timekeeping: cleanup _fast_ variety of functions
Subsystems (16)
cover timekeeping: cleanup _fast_ variety of functions
Hi Thomas,
When Arnd and I discussed this prior, he thought it best that I separate these two commits out into a separate patchset, because they might require additional discussion or consideration from you. They seem straightforward enough to me, but if deliberations require me to make some tweaks, I'm happy to do so.
Thanks,
Jason
Jason A. Donenfeld (2):
timekeeping: add missing non-_ns functions for fast accessors
timekeeping: rename fast getter functions to be consistent
Documentation/core-api/timekeeping.rst | 15 ++++--- Documentation/trace/ftrace.rst | 2 +- drivers/base/power/runtime.c | 12 +++--- drivers/gpu/drm/i915/i915_perf.c | 2 +- fs/pstore/platform.c | 2 +- include/linux/pm_runtime.h | 2 +- include/linux/timekeeping.h | 28 +++++++++++-- kernel/bpf/helpers.c | 2 +- kernel/debug/kdb/kdb_main.c | 2 +- kernel/events/core.c | 4 +- kernel/rcu/rcuperf.c | 6 +-- kernel/rcu/srcutree.c | 6 +-- kernel/time/timekeeping.c | 56 +++++++++++++------------- kernel/trace/trace.c | 6 +-- kernel/watchdog_hld.c | 2 +- 15 files changed, 85 insertions(+), 62 deletions(-)
Signature · 1 line
2.21.0
PATCH 1/2 timekeeping: add missing non-_ns functions for fast accessors
Previously there was no analogue to get proper ktime_t versions of the fast variety of ktime invocations. This commit makes the interface uniform with the other accessors.
--- Documentation/core-api/timekeeping.rst | 7 +++- include/linux/timekeeping.h | 28 ++++++++++++-- kernel/time/timekeeping.c | 52 +++++++++++++------------- 3 files changed, 55 insertions(+), 32 deletions(-)
Documentation/core-api/timekeeping.rst +6 -1
@@ -94,7 +94,7 @@ different format depending on what is required by the user: down the seconds to the full seconds of the last timer tick using the respective reference. -Coarse and fast_ns access +Coarse and fast access ------------------------- Some additional variants exist for more specialized cases:
@@ -129,6 +129,11 @@ Some additional variants exist for more specialized cases: up to several microseconds on older hardware with an external clocksource. +.. c:function:: ktime_t ktime_get_mono_fast( void ) + ktime_t ktime_get_raw_fast( void ) + ktime_t ktime_get_boottime_fast( void ) + ktime_t ktime_get_real_fast( void ) + .. c:function:: u64 ktime_get_mono_fast_ns( void ) u64 ktime_get_raw_fast_ns( void ) u64 ktime_get_boot_fast_ns( void )
include/linux/timekeeping.h +24 -4
@@ -174,10 +174,30 @@ static inline u64 ktime_get_raw_ns(void)return ktime_to_ns (ktime_get_raw ()); } -extern u64 ktime_get_mono_fast_ns (void ); -extern u64 ktime_get_raw_fast_ns (void ); -extern u64 ktime_get_boot_fast_ns (void ); -extern u64 ktime_get_real_fast_ns (void ); +extern ktime_t ktime_get_mono_fast (void ); +extern ktime_t ktime_get_raw_fast (void ); +extern ktime_t ktime_get_boottime_fast (void ); +extern ktime_t ktime_get_real_fast (void ); + +static inline u64 ktime_get_mono_fast_ns (void ) +{ +return ktime_to_ns (ktime_get_mono_fast ()); +} + +static inline u64 ktime_get_raw_fast_ns (void ) +{ +return ktime_to_ns (ktime_get_raw_fast ()); +} + +static inline u64 ktime_get_boot_fast_ns (void ) +{ +return ktime_to_ns (ktime_get_boottime_fast ()); +} + +static inline u64 ktime_get_real_fast_ns (void ) +{ +return ktime_to_ns (ktime_get_real_fast ()); +}/* * timespec64/time64_t interfaces utilizing the ktime based ones
kernel/time/timekeeping.c +25 -27
113 lines not on this page · raw message
Signature · 1 line
2.21.0
Randy Dunlap
reviewerRe: [PATCH 1/2] timekeeping: add missing non-_ns functions for fast accessors
Quoted 4 lines · Jason A. Donenfeld
Previously there was no analogue to get proper
ktime_tversions of the fast variety of ktime invocations. This commit makes the interface uniform with the other accessors.
Quoted 9 lines
Signed-off-by Jason A. Donenfeld <[email protected]>Cc Arnd Bergmann <[email protected]>Cc Thomas Gleixner <[email protected]>--- Documentation/core-api/timekeeping.rst | 7 +++- include/linux/timekeeping.h | 28 ++++++++++++-- kernel/time/timekeeping.c | 52 +++++++++++++------------- 3 files changed, 55 insertions(+), 32 deletions(-)
Hi,
There are a few minor documentation problems here. See below.
Documentation/core-api/timekeeping.rst different format depending on what is required by the user: · line 98
Quoted 11 lines
Documentation/core-api/timekeeping.rst +1
-1@@ -94,7 +94,7 @@ different format depending on what is required by the user: down the seconds to the full seconds of the last timer tick using the respective reference. -Coarse and fast_ns access +Coarse and fast access -------------------------
The "underline ---" length should match the heading length.
Quoted 47 lines
Documentation/core-api/timekeeping.rst +0
-0Some additional variants exist for more specialized cases:kernel/time/timekeeping.c +8
-9@@ -497,47 +496,46 @@ EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); * partially updated. Since the tk->offs_boot update is a rare event, this * should be a rare occurrence which postprocessing should be able to handle. */ -u64 notrace ktime_get_boot_fast_ns(void) +ktime_t notrace ktime_get_boottime_fast(void) { struct timekeeper *tk = &tk_core.timekeeper; - return (ktime_get_mono_fast_ns() + ktime_to_ns(tk->offs_boot)); + return ktime_add(ktime_get_mono_fast(), tk->offs_boot); } -EXPORT_SYMBOL_GPL(ktime_get_boot_fast_ns); +EXPORT_SYMBOL_GPL(ktime_get_boottime_fast); /* - * See comment for __ktime_get_fast_ns() vs. timestamp ordering + * See comment for __ktime_get_fast() vs. timestamp ordering */ -static __always_inline u64 __ktime_get_real_fast_ns(struct tk_fast *tkf) +static __always_inline ktime_t __ktime_get_real_fast(struct tk_fast *tkf) { struct tk_read_base *tkr; unsigned int seq; - u64 now; + ktime_t now; do { seq = raw_read_seqcount_latch(&tkf->seq); tkr = tkf->base + (seq & 0x01); - now = ktime_to_ns(tkr->base_real); - - now += timekeeping_delta_to_ns(tkr, + now = ktime_add_ns(tkr->base_real, + timekeeping_delta_to_ns(tkr, clocksource_delta( tk_clock_read(tkr), tkr->cycle_last,
kernel/time/timekeeping.c EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); · line 532
Quoted 10 lines
kernel/time/timekeeping.c +2
-2- tkr->mask)); + tkr->mask))); }while (read_seqcount_retry (&tkf->seq, seq));return now; }/** -* ktime_get_real_fast_ns: - NMI safe and fast access to clock realtime. +* ktime_get_real_fast: - NMI safe and fast access to clock realtime.
The function name line should not contain a ':' character. (I know, you didn't change that.)
kernel/time/timekeeping.c EXPORT_SYMBOL_GPL(ktime_get_raw_fast_ns); · line 541
Quoted 13 lines
kernel/time/timekeeping.c +3
-3*/ -u64 ktime_get_real_fast_ns (void ) +ktime_t ktime_get_real_fast (void ) { -return __ktime_get_real_fast_ns (&tk_fast_mono); +return __ktime_get_real_fast (&tk_fast_mono); } -EXPORT_SYMBOL_GPL (ktime_get_real_fast_ns); +EXPORT_SYMBOL_GPL (ktime_get_real_fast);/** * halt_fast_timekeeper - Prevent fast timekeeper from accessing clocksource.
thnx.
Signature · 1 line
~Randy
PATCH 2/2 timekeeping: rename fast getter functions to be consistent
In reply to Jason A. DonenfeldThe _fast_ family of functions put the _fast_ modifier in the wrong place and confuse tai with clocktai and boot with boottime. Also, no other functions use _mono, making that the default unlabeled one. This commit normalizes these to match the rest of timekeeping.h.
--- Documentation/core-api/timekeeping.rst | 18 ++++++++-------- Documentation/trace/ftrace.rst | 2 +- drivers/base/power/runtime.c | 12 +++++------ drivers/gpu/drm/i915/i915_perf.c | 2 +- fs/pstore/platform.c | 2 +- include/linux/pm_runtime.h | 2 +- include/linux/timekeeping.h | 24 ++++++++++----------- kernel/bpf/helpers.c | 2 +- kernel/debug/kdb/kdb_main.c | 2 +- kernel/events/core.c | 4 ++-- kernel/rcu/rcuperf.c | 6 +++--- kernel/rcu/srcutree.c | 6 +++--- kernel/time/timekeeping.c | 30 +++++++++++++------------- kernel/trace/trace.c | 6 +++--- kernel/watchdog_hld.c | 2 +- 15 files changed, 60 insertions(+), 60 deletions(-)
Documentation/core-api/timekeeping.rst +9 -9
@@ -129,15 +129,15 @@ Some additional variants exist for more specialized cases: up to several microseconds on older hardware with an external clocksource. -.. c:function:: ktime_t ktime_get_mono_fast( void ) - ktime_t ktime_get_raw_fast( void ) - ktime_t ktime_get_boottime_fast( void ) - ktime_t ktime_get_real_fast( void ) - -.. c:function:: u64 ktime_get_mono_fast_ns( void ) - u64 ktime_get_raw_fast_ns( void ) - u64 ktime_get_boot_fast_ns( void ) - u64 ktime_get_real_fast_ns( void ) +.. c:function:: ktime_t ktime_get_fast( void ) + ktime_t ktime_get_fast_raw( void ) + ktime_t ktime_get_fast_boottime( void ) + ktime_t ktime_get_fast_real( void ) + +.. c:function:: u64 ktime_get_fast_ns( void ) + u64 ktime_get_fast_raw_ns( void ) + u64 ktime_get_fast_boottime_ns( void ) + u64 ktime_get_fast_real_ns( void ) These variants are safe to call from any context, including from a non-maskable interrupt (NMI) during a timekeeper update, and
Documentation/trace/ftrace.rst +1 -1
@@ -482,7 +482,7 @@ of ftrace. Here is a list of some of the key files: Also on 32-bit systems, it's possible that the 64-bit boot offset sees a partial update. These effects are rare and post processing should be able to handle them. See comments in the - ktime_get_boot_fast_ns() function for more information. + ktime_get_fast_boottime_ns() function for more information. To set a clock, simply echo the clock name into this file::
drivers/base/power/runtime.c +6 -6
@@ -71,11 +71,11 @@ static void update_pm_runtime_accounting(struct device *dev) last = dev->power.accounting_timestamp; - now =ktime_get_mono_fast_ns (); + now =ktime_get_fast_ns (); dev->power.accounting_timestamp = now;/* -* Because ktime_get_mono_fast_ns() is not monotonic during +* Because ktime_get_fast_ns() is not monotonic during * timekeeping updates, ensure that 'now' is after the last saved * timesptamp. */
@@ -174,7 +174,7 @@ u64 pm_runtime_autosuspend_expiration(struct device *dev) expires =READ_ONCE (dev->power.last_busy); expires += (u64 )autosuspend_delay *NSEC_PER_MSEC ; -if (expires >ktime_get_mono_fast_ns ()) +if (expires >ktime_get_fast_ns ())return expires;/* Expires in the future */ return 0 ;
@@ -938,7 +938,7 @@ static enum hrtimer_restart pm_suspend_timer_fn(struct hrtimer *timer)* If 'expires' is after the current time, we've been called * too early. */ -if (expires >0 && expires <ktime_get_mono_fast_ns ()) { +if (expires >0 && expires <ktime_get_fast_ns ()) { dev->power.timer_expires =0 ;rpm_suspend (dev, dev->power.timer_autosuspends ? (RPM_ASYNC |RPM_AUTO ) :RPM_ASYNC );
@@ -974,7 +974,7 @@ int pm_schedule_suspend(struct device *dev, unsigned int delay)/* Other scheduled or pending requests need to be canceled. */ pm_runtime_cancel_pending (dev); - expires =ktime_get_mono_fast_ns () + (u64 )delay *NSEC_PER_MSEC ; + expires =ktime_get_fast_ns () + (u64 )delay *NSEC_PER_MSEC ; dev->power.timer_expires = expires; dev->power.timer_autosuspends =0 ;hrtimer_start (&dev->power.suspend_timer, expires,HRTIMER_MODE_ABS );
@@ -1378,7 +1378,7 @@ void pm_runtime_enable(struct device *dev)/* About to enable runtime pm, set accounting_timestamp to now */ if (!dev->power.disable_depth) - dev->power.accounting_timestamp =ktime_get_mono_fast_ns (); + dev->power.accounting_timestamp =ktime_get_fast_ns (); }else {dev_warn (dev,"Unbalanced %s!\n" , __func__); }
drivers/gpu/drm/i915/i915_perf.c +1 -1
@@ -478,7 +478,7 @@ static bool oa_buffer_check_unlocked(struct drm_i915_private *dev_priv)*/ hw_tail &= ~(report_size -1 ); - now =ktime_get_mono_fast_ns (); + now =ktime_get_fast_ns ();/* Update the aged tail *
fs/pstore/platform.c +1 -1
@@ -378,7 +378,7 @@ void pstore_record_init(struct pstore_record *record, record->psi = psinfo;/* Report zeroed timestamp if called before timekeeping has resumed. */ - record->time =ns_to_timespec64 (ktime_get_real_fast_ns ()); + record->time =ns_to_timespec64 (ktime_get_fast_real_ns ()); }/*
include/linux/pm_runtime.h +1 -1
@@ -104,7 +104,7 @@ static inline bool pm_runtime_callbacks_present(struct device *dev)static inline void pm_runtime_mark_last_busy (struct device *dev) { -WRITE_ONCE (dev->power.last_busy,ktime_get_mono_fast_ns ()); +WRITE_ONCE (dev->power.last_busy,ktime_get_fast_ns ()); }static inline bool pm_runtime_is_irq_safe (struct device *dev)
include/linux/timekeeping.h +12 -12
@@ -174,29 +174,29 @@ static inline u64 ktime_get_raw_ns(void)return ktime_to_ns (ktime_get_raw ()); } -extern ktime_t ktime_get_mono_fast (void ); -extern ktime_t ktime_get_raw_fast (void ); -extern ktime_t ktime_get_boottime_fast (void ); -extern ktime_t ktime_get_real_fast (void ); +extern ktime_t ktime_get_fast (void ); +extern ktime_t ktime_get_fast_raw (void ); +extern ktime_t ktime_get_fast_boottime (void ); +extern ktime_t ktime_get_fast_real (void ); -static inline u64 ktime_get_mono_fast_ns (void ) +static inline u64 ktime_get_fast_ns (void ) { -return ktime_to_ns (ktime_get_mono_fast ()); +return ktime_to_ns (ktime_get_fast ()); } -static inline u64 ktime_get_raw_fast_ns (void ) +static inline u64 ktime_get_fast_raw_ns (void ) { -return ktime_to_ns (ktime_get_raw_fast ()); +return ktime_to_ns (ktime_get_fast_raw ()); } -static inline u64 ktime_get_boot_fast_ns (void ) +static inline u64 ktime_get_fast_boottime_ns (void ) { -return ktime_to_ns (ktime_get_boottime_fast ()); +return ktime_to_ns (ktime_get_fast_boottime ()); } -static inline u64 ktime_get_real_fast_ns (void ) +static inline u64 ktime_get_fast_real_ns (void ) { -return ktime_to_ns (ktime_get_real_fast ()); +return ktime_to_ns (ktime_get_fast_real ()); }/*
kernel/bpf/helpers.c +1 -1
@@ -143,7 +143,7 @@ const struct bpf_func_proto bpf_get_numa_node_id_proto = {BPF_CALL_0 (bpf_ktime_get_ns) {/* NMI safe access to clock monotonic */ -return ktime_get_mono_fast_ns (); +return ktime_get_fast_ns (); }const struct bpf_func_proto bpf_ktime_get_ns_proto = {
kernel/debug/kdb/kdb_main.c +1 -1
@@ -2491,7 +2491,7 @@ static int kdb_kill(int argc, const char **argv)*/ static void kdb_sysinfo (struct sysinfo *val) { -u64 uptime =ktime_get_mono_fast_ns (); +u64 uptime =ktime_get_fast_ns ();memset (val,0 ,sizeof (*val)); val->uptime =div_u64 (uptime,NSEC_PER_SEC );
kernel/events/core.c +2 -2
@@ -10666,12 +10666,12 @@ static int perf_event_set_clock(struct perf_event *event, clockid_t clk_id)switch (clk_id) {case CLOCK_MONOTONIC : - event->clock = &ktime_get_mono_fast_ns; + event->clock = &ktime_get_fast_ns; nmi_safe = true;break ;case CLOCK_MONOTONIC_RAW : - event->clock = &ktime_get_raw_fast_ns; + event->clock = &ktime_get_fast_raw_ns; nmi_safe = true;break ;
kernel/rcu/rcuperf.c +3 -3
@@ -375,7 +375,7 @@ rcu_perf_writer(void *arg)if (holdoff)schedule_timeout_uninterruptible (holdoff *HZ ); - t =ktime_get_mono_fast_ns (); + t =ktime_get_fast_ns ();if (atomic_inc_return (&n_rcu_perf_writer_started) >= nrealwriters) { t_rcu_perf_writer_started = t;if (gp_exp) {
@@ -390,7 +390,7 @@ rcu_perf_writer(void *arg)if (writer_holdoff)udelay (writer_holdoff); wdp = &wdpp[i]; - *wdp =ktime_get_mono_fast_ns (); + *wdp =ktime_get_fast_ns ();if (gp_async) { retry:if (!rhp)
@@ -415,7 +415,7 @@ rcu_perf_writer(void *arg) cur_ops->sync (); } rcu_perf_writer_state =RTWS_IDLE ; - t =ktime_get_mono_fast_ns (); + t =ktime_get_fast_ns (); *wdp = t - *wdp; i_max = i; if (!started &&
kernel/rcu/srcutree.c +3 -3
@@ -179,7 +179,7 @@ static int init_srcu_struct_fields(struct srcu_struct *ssp, bool is_static) ssp->sda = alloc_percpu(struct srcu_data);init_srcu_struct_nodes (ssp, is_static); ssp->srcu_gp_seq_needed_exp =0 ; - ssp->srcu_last_gp_end =ktime_get_mono_fast_ns (); + ssp->srcu_last_gp_end =ktime_get_fast_ns ();smp_store_release (&ssp->srcu_gp_seq_needed,0 );/* Init done. */ return ssp->sda ?0 : -ENOMEM ; }
@@ -530,7 +530,7 @@ static void srcu_gp_end(struct srcu_struct *ssp) idx =rcu_seq_state (ssp->srcu_gp_seq);WARN_ON_ONCE (idx !=SRCU_STATE_SCAN2 ); cbdelay =srcu_get_delay (ssp); - ssp->srcu_last_gp_end =ktime_get_mono_fast_ns (); + ssp->srcu_last_gp_end =ktime_get_fast_ns ();rcu_seq_end (&ssp->srcu_gp_seq); gpseq =rcu_seq_current (&ssp->srcu_gp_seq);if (ULONG_CMP_LT (ssp->srcu_gp_seq_needed_exp, gpseq))
@@ -779,7 +779,7 @@ static bool srcu_might_be_idle(struct srcu_struct *ssp)*/ /* First, see if enough time has passed since the last GP. */ - t =ktime_get_mono_fast_ns (); + t =ktime_get_fast_ns ();if (exp_holdoff ==0 ||time_in_range_open (t, ssp->srcu_last_gp_end, ssp->srcu_last_gp_end + exp_holdoff))
kernel/time/timekeeping.c +15 -15
@@ -391,7 +391,7 @@ static inline u64 timekeeping_cycles_to_ns(const struct tk_read_base *tkr, u64 c* So if a NMI hits the update of base[0] then it will use base[1] * which is still consistent. In the worst case this can result is a * slightly wrong timestamp (a few nanoseconds). See -* @ktime_get_mono_fast_ns. +* @ktime_get_fast_ns. */ static void update_fast_timekeeper (const struct tk_read_base *tkr,struct tk_fast *tkf)
@@ -412,7 +412,7 @@ static void update_fast_timekeeper(const struct tk_read_base *tkr, }/** -* ktime_get_mono_fast_ns - Fast NMI safe access to clock monotonic +* ktime_get_fast_ns - Fast NMI safe access to clock monotonic * * This timestamp is not guaranteed to be monotonic across an update. * The timestamp is calculated by:
@@ -463,20 +463,20 @@ static __always_inline ktime_t __ktime_get_fast(struct tk_fast *tkf)return now; } -ktime_t ktime_get_mono_fast (void ) +ktime_t ktime_get_fast (void ) {return __ktime_get_fast (&tk_fast_mono); } -EXPORT_SYMBOL_GPL (ktime_get_mono_fast); +EXPORT_SYMBOL_GPL (ktime_get_fast); -ktime_t ktime_get_raw_fast (void ) +ktime_t ktime_get_fast_raw (void ) {return __ktime_get_fast (&tk_fast_raw); } -EXPORT_SYMBOL_GPL (ktime_get_raw_fast); +EXPORT_SYMBOL_GPL (ktime_get_fast_raw);/** -* ktime_get_boottime_fast - NMI safe and fast access to boot clock. +* ktime_get_fast_boottime - NMI safe and fast access to boot clock. * * To keep it NMI safe since we're accessing from tracing, we're not using a * separate timekeeper with updates to monotonic clock and boot offset
@@ -496,19 +496,19 @@ EXPORT_SYMBOL_GPL(ktime_get_raw_fast);* partially updated. Since the tk->offs_boot update is a rare event, this * should be a rare occurrence which postprocessing should be able to handle. */ -ktime_t notrace ktime_get_boottime_fast(void) +ktime_t notrace ktime_get_fast_boottime(void) {struct timekeeper *tk = &tk_core.timekeeper; -return ktime_add (ktime_get_mono_fast (), tk->offs_boot); +return ktime_add (ktime_get_fast (), tk->offs_boot); } -EXPORT_SYMBOL_GPL (ktime_get_boottime_fast); +EXPORT_SYMBOL_GPL (ktime_get_fast_boottime);/* * See comment for __ktime_get_fast() vs. timestamp ordering */ -static __always_inline ktime_t __ktime_get_real_fast(struct tk_fast *tkf) +static __always_inline ktime_t __ktime_get_fast_real(struct tk_fast *tkf) {struct tk_read_base *tkr;unsigned int seq;
@@ -529,13 +529,13 @@ static __always_inline ktime_t __ktime_get_real_fast(struct tk_fast *tkf) }/** -* ktime_get_real_fast: - NMI safe and fast access to clock realtime. +* ktime_get_fast_real: - NMI safe and fast access to clock realtime. */ -ktime_t ktime_get_real_fast (void ) +ktime_t ktime_get_fast_real (void ) { -return __ktime_get_real_fast (&tk_fast_mono); +return __ktime_get_fast_real (&tk_fast_mono); } -EXPORT_SYMBOL_GPL (ktime_get_real_fast); +EXPORT_SYMBOL_GPL (ktime_get_fast_real);/** * halt_fast_timekeeper - Prevent fast timekeeper from accessing clocksource.
kernel/trace/trace.c +3 -3
@@ -1336,9 +1336,9 @@ static struct { { trace_clock_counter, "counter", 0 }, { trace_clock_jiffies, "uptime", 0 }, { trace_clock, "perf", 1 }, - { ktime_get_mono_fast_ns, "mono", 1 }, - { ktime_get_raw_fast_ns, "mono_raw", 1 }, - { ktime_get_boot_fast_ns, "boot", 1 }, + { ktime_get_fast_ns, "mono", 1 }, + { ktime_get_fast_raw_ns, "mono_raw", 1 }, + { ktime_get_fast_boottime_ns, "boot", 1 },ARCH_TRACE_CLOCKS };
kernel/watchdog_hld.c +1 -1
@@ -75,7 +75,7 @@ void watchdog_update_hrtimer_threshold(u64 period)static bool watchdog_check_timestamp (void ) { -ktime_t delta, now =ktime_get_mono_fast_ns (); +ktime_t delta, now =ktime_get_fast_ns (); delta = now -__this_cpu_read (last_timestamp);if (delta < watchdog_hrtimer_sample_threshold) {
Signature · 1 line
2.21.0
Quoted 8 lines · Jason A. Donenfeld
Hi Thomas,
When Arnd and I discussed this prior, he thought it best that I separate these two commits out into a separate patchset, because they might require additional discussion or consideration from you. They seem straightforward enough to me, but if deliberations require me to make some tweaks, I'm happy to do so.
One concern I had was whether we want to replace 'fast' with something else, such as 'in_nmi' that might be less confusing. The current naming might be easy to confuse between 'fast' and 'coarse'.
Another point might be whether we actually need more than one kind of accessor for each time domain, given how rarely these are used. In theory we could have the full set of combinations of fast:
monotonic/real/boottime/raw (but not clocktai) with ktime_t/ns/seconds/ts64 for 16 versions. We currently have four, and you are adding another four, but not the final eight. I'm not saying this is wrong, but it feels a bit arbitrary and could use an explanation why you feel that is the right subset.
For coarse, we have ktime_t and ts64. The _seconds() accessors are coarse by definition, but we probably don't want to add _ns(). We also don't have the combination of 'raw' with 'coarse' or 'seconds', as that seems to have no use case.
Arnd
Quoted 21 lines · Arnd Bergmann
Quoted 6 lines · Jason A. Donenfeld
When Arnd and I discussed this prior, he thought it best that I separate these two commits out into a separate patchset, because they might require additional discussion or consideration from you. They seem straightforward enough to me, but if deliberations require me to make some tweaks, I'm happy to do so.
One concern I had was whether we want to replace 'fast' with something else, such as '
in_nmi' that might be less confusing. The current naming might be easy to confuse between 'fast' and 'coarse'.Another point might be whether we actually need more than one kind of accessor for each time domain, given how rarely these are used. In theory we could have the full set of combinations of fast:
monotonic/real/boottime/raw (but not clocktai) with ktime_t/ns/seconds/ts64 for 16 versions. We currently have four, and you are adding another four, but not the final eight. I'm not saying this is wrong, but it feels a bit arbitrary and could use an explanation why you feel that is the right subset.
Quoted 4 lines
For coarse, we have
ktime_tand ts64. The_seconds()accessors are coarse by definition, but we probably don't want to add_ns(). We also don't have the combination of 'raw' with 'coarse' or 'seconds', as that seems to have no use case.
Can we please just add those which are actually needed. If new code misses something we can add them anytime later.
Thanks,
tglx