Broken process startup times after suspend (regression)
Subsystems (5)
Hi,
the commits
411187fb05cd11676b0979d9fbf3291db69dbce2 (GTOD: persistent clock support) c1d370e167d66b10bca3b602d3740405469383de (i386: use GTOD persistent clock
support)
caused a change in the way uptime is measured and introduced a regression such that it's no longer possible to obtain a correct process start time or the system boot time.
These two commits make the monotonic time not jump by hudreds of seconds after the kernel resumes from suspend, but they achieve it by moving the wall_to_monotonic offset, which is used for all boot time / uptime calculations. It's not possible to read the real boot time / process start time then.
I was thinking about a solution and I am posting the least intrusive one I found. I add a total_sleep_time variable which is to be added to wall_to_monotonic when one wants the proper boot time. I do by no means say it's the best one, I expect comments.
This requires a patch to procps that makes it use /proc/stat's btime instead of "now - uptime". This was written by Tomas Smetana and I'm attaching it as well. (It was posted upstream as it fixes another bug in ps as well.) Or we could just make the /proc/uptime contain "now - btime" like it used to before those two commits.
--- fs/proc/proc_misc.c | 2 +- include/linux/time.h | 1 + kernel/fork.c | 1 + kernel/timer.c | 6 ++++++ 4 files changed, 9 insertions(+), 1 deletions(-)
fs/proc/proc_misc.c +1 -1
@@ -456,7 +456,7 @@ static int show_stat(struct seq_file *p, void *v) user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; + jif = - (wall_to_monotonic.tv_sec + total_sleep_time);if (wall_to_monotonic.tv_nsec) --jif;
include/linux/time.h +1 -0
@@ -90,6 +90,7 @@ static inline struct timespec timespec_sub(struct timespec lhs,extern struct timespec xtime;extern struct timespec wall_to_monotonic; +extern unsigned long total_sleep_time; extern seqlock_t xtime_lock __attribute__((weak));extern unsigned long read_persistent_clock (void );
kernel/fork.c +1 -0
@@ -1056,6 +1056,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); + p->start_time.tv_sec += total_sleep_time; p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;
kernel/timer.c +6 -0
@@ -759,9 +759,13 @@ unsigned long next_timer_interrupt(void)* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* +* One needs to add total_sleep_time to wall_to_monotonic to get the real boot +* time. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +unsigned long total_sleep_time;EXPORT_SYMBOL (xtime);
@@ -975,6 +979,7 @@ void __init timekeeping_init(void) xtime.tv_nsec =0 ;set_normalized_timespec (&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); + total_sleep_time =0 ;write_sequnlock_irqrestore (&xtime_lock, flags); }
@@ -1004,6 +1009,7 @@ static int timekeeping_resume(struct sys_device *dev) xtime.tv_sec += sleep_length; wall_to_monotonic.tv_sec -= sleep_length; + total_sleep_time += sleep_length; }/* re-base the last cycle value */ clock->cycle_last =clocksource_read (clock);
Regards,
Signature · 1 line
Tomas Janousek, SW Engineer, Red Hat, Inc.
ps/output.c renamed from ps/output.c.jitter +0 -2
@@ -77,7 +77,6 @@static int wide_signals;/* true if we have room */ static unsigned long seconds_since_1970; -static unsigned long time_of_boot;static unsigned long page_shift;
@@ -1952,7 +1951,6 @@// available space: page_size*outbuf_pages-SPACE_AMOUNT seconds_since_1970 =time (NULL ); - time_of_boot = seconds_since_1970 - seconds_since_boot;meminfo ();
ps/common.h renamed from ps/common.h.jitter +1 -0
@@ -302,6 +302,7 @@extern int screen_cols;extern int screen_rows;extern unsigned long seconds_since_boot; +extern unsigned long time_of_boot;extern selection_node *selection_list;extern unsigned simple_select;extern sort_node *sort_list;
ps/global.c renamed from ps/global.c.jitter +18 -2
@@ -70,6 +70,7 @@int screen_cols =-1 ;int screen_rows =-1 ;unsigned long seconds_since_boot =-1 ; +unsigned long time_of_boot =-1 ;selection_node *selection_list = (selection_node *)0xdeadbeef ;unsigned simple_select =0xffffffff ;sort_node *sort_list = (sort_node *)0xdeadbeef ;/* ready-to-use sort list */
@@ -361,7 +362,23 @@look_up_our_self (&p);set_screen_size ();set_personality (); - +int fd; +char buf[BUFFSIZE ]; +const char *b; + +/* get boot time from /proc/stat */ + fd =open ("/proc/stat" ,O_RDONLY ,0 ); +if (fd !=-1 ) { + buf[BUFFSIZE -1 ] =0 ; +read (fd, buf,BUFFSIZE -1 ); + b =strstr (buf,"btime " ); +if (b) { +sscanf (b,"btime %lu" , &time_of_boot); + seconds_since_boot =time (0 ) - time_of_boot; + } +close (fd); + } + all_processes =0 ; bsd_c_option =0 ; bsd_e_option =0 ;
@@ -380,7 +397,6 @@ negate_selection =0 ; page_size =getpagesize (); running_only =0 ; - seconds_since_boot =uptime (0 ,0 ); selection_list =NULL ; simple_select =0 ; sort_list =NULL ;
Quoted 10 lines · Tomas Janousek
Hi,
the commits
411187fb05cd11676b0979d9fbf3291db69dbce2(GTOD: persistent clock support)c1d370e167d66b10bca3b602d3740405469383de(i386: use GTOD persistent clock
support)
caused a change in the way uptime is measured and introduced a regression such that it's no longer possible to obtain a correct process start time or the system boot time.
Quoted 5 lines
These two commits make the monotonic time not jump by hudreds of seconds after the kernel resumes from suspend, but they achieve it by moving the
wall_to_monotonicoffset, which is used for all boot time / uptime calculations. It's not possible to read the real boot time / process start time then.
Indeed. The monotonic clock's behavior around suspend and resume is poorly defined. When we increased it, folks didn't like the fact that uptime would increase while a system was suspended.
Quoted 4 lines
I was thinking about a solution and I am posting the least intrusive one I found. I add a
total_sleep_timevariable which is to be added towall_to_monotonicwhen one wants the proper boot time. I do by no means say it's the best one, I expect comments.
While I'd prefer wrapping it in a interface and keeping it in timekeeping (rather then exporting a collection of values that should be combined in varying ways), I think this approach is probably the best. Rather then forcing one behavior or the other we can provide both smooth monotonic and monotonic w/ sleep.
Quoted 18 lines
This requires a patch to procps that makes it use /proc/stat's btime instead of "now - uptime". This was written by Tomas Smetana and I'm attaching it as well. (It was posted upstream as it fixes another bug in ps as well.) Or we could just make the
/proc/uptimecontain "now - btime" like it used to before those two commits.--- fs/proc/proc_misc.c | 2 +- include/linux/time.h | 1 + kernel/fork.c | 1 + kernel/timer.c | 6 ++++++ 4 files changed, 9 insertions(+), 1 deletions(-)fs/proc/proc_misc.c +0
-0
fs/proc/proc_misc.c static int show_stat(struct seq_file *p, void *v) · line 461
Quoted 7 lines
fs/proc/proc_misc.c +1
-1@@ -456,7 +456,7 @@ static int show_stat(struct seq_file *p, void *v) user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; + jif = - (wall_to_monotonic.tv_sec + total_sleep_time);if (wall_to_monotonic.tv_nsec) --jif;
As I said, could we wrap this in a function?
Maybe something like get_monotonicwithsleep_time() (feel free to come up with a better name :).
Quoted 51 lines
include/linux/time.h +1
-0@@ -90,6 +90,7 @@ static inline struct timespec timespec_sub(struct timespec lhs,extern struct timespec xtime;extern struct timespec wall_to_monotonic; +extern unsigned long total_sleep_time; extern seqlock_t xtime_lock __attribute__((weak));extern unsigned long read_persistent_clock (void );kernel/fork.c +1
-0@@ -1056,6 +1056,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); + p->start_time.tv_sec += total_sleep_time; p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;kernel/timer.c +5
-0@@ -759,9 +759,13 @@ unsigned long next_timer_interrupt(void)* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* +* One needs to add total_sleep_time to wall_to_monotonic to get the real boot +* time. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +unsigned long total_sleep_time;EXPORT_SYMBOL (xtime);@@ -975,6 +979,7 @@ void __init timekeeping_init(void) xtime.tv_nsec =0 ;set_normalized_timespec (&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); + total_sleep_time =0 ;write_sequnlock_irqrestore (&xtime_lock, flags); }
kernel/timer.c static int timekeeping_resume(struct sys_device *dev) · line 1015
Quoted 7 lines
kernel/timer.c +1
-0@@ -1004,6 +1009,7 @@ static int timekeeping_resume(struct sys_device *dev) xtime.tv_sec += sleep_length; wall_to_monotonic.tv_sec -= sleep_length; + total_sleep_time += sleep_length; }/* re-base the last cycle value */ clock->cycle_last =clocksource_read (clock);
Otherwise it looks good.
Thanks for bringing this up!
-john
Hi,
Quoted 3 lines · john stultz
Indeed. The monotonic clock's behavior around suspend and resume is poorly defined. When we increased it, folks didn't like the fact that uptime would increase while a system was suspended.
Seems like your change also made accounting and OOM calculations behave more correctly. And my previous patch broke that completely, because they would use "your" uptime and "my" process start time, resulting in negative numbers when calculating the process uptime. Therefore, I had to create another start_time in the task_struct.
Quoted 10 lines
Quoted 4 lines
I was thinking about a solution and I am posting the least intrusive one I found. I add a
total_sleep_timevariable which is to be added towall_to_monotonicwhen one wants the proper boot time. I do by no means say it's the best one, I expect comments.While I'd prefer wrapping it in a interface and keeping it in timekeeping (rather then exporting a collection of values that should be combined in varying ways), I think this approach is probably the best. Rather then forcing one behavior or the other we can provide both smooth monotonic and monotonic w/ sleep.
Ok, thanks for the comment, I did so.
And here's new patch:
From: Tomas Janousek <[email protected]>
Subject: [PATCH] Fix boot time and process startup times after suspend
The commits
411187fb05cd11676b0979d9fbf3291db69dbce2 (GTOD: persistent clock support) c1d370e167d66b10bca3b602d3740405469383de (i386: use GTOD persistent clock
support)
changed the monotonic time so that it no longer jumps after resume, but it's not possible to use it for boot time and process start time calculations then.
I add a getboottime function to get the real boot time and a boot-based time concept. Because we need a process start time corresponding to the monotonic time, I add real_start_time to the task_struct which is then used in /proc/pid/stat. The /proc/stat's btime uses getboottime instead of the monotonic time offset as well.
--- fs/proc/array.c | 5 +++-- fs/proc/proc_misc.c | 8 +++++--- include/linux/ktime.h | 2 ++ include/linux/sched.h | 2 +- include/linux/time.h | 2 ++ kernel/fork.c | 1 + kernel/hrtimer.c | 27 +++++++++++++++++++++++++++ kernel/timer.c | 21 +++++++++++++++++++++ 8 files changed, 62 insertions(+), 6 deletions(-)
fs/proc/array.c +3 -2
@@ -405,8 +405,9 @@ static int do_task_stat(struct task_struct *task, char * buffer, int whole)/* Temporary variable needed for gcc-2.96 */ /* convert timespec -> nsec*/ - start_time = (unsigned long long )task->start_time.tv_sec *NSEC_PER_SEC - + task->start_time.tv_nsec; + start_time = + (unsigned long long )task->real_start_time.tv_sec *NSEC_PER_SEC + + task->real_start_time.tv_nsec;/* convert nsec -> ticks */ start_time =nsec_to_clock_t (start_time);
fs/proc/proc_misc.c +5 -3
@@ -453,12 +453,14 @@ static int show_stat(struct seq_file *p, void *v)unsigned long jif;cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;u64 sum =0 ; +struct timespec boottime; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; -if (wall_to_monotonic.tv_nsec) - --jif; +getboottime (&boottime); + jif = boottime.tv_sec; +if (boottime.tv_nsec) + ++jif;for_each_possible_cpu (i) {int j;
include/linux/ktime.h +2 -0
@@ -269,6 +269,8 @@ static inline s64 ktime_to_ns(const ktime_t kt)/* Get the monotonic time in timespec format: */ extern void ktime_get_ts (struct timespec *ts); +/* Get the boot based time in timespec format: */ +extern void bbtime_get_ts (struct timespec *ts);/* Get the real (wall-) time in timespec format: */ #define ktime_get_real_ts (ts) getnstimeofday(ts)
include/linux/sched.h +1 -1
@@ -884,7 +884,7 @@ struct task_struct {unsigned long rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; +struct timespec start_time, real_start_time;/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ unsigned long min_flt, maj_flt;
include/linux/time.h +2 -0
@@ -90,6 +90,7 @@ static inline struct timespec timespec_sub(struct timespec lhs,extern struct timespec xtime;extern struct timespec wall_to_monotonic; +extern unsigned long total_sleep_time; extern seqlock_t xtime_lock __attribute__((weak));extern unsigned long read_persistent_clock (void );
@@ -116,6 +117,7 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );
kernel/fork.c +1 -0
@@ -1056,6 +1056,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); +bbtime_get_ts (&p->real_start_time); p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;
kernel/hrtimer.c +27 -0
@@ -128,6 +128,33 @@ void ktime_get_ts(struct timespec *ts) }EXPORT_SYMBOL_GPL (ktime_get_ts); +/** +* bbtime_get_ts - get the boot based clock in timespec format +* @ts: pointer to timespec variable +* +* The function calculates the boot based clock from the realtime +* clock, the wall_to_monotonic offset and the total sleep time and +* stores the result in normalized timespec format in the variable +* pointed to by @ts. +*/ +void bbtime_get_ts (struct timespec *ts) +{ +struct timespec tobb; +unsigned long seq; + +do { + seq =read_seqbegin (&xtime_lock); +getnstimeofday (ts); + tobb = wall_to_monotonic; + tobb.tv_sec += total_sleep_time; + + }while (read_seqretry (&xtime_lock, seq)); + +set_normalized_timespec (ts, ts->tv_sec + tobb.tv_sec, + ts->tv_nsec + tobb.tv_nsec); +} +EXPORT_SYMBOL_GPL (bbtime_get_ts); +/* * Get the coarse grained time at the softirq based on xtime and * wall_to_monotonic.
kernel/timer.c +21 -0
@@ -759,9 +759,13 @@ unsigned long next_timer_interrupt(void)* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* +* We need to add total_sleep_time to wall_to_monotonic to get the real boot +* time. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +unsigned long total_sleep_time;EXPORT_SYMBOL (xtime);
@@ -832,6 +836,21 @@ void getnstimeofday(struct timespec *ts)EXPORT_SYMBOL (getnstimeofday);/** +* getboottime - Return the real time of system boot. +* @ts: pointer to the timespec to be set +* +* Returns the time of day in a timespec. +*/ +void getboottime (struct timespec *ts) +{ +set_normalized_timespec (ts, + - (wall_to_monotonic.tv_sec + total_sleep_time), + - wall_to_monotonic.tv_nsec); +} + +EXPORT_SYMBOL (getboottime); + +/** * do_gettimeofday - Returns the time of day in a timeval * @tv: pointer to the timeval to be set *
@@ -975,6 +994,7 @@ void __init timekeeping_init(void) xtime.tv_nsec =0 ;set_normalized_timespec (&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); + total_sleep_time =0 ;write_sequnlock_irqrestore (&xtime_lock, flags); }
@@ -1004,6 +1024,7 @@ static int timekeeping_resume(struct sys_device *dev) xtime.tv_sec += sleep_length; wall_to_monotonic.tv_sec -= sleep_length; + total_sleep_time += sleep_length; }/* re-base the last cycle value */ clock->cycle_last =clocksource_read (clock);
Signature · 1 line
TJ. (Brno, CZ), BaseOS, Red Hat
Quoted 18 lines · Tomas Janousek
Hi,
Quoted 3 lines · john stultz
Indeed. The monotonic clock's behavior around suspend and resume is poorly defined. When we increased it, folks didn't like the fact that uptime would increase while a system was suspended.
Seems like your change also made accounting and OOM calculations behave more correctly. And my previous patch broke that completely, because they would use "your" uptime and "my" process start time, resulting in negative numbers when calculating the process uptime. Therefore, I had to create another
start_timein thetask_struct.Quoted 5 lines
Quoted 4 lines
I was thinking about a solution and I am posting the least intrusive one I found. I add a
total_sleep_timevariable which is to be added towall_to_monotonicwhen one wants the proper boot time. I do by no means say it's the best one, I expect comments.
Quoted 7 lines
Quoted 5 lines
While I'd prefer wrapping it in a interface and keeping it in timekeeping (rather then exporting a collection of values that should be combined in varying ways), I think this approach is probably the best. Rather then forcing one behavior or the other we can provide both smooth monotonic and monotonic w/ sleep.
Ok, thanks for the comment, I did so.
Looks great! Just a few minor comments.
Quoted 121 lines
And here's new patch:
From: Tomas Janousek <[email protected]>
Subject: [PATCH] Fix boot time and process startup times after suspendThe commits
411187fb05cd11676b0979d9fbf3291db69dbce2(GTOD: persistent clock support)c1d370e167d66b10bca3b602d3740405469383de(i386: use GTOD persistent clock
support)
changed the monotonic time so that it no longer jumps after resume, but it's not possible to use it for boot time and process start time calculations then.I add a getboottime function to get the real boot time and a boot-based time concept. Because we need a process start time corresponding to the monotonic time, I add
real_start_timeto thetask_structwhich is then used in/proc/pid/stat. The /proc/stat's btime uses getboottime instead of the monotonic time offset as well.Signed-off-by Tomas Janousek <[email protected]>Cc Tomas Smetana <[email protected]>--- fs/proc/array.c | 5 +++-- fs/proc/proc_misc.c | 8 +++++--- include/linux/ktime.h | 2 ++ include/linux/sched.h | 2 +- include/linux/time.h | 2 ++ kernel/fork.c | 1 + kernel/hrtimer.c | 27 +++++++++++++++++++++++++++ kernel/timer.c | 21 +++++++++++++++++++++ 8 files changed, 62 insertions(+), 6 deletions(-)fs/proc/array.c +3
-2@@ -405,8 +405,9 @@ static int do_task_stat(struct task_struct *task, char * buffer, int whole)/* Temporary variable needed for gcc-2.96 */ /* convert timespec -> nsec*/ - start_time = (unsigned long long )task->start_time.tv_sec *NSEC_PER_SEC - + task->start_time.tv_nsec; + start_time = + (unsigned long long )task->real_start_time.tv_sec *NSEC_PER_SEC + + task->real_start_time.tv_nsec;/* convert nsec -> ticks */ start_time =nsec_to_clock_t (start_time);fs/proc/proc_misc.c +5
-3@@ -453,12 +453,14 @@ static int show_stat(struct seq_file *p, void *v)unsigned long jif;cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;u64 sum =0 ; +struct timespec boottime; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; -if (wall_to_monotonic.tv_nsec) - --jif; +getboottime (&boottime); + jif = boottime.tv_sec; +if (boottime.tv_nsec) + ++jif;for_each_possible_cpu (i) {int j;include/linux/ktime.h +2
-0@@ -269,6 +269,8 @@ static inline s64 ktime_to_ns(const ktime_t kt)/* Get the monotonic time in timespec format: */ extern void ktime_get_ts (struct timespec *ts); +/* Get the boot based time in timespec format: */ +extern void bbtime_get_ts (struct timespec *ts);/* Get the real (wall-) time in timespec format: */ #define ktime_get_real_ts (ts) getnstimeofday(ts)include/linux/sched.h +1
-1@@ -884,7 +884,7 @@ struct task_struct {unsigned long rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; +struct timespec start_time, real_start_time;/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ unsigned long min_flt, maj_flt;include/linux/time.h +2
-0@@ -90,6 +90,7 @@ static inline struct timespec timespec_sub(struct timespec lhs,extern struct timespec xtime;extern struct timespec wall_to_monotonic; +extern unsigned long total_sleep_time; extern seqlock_t xtime_lock __attribute__((weak));extern unsigned long read_persistent_clock (void );@@ -116,6 +117,7 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );kernel/fork.c +0
-0
kernel/fork.c static struct task_struct *copy_process(unsigned long clone_flags, · line 1062
Quoted 7 lines
kernel/fork.c +1
-0@@ -1056,6 +1056,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); +bbtime_get_ts (&p->real_start_time); p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;
Since both do_posix_clock_monotonic_gettime and bbtime_get_ts actually read the timekeeping hardware (which could cost around 1.5us each), we probably will want to optimize this down to a single read, or just avoid the hardware read and take lower-granularity xtime value.
However, I'm not sure how fine-grained the start_time values need to be.
Quoted 6 lines
kernel/hrtimer.c +0
-0@@ -128,6 +128,33 @@ void ktime_get_ts(struct timespec *ts) }
kernel/hrtimer.c void ktime_get_ts(struct timespec *ts) · line 138
Quoted 10 lines
kernel/hrtimer.c +8
-0EXPORT_SYMBOL_GPL (ktime_get_ts); +/** +* bbtime_get_ts - get the boot based clock in timespec format +* @ts: pointer to timespec variable +* +* The function calculates the boot based clock from the realtime +* clock, the wall_to_monotonic offset and the total sleep time and +* stores the result in normalized timespec format in the variable +* pointed to by @ts.
You might want to clarify this as being the "monotonic boot time", or the time the system has been running including sleep time and that calls to settimeofday() will not affect this value.
kernel/hrtimer.c void ktime_get_ts(struct timespec *ts) · line 140
Quoted 2 lines
kernel/hrtimer.c +2
-0+*/ +void bbtime_get_ts (struct timespec *ts)
Any reason you added this in hrtimer.c instead of timer.c? I think keeping the new functions closer together would make the subtle difference between them more clear, and would allow total_sleep_time to be static to one file.
Quoted 39 lines
kernel/hrtimer.c +17
-0+{ +struct timespec tobb; +unsigned long seq; + +do { + seq =read_seqbegin (&xtime_lock); +getnstimeofday (ts); + tobb = wall_to_monotonic; + tobb.tv_sec += total_sleep_time; + + }while (read_seqretry (&xtime_lock, seq)); + +set_normalized_timespec (ts, ts->tv_sec + tobb.tv_sec, + ts->tv_nsec + tobb.tv_nsec); +} +EXPORT_SYMBOL_GPL (bbtime_get_ts); +/* * Get the coarse grained time at the softirq based on xtime and * wall_to_monotonic. kernel/timer.c +4
-0@@ -759,9 +759,13 @@ unsigned long next_timer_interrupt(void)* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* +* We need to add total_sleep_time to wall_to_monotonic to get the real boot +* time. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +unsigned long total_sleep_time;EXPORT_SYMBOL (xtime);
kernel/timer.c void getnstimeofday(struct timespec *ts) · line 840
Quoted 5 lines
kernel/timer.c +2
-0@@ -832,6 +836,21 @@ void getnstimeofday(struct timespec *ts)EXPORT_SYMBOL (getnstimeofday);/** +* getboottime - Return the real time of system boot. +* @ts: pointer to the timespec to be set
Might want to clarify this as being the "boot time" based on the current CLOCK_REALTIME clock, noting that calls to settimeofday() would affect this value.
Quoted 24 lines
kernel/timer.c +13
-0+* Returns the time of day in a timespec. +*/ +void getboottime (struct timespec *ts) +{ +set_normalized_timespec (ts, + - (wall_to_monotonic.tv_sec + total_sleep_time), + - wall_to_monotonic.tv_nsec); +} + +EXPORT_SYMBOL (getboottime); + +/** * do_gettimeofday - Returns the time of day in a timeval * @tv: pointer to the timeval to be set * @@ -975,6 +994,7 @@ void __init timekeeping_init(void) xtime.tv_nsec =0 ;set_normalized_timespec (&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); + total_sleep_time =0 ;write_sequnlock_irqrestore (&xtime_lock, flags); }
kernel/timer.c static int timekeeping_resume(struct sys_device *dev) · line 1030
Quoted 7 lines
kernel/timer.c +1
-0@@ -1004,6 +1024,7 @@ static int timekeeping_resume(struct sys_device *dev) xtime.tv_sec += sleep_length; wall_to_monotonic.tv_sec -= sleep_length; + total_sleep_time += sleep_length; }/* re-base the last cycle value */ clock->cycle_last =clocksource_read (clock);
The only other gotcha is that this patch conflicts with a 2.6.22 queued patch in -mm that moves the majority of timekeeping code in kernel/timer.c to kernel/time/timekeeping.c
So it might be easiest to re-diff this against -mm and send it to Andrew for inclusion along with the cleanups. However, since this is a fix, it should have priority over cleanups. I just don't want to upset Andrew's tree too much :)
Your thoughts?
-john
Hi,
Quoted 18 lines · john stultz
kernel/fork.c static struct task_struct *copy_process(unsigned long clone_flags, · line 1062
Quoted 12 lines
kernel/fork.c +1
-0@@ -1056,6 +1056,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); +bbtime_get_ts (&p->real_start_time); p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;Since both
do_posix_clock_monotonic_gettimeandbbtime_get_tsactually read the timekeeping hardware (which could cost around 1.5us each), we probably will want to optimize this down to a single read, or just avoid the hardware read and take lower-granularity xtime value.
Quoted 1 line
However, I'm not sure how fine-grained the
start_timevalues need to be.
We can add a function like monotonic_to_bb probably? Nothing else crossed my mind.
Quoted 8 lines
Quoted 8 lines
kernel/hrtimer.c +0
-0@@ -128,6 +128,33 @@ void ktime_get_ts(struct timespec *ts) }EXPORT_SYMBOL_GPL (ktime_get_ts);
Quoted 12 lines
kernel/hrtimer.c void ktime_get_ts(struct timespec *ts) · line 138
Quoted 8 lines
kernel/hrtimer.c +8
-0+/** +* bbtime_get_ts - get the boot based clock in timespec format +* @ts: pointer to timespec variable +* +* The function calculates the boot based clock from the realtime +* clock, the wall_to_monotonic offset and the total sleep time and +* stores the result in normalized timespec format in the variable +* pointed to by @ts. You might want to clarify this as being the "monotonic boot time", or the time the system has been running including sleep time and that calls to
settimeofday()will not affect this value.
I just copied the text from the ktime_get_ts, but yes, why not.
Quoted 7 lines
kernel/hrtimer.c void ktime_get_ts(struct timespec *ts) · line 140
Quoted 2 lines
kernel/hrtimer.c +2
-0+*/ +void bbtime_get_ts (struct timespec *ts)Any reason you added this in
hrtimer.cinstead oftimer.c? I think keeping the new functions closer together would make the subtle difference between them more clear, and would allowtotal_sleep_timeto be static to one file.
The same as above, I copied the ktime_get_ts and put it under that. Will doing the diff against -mm solve this automagically? :)
Quoted 10 lines
kernel/hrtimer.c void getnstimeofday(struct timespec *ts) · line 840
Quoted 6 lines
kernel/hrtimer.c +2
-0@@ -832,6 +836,21 @@ void getnstimeofday(struct timespec *ts)EXPORT_SYMBOL (getnstimeofday);/** +* getboottime - Return the real time of system boot. +* @ts: pointer to the timespec to be set Might want to clarify this as being the "boot time" based on the current
CLOCK_REALTIMEclock, noting that calls tosettimeofday()would affect this value.
Ok.
Quoted 10 lines
The only other gotcha is that this patch conflicts with a 2.6.22 queued patch in -mm that moves the majority of timekeeping code in
kernel/timer.ctokernel/time/timekeeping.cSo it might be easiest to re-diff this against -mm and send it to Andrew for inclusion along with the cleanups. However, since this is a fix, it should have priority over cleanups. I just don't want to upset Andrew's tree too much :)
Your thoughts?
I'll look at that and will see.
Thanks for the comments,
Signature · 1 line
TJ. (Brno, CZ), BaseOS, Red Hat
PATCH 1/1 Use boot based time for process start time and boot time in /proc
In reply to Tomas JanousekCommit 411187fb05cd11676b0979d9fbf3291db69dbce2 caused boot time to move and process start times to become invalid after suspend. Using boot based time for those restores the old behaviour and fixes the issue.
--- fs/proc/array.c | 5 +++-- fs/proc/proc_misc.c | 8 +++++--- include/linux/sched.h | 2 +- kernel/fork.c | 2 ++ 4 files changed, 11 insertions(+), 6 deletions(-)
fs/proc/array.c +3 -2
@@ -405,8 +405,9 @@ static int do_task_stat(struct task_struct *task, char * buffer, int whole)/* Temporary variable needed for gcc-2.96 */ /* convert timespec -> nsec*/ - start_time = (unsigned long long )task->start_time.tv_sec *NSEC_PER_SEC - + task->start_time.tv_nsec; + start_time = + (unsigned long long )task->real_start_time.tv_sec *NSEC_PER_SEC + + task->real_start_time.tv_nsec;/* convert nsec -> ticks */ start_time =nsec_to_clock_t (start_time);
fs/proc/proc_misc.c +5 -3
@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v)unsigned long jif;cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;u64 sum =0 ; +struct timespec boottime; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; -if (wall_to_monotonic.tv_nsec) - --jif; +getboottime (&boottime); + jif = boottime.tv_sec; +if (boottime.tv_nsec) + ++jif;for_each_possible_cpu (i) {int j;
include/linux/sched.h +1 -1
@@ -918,7 +918,7 @@ struct task_struct {unsigned int rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; +struct timespec start_time, real_start_time;/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ unsigned long min_flt, maj_flt;
kernel/fork.c +2 -0
@@ -1059,6 +1059,8 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); + p->real_start_time = p->start_time; +monotonic_to_bootbased (&p->real_start_time); p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;
Signature · 5 lines
1.5.1.4 -- TJ. (Brno, CZ), BaseOS, Red Hat
PATCH 1/1 Use boot based time for uptime in /proc
In reply to Tomas JanousekCommit 411187fb05cd11676b0979d9fbf3291db69dbce2 caused uptime not to increase during suspend. This may cause confusion so I restore the old behaviour by using the boot based time instead of monotonic for uptime.
--- fs/proc/proc_misc.c | 1 + kernel/timer.c | 1 + 2 files changed, 2 insertions(+), 0 deletions(-)
fs/proc/proc_misc.c +1 -0
@@ -107,6 +107,7 @@ static int uptime_read_proc(char *page, char **start, off_t off,cputime_t idletime =cputime_add (init_task.utime, init_task.stime);do_posix_clock_monotonic_gettime (&uptime); +monotonic_to_bootbased (&uptime);cputime_to_timespec (idletime, &idle); len = sprintf(page,"%lu.%02lu %lu.%02lu\n", (unsigned long) uptime.tv_sec,
kernel/timer.c +1 -0
@@ -1106,6 +1106,7 @@ int do_sysinfo(struct sysinfo *info)getnstimeofday (&tp); tp.tv_sec += wall_to_monotonic.tv_sec; tp.tv_nsec += wall_to_monotonic.tv_nsec; +monotonic_to_bootbased (&tp);if (tp.tv_nsec -NSEC_PER_SEC >=0 ) { tp.tv_nsec = tp.tv_nsec -NSEC_PER_SEC ; tp.tv_sec++;
Signature · 5 lines
1.5.1.4 -- TJ. (Brno, CZ), BaseOS, Red Hat
PATCH 1/1 Introduce boot based time
In reply to john stultzThe commits
411187fb05cd11676b0979d9fbf3291db69dbce2 (GTOD: persistent clock support) c1d370e167d66b10bca3b602d3740405469383de (i386: use GTOD persistent clock
support)
changed the monotonic time so that it no longer jumps after resume, but it's not possible to use it for boot time and process start time calculations then. Also, the uptime no longer increases during suspend.
I add a variable to track the wall_to_monotonic changes, a function to get the real boot time and a function to get the boot based time from the monotonic one.
--- include/linux/time.h | 2 ++ kernel/time/timekeeping.c | 41 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 43 insertions(+), 0 deletions(-)
include/linux/time.h +2 -0
@@ -116,6 +116,8 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts); +extern void monotonic_to_bootbased (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );
kernel/time/timekeeping.c +41 -0
65 lines not on this page · raw message
Signature · 5 lines
1.5.1.4 -- TJ. (Brno, CZ), BaseOS, Red Hat
Re: [PATCH] Introduce boot based time
Quoted 8 lines · Tomas Janousek
The commits
411187fb05cd11676b0979d9fbf3291db69dbce2(GTOD: persistent clock support)c1d370e167d66b10bca3b602d3740405469383de(i386: use GTOD persistent clock
support)
changed the monotonic time so that it no longer jumps after resume, but it's not possible to use it for boot time and process start time calculations then. Also, the uptime no longer increases during suspend.
Quoted 7 lines
I add a variable to track the
wall_to_monotonicchanges, a function to get the real boot time and a function to get the boot based time from the monotonic one.Signed-off-by Tomas Janousek <[email protected]>Cc Tomas Smetana <[email protected]>Cc John Stultz <[email protected]>
Looks good! Thanks again for catching this!
Quoted 95 lines
--- include/linux/time.h | 2 ++ kernel/time/timekeeping.c | 41 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 43 insertions(+), 0 deletions(-)include/linux/time.h +2
-0@@ -116,6 +116,8 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts); +extern void monotonic_to_bootbased (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );kernel/time/timekeeping.c +41
-0@@ -36,9 +36,17 @@ EXPORT_SYMBOL(xtime_lock);* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* +* wall_to_monotonic is moved after resume from suspend for the monotonic +* time not to jump. We need to add total_sleep_time to wall_to_monotonic +* to get the real boot based time offset. +* +* - wall_to_monotonic is no longer the boot time, getboottime must be +* used instead. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +static unsigned long total_sleep_time;EXPORT_SYMBOL (xtime);@@ -251,6 +259,7 @@ void __init timekeeping_init(void) xtime.tv_nsec =0 ;set_normalized_timespec (&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); + total_sleep_time =0 ;write_sequnlock_irqrestore (&xtime_lock, flags); }@@ -280,6 +289,7 @@ static int timekeeping_resume(struct sys_device *dev) xtime.tv_sec += sleep_length; wall_to_monotonic.tv_sec -= sleep_length; + total_sleep_time += sleep_length; }/* re-base the last cycle value */ clock->cycle_last =clocksource_read (clock);@@ -474,3 +484,34 @@ void update_wall_time(void)change_clocksource ();update_vsyscall (&xtime, clock); } + +/** +* getboottime - Return the real time of system boot. +* @ts: pointer to the timespec to be set +* +* Returns the time of day in a timespec. +* +* This is based on the wall_to_monotonic offset and the total suspend +* time. Calls to settimeofday will affect the value returned (which +* basically means that however wrong your real time clock is at boot time, +* you get the right time here). +*/ +void getboottime (struct timespec *ts) +{ +set_normalized_timespec (ts, + - (wall_to_monotonic.tv_sec + total_sleep_time), + - wall_to_monotonic.tv_nsec); +} + +EXPORT_SYMBOL (getboottime); + +/** +* monotonic_to_bootbased - Convert the monotonic time to boot based. +* @ts: pointer to the timespec to be converted +*/ +void monotonic_to_bootbased (struct timespec *ts) +{ + ts->tv_sec += total_sleep_time; +} + +EXPORT_SYMBOL (monotonic_to_bootbased);Signature · 1 line
1.5.1.4
Re: [PATCH] Use boot based time for process start time and boot time in /proc
In reply to Tomas JanousekQuoted 7 lines · Tomas Janousek
Commit
411187fb05cd11676b0979d9fbf3291db69dbce2caused boot time to move and process start times to become invalid after suspend. Using boot based time for those restores the old behaviour and fixes the issue.Signed-off-by Tomas Janousek <[email protected]>Cc Tomas Smetana <[email protected]>Cc John Stultz <[email protected]>
Quoted 75 lines
--- fs/proc/array.c | 5 +++-- fs/proc/proc_misc.c | 8 +++++--- include/linux/sched.h | 2 +- kernel/fork.c | 2 ++ 4 files changed, 11 insertions(+), 6 deletions(-)fs/proc/array.c +3
-2@@ -405,8 +405,9 @@ static int do_task_stat(struct task_struct *task, char * buffer, int whole)/* Temporary variable needed for gcc-2.96 */ /* convert timespec -> nsec*/ - start_time = (unsigned long long )task->start_time.tv_sec *NSEC_PER_SEC - + task->start_time.tv_nsec; + start_time = + (unsigned long long )task->real_start_time.tv_sec *NSEC_PER_SEC + + task->real_start_time.tv_nsec;/* convert nsec -> ticks */ start_time =nsec_to_clock_t (start_time);fs/proc/proc_misc.c +5
-3@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v)unsigned long jif;cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;u64 sum =0 ; +struct timespec boottime; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; -if (wall_to_monotonic.tv_nsec) - --jif; +getboottime (&boottime); + jif = boottime.tv_sec; +if (boottime.tv_nsec) + ++jif;for_each_possible_cpu (i) {int j;include/linux/sched.h +1
-1@@ -918,7 +918,7 @@ struct task_struct {unsigned int rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; +struct timespec start_time, real_start_time;/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ unsigned long min_flt, maj_flt;kernel/fork.c +2
-0@@ -1059,6 +1059,8 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->lock_depth =-1 ;/* -1 = no lock */ do_posix_clock_monotonic_gettime (&p->start_time); + p->real_start_time = p->start_time; +monotonic_to_bootbased (&p->real_start_time); p->security =NULL ; p->io_context =NULL ; p->io_wait =NULL ;Signature · 1 line
1.5.1.4
Re: [PATCH] Use boot based time for uptime in /proc
In reply to Tomas JanousekQuoted 5 lines · Tomas Janousek
Commit
411187fb05cd11676b0979d9fbf3291db69dbce2caused uptime not to increase during suspend. This may cause confusion so I restore the old behaviour by using the boot based time instead of monotonic for uptime.Signed-off-by Tomas Janousek <[email protected]>
Quoted 33 lines
--- fs/proc/proc_misc.c | 1 + kernel/timer.c | 1 + 2 files changed, 2 insertions(+), 0 deletions(-)fs/proc/proc_misc.c +1
-0@@ -107,6 +107,7 @@ static int uptime_read_proc(char *page, char **start, off_t off,cputime_t idletime =cputime_add (init_task.utime, init_task.stime);do_posix_clock_monotonic_gettime (&uptime); +monotonic_to_bootbased (&uptime);cputime_to_timespec (idletime, &idle); len = sprintf(page,"%lu.%02lu %lu.%02lu\n", (unsigned long) uptime.tv_sec,kernel/timer.c +1
-0@@ -1106,6 +1106,7 @@ int do_sysinfo(struct sysinfo *info)getnstimeofday (&tp); tp.tv_sec += wall_to_monotonic.tv_sec; tp.tv_nsec += wall_to_monotonic.tv_nsec; +monotonic_to_bootbased (&tp);if (tp.tv_nsec -NSEC_PER_SEC >=0 ) { tp.tv_nsec = tp.tv_nsec -NSEC_PER_SEC ; tp.tv_sec++;Signature · 1 line
1.5.1.4
Re: [PATCH] Introduce boot based time
In reply to Tomas JanousekHi Tomas,
Quoted 22 lines · Tomas Janousek
include/linux/time.h +2
-0@@ -116,6 +116,8 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts); +extern void monotonic_to_bootbased (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );kernel/time/timekeeping.c +1
-0@@ -36,9 +36,17 @@ EXPORT_SYMBOL(xtime_lock);* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +*
kernel/time/timekeeping.c EXPORT_SYMBOL(xtime_lock); · line 49
Quoted 10 lines
kernel/time/timekeeping.c +7
-0+* wall_to_monotonic is moved after resume from suspend for the monotonic +* time not to jump. We need to add total_sleep_time to wall_to_monotonic +* to get the real boot based time offset. +* +* - wall_to_monotonic is no longer the boot time, getboottime must be +* used instead. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +static unsigned long total_sleep_time;
Could you make that a ktime_t (or struct ktime)? There are machines, which sleep more than they are awake. Just imagine a surveillance camera triggered by door entrance.
Yes, these things might run Linux (e.g. on "cris" architecture).
Or your VCR.
Yes, these devices might sleep more than they are awake, if you are not a TV junkie :-)
Best regards
Ingo Oeser
Re: [PATCH] Introduce boot based time
Hi Ingo,
Quoted 11 lines · Ingo Oeser
Quoted 1 line
+static unsigned long
total_sleep_time;Could you make that a
ktime_t(or struct ktime)? There are machines, which sleep more than they are awake. Just imagine a surveillance camera triggered by door entrance.Yes, these things might run Linux (e.g. on "cris" architecture).
Or your VCR.
Yes, these devices might sleep more than they are awake, if you are not a TV junkie :-)
The value is in seconds. It can hold up to 136 years of sleep. I'd rather spend the money on beer than on a VCR that I'm gonna suspend for 137 years :)
(or do we have a 16-bit long anywhere?)
Regards,
Signature · 1 line
TJ. (Brno, CZ), BaseOS, Red Hat
Re: [PATCH] Introduce boot based time
In reply to Ingo OeserQuoted 40 lines · Ingo Oeser
Hi Tomas,
Quoted 22 lines · Tomas Janousek
include/linux/time.h +2
-0@@ -116,6 +116,8 @@ extern int do_setitimer(int which, struct itimerval *value,extern unsigned int alarm_setitimer (unsigned int seconds);extern int do_getitimer (int which,struct itimerval *value);extern void getnstimeofday (struct timespec *tv); +extern void getboottime (struct timespec *ts); +extern void monotonic_to_bootbased (struct timespec *ts);extern struct timespec timespec_trunc (struct timespec t,unsigned gran);extern int timekeeping_is_continuous (void );kernel/time/timekeeping.c +1
-0@@ -36,9 +36,17 @@ EXPORT_SYMBOL(xtime_lock);* at zero at system boot time, so wall_to_monotonic will be negative, * however, we will ALWAYS keep the tv_nsec part positive so we can use * the usual normalization. +* kernel/time/timekeeping.c EXPORT_SYMBOL(xtime_lock); · line 49
Quoted 10 lines
kernel/time/timekeeping.c +7
-0+* wall_to_monotonic is moved after resume from suspend for the monotonic +* time not to jump. We need to add total_sleep_time to wall_to_monotonic +* to get the real boot based time offset. +* +* - wall_to_monotonic is no longer the boot time, getboottime must be +* used instead. */ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); +static unsigned long total_sleep_time;Could you make that a
ktime_t(or struct ktime)? There are machines, which sleep more than they are awake. Just imagine a surveillance camera triggered by door entrance.
Quoted 5 lines
Yes, these things might run Linux (e.g. on "cris" architecture).
Or your VCR.
Yes, these devices might sleep more than they are awake, if you are not a TV junkie :-)
I'm not sure I follow this.
total_sleep_time stores seconds. So on 32bit systems that's 130some years, so it shouldn't be an issue.
Is the reason you want it to be a ktime is because you want a way to keep sub-second sleep granularity?
thanks
-john
Re: [PATCH] Introduce boot based time
Hi John and Tomas,
Quoted 7 lines · john stultz
I'm not sure I follow this.
total_sleep_timestores seconds. So on 32bit systems that's 130some years, so it shouldn't be an issue.Is the reason you want it to be a ktime is because you want a way to keep sub-second sleep granularity?
No, I'm just overworked and getting sloppy :-/
Sorry for the noise...
Best regards
Ingo Oeser
Andrew Morton
maintainerRe: [PATCH] Introduce boot based time
In reply to Tomas JanousekQuoted 11 lines · On Thu, 10 May 2007 19:10:25 +0200 Tomas Janousek <[email protected]> wrote:
The commits
411187fb05cd11676b0979d9fbf3291db69dbce2(GTOD: persistent clock support)c1d370e167d66b10bca3b602d3740405469383de(i386: use GTOD persistent clock
support)
changed the monotonic time so that it no longer jumps after resume, but it's not possible to use it for boot time and process start time calculations then. Also, the uptime no longer increases during suspend.I add a variable to track the
wall_to_monotonicchanges, a function to get the real boot time and a function to get the boot based time from the monotonic one.
From: Andrew Morton <[email protected]>
- I don't think those sybols are needed in modules.
- Document total_sleep_time units (would have been better to call it total_sleep_time_secs, perhaps).
---
kernel/time/timekeeping.c | 6 +----- 1 files changed, 1 insertion(+), 5 deletions(-)
diff -puN include/linux/time.h~introduce-boot-based-time-fix include/linux/time.h
kernel/time/timekeeping.c renamed from kernel/time/timekeeping.c~introduce-boot-based-time-fix +1 -5
@@ -46,7 +46,7 @@ EXPORT_SYMBOL(xtime_lock);*/ struct timespec xtime __attribute__ ((aligned (16))); struct timespec wall_to_monotonic __attribute__ ((aligned (16))); -static unsigned long total_sleep_time; +static unsigned long total_sleep_time;/* seconds */ EXPORT_SYMBOL (xtime);
@@ -503,8 +503,6 @@ void getboottime(struct timespec *ts) - wall_to_monotonic.tv_nsec); } -EXPORT_SYMBOL (getboottime); -/** * monotonic_to_bootbased - Convert the monotonic time to boot based. * @ts: pointer to the timespec to be converted
@@ -513,5 +511,3 @@ void monotonic_to_bootbased(struct times { ts->tv_sec += total_sleep_time; } - -EXPORT_SYMBOL (monotonic_to_bootbased);
_
Andrew Morton
maintainerRe: [PATCH] Use boot based time for process start time and boot time in /proc
In reply to Tomas JanousekQuoted 12 lines · On Thu, 10 May 2007 19:10:42 +0200 Tomas Janousek <[email protected]> wrote:
Commit
411187fb05cd11676b0979d9fbf3291db69dbce2caused boot time to move and process start times to become invalid after suspend. Using boot based time for those restores the old behaviour and fixes the issue...
fs/proc/proc_misc.c +1
-0@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v)unsigned long jif;cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;u64 sum =0 ; +struct timespec boottime;
fs/proc/proc_misc.c static int show_stat(struct seq_file *p, void *v) · line 456
Quoted 10 lines
fs/proc/proc_misc.c +4
-3user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; -if (wall_to_monotonic.tv_nsec) - --jif; +getboottime (&boottime); + jif = boottime.tv_sec; +if (boottime.tv_nsec) + ++jif;
Is the switch from --jif to ++jif a functional change? If so, how come?
include/linux/sched.h struct task_struct { · line 921
Quoted 12 lines
fs/proc/proc_misc.c +0
-0for_each_possible_cpu (i) {int j;include/linux/sched.h +1
-1@@ -918,7 +918,7 @@ struct task_struct {unsigned int rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; +struct timespec start_time, real_start_time;
no, please prefer to do
struct timespec start_time; struct timespec real_start_time;
which gives a nice place to add a comment documenting the field.
Please document fields.
What is the difference between start_time and real_start_time?
Re: [PATCH] Use boot based time for process start time and boot time in /proc
Hello,
Quoted 7 lines · Andrew Morton
Quoted 6 lines · Tomas Janousek
(quoted diff) +1
-0@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v) unsigned long jif; cputime64_t user, nice, system, idle, iowait, irq, softirq, steal; u64 sum = 0; + struct timespec boottime;
Quoted 12 lines
Review static int show_stat(struct seq_file *p, void *v) · line 456
Quoted 10 lines
(quoted diff) +4
-3user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; - if (wall_to_monotonic.tv_nsec) - --jif; + getboottime(&boottime); + jif = boottime.tv_sec; + if (boottime.tv_nsec) + ++jif;Is the switch from --jif to ++jif a functional change? If so, how come?
Yes, I'm afraid it was.
-----------------------
From: Tomas Janousek <[email protected]>
Date: Fri, 11 May 2007 10:34:55 +0200
Subject: [PATCH] Lower the boot time to seconds, not ceil
Restores the original behaviour of boot time calculation.
--- fs/proc/proc_misc.c | 2 -- 1 files changed, 0 insertions(+), 2 deletions(-)
fs/proc/proc_misc.c +0 -2
@@ -452,8 +452,6 @@ static int show_stat(struct seq_file *p, void *v) irq = softirq = steal = cputime64_zero;getboottime (&boottime); jif = boottime.tv_sec; -if (boottime.tv_nsec) - ++jif;for_each_possible_cpu (i) {int j;
Signature · 4 lines
1.5.1.4 -- Tomas Janousek, SW Engineer, Red Hat, Inc.
Re: [PATCH] Use boot based time for process start time and boot time in /proc
In reply to Andrew MortonHi,
Quoted 11 lines · Andrew Morton
Review struct task_struct { · line 921
Quoted 6 lines · On Thu, 10 May 2007 19:10:42 +0200 Tomas Janousek <[email protected]> wrote:
(quoted diff) +1
-1@@ -918,7 +918,7 @@ struct task_struct { unsigned int rt_priority; cputime_t utime, stime; unsigned long nvcsw, nivcsw; /* context switch counts */ - struct timespec start_time; + struct timespec start_time, real_start_time;no, please prefer to do
Quoted 8 lines
struct timespec start_time; struct timespec real_start_time;which gives a nice place to add a comment documenting the field.
Please document fields.
What is the difference between
start_timeandreal_start_time?
-------------------------
From: Tomas Janousek <[email protected]>
Date: Fri, 11 May 2007 10:38:47 +0200
Subject: [PATCH] Document start times in struct task_struct
--- include/linux/sched.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
include/linux/sched.h +2 -2
@@ -918,8 +918,8 @@ struct task_struct {unsigned int rt_priority;cputime_t utime, stime;unsigned long nvcsw, nivcsw;/* context switch counts */ -struct timespec start_time; -struct timespec real_start_time; +struct timespec start_time;/* monotonic time */ +struct timespec real_start_time;/* boot based time */ /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ unsigned long min_flt, maj_flt;
Signature · 5 lines
1.5.1.4 -- TJ. (Brno, CZ), BaseOS, Red Hat
Andrew Morton
maintainerRe: [PATCH] Use boot based time for process start time and boot time in /proc
In reply to Tomas JanousekQuoted 45 lines · On Fri, 11 May 2007 10:45:31 +0200 Tomas Janousek <[email protected]> wrote:
Hello,
Quoted 7 lines · Andrew Morton
Quoted 6 lines · Tomas Janousek
(quoted diff) +1
-0@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v) unsigned long jif; cputime64_t user, nice, system, idle, iowait, irq, softirq, steal; u64 sum = 0; + struct timespec boottime;Quoted 12 lines
Review static int show_stat(struct seq_file *p, void *v) · line 456
Quoted 10 lines
(quoted diff) +4
-3user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; - if (wall_to_monotonic.tv_nsec) - --jif; + getboottime(&boottime); + jif = boottime.tv_sec; + if (boottime.tv_nsec) + ++jif;Is the switch from --jif to ++jif a functional change? If so, how come?
Yes, I'm afraid it was.
-----------------------
From: Tomas Janousek <[email protected]>
Date: Fri, 11 May 2007 10:34:55 +0200
Subject: [PATCH] Lower the boot time to seconds, not ceilRestores the original behaviour of boot time calculation.
Signed-off-by Tomas Janousek <[email protected]>Cc Andrew Morton <[email protected]>Cc Tomas Smetana <[email protected]>Cc John Stultz <[email protected]>--- fs/proc/proc_misc.c | 2 -- 1 files changed, 0 insertions(+), 2 deletions(-)fs/proc/proc_misc.c +0
-0
fs/proc/proc_misc.c static int show_stat(struct seq_file *p, void *v) · line 457
Quoted 8 lines
fs/proc/proc_misc.c +0
-2@@ -452,8 +452,6 @@ static int show_stat(struct seq_file *p, void *v) irq = softirq = steal = cputime64_zero;getboottime (&boottime); jif = boottime.tv_sec; -if (boottime.tv_nsec) - ++jif;for_each_possible_cpu (i) {int j;
So we've gone from --jif to ++jif to no change at all.
Are you sure that this net removal of --jif is correct?
Re: [PATCH] Use boot based time for process start time and boot time in /proc
Hi,
Quoted 33 lines · Andrew Morton
Quoted 23 lines · On Fri, 11 May 2007 10:45:31 +0200 Tomas Janousek <[email protected]> wrote:
Hello,
Quoted 18 lines · Andrew Morton
Quoted 16 lines · Tomas Janousek
(quoted diff) +5
-3@@ -445,12 +445,14 @@ static int show_stat(struct seq_file *p, void *v) unsigned long jif; cputime64_t user, nice, system, idle, iowait, irq, softirq, steal; u64 sum = 0; + struct timespec boottime; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; - jif = - wall_to_monotonic.tv_sec; - if (wall_to_monotonic.tv_nsec) - --jif; + getboottime(&boottime); + jif = boottime.tv_sec; + if (boottime.tv_nsec) + ++jif;fs/proc/proc_misc.c static int show_stat(struct seq_file *p, void *v) · line 456
Quoted 4 lines
fs/proc/proc_misc.c +0
-2getboottime (&boottime); jif = boottime.tv_sec; -if (boottime.tv_nsec) - ++jif;So we've gone from --jif to ++jif to no change at all.
Quoted 1 line
Are you sure that this net removal of --jif is correct?
Yes.
Let's say wall_to_monotonic = { -10, 500000 } (which is { -9, -500000 }, and the original code would result in - (- 10) - 1 == 9).
The getboottime calls set_normalized_timespec on { - (-10), - (500000) } which results to { 10 - 1, - 500000 + 1000000 } = { 9, 500000 }.tv_sec == 9 => correct.
Signature · 1 line
TJ. (Brno, CZ), BaseOS, Red Hat