On Fri, Sep 13, 2024 at 11:59 AM John Stultz <jstultz@xxxxxxxxxx> wrote:
>
> On Fri, Sep 13, 2024 at 6:54 AM Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> >
> > For multigrain timestamps, we must keep track of the latest timestamp
> > that has ever been handed out, and never hand out a coarse time below
> > that value.
> >
> > Add a static singleton atomic64_t into timekeeper.c that we can use to
> > keep track of the latest fine-grained time ever handed out. This is
>
> Maybe drop "ever" and add "handed out through a specific interface",
> as timestamps can be accessed in a lot of ways that don't keep track
> of what was returned.
>
>
> > tracked as a monotonic ktime_t value to ensure that it isn't affected by
> > clock jumps.
> >
> > Add two new public interfaces:
> >
> > - ktime_get_coarse_real_ts64_mg() fills a timespec64 with the later of the
> > coarse-grained clock and the floor time
> >
> > - ktime_get_real_ts64_mg() gets the fine-grained clock value, and tries
> > to swap it into the floor. A timespec64 is filled with the result.
> >
> > Since the floor is global, we take great pains to avoid updating it
> > unless it's absolutely necessary. If we do the cmpxchg and find that the
> > value has been updated since we fetched it, then we discard the
> > fine-grained time that was fetched in favor of the recent update.
> >
> > To maximize the window of this occurring when multiple tasks are racing
> > to update the floor, ktime_get_coarse_real_ts64_mg returns a cookie
> > value that represents the state of the floor tracking word, and
> > ktime_get_real_ts64_mg accepts a cookie value that it uses as the "old"
> > value when calling cmpxchg().
> >
> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > ---
> > include/linux/timekeeping.h | 4 +++
> > kernel/time/timekeeping.c | 81 +++++++++++++++++++++++++++++++++++++++++++++
> > 2 files changed, 85 insertions(+)
> >
> > diff --git a/include/linux/timekeeping.h b/include/linux/timekeeping.h
> > index fc12a9ba2c88..cf2293158c65 100644
> > --- a/include/linux/timekeeping.h
> > +++ b/include/linux/timekeeping.h
> > @@ -45,6 +45,10 @@ extern void ktime_get_real_ts64(struct timespec64 *tv);
> > extern void ktime_get_coarse_ts64(struct timespec64 *ts);
> > extern void ktime_get_coarse_real_ts64(struct timespec64 *ts);
> >
> > +/* Multigrain timestamp interfaces */
> > +extern u64 ktime_get_coarse_real_ts64_mg(struct timespec64 *ts);
> > +extern void ktime_get_real_ts64_mg(struct timespec64 *ts, u64 cookie);
> > +
> > void getboottime64(struct timespec64 *ts);
> >
> > /*
> > diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
> > index 5391e4167d60..ee11006a224f 100644
> > --- a/kernel/time/timekeeping.c
> > +++ b/kernel/time/timekeeping.c
> > @@ -114,6 +114,13 @@ static struct tk_fast tk_fast_raw ____cacheline_aligned = {
> > .base[1] = FAST_TK_INIT,
> > };
> >
> > +/*
> > + * This represents the latest fine-grained time that we have handed out as a
> > + * timestamp on the system. Tracked as a monotonic ktime_t, and converted to the
> > + * realtime clock on an as-needed basis.
> > + */
> > +static __cacheline_aligned_in_smp atomic64_t mg_floor;
> > +
> > static inline void tk_normalize_xtime(struct timekeeper *tk)
> > {
> > while (tk->tkr_mono.xtime_nsec >= ((u64)NSEC_PER_SEC << tk->tkr_mono.shift)) {
> > @@ -2394,6 +2401,80 @@ void ktime_get_coarse_real_ts64(struct timespec64 *ts)
> > }
> > EXPORT_SYMBOL(ktime_get_coarse_real_ts64);
> >
> > +/**
> > + * ktime_get_coarse_real_ts64_mg - get later of coarse grained time or floor
> > + * @ts: timespec64 to be filled
> > + *
> > + * Adjust floor to realtime and compare it to the coarse time. Fill
> > + * @ts with the latest one. Returns opaque cookie suitable for passing
> > + * to ktime_get_real_ts64_mg().
> > + */
> > +u64 ktime_get_coarse_real_ts64_mg(struct timespec64 *ts)
> > +{
> > + struct timekeeper *tk = &tk_core.timekeeper;
> > + u64 floor = atomic64_read(&mg_floor);
> > + ktime_t f_real, offset, coarse;
> > + unsigned int seq;
> > +
> > + WARN_ON(timekeeping_suspended);
> > +
> > + do {
> > + seq = read_seqcount_begin(&tk_core.seq);
> > + *ts = tk_xtime(tk);
> > + offset = *offsets[TK_OFFS_REAL];
> > + } while (read_seqcount_retry(&tk_core.seq, seq));
> > +
> > + coarse = timespec64_to_ktime(*ts);
> > + f_real = ktime_add(floor, offset);
> > + if (ktime_after(f_real, coarse))
> > + *ts = ktime_to_timespec64(f_real);
> > + return floor;
> > +}
> > +EXPORT_SYMBOL_GPL(ktime_get_coarse_real_ts64_mg);
> > +
> > +/**
> > + * ktime_get_real_ts64_mg - attempt to update floor value and return result
> > + * @ts: pointer to the timespec to be set
> > + * @cookie: opaque cookie from earlier call to ktime_get_coarse_real_ts64_mg()
> > + *
> > + * Get a current monotonic fine-grained time value and attempt to swap
> > + * it into the floor using @cookie as the "old" value. @ts will be
> > + * filled with the resulting floor value, regardless of the outcome of
> > + * the swap.
>
> I'd add more detail here to clarify that this can return a coarse
> floor value if the cookie is stale.
Additionally, for these two new interfaces, since they are so
specifically tuned to this particular need in the vfs, it might be
good to add a comments in the kerneldoc here that they are special
case interfaces for the vfs and should be avoided outside that space.
That probably would alleviate my main worries, and we can polish the
details around cookie or no cookie later if needed.
thanks
-john
