argv)
 {
@@ -1150,6 +1184,10 @@ kvm_events_record(struct perf_kvm_stat *kvm, int argc, const char **argv)
 		"-c", "1",
 	};
 	const char * const *events_tp;
+
+	if (argc == 1 || (argc > 1 && is_help_option(argc, argv)))
+		print_record_help();
+
 	events_tp_size = 0;
 
 	for (events_tp = kvm_events_tp; *events_tp; events_tp++)
Signature · 1 line
1.9.1
#2 lore raw
Quoted 13 lines · Alexander Yarygin
tools/perf/builtin-kvm.c +13 -0
+static void print_record_help(void)
+{
+	fprintf(stderr, "\n usage: perf kvm stat record [<options>]\n\n");
+
+	fprintf(stderr, "    -p, --pid <pid>      record events on existing process id\n");
+	fprintf(stderr, "    -t, --tid <tid>      record events on existing thread id\n");
+	fprintf(stderr, "    -r, --realtime <n>   collect data with this RT SCHED_FIFO priority\n");
+	fprintf(stderr, "        --no-buffering   collect data without buffering\n");
+	fprintf(stderr, "    -a, --all-cpus       system-wide collection from all CPUs\n");
+	fprintf(stderr, "    -C, --cpu <cpu>      list of cpus to monitor\n");
+	fprintf(stderr, "    -c, --count <n>      event period to sample\n");
+	fprintf(stderr, "    -o, --output <file>  output file name\n");
+	fprintf(stderr, "    -i, --no-inherit     child tasks do not inherit counters\n");

tools/perf/builtin-kvm.c exit: · line 1163

Quoted 10 lines
tools/perf/builtin-kvm.c +10 -0
+	fprintf(stderr, "    -m, --mmap-pages <pages>\n");
+	fprintf(stderr, "                         number of mmap data pages\n");
+	fprintf(stderr, "    -v, --verbose        be more verbose (show counter open errors, etc)\n");
+	fprintf(stderr, "    -q, --quiet          don't print any message\n");
+	fprintf(stderr, "    -D, --delay <n>      ms to wait before starting measurement after program start\n");
+	fprintf(stderr, "    -u, --uid <user>     user to profile\n");
+	fprintf(stderr, "        --per-thread     use per-thread mmaps\n");
+
+	fprintf(stderr, "\nOtherwise, it is the alias of 'perf record:'\n");
+}

Those are all perf-record options, not perf-kvm-stat-record options. This usage should not repeat perf-record.

David

#3 lore raw

Em Tue, Oct 07, 2014 at 09:10:46AM -0600, David Ahern escreveu:

Quoted 20 lines · Alexander Yarygin
Quoted 19 lines
tools/perf/builtin-kvm.c +18 -0
+static void print_record_help(void)
+{
+	fprintf(stderr, "\n usage: perf kvm stat record [<options>]\n\n");
+
+	fprintf(stderr, "    -p, --pid <pid>      record events on existing process id\n");
+	fprintf(stderr, "    -t, --tid <tid>      record events on existing thread id\n");
+	fprintf(stderr, "    -r, --realtime <n>   collect data with this RT SCHED_FIFO priority\n");
+	fprintf(stderr, "        --no-buffering   collect data without buffering\n");
+	fprintf(stderr, "    -a, --all-cpus       system-wide collection from all CPUs\n");
+	fprintf(stderr, "    -C, --cpu <cpu>      list of cpus to monitor\n");
+	fprintf(stderr, "    -c, --count <n>      event period to sample\n");
+	fprintf(stderr, "    -o, --output <file>  output file name\n");
+	fprintf(stderr, "    -i, --no-inherit     child tasks do not inherit counters\n");
+	fprintf(stderr, "    -m, --mmap-pages <pages>\n");
+	fprintf(stderr, "                         number of mmap data pages\n");
+	fprintf(stderr, "    -v, --verbose        be more verbose (show counter open errors, etc)\n");
+	fprintf(stderr, "    -q, --quiet          don't print any message\n");
+	fprintf(stderr, "    -D, --delay <n>      ms to wait before starting measurement after program start\n");
Quoted 8 lines

tools/perf/builtin-kvm.c exit: · line 1163

Quoted 5 lines
tools/perf/builtin-kvm.c +5 -0
+	fprintf(stderr, "    -u, --uid <user>     user to profile\n");
+	fprintf(stderr, "        --per-thread     use per-thread mmaps\n");
+
+	fprintf(stderr, "\nOtherwise, it is the alias of 'perf record:'\n");
+}

Those are all perf-record options, not perf-kvm-stat-record options. This usage should not repeat perf-record.

Right, and from a quick look, the way Namhyung is tackling this problem is better.

- Arnaldo