rtla: Add --house-keeping option

To avoid having rtla interfering with the measurement threads, add an
option for the user to set the CPUs in which rtla should run. For
instance:

  # rtla timerlat top -H 0 -c 1-7

Will place rtla in the CPU 0, while running the measurement threads in
the CPU 1-7.

Link: https://lkml.kernel.org/r/6a6c78a579a96ba8b02ae67ee1e0ba2cb5e03c4a.1686066600.git.bristot@kernel.org

Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Suggested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

Author: Daniel Bristot de Oliveira

Documentation/tools/rtla/common_options.rst

@@ -2,6 +2,10 @@
 
         Set the osnoise tracer to run the sample threads in the cpu-list.
 
+**-H**, **--house-keeping** *cpu-list*
+
+        Run rtla control threads only on the given cpu-list.
+
 **-d**, **--duration** *time[s|m|h|d]*
 
         Set the duration of the session.

tools/tracing/rtla/src/osnoise_hist.c

@@ -3,6 +3,7 @@
  * Copyright (C) 2021 Red Hat Inc, Daniel Bristot de Oliveira <bristot@kernel.org>
  */
 
+#define _GNU_SOURCE
 #include <getopt.h>
 #include <stdlib.h>
 #include <string.h>
@@ -11,6 +12,7 @@
 #include <errno.h>
 #include <stdio.h>
 #include <time.h>
+#include <sched.h>
 
 #include "utils.h"
 #include "osnoise.h"
@@ -30,6 +32,8 @@ struct osnoise_hist_params {
 	int			set_sched;
 	int			output_divisor;
 	int			cgroup;
+	int			hk_cpus;
+	cpu_set_t		hk_cpu_set;
 	struct sched_attr	sched_param;
 	struct trace_events	*events;
 
@@ -434,8 +438,8 @@ static void osnoise_hist_usage(char *usage)
 		"",
 		"  usage: rtla osnoise hist [-h] [-D] [-d s] [-a us] [-p us] [-r us] [-s us] [-S us] \\",
 		"	  [-T us] [-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] \\",
-		"	  [-c cpu-list] [-P priority] [-b N] [-E N] [--no-header] [--no-summary] [--no-index] \\",
-		"	  [--with-zeros] [-C[=cgroup_name]]",
+		"	  [-c cpu-list] [-H cpu-list] [-P priority] [-b N] [-E N] [--no-header] [--no-summary] \\",
+		"	  [--no-index] [--with-zeros] [-C[=cgroup_name]]",
 		"",
 		"	  -h/--help: print this menu",
 		"	  -a/--auto: set automatic trace mode, stopping the session if argument in us sample is hit",
@@ -445,6 +449,7 @@ static void osnoise_hist_usage(char *usage)
 		"	  -S/--stop-total us: stop trace if the total sample is higher than the argument in us",
 		"	  -T/--threshold us: the minimum delta to be considered a noise",
 		"	  -c/--cpus cpu-list: list of cpus to run osnoise threads",
+		"	  -H/--house-keeping cpus: run rtla control threads only on the given cpus",
 		"	  -C/--cgroup[=cgroup_name]: set cgroup, if no cgroup_name is passed, the rtla's cgroup will be inherited",
 		"	  -d/--duration time[s|m|h|d]: duration of the session",
 		"	  -D/--debug: print debug info",
@@ -507,6 +512,7 @@ static struct osnoise_hist_params
 			{"cgroup",		optional_argument,	0, 'C'},
 			{"debug",		no_argument,		0, 'D'},
 			{"duration",		required_argument,	0, 'd'},
+			{"house-keeping",	required_argument,		0, 'H'},
 			{"help",		no_argument,		0, 'h'},
 			{"period",		required_argument,	0, 'p'},
 			{"priority",		required_argument,	0, 'P'},
@@ -528,7 +534,7 @@ static struct osnoise_hist_params
 		/* getopt_long stores the option index here. */
 		int option_index = 0;
 
-		c = getopt_long(argc, argv, "a:c:C::b:d:e:E:Dhp:P:r:s:S:t::T:01234:5:",
+		c = getopt_long(argc, argv, "a:c:C::b:d:e:E:DhH:p:P:r:s:S:t::T:01234:5:",
 				 long_options, &option_index);
 
 		/* detect the end of the options. */
@@ -597,6 +603,14 @@ static struct osnoise_hist_params
 		case '?':
 			osnoise_hist_usage(NULL);
 			break;
+		case 'H':
+			params->hk_cpus = 1;
+			retval = parse_cpu_set(optarg, &params->hk_cpu_set);
+			if (retval) {
+				err_msg("Error parsing house keeping CPUs\n");
+				exit(EXIT_FAILURE);
+			}
+			break;
 		case 'p':
 			params->period = get_llong_from_str(optarg);
 			if (params->period > 10000000)
@@ -732,6 +746,15 @@ osnoise_hist_apply_config(struct osnoise_tool *tool, struct osnoise_hist_params
 		}
 	}
 
+	if (params->hk_cpus) {
+		retval = sched_setaffinity(getpid(), sizeof(params->hk_cpu_set),
+					   &params->hk_cpu_set);
+		if (retval == -1) {
+			err_msg("Failed to set rtla to the house keeping CPUs\n");
+			goto out_err;
+		}
+	}
+
 	return 0;
 
 out_err:

tools/tracing/rtla/src/osnoise_top.c

@@ -3,13 +3,15 @@
  * Copyright (C) 2021 Red Hat Inc, Daniel Bristot de Oliveira <bristot@kernel.org>
  */
 
+#define _GNU_SOURCE
 #include <getopt.h>
 #include <stdlib.h>
 #include <string.h>
 #include <signal.h>
 #include <unistd.h>
 #include <stdio.h>
 #include <time.h>
+#include <sched.h>
 
 #include "osnoise.h"
 #include "utils.h"
@@ -37,6 +39,8 @@ struct osnoise_top_params {
 	int			quiet;
 	int			set_sched;
 	int			cgroup;
+	int			hk_cpus;
+	cpu_set_t		hk_cpu_set;
 	struct sched_attr	sched_param;
 	struct trace_events	*events;
 	enum osnoise_mode	mode;
@@ -278,7 +282,7 @@ static void osnoise_top_usage(struct osnoise_top_params *params, char *usage)
 	static const char * const msg[] = {
 		" [-h] [-q] [-D] [-d s] [-a us] [-p us] [-r us] [-s us] [-S us] \\",
 		"	  [-T us] [-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] \\",
-		"	  [-c cpu-list] [-P priority] [-C[=cgroup_name]]",
+		"	  [-c cpu-list] [-H cpu-list] [-P priority] [-C[=cgroup_name]]",
 		"",
 		"	  -h/--help: print this menu",
 		"	  -a/--auto: set automatic trace mode, stopping the session if argument in us sample is hit",
@@ -288,6 +292,7 @@ static void osnoise_top_usage(struct osnoise_top_params *params, char *usage)
 		"	  -S/--stop-total us: stop trace if the total sample is higher than the argument in us",
 		"	  -T/--threshold us: the minimum delta to be considered a noise",
 		"	  -c/--cpus cpu-list: list of cpus to run osnoise threads",
+		"	  -H/--house-keeping cpus: run rtla control threads only on the given cpus",
 		"	  -C/--cgroup[=cgroup_name]: set cgroup, if no cgroup_name is passed, the rtla's cgroup will be inherited",
 		"	  -d/--duration time[s|m|h|d]: duration of the session",
 		"	  -D/--debug: print debug info",
@@ -354,6 +359,7 @@ struct osnoise_top_params *osnoise_top_parse_args(int argc, char **argv)
 			{"debug",		no_argument,		0, 'D'},
 			{"duration",		required_argument,	0, 'd'},
 			{"event",		required_argument,	0, 'e'},
+			{"house-keeping",	required_argument,	0, 'H'},
 			{"help",		no_argument,		0, 'h'},
 			{"period",		required_argument,	0, 'p'},
 			{"priority",		required_argument,	0, 'P'},
@@ -371,7 +377,7 @@ struct osnoise_top_params *osnoise_top_parse_args(int argc, char **argv)
 		/* getopt_long stores the option index here. */
 		int option_index = 0;
 
-		c = getopt_long(argc, argv, "a:c:C::d:De:hp:P:qr:s:S:t::T:0:1:",
+		c = getopt_long(argc, argv, "a:c:C::d:De:hH:p:P:qr:s:S:t::T:0:1:",
 				 long_options, &option_index);
 
 		/* Detect the end of the options. */
@@ -430,6 +436,14 @@ struct osnoise_top_params *osnoise_top_parse_args(int argc, char **argv)
 		case '?':
 			osnoise_top_usage(params, NULL);
 			break;
+		case 'H':
+			params->hk_cpus = 1;
+			retval = parse_cpu_set(optarg, &params->hk_cpu_set);
+			if (retval) {
+				err_msg("Error parsing house keeping CPUs\n");
+				exit(EXIT_FAILURE);
+			}
+			break;
 		case 'p':
 			params->period = get_llong_from_str(optarg);
 			if (params->period > 10000000)
@@ -561,6 +575,15 @@ osnoise_top_apply_config(struct osnoise_tool *tool, struct osnoise_top_params *p
 		}
 	}
 
+	if (params->hk_cpus) {
+		retval = sched_setaffinity(getpid(), sizeof(params->hk_cpu_set),
+					   &params->hk_cpu_set);
+		if (retval == -1) {
+			err_msg("Failed to set rtla to the house keeping CPUs\n");
+			goto out_err;
+		}
+	}
+
 	return 0;
 
 out_err:

tools/tracing/rtla/src/timerlat_hist.c

@@ -3,13 +3,15 @@
  * Copyright (C) 2021 Red Hat Inc, Daniel Bristot de Oliveira <bristot@kernel.org>
  */
 
+#define _GNU_SOURCE
 #include <getopt.h>
 #include <stdlib.h>
 #include <string.h>
 #include <signal.h>
 #include <unistd.h>
 #include <stdio.h>
 #include <time.h>
+#include <sched.h>
 
 #include "utils.h"
 #include "osnoise.h"
@@ -31,6 +33,8 @@ struct timerlat_hist_params {
 	int			set_sched;
 	int			dma_latency;
 	int			cgroup;
+	int			hk_cpus;
+	cpu_set_t		hk_cpu_set;
 	struct sched_attr	sched_param;
 	struct trace_events	*events;
 	char			no_irq;
@@ -432,7 +436,7 @@ static void timerlat_hist_usage(char *usage)
 	char *msg[] = {
 		"",
 		"  usage: [rtla] timerlat hist [-h] [-q] [-d s] [-D] [-n] [-a us] [-p us] [-i us] [-T us] [-s us] \\",
-		"         [-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] [-c cpu-list] \\",
+		"         [-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] [-c cpu-list] [-H cpu-list]\\",
 		"	  [-P priority] [-E N] [-b N] [--no-irq] [--no-thread] [--no-header] [--no-summary] \\",
 		"	  [--no-index] [--with-zeros] [--dma-latency us] [-C[=cgroup_name]]",
 		"",
@@ -443,6 +447,7 @@ static void timerlat_hist_usage(char *usage)
 		"	  -T/--thread us: stop trace if the thread latency is higher than the argument in us",
 		"	  -s/--stack us: save the stack trace at the IRQ if a thread latency is higher than the argument in us",
 		"	  -c/--cpus cpus: run the tracer only on the given cpus",
+		"	  -H/--house-keeping cpus: run rtla control threads only on the given cpus",
 		"	  -C/--cgroup[=cgroup_name]: set cgroup, if no cgroup_name is passed, the rtla's cgroup will be inherited",
 		"	  -d/--duration time[m|h|d]: duration of the session in seconds",
 		"	  -D/--debug: print debug info",
@@ -513,6 +518,7 @@ static struct timerlat_hist_params
 			{"debug",		no_argument,		0, 'D'},
 			{"entries",		required_argument,	0, 'E'},
 			{"duration",		required_argument,	0, 'd'},
+			{"house-keeping",	required_argument,	0, 'H'},
 			{"help",		no_argument,		0, 'h'},
 			{"irq",			required_argument,	0, 'i'},
 			{"nano",		no_argument,		0, 'n'},
@@ -537,7 +543,7 @@ static struct timerlat_hist_params
 		/* getopt_long stores the option index here. */
 		int option_index = 0;
 
-		c = getopt_long(argc, argv, "a:c:C::b:d:e:E:Dhi:np:P:s:t::T:0123456:7:8:",
+		c = getopt_long(argc, argv, "a:c:C::b:d:e:E:DhH:i:np:P:s:t::T:0123456:7:8:",
 				 long_options, &option_index);
 
 		/* detect the end of the options. */
@@ -608,6 +614,14 @@ static struct timerlat_hist_params
 		case '?':
 			timerlat_hist_usage(NULL);
 			break;
+		case 'H':
+			params->hk_cpus = 1;
+			retval = parse_cpu_set(optarg, &params->hk_cpu_set);
+			if (retval) {
+				err_msg("Error parsing house keeping CPUs\n");
+				exit(EXIT_FAILURE);
+			}
+			break;
 		case 'i':
 			params->stop_us = get_llong_from_str(optarg);
 			break;
@@ -755,6 +769,15 @@ timerlat_hist_apply_config(struct osnoise_tool *tool, struct timerlat_hist_param
 		}
 	}
 
+	if (params->hk_cpus) {
+		retval = sched_setaffinity(getpid(), sizeof(params->hk_cpu_set),
+					   &params->hk_cpu_set);
+		if (retval == -1) {
+			err_msg("Failed to set rtla to the house keeping CPUs\n");
+			goto out_err;
+		}
+	}
+
 	return 0;
 
 out_err:

tools/tracing/rtla/src/timerlat_top.c

@@ -3,6 +3,7 @@
  * Copyright (C) 2021 Red Hat Inc, Daniel Bristot de Oliveira <bristot@kernel.org>
  */
 
+#define _GNU_SOURCE
 #include <getopt.h>
 #include <stdlib.h>
 #include <string.h>
@@ -11,6 +12,7 @@
 #include <stdio.h>
 #include <time.h>
 #include <errno.h>
+#include <sched.h>
 
 #include "utils.h"
 #include "osnoise.h"
@@ -37,6 +39,8 @@ struct timerlat_top_params {
 	int			aa_only;
 	int			dump_tasks;
 	int			cgroup;
+	int			hk_cpus;
+	cpu_set_t		hk_cpu_set;
 	struct sched_attr	sched_param;
 	struct trace_events	*events;
 };
@@ -286,7 +290,7 @@ static void timerlat_top_usage(char *usage)
 	static const char *const msg[] = {
 		"",
 		"  usage: rtla timerlat [top] [-h] [-q] [-a us] [-d s] [-D] [-n] [-p us] [-i us] [-T us] [-s us] \\",
-		"	  [[-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] [-c cpu-list] \\",
+		"	  [[-t[=file]] [-e sys[:event]] [--filter <filter>] [--trigger <trigger>] [-c cpu-list] [-H cpu-list]\\",
 		"	  [-P priority] [--dma-latency us] [--aa-only us] [-C[=cgroup_name]]",
 		"",
 		"	  -h/--help: print this menu",
@@ -297,6 +301,7 @@ static void timerlat_top_usage(char *usage)
 		"	  -T/--thread us: stop trace if the thread latency is higher than the argument in us",
 		"	  -s/--stack us: save the stack trace at the IRQ if a thread latency is higher than the argument in us",
 		"	  -c/--cpus cpus: run the tracer only on the given cpus",
+		"	  -H/--house-keeping cpus: run rtla control threads only on the given cpus",
 		"	  -C/--cgroup[=cgroup_name]: set cgroup, if no cgroup_name is passed, the rtla's cgroup will be inherited",
 		"	  -d/--duration time[m|h|d]: duration of the session in seconds",
 		"	  -D/--debug: print debug info",
@@ -360,6 +365,7 @@ static struct timerlat_top_params
 			{"duration",		required_argument,	0, 'd'},
 			{"event",		required_argument,	0, 'e'},
 			{"help",		no_argument,		0, 'h'},
+			{"house-keeping",	required_argument,	0, 'H'},
 			{"irq",			required_argument,	0, 'i'},
 			{"nano",		no_argument,		0, 'n'},
 			{"period",		required_argument,	0, 'p'},
@@ -380,7 +386,7 @@ static struct timerlat_top_params
 		/* getopt_long stores the option index here. */
 		int option_index = 0;
 
-		c = getopt_long(argc, argv, "a:c:C::d:De:hi:np:P:qs:t::T:0:1:2:345:",
+		c = getopt_long(argc, argv, "a:c:C::d:De:hH:i:np:P:qs:t::T:0:1:2:345:",
 				 long_options, &option_index);
 
 		/* detect the end of the options. */
@@ -454,6 +460,14 @@ static struct timerlat_top_params
 		case '?':
 			timerlat_top_usage(NULL);
 			break;
+		case 'H':
+			params->hk_cpus = 1;
+			retval = parse_cpu_set(optarg, &params->hk_cpu_set);
+			if (retval) {
+				err_msg("Error parsing house keeping CPUs\n");
+				exit(EXIT_FAILURE);
+			}
+			break;
 		case 'i':
 			params->stop_us = get_llong_from_str(optarg);
 			break;
@@ -598,6 +612,15 @@ timerlat_top_apply_config(struct osnoise_tool *top, struct timerlat_top_params *
 		}
 	}
 
+	if (params->hk_cpus) {
+		retval = sched_setaffinity(getpid(), sizeof(params->hk_cpu_set),
+					   &params->hk_cpu_set);
+		if (retval == -1) {
+			err_msg("Failed to set rtla to the house keeping CPUs\n");
+			goto out_err;
+		}
+	}
+
 	return 0;
 
 out_err: