[PATCH 0/4] Time fixes for 3.6

[PATCH 0/4] Time fixes for 3.6

[John Stultz]

Thomas, Ingo,
	Here are four small time fixes for 3.6
that apply against tip/timers/urgent.

One patch resolves the suspend issue Andreas saw.

The next two are casting fixes for potential overflows
on 32bits, that Andreas found, but haven't been observed
in the wild.

Finally the last one fixes an issue I found in my own
testing w/ the _COARSE clocksources.

Please take a look at them and let me know if you have
any thoughts or objections.

Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andreas Schwab <schwab@linux-m68k.org>

Andreas Schwab (1):
  time: Fix casting issue in timekeeping_forward_now

John Stultz (3):
  time: Ensure we normalize the timekeeper in tk_xtime_add
  time: Avoid potential shift-overflow with large shift values
  time: Avoid making adjustments if we havne't accumulated anything

 kernel/time/timekeeping.c |   11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

-- 
1.7.9.5

[PATCH 1/4] time: Ensure we normalize the timekeeper in tk_xtime_add

[John Stultz]

Andreas noticed problems with resume on specific hardware
after commit 1e75fa8be9fb61e1af46b5b3b176347a4c958ca1
combined with commit b44d50dcacea0d485ca2ff9140f8cc28ee22f28d

After some digging I realized we aren't normalizing
the timekeeper after the add. This patch to correct this
resolved the issue.

Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Reported-by: Andreas Schwab <schwab@linux-m68k.org>
Tested-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/timekeeping.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 898bef0..258164a 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -115,6 +115,7 @@ static void tk_xtime_add(struct timekeeper *tk, const struct timespec *ts)
 {
 	tk->xtime_sec += ts->tv_sec;
 	tk->xtime_nsec += (u64)ts->tv_nsec << tk->shift;
+	tk_normalize_xtime(tk);
 }
 
 static void tk_set_wall_to_mono(struct timekeeper *tk, struct timespec wtm)
-- 
1.7.9.5

[PATCH 2/4] time: Fix casting issue in timekeeping_forward_now

[John Stultz]

From: Andreas Schwab <schwab@linux-m68k.org>

arch_gettimeoffset returns a u32 value which when shifted by tk->shift can
overflow.  Cast it to u64 first.

This issue was introduced with 1e75fa8be9fb61e1af46b5b3b176347a4c958ca1

Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/timekeeping.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 258164a..1dbf80e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -277,7 +277,7 @@ static void timekeeping_forward_now(struct timekeeper *tk)
 	tk->xtime_nsec += cycle_delta * tk->mult;
 
 	/* If arch requires, add in gettimeoffset() */
-	tk->xtime_nsec += arch_gettimeoffset() << tk->shift;
+	tk->xtime_nsec += (u64)arch_gettimeoffset() << tk->shift;
 
 	tk_normalize_xtime(tk);
 
-- 
1.7.9.5

[PATCH 3/4] time: Avoid potential shift-overflow with large shift values

[John Stultz]

Andreas Schwab noted that the 1 << tk->shift could overflow
if the shift value was greater then 30, since 1 would be
a 32bit long on 32bit architectures.

This patch uses 1ULL instead to ensure we don't overflow on
the shift.

This issue was introduced by 1e75fa8be9fb61e1af46b5b3b176347a4c958ca1

Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Reported-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/timekeeping.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 1dbf80e..a5a9389 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1184,9 +1184,9 @@ static void update_wall_time(void)
 	* the vsyscall implementations are converted to use xtime_nsec
 	* (shifted nanoseconds), this can be killed.
 	*/
-	remainder = tk->xtime_nsec & ((1 << tk->shift) - 1);
+	remainder = tk->xtime_nsec & ((1ULL << tk->shift) - 1);
 	tk->xtime_nsec -= remainder;
-	tk->xtime_nsec += 1 << tk->shift;
+	tk->xtime_nsec += 1ULL << tk->shift;
 	tk->ntp_error += remainder << tk->ntp_error_shift;
 
 	/*
-- 
1.7.9.5

[PATCH 4/4] time: Avoid making adjustments if we havne't accumulated anything

[John Stultz]

If update_wall_time() is called and the current offset isn't
large enough to accumulate, avoid re-calling timekeeping_adjust
which may change the clock freq and can cause 1ns inconsistencies
with CLOCK_REALTIME_COARSE/CLOCK_MONOTONIC_COARSE.

Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Signed-off-by: John Stultz <john.stultz@linaro.org>
---
 kernel/time/timekeeping.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index a5a9389..0c1485e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1152,6 +1152,10 @@ static void update_wall_time(void)
 	offset = (clock->read(clock) - clock->cycle_last) & clock->mask;
 #endif
 
+	/* Check if there's really nothing to do */
+	if (offset < tk->cycle_interval)
+		goto out;
+
 	/*
 	 * With NO_HZ we may have to accumulate many cycle_intervals
 	 * (think "ticks") worth of time at once. To do this efficiently,
-- 
1.7.9.5

Re: [PATCH 3/4] time: Avoid potential shift-overflow with large shift values

[Geert Uytterhoeven]

Hi John,

On Wed, Aug 22, 2012 at 2:30 AM, John Stultz <john.stultz@linaro.org> wrote:
> Andreas Schwab noted that the 1 << tk->shift could overflow
> if the shift value was greater then 30, since 1 would be
> a 32bit long on 32bit architectures.

This comment is not entirely correct: "1" is not a long, but an
(32-bit signed) int,
so it can overflow on 64-bit platforms, too.

> This patch uses 1ULL instead to ensure we don't overflow on
> the shift.
>
> This issue was introduced by 1e75fa8be9fb61e1af46b5b3b176347a4c958ca1
>
> Cc: Ingo Molnar <mingo@kernel.org>
> Cc: Prarit Bhargava <prarit@redhat.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Andreas Schwab <schwab@linux-m68k.org>
> Reported-by: Andreas Schwab <schwab@linux-m68k.org>
> Signed-off-by: John Stultz <john.stultz@linaro.org>
> ---
>  kernel/time/timekeeping.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
> index 1dbf80e..a5a9389 100644
> --- a/kernel/time/timekeeping.c
> +++ b/kernel/time/timekeeping.c
> @@ -1184,9 +1184,9 @@ static void update_wall_time(void)
>         * the vsyscall implementations are converted to use xtime_nsec
>         * (shifted nanoseconds), this can be killed.
>         */
> -       remainder = tk->xtime_nsec & ((1 << tk->shift) - 1);
> +       remainder = tk->xtime_nsec & ((1ULL << tk->shift) - 1);
>         tk->xtime_nsec -= remainder;
> -       tk->xtime_nsec += 1 << tk->shift;
> +       tk->xtime_nsec += 1ULL << tk->shift;
>         tk->ntp_error += remainder << tk->ntp_error_shift;
>
>         /*

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[tip:timers/urgent] time: Ensure we normalize the timekeeper in tk_xtime_add

[tip-bot for John Stultz]

Commit-ID:  784ffcbb96c3a97b4c64fd48b1dfe12ef3fcbcda
Gitweb:     http://git.kernel.org/tip/784ffcbb96c3a97b4c64fd48b1dfe12ef3fcbcda
Author:     John Stultz <john.stultz@linaro.org>
AuthorDate: Tue, 21 Aug 2012 20:30:46 -0400
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 22 Aug 2012 10:42:12 +0200

time: Ensure we normalize the timekeeper in tk_xtime_add

Andreas noticed problems with resume on specific hardware after commit
1e75fa8b (time: Condense timekeeper.xtime into xtime_sec) combined
with commit b44d50dca (time: Fix casting issue in tk_set_xtime and
tk_xtime_add)

After some digging I realized we aren't normalizing the timekeeper
after the add. Add the missing normalize call.

Reported-by: Andreas Schwab <schwab@linux-m68k.org>
Tested-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1345595449-34965-2-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 kernel/time/timekeeping.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 898bef0..258164a 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -115,6 +115,7 @@ static void tk_xtime_add(struct timekeeper *tk, const struct timespec *ts)
 {
 	tk->xtime_sec += ts->tv_sec;
 	tk->xtime_nsec += (u64)ts->tv_nsec << tk->shift;
+	tk_normalize_xtime(tk);
 }
 
 static void tk_set_wall_to_mono(struct timekeeper *tk, struct timespec wtm)

[tip:timers/urgent] time: Fix casting issue in timekeeping_forward_now

[tip-bot for Andreas Schwab]

Commit-ID:  85dc8f05c93c8105987de9d7e7cebf15a72ff4ec
Gitweb:     http://git.kernel.org/tip/85dc8f05c93c8105987de9d7e7cebf15a72ff4ec
Author:     Andreas Schwab <schwab@linux-m68k.org>
AuthorDate: Tue, 21 Aug 2012 20:30:47 -0400
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 22 Aug 2012 10:42:13 +0200

time: Fix casting issue in timekeeping_forward_now

arch_gettimeoffset returns a u32 value which when shifted by tk->shift
can overflow. This issue was introduced with 1e75fa8be (time: Condense
timekeeper.xtime into xtime_sec)

Cast it to u64 first.

Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1345595449-34965-3-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 kernel/time/timekeeping.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 258164a..1dbf80e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -277,7 +277,7 @@ static void timekeeping_forward_now(struct timekeeper *tk)
 	tk->xtime_nsec += cycle_delta * tk->mult;
 
 	/* If arch requires, add in gettimeoffset() */
-	tk->xtime_nsec += arch_gettimeoffset() << tk->shift;
+	tk->xtime_nsec += (u64)arch_gettimeoffset() << tk->shift;
 
 	tk_normalize_xtime(tk);
 

[tip:timers/urgent] time: Avoid potential shift overflow with large shift values

[tip-bot for John Stultz]

Commit-ID:  6ea565a9be32a3c8d1092017686f183b6d8c4514
Gitweb:     http://git.kernel.org/tip/6ea565a9be32a3c8d1092017686f183b6d8c4514
Author:     John Stultz <john.stultz@linaro.org>
AuthorDate: Tue, 21 Aug 2012 20:30:48 -0400
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 22 Aug 2012 10:42:13 +0200

time: Avoid potential shift overflow with large shift values

Andreas Schwab noticed that the 1 << tk->shift could overflow if the
shift value was greater than 30, since 1 would be a 32bit long on
32bit architectures. This issue was introduced by 1e75fa8be (time:
Condense timekeeper.xtime into xtime_sec)

Use 1ULL instead to ensure we don't overflow on the shift.

Reported-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1345595449-34965-4-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 kernel/time/timekeeping.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 1dbf80e..a5a9389 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1184,9 +1184,9 @@ static void update_wall_time(void)
 	* the vsyscall implementations are converted to use xtime_nsec
 	* (shifted nanoseconds), this can be killed.
 	*/
-	remainder = tk->xtime_nsec & ((1 << tk->shift) - 1);
+	remainder = tk->xtime_nsec & ((1ULL << tk->shift) - 1);
 	tk->xtime_nsec -= remainder;
-	tk->xtime_nsec += 1 << tk->shift;
+	tk->xtime_nsec += 1ULL << tk->shift;
 	tk->ntp_error += remainder << tk->ntp_error_shift;
 
 	/*

[tip:timers/urgent] time: Avoid making adjustments if we haven' t accumulated anything

[tip-bot for John Stultz]

Commit-ID:  bf2ac312195155511a0f79325515cbb61929898a
Gitweb:     http://git.kernel.org/tip/bf2ac312195155511a0f79325515cbb61929898a
Author:     John Stultz <john.stultz@linaro.org>
AuthorDate: Tue, 21 Aug 2012 20:30:49 -0400
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 22 Aug 2012 10:42:13 +0200

time: Avoid making adjustments if we haven't accumulated anything

If update_wall_time() is called and the current offset isn't large
enough to accumulate, avoid re-calling timekeeping_adjust which may
change the clock freq and can cause 1ns inconsistencies with
CLOCK_REALTIME_COARSE/CLOCK_MONOTONIC_COARSE.

Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1345595449-34965-5-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 kernel/time/timekeeping.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index a5a9389..0c1485e 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1152,6 +1152,10 @@ static void update_wall_time(void)
 	offset = (clock->read(clock) - clock->cycle_last) & clock->mask;
 #endif
 
+	/* Check if there's really nothing to do */
+	if (offset < tk->cycle_interval)
+		goto out;
+
 	/*
 	 * With NO_HZ we may have to accumulate many cycle_intervals
 	 * (think "ticks") worth of time at once. To do this efficiently,