From: Daniel Bristot de Oliveira <bristot@kernel.org>
To: linux-kernel@vger.kernel.org, linux-trace-kernel@vger.kernel.org,
Steven Rostedt <rostedt@goodmis.org>
Cc: linux-doc@vger.kernel.org, Juri Lelli <juri.lelli@redhat.com>,
William White <chwhite@redhat.com>,
Daniel Bristot de Oliveira <bristot@kernel.org>,
Masami Hiramatsu <mhiramat@kernel.org>,
Jonathan Corbet <corbet@lwn.net>
Subject: [PATCH V3 04/11] rtla: Automatically move rtla to a house-keeping cpu
Date: Tue, 6 Jun 2023 18:12:18 +0200 [thread overview]
Message-ID: <c54304d90c777310fb85a3e658d1449173759aab.1686066600.git.bristot@kernel.org> (raw)
In-Reply-To: <cover.1686066600.git.bristot@kernel.org>
When the user sets -c <cpu-list> try to move rtla out of the <cpu-list>,
even without an -H option. This is useful to avoid having rtla
interfering with the workload.
This works by removing <cpu-list> from rtla's current affinity.
If rtla fails to move itself away it is not that of a problem as this
is an automatic measure.
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
---
tools/tracing/rtla/src/osnoise_hist.c | 9 +++++
tools/tracing/rtla/src/osnoise_top.c | 9 +++++
tools/tracing/rtla/src/timerlat_hist.c | 9 +++++
tools/tracing/rtla/src/timerlat_top.c | 9 +++++
tools/tracing/rtla/src/utils.c | 50 ++++++++++++++++++++++++++
tools/tracing/rtla/src/utils.h | 1 +
6 files changed, 87 insertions(+)
diff --git a/tools/tracing/rtla/src/osnoise_hist.c b/tools/tracing/rtla/src/osnoise_hist.c
index b616a72d5c0a..dfbcb5ca7ecb 100644
--- a/tools/tracing/rtla/src/osnoise_hist.c
+++ b/tools/tracing/rtla/src/osnoise_hist.c
@@ -753,6 +753,15 @@ osnoise_hist_apply_config(struct osnoise_tool *tool, struct osnoise_hist_params
err_msg("Failed to set rtla to the house keeping CPUs\n");
goto out_err;
}
+ } else if (params->cpus) {
+ /*
+ * Even if the user do not set a house-keeping CPU, try to
+ * move rtla to a CPU set different to the one where the user
+ * set the workload to run.
+ *
+ * No need to check results as this is an automatic attempt.
+ */
+ auto_house_keeping(¶ms->monitored_cpus);
}
return 0;
diff --git a/tools/tracing/rtla/src/osnoise_top.c b/tools/tracing/rtla/src/osnoise_top.c
index 0e2f3b216d34..0833537bb2eb 100644
--- a/tools/tracing/rtla/src/osnoise_top.c
+++ b/tools/tracing/rtla/src/osnoise_top.c
@@ -582,6 +582,15 @@ osnoise_top_apply_config(struct osnoise_tool *tool, struct osnoise_top_params *p
err_msg("Failed to set rtla to the house keeping CPUs\n");
goto out_err;
}
+ } else if (params->cpus) {
+ /*
+ * Even if the user do not set a house-keeping CPU, try to
+ * move rtla to a CPU set different to the one where the user
+ * set the workload to run.
+ *
+ * No need to check results as this is an automatic attempt.
+ */
+ auto_house_keeping(¶ms->monitored_cpus);
}
return 0;
diff --git a/tools/tracing/rtla/src/timerlat_hist.c b/tools/tracing/rtla/src/timerlat_hist.c
index 00287e96f22e..1675d54cae81 100644
--- a/tools/tracing/rtla/src/timerlat_hist.c
+++ b/tools/tracing/rtla/src/timerlat_hist.c
@@ -776,6 +776,15 @@ timerlat_hist_apply_config(struct osnoise_tool *tool, struct timerlat_hist_param
err_msg("Failed to set rtla to the house keeping CPUs\n");
goto out_err;
}
+ } else if (params->cpus) {
+ /*
+ * Even if the user do not set a house-keeping CPU, try to
+ * move rtla to a CPU set different to the one where the user
+ * set the workload to run.
+ *
+ * No need to check results as this is an automatic attempt.
+ */
+ auto_house_keeping(¶ms->monitored_cpus);
}
return 0;
diff --git a/tools/tracing/rtla/src/timerlat_top.c b/tools/tracing/rtla/src/timerlat_top.c
index 920f2f6ef842..f0c6d9735e2a 100644
--- a/tools/tracing/rtla/src/timerlat_top.c
+++ b/tools/tracing/rtla/src/timerlat_top.c
@@ -619,6 +619,15 @@ timerlat_top_apply_config(struct osnoise_tool *top, struct timerlat_top_params *
err_msg("Failed to set rtla to the house keeping CPUs\n");
goto out_err;
}
+ } else if (params->cpus) {
+ /*
+ * Even if the user do not set a house-keeping CPU, try to
+ * move rtla to a CPU set different to the one where the user
+ * set the workload to run.
+ *
+ * No need to check results as this is an automatic attempt.
+ */
+ auto_house_keeping(¶ms->monitored_cpus);
}
return 0;
diff --git a/tools/tracing/rtla/src/utils.c b/tools/tracing/rtla/src/utils.c
index 8f9ad8f01e0f..3e25f0277fb9 100644
--- a/tools/tracing/rtla/src/utils.c
+++ b/tools/tracing/rtla/src/utils.c
@@ -709,3 +709,53 @@ int set_comm_cgroup(const char *comm_prefix, const char *cgroup)
close(cg_fd);
return 0;
}
+
+/**
+ * auto_house_keeping - Automatically move rtla out of measurement threads
+ *
+ * Try to move rtla away from the tracer, if possible.
+ *
+ * Returns 1 on success, 0 otherwise.
+ */
+int auto_house_keeping(cpu_set_t *monitored_cpus)
+{
+ cpu_set_t rtla_cpus, house_keeping_cpus;
+ int retval;
+
+ /* first get the CPUs in which rtla can actually run. */
+ retval = sched_getaffinity(getpid(), sizeof(rtla_cpus), &rtla_cpus);
+ if (retval == -1) {
+ debug_msg("Could not get rtla affinity, rtla might run with the threads!\n");
+ return 0;
+ }
+
+ /* then check if the existing setup is already good. */
+ CPU_AND(&house_keeping_cpus, &rtla_cpus, monitored_cpus);
+ if (!CPU_COUNT(&house_keeping_cpus)) {
+ debug_msg("rtla and the monitored CPUs do not share CPUs.");
+ debug_msg("Skipping auto house-keeping\n");
+ return 1;
+ }
+
+ /* remove the intersection */
+ CPU_XOR(&house_keeping_cpus, &rtla_cpus, monitored_cpus);
+
+ /* get only those that rtla can run */
+ CPU_AND(&house_keeping_cpus, &house_keeping_cpus, &rtla_cpus);
+
+ /* is there any cpu left? */
+ if (!CPU_COUNT(&house_keeping_cpus)) {
+ debug_msg("Could not find any CPU for auto house-keeping\n");
+ return 0;
+ }
+
+ retval = sched_setaffinity(getpid(), sizeof(house_keeping_cpus), &house_keeping_cpus);
+ if (retval == -1) {
+ debug_msg("Could not set affinity for auto house-keeping\n");
+ return 0;
+ }
+
+ debug_msg("rtla automatically moved to an auto house-keeping cpu set\n");
+
+ return 1;
+}
diff --git a/tools/tracing/rtla/src/utils.h b/tools/tracing/rtla/src/utils.h
index 9ab2f0d7bc1c..dec59163cfbc 100644
--- a/tools/tracing/rtla/src/utils.h
+++ b/tools/tracing/rtla/src/utils.h
@@ -60,6 +60,7 @@ int parse_cpu_set(char *cpu_list, cpu_set_t *set);
int set_comm_sched_attr(const char *comm_prefix, struct sched_attr *attr);
int set_comm_cgroup(const char *comm_prefix, const char *cgroup);
int set_cpu_dma_latency(int32_t latency);
+int auto_house_keeping(cpu_set_t *monitored_cpus);
#define ns_to_usf(x) (((double)x/1000))
#define ns_to_per(total, part) ((part * 100) / (double)total)
--
2.38.1
next prev parent reply other threads:[~2023-06-06 16:12 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-06 16:12 [PATCH V3 00/11] rtla improvements Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 01/11] rtla: Add -C cgroup support Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 02/11] rtla: Add --house-keeping option Daniel Bristot de Oliveira
2023-06-12 14:06 ` Masami Hiramatsu
2023-06-12 14:21 ` Daniel Bristot de Oliveira
2023-06-13 20:27 ` Steven Rostedt
2023-06-06 16:12 ` [PATCH V3 03/11] rtla: Change monitored_cpus from char * to cpu_set_t Daniel Bristot de Oliveira
2023-06-06 16:12 ` Daniel Bristot de Oliveira [this message]
2023-06-06 16:12 ` [PATCH V3 05/11] rtla/timerlat: Give timerlat auto analysis its own instance Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 06/11] rtla/timerlat_hist: Add auto-analysis support Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 07/11] rtla: Start the tracers after creating all instances Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 08/11] rtla/hwnoise: Reduce runtime to 75% Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 09/11] rtla/timerlat_top: Add timerlat user-space support Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 10/11] rtla/timerlat_hist: " Daniel Bristot de Oliveira
2023-06-06 16:12 ` [PATCH V3 11/11] Documentation: Add tools/rtla timerlat -u option documentation Daniel Bristot de Oliveira
2023-06-12 9:41 ` [PATCH V3 00/11] rtla improvements Juri Lelli
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c54304d90c777310fb85a3e658d1449173759aab.1686066600.git.bristot@kernel.org \
--to=bristot@kernel.org \
--cc=chwhite@redhat.com \
--cc=corbet@lwn.net \
--cc=juri.lelli@redhat.com \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-trace-kernel@vger.kernel.org \
--cc=mhiramat@kernel.org \
--cc=rostedt@goodmis.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).