linux-kernel - [PATCH] add miss time to CPUTIME_USER when kernel boots with highres=off

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20220730091456.3741503-1-liuchao173@huawei.com>
Date:   Sat, 30 Jul 2022 17:14:56 +0800
From:   Liu Chao <liuchao173@...wei.com>
To:     <linux-kernel@...r.kernel.org>, <tglx@...utronix.de>
CC:     <hewenliang4@...wei.com>, <linfeilong@...wei.com>
Subject: [PATCH] add miss time to CPUTIME_USER when kernel boots with highres=off

When kernel boots with highres=off and the cpu load isn't full, almost
every tick will account towards CPUTIME_IDLE. But in /proc/stat, idle
time is obtained from get_idle_time.
So the sum time will lose the time of user tasks.

For example:
 # stress-ng --cpu 2 --taskset 66 --timeout 600 --cpu-load 50 &
 # cat /proc/stat | grep "cpu66";sleep 10; cat /proc/stat | grep "cpu66"
 cpu66 1038 0 72 50124 75 17 1 0 0 0
 cpu66 1038 0 72 50569 75 18 2 0 0 0

the detla of sum time is 447, which should be 1000 actually. It will
casue the data in top is imprecise.

Signed-off-by: Liu Chao <liuchao173@...wei.com>
---
 fs/proc/stat.c          | 5 +++++
 include/linux/hrtimer.h | 2 ++
 kernel/time/hrtimer.c   | 2 +-
 3 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 4fb8729a68d4..c5b9116e74ad 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -141,6 +141,9 @@ static int show_stat(struct seq_file *p, void *v)
 		sum		+= kstat_cpu_irqs_sum(i);
 		sum		+= arch_irq_stat_cpu(i);
 
+		if (hrtimer_is_hres_enabled() == 0)
+			user	+= cpustat[CPUTIME_IDLE] - get_idle_time(&kcpustat, i);
+
 		for (j = 0; j < NR_SOFTIRQS; j++) {
 			unsigned int softirq_stat = kstat_softirqs_cpu(j, i);
 
@@ -179,6 +182,8 @@ static int show_stat(struct seq_file *p, void *v)
 		steal		= cpustat[CPUTIME_STEAL];
 		guest		= cpustat[CPUTIME_GUEST];
 		guest_nice	= cpustat[CPUTIME_GUEST_NICE];
+		if (hrtimer_is_hres_enabled() == 0)
+			user	+=  cpustat[CPUTIME_IDLE] - idle;
 		seq_printf(p, "cpu%d", i);
 		seq_put_decimal_ull(p, " ", nsec_to_clock_t(user));
 		seq_put_decimal_ull(p, " ", nsec_to_clock_t(nice));
diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index 0ee140176f10..4fa1bbe86287 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -326,6 +326,8 @@ extern unsigned int hrtimer_resolution;
 
 #endif
 
+extern int hrtimer_is_hres_enabled(void);
+
 static inline ktime_t
 __hrtimer_expires_remaining_adjusted(const struct hrtimer *timer, ktime_t now)
 {
diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 0ea8702eb516..3ba2d9fe492c 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -722,7 +722,7 @@ __setup("highres=", setup_hrtimer_hres);
 /*
  * hrtimer_high_res_enabled - query, if the highres mode is enabled
  */
-static inline int hrtimer_is_hres_enabled(void)
+int hrtimer_is_hres_enabled(void)
 {
 	return hrtimer_hres_enabled;
 }
-- 
2.23.0