timekeeping: Add persistent_clock_exist flag
PATCH v2 1/3 timekeeping: Add persistent_clock_exist flag
In current kernel, there are several places which need to check whether there is a persistent clock for the platform. Current check is done by calling the read_persistent_clock() and validating its return value.
So one optimization is to do the check only once in timekeeping_init(), and use a flag persistent_clock_exist to record it.
v2: Add a has_persistent_clock() helper function, as suggested by John.
--- include/linux/time.h | 6 ++++++ kernel/time/timekeeping.c | 16 +++++++++++----- 2 files changed, 17 insertions(+), 5 deletions(-)
include/linux/time.h +6 -0
@@ -115,6 +115,12 @@ static inline bool timespec_valid_strict(const struct timespec *ts)return true; } +extern bool persistent_clock_exist; +static inline bool has_persistent_clock (void ) +{ +return persistent_clock_exist; +} +extern void read_persistent_clock (struct timespec *ts);extern void read_boot_clock (struct timespec *ts);extern int update_persistent_clock (struct timespec now);
kernel/time/timekeeping.c +11 -5
@@ -29,6 +29,9 @@ static struct timekeeper timekeeper;/* flag for if timekeeping is suspended */ int __read_mostly timekeeping_suspended; +/* Flag for if there is a persistent clock on this platform */ +bool __read_mostly persistent_clock_exist = false; +static inline void tk_normalize_xtime (struct timekeeper *tk) {while (tk->xtime_nsec >= ((u64 )NSEC_PER_SEC << tk->shift)) {
@@ -640,12 +643,14 @@ void __init timekeeping_init(void)struct timespec now, boot, tmp;read_persistent_clock (&now); +if (!timespec_valid_strict (&now)) {pr_warn ("WARNING: Persistent clock returned invalid value!\n" " Check your CMOS/BIOS settings.\n" ); now.tv_sec =0 ; now.tv_nsec =0 ; - } + }else if (now.tv_sec || now.tv_nsec) + persistent_clock_exist = true;read_boot_clock (&boot);if (!timespec_valid_strict (&boot)) {
@@ -718,11 +723,12 @@ void timekeeping_inject_sleeptime(struct timespec *delta) {struct timekeeper *tk = &timekeeper;unsigned long flags; -struct timespec ts; -/* Make sure we don't set the clock twice */ -read_persistent_clock (&ts); -if (!(ts.tv_sec ==0 && ts.tv_nsec ==0 )) +/* +* Make sure we don't set the clock twice, as timekeeping_resume() +* already did it +*/ +if (has_persistent_clock ())return ;write_seqlock_irqsave (&tk->lock, flags);
Signature · 1 line
1.7.9.5
PATCH v2 2/3 rtc: Skip the suspend/resume handling if persistent clock exist
All the RTC suspend and resume functions are to compensate the sleep time, but this is already done in timekeeping.c if persistent clock exist.
--- drivers/rtc/class.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/rtc/class.c +7 -0
@@ -50,6 +50,10 @@ static int rtc_suspend(struct device *dev, pm_message_t mesg)struct rtc_device *rtc =to_rtc_device (dev);struct rtc_time tm;struct timespec delta, delta_delta; + +if (has_persistent_clock ()) +return 0 ; +if (strcmp (dev_name (&rtc->dev),CONFIG_RTC_HCTOSYS_DEVICE ) !=0 )return 0 ;
@@ -88,6 +92,9 @@ static int rtc_resume(struct device *dev)struct timespec new_system, new_rtc;struct timespec sleep_time; +if (has_persistent_clock ()) +return 0 ; + rtc_hctosys_ret = -ENODEV ;if (strcmp (dev_name (&rtc->dev),CONFIG_RTC_HCTOSYS_DEVICE ) !=0 )return 0 ;
Signature · 1 line
1.7.9.5
John Stultz
AppliedRe: [PATCH v2 2/3] rtc: Skip the suspend/resume handling if persistent clock exist
Quoted 4 lines · Feng Tang
All the RTC suspend and resume functions are to compensate the sleep time, but this is already done in
timekeeping.cif persistent clock exist.
Applied. Thanks!
-john
PATCH v2 3/3 timekeeping: Add CONFIG_HAS_PERSISTENT_CLOCK option
In reply to Feng TangMake the persistent clock check a kernel config option, so that some platform can explicitely select it, also make CONFIG_RTC_HCTOSYS depends on its non-existence, which could prevent the persistent clock and RTC code from doing similar thing twice during system's init/suspend/resume phases.
If the CONFIG_HAS_PERSISTENT_CLOCK=n, then no change happens for kernel which still does the persistent clock check in timekeeping_init().
--- drivers/rtc/Kconfig | 1 + include/linux/time.h | 5 +++++ kernel/time/Kconfig | 5 +++++ 3 files changed, 11 insertions(+)
drivers/rtc/Kconfig +1 -0
@@ -20,6 +20,7 @@ if RTC_CLASS config RTC_HCTOSYS bool "Set system time from RTC on startup and resume" default y + depends on !HAS_PERSISTENT_CLOCK help If you say yes here, the system time (wall clock) will be set using the value read from a specified RTC device. This is useful to avoid
include/linux/time.h +5 -0
@@ -116,10 +116,15 @@ static inline bool timespec_valid_strict(const struct timespec *ts) }extern bool persistent_clock_exist; + +#ifdef CONFIG_HAS_PERSISTENT_CLOCK +#define has_persistent_clock () true +#else static inline bool has_persistent_clock (void ) {return persistent_clock_exist; } +#endif extern void read_persistent_clock (struct timespec *ts);extern void read_boot_clock (struct timespec *ts);
kernel/time/Kconfig +5 -0
@@ -12,6 +12,11 @@ config CLOCKSOURCE_WATCHDOG config ARCH_CLOCKSOURCE_DATA bool +# Platforms has a persistent clock +config HAS_PERSISTENT_CLOCK + bool + default n + # Timekeeping vsyscall support config GENERIC_TIME_VSYSCALL bool
Signature · 1 line
1.7.9.5
John Stultz
AppliedRe: [PATCH v2 3/3] timekeeping: Add CONFIG_HAS_PERSISTENT_CLOCK option
Quoted 6 lines · Feng Tang
Make the persistent clock check a kernel config option, so that some platform can explicitely select it, also make
CONFIG_RTC_HCTOSYSdepends on its non-existence, which could prevent the persistent clock and RTC code from doing similar thing twice during system'sinit/suspend/resumephases.
Quoted 6 lines
If the
CONFIG_HAS_PERSISTENT_CLOCK=n, then no change happens for kernel which still does the persistent clock check intimekeeping_init().Cc Thomas Gleixner <[email protected]>Suggested-by John Stultz <[email protected]>Signed-off-by Feng Tang <[email protected]>
Applied. I also added a dependency for Jason's CONFIG_RTC_SYSTOHC.
thanks
-john
Re: [PATCH v2 3/3] timekeeping: Add CONFIG_HAS_PERSISTENT_CLOCK option
Quoted 7 lines · John Stultz
Quoted 6 lines · Feng Tang
Make the persistent clock check a kernel config option, so that some platform can explicitely select it, also make
CONFIG_RTC_HCTOSYSdepends on its non-existence, which could prevent the persistent clock and RTC code from doing similar thing twice during system'sinit/suspend/resumephases.
Quoted 8 lines
Quoted 6 lines
If the
CONFIG_HAS_PERSISTENT_CLOCK=n, then no change happens for kernel which still does the persistent clock check intimekeeping_init().Cc Thomas Gleixner <[email protected]>Suggested-by John Stultz <[email protected]>Signed-off-by Feng Tang <[email protected]>Applied. I also added a dependency for Jason's
CONFIG_RTC_SYSTOHC.
Sort of an ugly config name, since I gather ARM should always set this to 'n'...
CONFIG_USE_ONLY_PERSISTENT_CLOCK ?
Cheers,
Jason
Re: [PATCH v2 3/3] timekeeping: Add CONFIG_HAS_PERSISTENT_CLOCK option
Quoted 11 lines · Jason Gunthorpe
Quoted 10 lines · John Stultz
Quoted 9 lines · Feng Tang
Make the persistent clock check a kernel config option, so that some platform can explicitely select it, also make
CONFIG_RTC_HCTOSYSdepends on its non-existence, which could prevent the persistent clock and RTC code from doing similar thing twice during system'sinit/suspend/resumephases.If the
CONFIG_HAS_PERSISTENT_CLOCK=n, then no change happens for kernel which still does the persistent clock check intimekeeping_init().
Quoted 8 lines
Quoted 4 lines
Quoted 3 lines
Cc Thomas Gleixner <[email protected]>Suggested-by John Stultz <[email protected]>Signed-off-by Feng Tang <[email protected]>Applied. I also added a dependency for Jason's
CONFIG_RTC_SYSTOHC.Sort of an ugly config name, since I gather ARM should always set this to 'n'...
CONFIG_USE_ONLY_PERSISTENT_CLOCK?
(Sigh. I got this seemingly microseconds after I sent the pull request :)
So yea, fair point, there could be some confusion. But ONLY_PERSISTENT_CLOCK isn't quite right either, more like CONFIG_HAS_PERSISTENT_CLOCK_ALWAYS or something.
Hrm. Let me think on it for a bit, and feel free to suggest further improvements.
thanks
-john
Re: [PATCH v2 3/3] timekeeping: Add CONFIG_HAS_PERSISTENT_CLOCK option
Quoted 15 lines · John Stultz
Quoted 14 lines · Jason Gunthorpe
Quoted 13 lines · John Stultz
Quoted 12 lines · Feng Tang
Make the persistent clock check a kernel config option, so that some platform can explicitely select it, also make
CONFIG_RTC_HCTOSYSdepends on its non-existence, which could prevent the persistent clock and RTC code from doing similar thing twice during system'sinit/suspend/resumephases.If the
CONFIG_HAS_PERSISTENT_CLOCK=n, then no change happens for kernel which still does the persistent clock check intimekeeping_init().Cc Thomas Gleixner <[email protected]>Suggested-by John Stultz <[email protected]>Signed-off-by Feng Tang <[email protected]>
Quoted 9 lines
Quoted 5 lines
Quoted 1 line
Applied. I also added a dependency for Jason's
CONFIG_RTC_SYSTOHC.Sort of an ugly config name, since I gather ARM should always set this to 'n'...
CONFIG_USE_ONLY_PERSISTENT_CLOCK?(Sigh. I got this seemingly microseconds after I sent the pull request :)
So yea, fair point, there could be some confusion. But
ONLY_PERSISTENT_CLOCKisn't quite right either, more likeCONFIG_HAS_PERSISTENT_CLOCK_ALWAYSor something.
Decided upon CONFIG_ALWAYS_USE_PERSISTENT_CLOCK which I think is clear enough.
Let me know if you object or have a better idea.
thanks
-john
John Stultz
AppliedQuoted 9 lines · Feng Tang
In current kernel, there are several places which need to check whether there is a persistent clock for the platform. Current check is done by calling the
read_persistent_clock()and validating its return value.So one optimization is to do the check only once in
timekeeping_init(), and use a flagpersistent_clock_existto record it.v2: Add a
has_persistent_clock()helper function, as suggested by John.
Applied! Thanks
-john