linux-kernel - [tip: timers/urgent] vdso/vsyscall: Avoid slow division loop in auxiliary clock update

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <175689353303.1920.4835241898909824126.tip-bot2@tip-bot2>
Date: Wed, 03 Sep 2025 09:58:53 -0000
From: tip-bot2 for Thomas Weißschuh <tip-bot2@...utronix.de>
To: linux-tip-commits@...r.kernel.org
Cc: Miroslav Lichvar <mlichvar@...hat.com>, thomas.weissschuh@...utronix.de,
 Thomas Gleixner <tglx@...utronix.de>, x86@...nel.org,
 linux-kernel@...r.kernel.org
Subject: [tip: timers/urgent] vdso/vsyscall: Avoid slow division loop in
 auxiliary clock update

The following commit has been merged into the timers/urgent branch of tip:

Commit-ID:     762af5a2aa0ad18da1316666dae30d369268d44c
Gitweb:        https://git.kernel.org/tip/762af5a2aa0ad18da1316666dae30d369268d44c
Author:        Thomas Weißschuh <thomas.weissschuh@...utronix.de>
AuthorDate:    Mon, 25 Aug 2025 15:26:35 +02:00
Committer:     Thomas Gleixner <tglx@...utronix.de>
CommitterDate: Wed, 03 Sep 2025 11:55:11 +02:00

vdso/vsyscall: Avoid slow division loop in auxiliary clock update

The call to __iter_div_u64_rem() in vdso_time_update_aux() is a wrapper
around subtraction. It cannot be used to divide large numbers, as that
introduces long, computationally expensive delays.  A regular u64 division
is also not possible in the timekeeper update path as it can be too slow.

Instead of splitting the ktime_t offset into into second and subsecond
components during the timekeeper update fast-path, do it together with the
adjustment of tk->offs_aux in the slow-path. Equivalent to the handling of
offs_boot and monotonic_to_boot.

Reuse the storage of monotonic_to_boot for the new field, as it is not used
by auxiliary timekeepers.

Fixes: 380b84e168e5 ("vdso/vsyscall: Update auxiliary clock data in the datapage")
Reported-by: Miroslav Lichvar <mlichvar@...hat.com>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@...utronix.de>
Signed-off-by: Thomas Gleixner <tglx@...utronix.de>
Link: https://lore.kernel.org/all/20250825-vdso-auxclock-division-v1-1-a1d32a16a313@linutronix.de
Closes: https://lore.kernel.org/lkml/aKwsNNWsHJg8IKzj@localhost/
---
 include/linux/timekeeper_internal.h |  9 ++++++++-
 kernel/time/timekeeping.c           | 10 ++++++++--
 kernel/time/vsyscall.c              |  4 ++--
 3 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/include/linux/timekeeper_internal.h b/include/linux/timekeeper_internal.h
index c27aac6..b8ae89e 100644
--- a/include/linux/timekeeper_internal.h
+++ b/include/linux/timekeeper_internal.h
@@ -76,6 +76,7 @@ struct tk_read_base {
  * @cs_was_changed_seq:		The sequence number of clocksource change events
  * @clock_valid:		Indicator for valid clock
  * @monotonic_to_boot:		CLOCK_MONOTONIC to CLOCK_BOOTTIME offset
+ * @monotonic_to_aux:		CLOCK_MONOTONIC to CLOCK_AUX offset
  * @cycle_interval:		Number of clock cycles in one NTP interval
  * @xtime_interval:		Number of clock shifted nano seconds in one NTP
  *				interval.
@@ -117,6 +118,9 @@ struct tk_read_base {
  * @offs_aux is used by the auxiliary timekeepers which do not utilize any
  * of the regular timekeeper offset fields.
  *
+ * @monotonic_to_aux is a timespec64 representation of @offs_aux to
+ * accelerate the VDSO update for CLOCK_AUX.
+ *
  * The cacheline ordering of the structure is optimized for in kernel usage of
  * the ktime_get() and ktime_get_ts64() family of time accessors. Struct
  * timekeeper is prepended in the core timekeeping code with a sequence count,
@@ -159,7 +163,10 @@ struct timekeeper {
 	u8			cs_was_changed_seq;
 	u8			clock_valid;
 
-	struct timespec64	monotonic_to_boot;
+	union {
+		struct timespec64	monotonic_to_boot;
+		struct timespec64	monotonic_to_aux;
+	};
 
 	u64			cycle_interval;
 	u64			xtime_interval;
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 059fa8b..b6974fc 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -83,6 +83,12 @@ static inline bool tk_is_aux(const struct timekeeper *tk)
 }
 #endif
 
+static inline void tk_update_aux_offs(struct timekeeper *tk, ktime_t offs)
+{
+	tk->offs_aux = offs;
+	tk->monotonic_to_aux = ktime_to_timespec64(offs);
+}
+
 /* flag for if timekeeping is suspended */
 int __read_mostly timekeeping_suspended;
 
@@ -1506,7 +1512,7 @@ static int __timekeeping_inject_offset(struct tk_data *tkd, const struct timespe
 			timekeeping_restore_shadow(tkd);
 			return -EINVAL;
 		}
-		tks->offs_aux = offs;
+		tk_update_aux_offs(tks, offs);
 	}
 
 	timekeeping_update_from_shadow(tkd, TK_UPDATE_ALL);
@@ -2937,7 +2943,7 @@ static int aux_clock_set(const clockid_t id, const struct timespec64 *tnew)
 	 * xtime ("realtime") is not applicable for auxiliary clocks and
 	 * kept in sync with "monotonic".
 	 */
-	aux_tks->offs_aux = ktime_sub(timespec64_to_ktime(*tnew), tnow);
+	tk_update_aux_offs(aux_tks, ktime_sub(timespec64_to_ktime(*tnew), tnow));
 
 	timekeeping_update_from_shadow(aux_tkd, TK_UPDATE_ALL);
 	return 0;
diff --git a/kernel/time/vsyscall.c b/kernel/time/vsyscall.c
index 8ba8b0d..aa59919 100644
--- a/kernel/time/vsyscall.c
+++ b/kernel/time/vsyscall.c
@@ -159,10 +159,10 @@ void vdso_time_update_aux(struct timekeeper *tk)
 	if (clock_mode != VDSO_CLOCKMODE_NONE) {
 		fill_clock_configuration(vc, &tk->tkr_mono);
 
-		vdso_ts->sec	= tk->xtime_sec;
+		vdso_ts->sec = tk->xtime_sec + tk->monotonic_to_aux.tv_sec;
 
 		nsec = tk->tkr_mono.xtime_nsec >> tk->tkr_mono.shift;
-		nsec += tk->offs_aux;
+		nsec += tk->monotonic_to_aux.tv_nsec;
 		vdso_ts->sec += __iter_div_u64_rem(nsec, NSEC_PER_SEC, &nsec);
 		nsec = nsec << tk->tkr_mono.shift;
 		vdso_ts->nsec = nsec;