diff mbox series

[RFC,3/7] timekeeping: Fix cross-timestamp interpolation for non-x86

Message ID 20230630171052.985577-4-peter.hilber@opensynergy.com
State New
Headers
Series [RFC,1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap |

Commit Message

Peter Hilber June 30, 2023, 5:10 p.m. UTC 
  So far, get_device_system_crosststamp() unconditionally passes
system_counterval.cycles to timekeeping_cycles_to_ns(). But when
interpolating system time (do_interp == true), system_counterval.cycles is
before tkr_mono.cycle_last, contrary to the timekeeping_cycles_to_ns()
expectations.

On x86, CONFIG_CLOCKSOURCE_VALIDATE_LAST_CYCLE will mitigate on
interpolating, setting delta to 0. With delta == 0, xtstamp->sys_monoraw
and xtstamp->sys_realtime are then set to the last update time, as
implicitly expected by adjust_historical_crosststamp(). On other
architectures, the resulting nonsense xtstamp->sys_monoraw and
xtstamp->sys_realtime corrupt the xtstamp (ts) adjustment in
adjust_historical_crosststamp().

Fix this by always setting the delta to 0 when interpolating.

Fixes: 2c756feb18d9 ("time: Add history to cross timestamp interface supporting slower devices")
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
---
 kernel/time/timekeeping.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

Comments

John Stultz July 7, 2023, 11:31 p.m. UTC | #1 
On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
<peter.hilber@opensynergy.com> wrote:
>
> So far, get_device_system_crosststamp() unconditionally passes
> system_counterval.cycles to timekeeping_cycles_to_ns(). But when
> interpolating system time (do_interp == true), system_counterval.cycles is
> before tkr_mono.cycle_last, contrary to the timekeeping_cycles_to_ns()
> expectations.
>
> On x86, CONFIG_CLOCKSOURCE_VALIDATE_LAST_CYCLE will mitigate on
> interpolating, setting delta to 0. With delta == 0, xtstamp->sys_monoraw
> and xtstamp->sys_realtime are then set to the last update time, as
> implicitly expected by adjust_historical_crosststamp(). On other
> architectures, the resulting nonsense xtstamp->sys_monoraw and
> xtstamp->sys_realtime corrupt the xtstamp (ts) adjustment in
> adjust_historical_crosststamp().
>
> Fix this by always setting the delta to 0 when interpolating.
>
> Fixes: 2c756feb18d9 ("time: Add history to cross timestamp interface supporting slower devices")
> Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
> ---
>  kernel/time/timekeeping.c | 13 +++++++++----
>  1 file changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
> index 7e86d5cd784d..7ccc2377c319 100644
> --- a/kernel/time/timekeeping.c
> +++ b/kernel/time/timekeeping.c
> @@ -1259,10 +1259,15 @@ int get_device_system_crosststamp(int (*get_time_fn)
>                                       tk_core.timekeeper.offs_real);
>                 base_raw = tk->tkr_raw.base;
>
> -               nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono,
> -                                                    system_counterval.cycles);
> -               nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw,
> -                                                   system_counterval.cycles);
> +               if (do_interp) {
> +                       nsec_real = timekeeping_delta_to_ns(&tk->tkr_mono, 0);
> +                       nsec_raw = timekeeping_delta_to_ns(&tk->tkr_raw, 0);
> +               } else {
> +                       nsec_real = timekeeping_cycles_to_ns(
> +                               &tk->tkr_mono, system_counterval.cycles);
> +                       nsec_raw = timekeeping_cycles_to_ns(
> +                               &tk->tkr_raw, system_counterval.cycles);
> +               }

Rather than adding another conditional branch here to go through, why
not just use "cycles" instead of system_counterval.cycles as it seems
to be set properly already?

thanks
-john
Peter Hilber July 27, 2023, 10:21 a.m. UTC | #2 
On 08.07.23 01:31, John Stultz wrote:
> On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
> <peter.hilber@opensynergy.com> wrote:
>>
>> So far, get_device_system_crosststamp() unconditionally passes
>> system_counterval.cycles to timekeeping_cycles_to_ns(). But when
>> interpolating system time (do_interp == true), system_counterval.cycles is
>> before tkr_mono.cycle_last, contrary to the timekeeping_cycles_to_ns()
>> expectations.
>>
>> On x86, CONFIG_CLOCKSOURCE_VALIDATE_LAST_CYCLE will mitigate on
>> interpolating, setting delta to 0. With delta == 0, xtstamp->sys_monoraw
>> and xtstamp->sys_realtime are then set to the last update time, as
>> implicitly expected by adjust_historical_crosststamp(). On other
>> architectures, the resulting nonsense xtstamp->sys_monoraw and
>> xtstamp->sys_realtime corrupt the xtstamp (ts) adjustment in
>> adjust_historical_crosststamp().
>>
>> Fix this by always setting the delta to 0 when interpolating.
>>
>> Fixes: 2c756feb18d9 ("time: Add history to cross timestamp interface supporting slower devices")
>> Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
>> ---
>>  kernel/time/timekeeping.c | 13 +++++++++----
>>  1 file changed, 9 insertions(+), 4 deletions(-)
>>
>> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
>> index 7e86d5cd784d..7ccc2377c319 100644
>> --- a/kernel/time/timekeeping.c
>> +++ b/kernel/time/timekeeping.c
>> @@ -1259,10 +1259,15 @@ int get_device_system_crosststamp(int (*get_time_fn)
>>                                       tk_core.timekeeper.offs_real);
>>                 base_raw = tk->tkr_raw.base;
>>
>> -               nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono,
>> -                                                    system_counterval.cycles);
>> -               nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw,
>> -                                                   system_counterval.cycles);
>> +               if (do_interp) {
>> +                       nsec_real = timekeeping_delta_to_ns(&tk->tkr_mono, 0);
>> +                       nsec_raw = timekeeping_delta_to_ns(&tk->tkr_raw, 0);
>> +               } else {
>> +                       nsec_real = timekeeping_cycles_to_ns(
>> +                               &tk->tkr_mono, system_counterval.cycles);
>> +                       nsec_raw = timekeeping_cycles_to_ns(
>> +                               &tk->tkr_raw, system_counterval.cycles);
>> +               }
> 
> Rather than adding another conditional branch here to go through, why
> not just use "cycles" instead of system_counterval.cycles as it seems
> to be set properly already?

OK. Thanks for the review and suggestion!
diff mbox series

Patch

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 7e86d5cd784d..7ccc2377c319 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1259,10 +1259,15 @@  int get_device_system_crosststamp(int (*get_time_fn)
 				      tk_core.timekeeper.offs_real);
 		base_raw = tk->tkr_raw.base;
 
-		nsec_real = timekeeping_cycles_to_ns(&tk->tkr_mono,
-						     system_counterval.cycles);
-		nsec_raw = timekeeping_cycles_to_ns(&tk->tkr_raw,
-						    system_counterval.cycles);
+		if (do_interp) {
+			nsec_real = timekeeping_delta_to_ns(&tk->tkr_mono, 0);
+			nsec_raw = timekeeping_delta_to_ns(&tk->tkr_raw, 0);
+		} else {
+			nsec_real = timekeeping_cycles_to_ns(
+				&tk->tkr_mono, system_counterval.cycles);
+			nsec_raw = timekeeping_cycles_to_ns(
+				&tk->tkr_raw, system_counterval.cycles);
+		}
 	} while (read_seqcount_retry(&tk_core.seq, seq));
 
 	xtstamp->sys_realtime = ktime_add_ns(base_real, nsec_real);