Linux Trace Kernel
 help / color / mirror / Atom feed
From: tglozar@redhat.com
To: rostedt@goodmis.org
Cc: linux-trace-kernel@vger.kernel.org, linux-kernel@vger.kernel.org,
	jkacur@redhat.com, Tomas Glozar <tglozar@redhat.com>
Subject: [PATCH v2] rtla/timerlat: Fix histogram ALL for zero samples
Date: Wed, 27 Nov 2024 14:41:30 +0100	[thread overview]
Message-ID: <20241127134130.51171-1-tglozar@redhat.com> (raw)

From: Tomas Glozar <tglozar@redhat.com>

rtla timerlat hist currently computers the minimum, maximum and average
latency even in cases when there are zero samples. This leads to
nonsensical values being calculated for maximum and minimum, and to
divide by zero for average.

A similar bug is fixed by 01b05fc0e5f3 ("rtla/timerlat: Fix histogram
report when a cpu count is 0") but the bug still remains for printing
the sum over all CPUs in timerlat_print_stats_all.

The issue can be reproduced with this command:

$ rtla timerlat hist -U -d 1s
Index
over:
count:
min:
avg:
max:
Floating point exception (core dumped)

(There are always no samples with -U unless the user workload is
created.)

Fix the bug by omitting max/min/avg when sample count is zero,
displaying a dash instead, just like we already do for the individual
CPUs. The logic is moved into a new function called
format_summary_value, which is used for both the individual CPUs
and for the overall summary.

Cc: stable@vger.kernel.org
Fixes: 1462501c7a8 ("rtla/timerlat: Add a summary for hist mode")
Signed-off-by: Tomas Glozar <tglozar@redhat.com>
---
 tools/tracing/rtla/src/timerlat_hist.c | 177 ++++++++++++++-----------
 1 file changed, 96 insertions(+), 81 deletions(-)

diff --git a/tools/tracing/rtla/src/timerlat_hist.c b/tools/tracing/rtla/src/timerlat_hist.c
index 8b66387e5f35..4403cc4eba30 100644
--- a/tools/tracing/rtla/src/timerlat_hist.c
+++ b/tools/tracing/rtla/src/timerlat_hist.c
@@ -281,6 +281,21 @@ static void timerlat_hist_header(struct osnoise_tool *tool)
 	trace_seq_reset(s);
 }
 
+/*
+ * format_summary_value - format a line of summary value (min, max or avg)
+ * of hist data
+ */
+static void format_summary_value(struct trace_seq *seq,
+				 int count,
+				 unsigned long long val,
+				 bool avg)
+{
+	if (count)
+		trace_seq_printf(seq, "%9llu ", avg ? val / count : val);
+	else
+		trace_seq_printf(seq, "%9c ", '-');
+}
+
 /*
  * timerlat_print_summary - print the summary of the hist data to the output
  */
@@ -328,29 +343,23 @@ timerlat_print_summary(struct timerlat_hist_params *params,
 		if (!data->hist[cpu].irq_count && !data->hist[cpu].thread_count)
 			continue;
 
-		if (!params->no_irq) {
-			if (data->hist[cpu].irq_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						data->hist[cpu].min_irq);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_irq)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].irq_count,
+					     data->hist[cpu].min_irq,
+					     false);
 
-		if (!params->no_thread) {
-			if (data->hist[cpu].thread_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						data->hist[cpu].min_thread);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_thread)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].thread_count,
+					     data->hist[cpu].min_thread,
+					     false);
 
-		if (params->user_hist) {
-			if (data->hist[cpu].user_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						data->hist[cpu].min_user);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (params->user_hist)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].user_count,
+					     data->hist[cpu].min_user,
+					     false);
 	}
 	trace_seq_printf(trace->seq, "\n");
 
@@ -364,29 +373,23 @@ timerlat_print_summary(struct timerlat_hist_params *params,
 		if (!data->hist[cpu].irq_count && !data->hist[cpu].thread_count)
 			continue;
 
-		if (!params->no_irq) {
-			if (data->hist[cpu].irq_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						 data->hist[cpu].sum_irq / data->hist[cpu].irq_count);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_irq)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].irq_count,
+					     data->hist[cpu].sum_irq,
+					     true);
 
-		if (!params->no_thread) {
-			if (data->hist[cpu].thread_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						 data->hist[cpu].sum_thread / data->hist[cpu].thread_count);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_thread)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].thread_count,
+					     data->hist[cpu].sum_thread,
+					     true);
 
-		if (params->user_hist) {
-			if (data->hist[cpu].user_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						 data->hist[cpu].sum_user / data->hist[cpu].user_count);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (params->user_hist)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].user_count,
+					     data->hist[cpu].sum_user,
+					     true);
 	}
 	trace_seq_printf(trace->seq, "\n");
 
@@ -400,29 +403,23 @@ timerlat_print_summary(struct timerlat_hist_params *params,
 		if (!data->hist[cpu].irq_count && !data->hist[cpu].thread_count)
 			continue;
 
-		if (!params->no_irq) {
-			if (data->hist[cpu].irq_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						 data->hist[cpu].max_irq);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_irq)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].irq_count,
+					     data->hist[cpu].max_irq,
+					     false);
 
-		if (!params->no_thread) {
-			if (data->hist[cpu].thread_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						data->hist[cpu].max_thread);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (!params->no_thread)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].thread_count,
+					     data->hist[cpu].max_thread,
+					     false);
 
-		if (params->user_hist) {
-			if (data->hist[cpu].user_count)
-				trace_seq_printf(trace->seq, "%9llu ",
-						data->hist[cpu].max_user);
-			else
-				trace_seq_printf(trace->seq, "        - ");
-		}
+		if (params->user_hist)
+			format_summary_value(trace->seq,
+					     data->hist[cpu].user_count,
+					     data->hist[cpu].max_user,
+					     false);
 	}
 	trace_seq_printf(trace->seq, "\n");
 	trace_seq_do_printf(trace->seq);
@@ -506,16 +503,22 @@ timerlat_print_stats_all(struct timerlat_hist_params *params,
 		trace_seq_printf(trace->seq, "min:  ");
 
 	if (!params->no_irq)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.min_irq);
+		format_summary_value(trace->seq,
+				     sum.irq_count,
+				     sum.min_irq,
+				     false);
 
 	if (!params->no_thread)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.min_thread);
+		format_summary_value(trace->seq,
+				     sum.thread_count,
+				     sum.min_thread,
+				     false);
 
 	if (params->user_hist)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.min_user);
+		format_summary_value(trace->seq,
+				     sum.user_count,
+				     sum.min_user,
+				     false);
 
 	trace_seq_printf(trace->seq, "\n");
 
@@ -523,16 +526,22 @@ timerlat_print_stats_all(struct timerlat_hist_params *params,
 		trace_seq_printf(trace->seq, "avg:  ");
 
 	if (!params->no_irq)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.sum_irq / sum.irq_count);
+		format_summary_value(trace->seq,
+				     sum.irq_count,
+				     sum.sum_irq,
+				     true);
 
 	if (!params->no_thread)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.sum_thread / sum.thread_count);
+		format_summary_value(trace->seq,
+				     sum.thread_count,
+				     sum.sum_thread,
+				     true);
 
 	if (params->user_hist)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.sum_user / sum.user_count);
+		format_summary_value(trace->seq,
+				     sum.user_count,
+				     sum.sum_user,
+				     true);
 
 	trace_seq_printf(trace->seq, "\n");
 
@@ -540,16 +549,22 @@ timerlat_print_stats_all(struct timerlat_hist_params *params,
 		trace_seq_printf(trace->seq, "max:  ");
 
 	if (!params->no_irq)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.max_irq);
+		format_summary_value(trace->seq,
+				     sum.irq_count,
+				     sum.max_irq,
+				     false);
 
 	if (!params->no_thread)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.max_thread);
+		format_summary_value(trace->seq,
+				     sum.thread_count,
+				     sum.max_thread,
+				     false);
 
 	if (params->user_hist)
-		trace_seq_printf(trace->seq, "%9llu ",
-				 sum.max_user);
+		format_summary_value(trace->seq,
+				     sum.user_count,
+				     sum.max_user,
+				     false);
 
 	trace_seq_printf(trace->seq, "\n");
 	trace_seq_do_printf(trace->seq);
-- 
2.47.0


             reply	other threads:[~2024-11-27 13:42 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-27 13:41 tglozar [this message]
2024-11-27 13:50 ` [PATCH v2] rtla/timerlat: Fix histogram ALL for zero samples Tomas Glozar

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=20241127134130.51171-1-tglozar@redhat.com \
    --to=tglozar@redhat.com \
    --cc=jkacur@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-trace-kernel@vger.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