@@ -722,6 +722,18 @@ static inline void tk_update_leap_state(struct timekeeper *tk)
tk->next_leap_ktime = ktime_sub(tk->next_leap_ktime, tk->offs_real);
}
+/*
+ * Leap state update for both shadow and the real timekeeper
+ * Separate to spare a full memcpy() of the timekeeper.
+ */
+static void tk_update_leap_state_all(struct tk_data *tkd)
+{
+ write_seqcount_begin(&tkd->seq);
+ tk_update_leap_state(&tkd->shadow_timekeeper);
+ tkd->timekeeper.next_leap_ktime = tkd->shadow_timekeeper.next_leap_ktime;
+ write_seqcount_end(&tkd->seq);
+}
+
/*
* Update the ktime_t based scalar nsec members of the timekeeper
*/
@@ -2537,12 +2549,10 @@ EXPORT_SYMBOL_GPL(random_get_entropy_fallback);
*/
int do_adjtimex(struct __kernel_timex *txc)
{
- struct timekeeper *tk = &tk_core.timekeeper;
+ struct timekeeper *tk = &tk_core.shadow_timekeeper;
struct audit_ntp_data ad;
bool clock_set = false;
struct timespec64 ts;
- unsigned long flags;
- s32 orig_tai, tai;
int ret;
/* Validate the data before disabling interrupts */
@@ -2569,23 +2579,21 @@ int do_adjtimex(struct __kernel_timex *txc)
ktime_get_real_ts64(&ts);
add_device_randomness(&ts, sizeof(ts));
- raw_spin_lock_irqsave(&tk_core.lock, flags);
- write_seqcount_begin(&tk_core.seq);
+ scoped_guard (raw_spinlock_irqsave, &tk_core.lock) {
+ s32 orig_tai, tai;
- orig_tai = tai = tk->tai_offset;
- ret = __do_adjtimex(txc, &ts, &tai, &ad);
+ orig_tai = tai = tk->tai_offset;
+ ret = __do_adjtimex(txc, &ts, &tai, &ad);
- if (tai != orig_tai) {
- __timekeeping_set_tai_offset(tk, tai);
- timekeeping_update(&tk_core, tk, TK_MIRROR | TK_CLOCK_WAS_SET);
- clock_set = true;
- } else {
- tk_update_leap_state(tk);
+ if (tai != orig_tai) {
+ __timekeeping_set_tai_offset(tk, tai);
+ timekeeping_update_staged(&tk_core, TK_CLOCK_WAS_SET);
+ clock_set = true;
+ } else {
+ tk_update_leap_state_all(&tk_core);
+ }
}
- write_seqcount_end(&tk_core.seq);
- raw_spin_unlock_irqrestore(&tk_core.lock, flags);
-
audit_ntp_log(&ad);
/* Update the multiplier immediately if frequency was set directly */
Updates of the timekeeper can be done by operating on the shadow timekeeper and afterwards copying the result into the real timekeeper. This has the advantage, that the sequence count write protected region is kept as small as possible. Convert do_adjtimex() to use this scheme and take the opportunity to use a scoped_guard() for locking. That requires to have a separate function for updating the leap state so that the update is protected by the sequence count. This also brings the timekeeper and the shadow timekeeper in sync for this state, which was not the case so far. That's not a correctness problem as the state is only used at the read sides which use the real timekeeper, but it's inconsistent nevertheless. Signed-off-by: Anna-Maria Behnsen <anna-maria@linutronix.de> --- kernel/time/timekeeping.c | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-)