* [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-07-07 22:51 ` John Stultz
2023-06-30 17:10 ` [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision Peter Hilber
` (5 subsequent siblings)
6 siblings, 1 reply; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: linux-kernel
Cc: Peter Hilber, John Stultz, Thomas Gleixner, Stephen Boyd,
Christopher S. Hall
cycle_between() decides whether get_device_system_crosststamp() will
interpolate for older counter readings.
cycle_between() yields wrong results for a counter wrap-around where after
< before < test, and for the case after < test < before.
Fix the comparison logic.
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 | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 266d02809dbb..8f35455b6250 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1186,7 +1186,7 @@ static bool cycle_between(u64 before, u64 test, u64 after)
{
if (test > before && test < after)
return true;
- if (test < before && before > after)
+ if (before > after && (test > before || test < after))
return true;
return false;
}
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap
2023-06-30 17:10 ` [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap Peter Hilber
@ 2023-07-07 22:51 ` John Stultz
2023-07-27 10:21 ` Peter Hilber
0 siblings, 1 reply; 18+ messages in thread
From: John Stultz @ 2023-07-07 22:51 UTC (permalink / raw)
To: Peter Hilber
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
<peter.hilber@opensynergy.com> wrote:
>
> cycle_between() decides whether get_device_system_crosststamp() will
> interpolate for older counter readings.
>
> cycle_between() yields wrong results for a counter wrap-around where after
> < before < test, and for the case after < test < before.
>
> Fix the comparison logic.
>
> 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 | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
> index 266d02809dbb..8f35455b6250 100644
> --- a/kernel/time/timekeeping.c
> +++ b/kernel/time/timekeeping.c
> @@ -1186,7 +1186,7 @@ static bool cycle_between(u64 before, u64 test, u64 after)
> {
> if (test > before && test < after)
> return true;
> - if (test < before && before > after)
> + if (before > after && (test > before || test < after))
> return true;
> return false;
> }
Thanks for catching this and sending it in.
Looks good to me. Curious: Did you actually hit such a wrap around with u64s?
Acked-by: John Stultz <jstultz@google.com>
thanks
-john
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap
2023-07-07 22:51 ` John Stultz
@ 2023-07-27 10:21 ` Peter Hilber
0 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-07-27 10:21 UTC (permalink / raw)
To: John Stultz
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
On 08.07.23 00:51, John Stultz wrote:
> On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
> <peter.hilber@opensynergy.com> wrote:
>>
>> cycle_between() decides whether get_device_system_crosststamp() will
>> interpolate for older counter readings.
>>
>> cycle_between() yields wrong results for a counter wrap-around where after
>> < before < test, and for the case after < test < before.
>>
>> Fix the comparison logic.
>>
>> 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 | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
>> index 266d02809dbb..8f35455b6250 100644
>> --- a/kernel/time/timekeeping.c
>> +++ b/kernel/time/timekeeping.c
>> @@ -1186,7 +1186,7 @@ static bool cycle_between(u64 before, u64 test, u64 after)
>> {
>> if (test > before && test < after)
>> return true;
>> - if (test < before && before > after)
>> + if (before > after && (test > before || test < after))
>> return true;
>> return false;
>> }
>
> Thanks for catching this and sending it in.
> Looks good to me. Curious: Did you actually hit such a wrap around with u64s?
No, I just saw this when fixing the bug in the next patch.
Thanks,
Peter
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-07-07 23:02 ` John Stultz
2023-06-30 17:10 ` [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86 Peter Hilber
` (4 subsequent siblings)
6 siblings, 1 reply; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: linux-kernel
Cc: Peter Hilber, John Stultz, Thomas Gleixner, Stephen Boyd,
Christopher S. Hall
cycle_between() decides whether get_device_system_crosststamp() will
interpolate for older counter readings. So far, cycle_between() checks if
parameter test is in the open interval (before, after), when disregarding
the special case before > after.
The only cycle_between() user, get_device_system_crosststamp(), has the
following problem with this: If interval_start == cycles,
cycle_between(interval_start, cycles, now) returns false. If a
history_begin was supplied to get_device_system_crosststamp(), it will
later call cycle_between() again, with effective argument values
cycle_between(history_begin->cycles, cycles, cycles). Due to the test
against the open interval, cycle_between() returns false again, and
get_device_system_crosststamp() returns -EINVAL, when it could have
succeeded.
Fix this by testing against the closed interval in cycle_between(). This
disables interpolation if interval_start == cycles. For the special case
before > after, similar arguments hold. Fix this in a similar way.
At the second cycle_between() call site, add an extra condition in order to
effectively check a half-open interval, which keeps the condition
documented above the call site satisfied.
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 | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 8f35455b6250..7e86d5cd784d 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1180,13 +1180,13 @@ static int adjust_historical_crosststamp(struct system_time_snapshot *history,
}
/*
- * cycle_between - true if test occurs chronologically between before and after
+ * cycle_between - true if test occurs chronologically in [before, after]
*/
static bool cycle_between(u64 before, u64 test, u64 after)
{
- if (test > before && test < after)
+ if (test >= before && test <= after)
return true;
- if (before > after && (test > before || test < after))
+ if (before > after && (test >= before || test <= after))
return true;
return false;
}
@@ -1282,6 +1282,7 @@ int get_device_system_crosststamp(int (*get_time_fn)
* clocksource change
*/
if (!history_begin ||
+ history_begin->cycles == system_counterval.cycles ||
!cycle_between(history_begin->cycles,
system_counterval.cycles, cycles) ||
history_begin->cs_was_changed_seq != cs_was_changed_seq)
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision
2023-06-30 17:10 ` [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision Peter Hilber
@ 2023-07-07 23:02 ` John Stultz
2023-07-27 10:21 ` Peter Hilber
0 siblings, 1 reply; 18+ messages in thread
From: John Stultz @ 2023-07-07 23:02 UTC (permalink / raw)
To: Peter Hilber
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
<peter.hilber@opensynergy.com> wrote:
>
> cycle_between() decides whether get_device_system_crosststamp() will
> interpolate for older counter readings. So far, cycle_between() checks if
> parameter test is in the open interval (before, after), when disregarding
> the special case before > after.
>
> The only cycle_between() user, get_device_system_crosststamp(), has the
> following problem with this: If interval_start == cycles,
> cycle_between(interval_start, cycles, now) returns false. If a
> history_begin was supplied to get_device_system_crosststamp(), it will
> later call cycle_between() again, with effective argument values
> cycle_between(history_begin->cycles, cycles, cycles). Due to the test
> against the open interval, cycle_between() returns false again, and
> get_device_system_crosststamp() returns -EINVAL, when it could have
> succeeded.
>
> Fix this by testing against the closed interval in cycle_between(). This
> disables interpolation if interval_start == cycles. For the special case
> before > after, similar arguments hold. Fix this in a similar way.
>
> At the second cycle_between() call site, add an extra condition in order to
> effectively check a half-open interval, which keeps the condition
> documented above the call site satisfied.
I'm having a little bit of a hard time following this commit message.
Do you think you might be able to take another swing at it to make it
a bit clearer?
I get you're going from exclusive to inclusive intervals, but it's not
very clear why this change is needed.
> 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 | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
> index 8f35455b6250..7e86d5cd784d 100644
> --- a/kernel/time/timekeeping.c
> +++ b/kernel/time/timekeeping.c
> @@ -1180,13 +1180,13 @@ static int adjust_historical_crosststamp(struct system_time_snapshot *history,
> }
>
> /*
> - * cycle_between - true if test occurs chronologically between before and after
> + * cycle_between - true if test occurs chronologically in [before, after]
> */
> static bool cycle_between(u64 before, u64 test, u64 after)
> {
> - if (test > before && test < after)
> + if (test >= before && test <= after)
> return true;
> - if (before > after && (test > before || test < after))
> + if (before > after && (test >= before || test <= after))
> return true;
> return false;
> }
I'm with you here.
> @@ -1282,6 +1282,7 @@ int get_device_system_crosststamp(int (*get_time_fn)
> * clocksource change
> */
> if (!history_begin ||
> + history_begin->cycles == system_counterval.cycles ||
> !cycle_between(history_begin->cycles,
> system_counterval.cycles, cycles) ||
> history_begin->cs_was_changed_seq != cs_was_changed_seq)
> --
Roughly I see you're trying to preserve the behavior here for the case
a == b, which used to fail with cycles_between(a, b, c) but now
passes.
But it's unclear *why* we're making the change to begin with.
thanks
-john
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision
2023-07-07 23:02 ` John Stultz
@ 2023-07-27 10:21 ` Peter Hilber
0 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-07-27 10:21 UTC (permalink / raw)
To: John Stultz
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
On 08.07.23 01:02, John Stultz wrote:
> On Fri, Jun 30, 2023 at 10:12 AM Peter Hilber
> <peter.hilber@opensynergy.com> wrote:
>>
>> cycle_between() decides whether get_device_system_crosststamp() will
>> interpolate for older counter readings. So far, cycle_between() checks if
>> parameter test is in the open interval (before, after), when disregarding
>> the special case before > after.
>>
>> The only cycle_between() user, get_device_system_crosststamp(), has the
>> following problem with this: If interval_start == cycles,
>> cycle_between(interval_start, cycles, now) returns false. If a
>> history_begin was supplied to get_device_system_crosststamp(), it will
>> later call cycle_between() again, with effective argument values
>> cycle_between(history_begin->cycles, cycles, cycles). Due to the test
>> against the open interval, cycle_between() returns false again, and
>> get_device_system_crosststamp() returns -EINVAL, when it could have
>> succeeded.
>>
>> Fix this by testing against the closed interval in cycle_between(). This
>> disables interpolation if interval_start == cycles. For the special case
>> before > after, similar arguments hold. Fix this in a similar way.
>>
>> At the second cycle_between() call site, add an extra condition in order to
>> effectively check a half-open interval, which keeps the condition
>> documented above the call site satisfied.
>
> I'm having a little bit of a hard time following this commit message.
> Do you think you might be able to take another swing at it to make it
> a bit clearer?
>
> I get you're going from exclusive to inclusive intervals, but it's not
> very clear why this change is needed.
>
Thanks for the feedback, I'll post v2 soon and will try to come up with
a better commit message.
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 1/7] timekeeping: Fix cross-timestamp interpolation on counter wrap Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 2/7] timekeeping: Fix cross-timestamp interpolation corner case decision Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-07-07 23:31 ` John Stultz
2023-06-30 17:10 ` [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource Peter Hilber
` (3 subsequent siblings)
6 siblings, 1 reply; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: linux-kernel
Cc: Peter Hilber, John Stultz, Thomas Gleixner, Stephen Boyd,
Christopher S. Hall
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);
+ }
} while (read_seqcount_retry(&tk_core.seq, seq));
xtstamp->sys_realtime = ktime_add_ns(base_real, nsec_real);
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86
2023-06-30 17:10 ` [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86 Peter Hilber
@ 2023-07-07 23:31 ` John Stultz
2023-07-27 10:21 ` Peter Hilber
0 siblings, 1 reply; 18+ messages in thread
From: John Stultz @ 2023-07-07 23:31 UTC (permalink / raw)
To: Peter Hilber
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
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
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86
2023-07-07 23:31 ` John Stultz
@ 2023-07-27 10:21 ` Peter Hilber
0 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-07-27 10:21 UTC (permalink / raw)
To: John Stultz
Cc: linux-kernel, Thomas Gleixner, Stephen Boyd, Christopher S. Hall
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!
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
` (2 preceding siblings ...)
2023-06-30 17:10 ` [RFC PATCH 3/7] timekeeping: Fix cross-timestamp interpolation for non-x86 Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-07-03 8:13 ` Marc Zyngier
2023-06-30 17:10 ` [RFC PATCH 5/7] virtio_rtc: Add module and driver core Peter Hilber
` (2 subsequent siblings)
6 siblings, 1 reply; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: linux-arm-kernel
Cc: Peter Hilber, linux-kernel, Mark Rutland, Marc Zyngier,
Daniel Lezcano, Thomas Gleixner
Export helper functions to allow other code to
- determine the counter type in use (virtual or physical, CP15 or memory),
- get a pointer to the arm_arch_timer clocksource, which can be compared
with the current clocksource.
The virtio_rtc driver will require the clocksource pointer when using
get_device_system_crosststamp(), and should communicate the actual Arm
counter type to the Virtio RTC device (cf. spec draft [1]).
[1] https://lists.oasis-open.org/archives/virtio-comment/202306/msg00592.html
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
---
drivers/clocksource/arm_arch_timer.c | 16 ++++++++++++++++
include/clocksource/arm_arch_timer.h | 19 +++++++++++++++++++
2 files changed, 35 insertions(+)
diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
index e733a2a1927a..cebdc1b2db4c 100644
--- a/drivers/clocksource/arm_arch_timer.c
+++ b/drivers/clocksource/arm_arch_timer.c
@@ -92,6 +92,7 @@ static enum vdso_clock_mode vdso_default = VDSO_CLOCKMODE_ARCHTIMER;
#else
static enum vdso_clock_mode vdso_default = VDSO_CLOCKMODE_NONE;
#endif /* CONFIG_GENERIC_GETTIMEOFDAY */
+static enum arch_timer_counter_type arch_counter_type __ro_after_init = ARCH_COUNTER_CP15_VIRT;
static cpumask_t evtstrm_available = CPU_MASK_NONE;
static bool evtstrm_enable __ro_after_init = IS_ENABLED(CONFIG_ARM_ARCH_TIMER_EVTSTREAM);
@@ -1109,6 +1110,7 @@ static void __init arch_counter_register(unsigned type)
rd = arch_counter_get_cntvct;
scr = arch_counter_get_cntvct;
}
+ arch_counter_type = ARCH_COUNTER_CP15_VIRT;
} else {
if (arch_timer_counter_has_wa()) {
rd = arch_counter_get_cntpct_stable;
@@ -1117,6 +1119,7 @@ static void __init arch_counter_register(unsigned type)
rd = arch_counter_get_cntpct;
scr = arch_counter_get_cntpct;
}
+ arch_counter_type = ARCH_COUNTER_CP15_PHYS;
}
arch_timer_read_counter = rd;
@@ -1124,6 +1127,7 @@ static void __init arch_counter_register(unsigned type)
} else {
arch_timer_read_counter = arch_counter_get_cntvct_mem;
scr = arch_counter_get_cntvct_mem;
+ arch_counter_type = ARCH_COUNTER_MEM_VIRT;
}
width = arch_counter_get_width();
@@ -1777,6 +1781,18 @@ static int __init arch_timer_acpi_init(struct acpi_table_header *table)
TIMER_ACPI_DECLARE(arch_timer, ACPI_SIG_GTDT, arch_timer_acpi_init);
#endif
+enum arch_timer_counter_type arch_timer_counter_get_type(void)
+{
+ return arch_counter_type;
+}
+EXPORT_SYMBOL_GPL(arch_timer_counter_get_type);
+
+struct clocksource *arch_timer_get_cs(void)
+{
+ return &clocksource_counter;
+}
+EXPORT_SYMBOL_GPL(arch_timer_get_cs);
+
int kvm_arch_ptp_get_crosststamp(u64 *cycle, struct timespec64 *ts,
struct clocksource **cs)
{
diff --git a/include/clocksource/arm_arch_timer.h b/include/clocksource/arm_arch_timer.h
index cbbc9a6dc571..b442db0b5ca0 100644
--- a/include/clocksource/arm_arch_timer.h
+++ b/include/clocksource/arm_arch_timer.h
@@ -43,6 +43,13 @@ enum arch_timer_spi_nr {
ARCH_TIMER_MAX_TIMER_SPI
};
+enum arch_timer_counter_type {
+ ARCH_COUNTER_CP15_VIRT,
+ ARCH_COUNTER_CP15_PHYS,
+ ARCH_COUNTER_MEM_VIRT,
+ ARCH_COUNTER_MEM_PHYS,
+};
+
#define ARCH_TIMER_PHYS_ACCESS 0
#define ARCH_TIMER_VIRT_ACCESS 1
#define ARCH_TIMER_MEM_PHYS_ACCESS 2
@@ -89,6 +96,8 @@ extern u32 arch_timer_get_rate(void);
extern u64 (*arch_timer_read_counter)(void);
extern struct arch_timer_kvm_info *arch_timer_get_kvm_info(void);
extern bool arch_timer_evtstrm_available(void);
+extern enum arch_timer_counter_type arch_timer_counter_get_type(void);
+extern struct clocksource *arch_timer_get_cs(void);
#else
@@ -107,6 +116,16 @@ static inline bool arch_timer_evtstrm_available(void)
return false;
}
+static inline enum arch_timer_counter_type arch_timer_counter_get_type(void)
+{
+ return ARCH_COUNTER_CP15_VIRT;
+}
+
+static inline struct clocksource *arch_timer_get_cs(void)
+{
+ return NULL;
+}
+
#endif
#endif
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource
2023-06-30 17:10 ` [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource Peter Hilber
@ 2023-07-03 8:13 ` Marc Zyngier
2023-07-27 10:22 ` Peter Hilber
0 siblings, 1 reply; 18+ messages in thread
From: Marc Zyngier @ 2023-07-03 8:13 UTC (permalink / raw)
To: Peter Hilber
Cc: linux-arm-kernel, linux-kernel, Mark Rutland, Daniel Lezcano,
Thomas Gleixner
On Fri, 30 Jun 2023 18:10:47 +0100,
Peter Hilber <peter.hilber@opensynergy.com> wrote:
>
> Export helper functions to allow other code to
>
> - determine the counter type in use (virtual or physical, CP15 or memory),
>
> - get a pointer to the arm_arch_timer clocksource, which can be compared
> with the current clocksource.
>
> The virtio_rtc driver will require the clocksource pointer when using
> get_device_system_crosststamp(), and should communicate the actual Arm
> counter type to the Virtio RTC device (cf. spec draft [1]).
I really don't see why you should poke at the clocksource backend:
- the MMIO clocksource is only used in PM situations, which a virtio
driver has no business being involved with
- only the virtual counter is relevant -- it is always at a 0-offset
from the physical one when userspace has an opportunity to run
So it really looks that out of the four combinations, only one is
relevant.
I'm not Cc'd on the rest of the series, so I can't even see in which
context this is used. But as it is, the approach looks wrong.
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource
2023-07-03 8:13 ` Marc Zyngier
@ 2023-07-27 10:22 ` Peter Hilber
2023-07-28 8:11 ` Marc Zyngier
0 siblings, 1 reply; 18+ messages in thread
From: Peter Hilber @ 2023-07-27 10:22 UTC (permalink / raw)
To: Marc Zyngier
Cc: linux-arm-kernel, linux-kernel, Mark Rutland, Daniel Lezcano,
Thomas Gleixner
On 03.07.23 10:13, Marc Zyngier wrote:
> On Fri, 30 Jun 2023 18:10:47 +0100,
> Peter Hilber <peter.hilber@opensynergy.com> wrote:
>>
>> Export helper functions to allow other code to
>>
>> - determine the counter type in use (virtual or physical, CP15 or memory),
>>
>> - get a pointer to the arm_arch_timer clocksource, which can be compared
>> with the current clocksource.
>>
>> The virtio_rtc driver will require the clocksource pointer when using
>> get_device_system_crosststamp(), and should communicate the actual Arm
>> counter type to the Virtio RTC device (cf. spec draft [1]).
>
> I really don't see why you should poke at the clocksource backend:
>
> - the MMIO clocksource is only used in PM situations, which a virtio
> driver has no business being involved with
>
> - only the virtual counter is relevant -- it is always at a 0-offset
> from the physical one when userspace has an opportunity to run
>
> So it really looks that out of the four combinations, only one is
> relevant.
Thanks for the explanation. Dropping arch_timer_counter_get_type() and
assuming that the CP15 virtual counter is in use should also work for
now. With the physical/virtual counter distinction, I tried to be
future-proof, as per the following considerations:
The intended consumer of arch_timer_counter_get_type() is the Virtio RTC
device (draft spec [2], patch series [1]). If the Virtio device has
optional cross-timestamp support, it must know the current Linux kernel
view of the current clocksource counter. The Virtio driver tells the
Virtio device the current counter type (in the Arm case, CNTVCT_EL0 or
CNTPCT_EL0). It is intentionally left unspecified how the Virtio device
would know the counter value. AFAIU, if the Linux kernel were a
virtualization host itself, it would be better for the Virtio device to
look at the physical counter, since the virtual counter could be set for
a guest. And in other cases, the guest OSes use a virtual counter with
an offset.
This was the rationale to come up with the physical/virtual counter
distinction for the Virtio RTC device. Looking at extensions such as
FEAT_ECV, where the CNTPCT_EL0 value can depend on the EL, or FEAT_NV*,
it might be a bit simplistic.
Does this physical/virtual counter distinction sound like a good idea?
Otherwise I would drop the arch_timer_counter_get_type() in the next
iteration.
>
> I'm not Cc'd on the rest of the series, so I can't even see in which
> context this is used. But as it is, the approach looks wrong.
>
Sorry, I will Cc you on all relevant patches in the next iteration,
which I will send out soon.
The first patch series can be found at [1]. I think the second helper
function in this patch, arch_timer_get_cs(), would still be needed, in
order to supply the clocksource to get_device_system_crosststamp().
Thanks for the comments,
Peter
[1] https://lore.kernel.org/lkml/20230630171052.985577-1-peter.hilber@opensynergy.com/T/#me7df2d4db4fe1119d821dc9c4054b9404c15b02d
[2] https://lists.oasis-open.org/archives/virtio-comment/202306/msg00592.html
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource
2023-07-27 10:22 ` Peter Hilber
@ 2023-07-28 8:11 ` Marc Zyngier
2023-07-31 16:15 ` Peter Hilber
0 siblings, 1 reply; 18+ messages in thread
From: Marc Zyngier @ 2023-07-28 8:11 UTC (permalink / raw)
To: Peter Hilber
Cc: linux-arm-kernel, linux-kernel, Mark Rutland, Daniel Lezcano,
Thomas Gleixner
On Thu, 27 Jul 2023 11:22:11 +0100,
Peter Hilber <peter.hilber@opensynergy.com> wrote:
>
> On 03.07.23 10:13, Marc Zyngier wrote:
> > On Fri, 30 Jun 2023 18:10:47 +0100,
> > Peter Hilber <peter.hilber@opensynergy.com> wrote:
> >>
> >> Export helper functions to allow other code to
> >>
> >> - determine the counter type in use (virtual or physical, CP15 or memory),
> >>
> >> - get a pointer to the arm_arch_timer clocksource, which can be compared
> >> with the current clocksource.
> >>
> >> The virtio_rtc driver will require the clocksource pointer when using
> >> get_device_system_crosststamp(), and should communicate the actual Arm
> >> counter type to the Virtio RTC device (cf. spec draft [1]).
> >
> > I really don't see why you should poke at the clocksource backend:
> >
> > - the MMIO clocksource is only used in PM situations, which a virtio
> > driver has no business being involved with
> >
> > - only the virtual counter is relevant -- it is always at a 0-offset
> > from the physical one when userspace has an opportunity to run
> >
> > So it really looks that out of the four combinations, only one is
> > relevant.
>
> Thanks for the explanation. Dropping arch_timer_counter_get_type() and
> assuming that the CP15 virtual counter is in use should also work for
> now. With the physical/virtual counter distinction, I tried to be
> future-proof, as per the following considerations:
>
> The intended consumer of arch_timer_counter_get_type() is the Virtio RTC
> device (draft spec [2], patch series [1]). If the Virtio device has
> optional cross-timestamp support, it must know the current Linux kernel
> view of the current clocksource counter. The Virtio driver tells the
> Virtio device the current counter type (in the Arm case, CNTVCT_EL0 or
> CNTPCT_EL0). It is intentionally left unspecified how the Virtio device
> would know the counter value. AFAIU, if the Linux kernel were a
> virtualization host itself, it would be better for the Virtio device to
> look at the physical counter, since the virtual counter could be set for
> a guest. And in other cases, the guest OSes use a virtual counter with
> an offset.
The physical counter can equally be offset (and KVM does offset it),
just like the virtual counter. With NV, the offsets themselves are
partially under control of the guest itself.
So either counters *as seen from the guest* are absolutely pointless
to an observer on the host. That observer sees a virtual counter that
is strictly equal to the physical counter.
> This was the rationale to come up with the physical/virtual counter
> distinction for the Virtio RTC device. Looking at extensions such as
> FEAT_ECV, where the CNTPCT_EL0 value can depend on the EL, or FEAT_NV*,
> it might be a bit simplistic.
Not just simplistic. It doesn't make sense. For this to work, you'd
need to know the global offset that KVM applies to the global counter,
plus the *virtualised* CNTPOFF/CNTVOFF values that the guest can
change at any time on a *per-CPU* basis. None of that is available
outside of KVM, nor would it make any sense anyway.
> Does this physical/virtual counter distinction sound like a good idea?
> Otherwise I would drop the arch_timer_counter_get_type() in the next
> iteration.
My take on this is that only the global counter value makes any sense.
That value is already available from the host as the virtual counter,
because we guarantee that CNTVOFF is 0 when outside of the guest
(well, technically, outside of the vcpu_load/vcpu_put section).
>
> >
> > I'm not Cc'd on the rest of the series, so I can't even see in which
> > context this is used. But as it is, the approach looks wrong.
> >
>
> Sorry, I will Cc you on all relevant patches in the next iteration,
> which I will send out soon.
>
> The first patch series can be found at [1]. I think the second helper
> function in this patch, arch_timer_get_cs(), would still be needed, in
> order to supply the clocksource to get_device_system_crosststamp().
We already have to deal with the kvm_arch_ptp_get_crosststamp()
monstrosity (which I will forever regret having merged). Surely you
can reuse some of that?
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource
2023-07-28 8:11 ` Marc Zyngier
@ 2023-07-31 16:15 ` Peter Hilber
0 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-07-31 16:15 UTC (permalink / raw)
To: Marc Zyngier
Cc: linux-arm-kernel, linux-kernel, Mark Rutland, Daniel Lezcano,
Thomas Gleixner
On 28.07.23 10:11, Marc Zyngier wrote:
> On Thu, 27 Jul 2023 11:22:11 +0100,
> Peter Hilber <peter.hilber@opensynergy.com> wrote:
>>
>> On 03.07.23 10:13, Marc Zyngier wrote:
>>> On Fri, 30 Jun 2023 18:10:47 +0100,
>>> Peter Hilber <peter.hilber@opensynergy.com> wrote:
>>>>
>>>> Export helper functions to allow other code to
>>>>
>>>> - determine the counter type in use (virtual or physical, CP15 or memory),
>>>>
>>>> - get a pointer to the arm_arch_timer clocksource, which can be compared
>>>> with the current clocksource.
>>>>
>>>> The virtio_rtc driver will require the clocksource pointer when using
>>>> get_device_system_crosststamp(), and should communicate the actual Arm
>>>> counter type to the Virtio RTC device (cf. spec draft [1]).
>>>
>>> I really don't see why you should poke at the clocksource backend:
>>>
>>> - the MMIO clocksource is only used in PM situations, which a virtio
>>> driver has no business being involved with
>>>
>>> - only the virtual counter is relevant -- it is always at a 0-offset
>>> from the physical one when userspace has an opportunity to run
>>>
>>> So it really looks that out of the four combinations, only one is
>>> relevant.
>>
>> Thanks for the explanation. Dropping arch_timer_counter_get_type() and
>> assuming that the CP15 virtual counter is in use should also work for
>> now. With the physical/virtual counter distinction, I tried to be
>> future-proof, as per the following considerations:
>>
>> The intended consumer of arch_timer_counter_get_type() is the Virtio RTC
>> device (draft spec [2], patch series [1]). If the Virtio device has
>> optional cross-timestamp support, it must know the current Linux kernel
>> view of the current clocksource counter. The Virtio driver tells the
>> Virtio device the current counter type (in the Arm case, CNTVCT_EL0 or
>> CNTPCT_EL0). It is intentionally left unspecified how the Virtio device
>> would know the counter value. AFAIU, if the Linux kernel were a
>> virtualization host itself, it would be better for the Virtio device to
>> look at the physical counter, since the virtual counter could be set for
>> a guest. And in other cases, the guest OSes use a virtual counter with
>> an offset.
>
> The physical counter can equally be offset (and KVM does offset it),
> just like the virtual counter. With NV, the offsets themselves are
> partially under control of the guest itself.
>
> So either counters *as seen from the guest* are absolutely pointless
> to an observer on the host. That observer sees a virtual counter that
> is strictly equal to the physical counter.
>
>> This was the rationale to come up with the physical/virtual counter
>> distinction for the Virtio RTC device. Looking at extensions such as
>> FEAT_ECV, where the CNTPCT_EL0 value can depend on the EL, or FEAT_NV*,
>> it might be a bit simplistic.
>
> Not just simplistic. It doesn't make sense. For this to work, you'd
> need to know the global offset that KVM applies to the global counter,
> plus the *virtualised* CNTPOFF/CNTVOFF values that the guest can
> change at any time on a *per-CPU* basis. None of that is available
> outside of KVM, nor would it make any sense anyway.
>
>> Does this physical/virtual counter distinction sound like a good idea?
>> Otherwise I would drop the arch_timer_counter_get_type() in the next
>> iteration.
>
> My take on this is that only the global counter value makes any sense.
> That value is already available from the host as the virtual counter,
> because we guarantee that CNTVOFF is 0 when outside of the guest
> (well, technically, outside of the vcpu_load/vcpu_put section).
>
OK, I agree that a virtual/physical counter distinction doesn't make
sense on Linux (unless one wants to abuse it to distinguish very special
use cases with and without Linux).
Probably I'll also remove the distinction from the spec draft.
>>
>>>
>>> I'm not Cc'd on the rest of the series, so I can't even see in which
>>> context this is used. But as it is, the approach looks wrong.
>>>
>>
>> Sorry, I will Cc you on all relevant patches in the next iteration,
>> which I will send out soon.
>>
>> The first patch series can be found at [1]. I think the second helper
>> function in this patch, arch_timer_get_cs(), would still be needed, in
>> order to supply the clocksource to get_device_system_crosststamp().
>
> We already have to deal with the kvm_arch_ptp_get_crosststamp()
> monstrosity (which I will forever regret having merged). Surely you
> can reuse some of that?
>
I'm not sure what this is hinting at.
From the virtio_rtc perspective, the only behavior shared with
kvm_arch_ptp_get_crosststamp() would be exposing &clocksource_counter
(and distinguishing virtual/physical counter, but virtio_rtc should stop
doing this). Exposing &clocksource_counter is also the only thing
arch_timer_get_cs() does.
If &clocksource_counter should not be exposed, then I can see two
alternatives:
Alternative 1: Put a function of type
int (*get_time_fn) (ktime_t *device_time,
struct system_counterval_t *sys_counterval,
void *ctx)
into arm_arch_timer.c, as required by get_device_system_crosststamp()
(and include a virtio_rtc header).
Alternative 2: Change get_device_system_crosststamp(), resp. struct
system_counterval_t, to use enum clocksource_ids to identify a
clocksource, instead of using struct clocksource *. Then there would be
no need to change arm_arch_timer.
Thanks for the feedback,
Peter
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH 5/7] virtio_rtc: Add module and driver core
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
` (3 preceding siblings ...)
2023-06-30 17:10 ` [RFC PATCH 4/7] clocksource: arm_arch_timer: Export counter type, clocksource Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 6/7] virtio_rtc: Add PTP clocks Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 7/7] virtio_rtc: Add Arm Generic Timer cross-timestamping Peter Hilber
6 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: virtualization, virtio-dev
Cc: Peter Hilber, linux-kernel, Michael S. Tsirkin, Jason Wang,
Xuan Zhuo, Richard Cochran, netdev
Add the virtio_rtc module and driver core. The virtio_rtc module implements
a driver compatible with the proposed Virtio RTC device specification [1].
The Virtio RTC (Real Time Clock) device provides information about current
time. The device can provide different clocks, e.g. for the UTC or TAI time
standards, or for physical time elapsed since some past epoch. The driver
can read the clocks with simple or more accurate methods.
Implement the core, which interacts with the Virtio RTC device. Apart from
this, the core does not expose functionality outside of the virtio_rtc
module. A follow-up patch will expose PTP clocks.
Provide synchronous messaging, which is enough for the expected time
synchronization use cases through PTP clocks (similar to ptp_kvm) or RTC
Class driver.
[1] https://lists.oasis-open.org/archives/virtio-comment/202306/msg00592.html
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
---
MAINTAINERS | 7 +
drivers/virtio/Kconfig | 14 +
drivers/virtio/Makefile | 2 +
drivers/virtio/virtio_rtc_driver.c | 736 +++++++++++++++++++++++++++
drivers/virtio/virtio_rtc_internal.h | 23 +
include/uapi/linux/virtio_rtc.h | 159 ++++++
6 files changed, 941 insertions(+)
create mode 100644 drivers/virtio/virtio_rtc_driver.c
create mode 100644 drivers/virtio/virtio_rtc_internal.h
create mode 100644 include/uapi/linux/virtio_rtc.h
diff --git a/MAINTAINERS b/MAINTAINERS
index cd5388a33410..4dcdb98146be 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -22661,6 +22661,13 @@ S: Maintained
F: drivers/nvdimm/nd_virtio.c
F: drivers/nvdimm/virtio_pmem.c
+VIRTIO RTC DRIVER
+M: Peter Hilber <peter.hilber@opensynergy.com>
+L: virtualization@lists.linux-foundation.org
+S: Maintained
+F: drivers/virtio/virtio_rtc_*
+F: include/uapi/linux/virtio_rtc.h
+
VIRTIO SOUND DRIVER
M: Anton Yakovlev <anton.yakovlev@opensynergy.com>
M: "Michael S. Tsirkin" <mst@redhat.com>
diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 0a53a61231c2..e3dbf16fa977 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -173,4 +173,18 @@ config VIRTIO_DMA_SHARED_BUFFER
This option adds a flavor of dma buffers that are backed by
virtio resources.
+config VIRTIO_RTC
+ tristate "Virtio RTC driver"
+ depends on VIRTIO
+ depends on PTP_1588_CLOCK_OPTIONAL
+ help
+ This driver provides current time from a Virtio RTC device. The driver
+ provides the time through one or more clocks. The driver sub-option
+ VIRTIO_RTC_PTP must be enabled to expose the clocks to userspace.
+
+ To compile this code as a module, choose M here: the module will be
+ called virtio_rtc.
+
+ If unsure, say M.
+
endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 8e98d24917cc..f760414ed6ab 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -12,3 +12,5 @@ obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o
obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o
obj-$(CONFIG_VIRTIO_MEM) += virtio_mem.o
obj-$(CONFIG_VIRTIO_DMA_SHARED_BUFFER) += virtio_dma_buf.o
+obj-$(CONFIG_VIRTIO_RTC) += virtio_rtc.o
+virtio_rtc-y := virtio_rtc_driver.o
diff --git a/drivers/virtio/virtio_rtc_driver.c b/drivers/virtio/virtio_rtc_driver.c
new file mode 100644
index 000000000000..424500d2c4f7
--- /dev/null
+++ b/drivers/virtio/virtio_rtc_driver.c
@@ -0,0 +1,736 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * virtio_rtc driver core
+ *
+ * Copyright (C) 2022-2023 OpenSynergy GmbH
+ */
+
+#include <linux/completion.h>
+#include <linux/virtio.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_config.h>
+#include <linux/module.h>
+
+#include <uapi/linux/virtio_rtc.h>
+
+#include "virtio_rtc_internal.h"
+
+/* virtqueue order */
+enum {
+ VIORTC_READQ,
+ VIORTC_CONTROLQ,
+ VIORTC_MAX_NR_QUEUES,
+};
+
+/**
+ * struct viortc_vq - virtqueue abstraction
+ * @vq: virtqueue
+ * @lock: protects access to vq
+ */
+struct viortc_vq {
+ struct virtqueue *vq;
+ spinlock_t lock;
+};
+
+/**
+ * struct viortc_dev - virtio_rtc device data
+ * @vdev: virtio device
+ * @vqs: virtqueues
+ * @num_clocks: # of virtio_rtc clocks
+ */
+struct viortc_dev {
+ struct virtio_device *vdev;
+ struct viortc_vq vqs[VIORTC_MAX_NR_QUEUES];
+ u16 num_clocks;
+};
+
+/**
+ * struct viortc_msg - Message requested by driver, responded by device.
+ * @viortc: device data
+ * @req: request buffer
+ * @resp: response buffer
+ * @responded: vqueue callback signals response reception
+ * @refcnt: Message reference count, message and buffers will be deallocated
+ * once 0. refcnt is decremented in the vqueue callback and in the
+ * thread waiting on the responded completion.
+ * If a message response wait function times out, the message will be
+ * freed upon late reception (refcnt will reach 0 in the callback), or
+ * device removal.
+ * @req_size: size of request in bytes
+ * @resp_cap: maximum size of response in bytes
+ * @resp_actual_size: actual size of response
+ */
+struct viortc_msg {
+ struct viortc_dev *viortc;
+ void *req;
+ void *resp;
+ struct completion responded;
+ refcount_t refcnt;
+ unsigned int req_size;
+ unsigned int resp_cap;
+ unsigned int resp_actual_size;
+};
+
+/**
+ * viortc_msg_init() - Allocate and initialize message.
+ * @viortc: device data
+ * @msg_type: virtio_rtc message type
+ * @req_size: size of request buffer to be allocated
+ * @resp_cap: size of response buffer to be allocated
+ *
+ * Initializes the message refcnt to 2. The refcnt will be decremented once in
+ * the virtqueue callback, and once in the thread waiting on the message (on
+ * completion or timeout).
+ *
+ * Context: Process context.
+ * Return: non-NULL on success.
+ */
+static struct viortc_msg *viortc_msg_init(struct viortc_dev *viortc,
+ u16 msg_type, unsigned int req_size,
+ unsigned int resp_cap)
+{
+ struct viortc_msg *msg;
+ struct device *dev = &viortc->vdev->dev;
+ struct virtio_rtc_req_head *req_head;
+
+ msg = devm_kzalloc(dev, sizeof(*msg), GFP_KERNEL);
+ if (!msg)
+ return NULL;
+
+ init_completion(&msg->responded);
+
+ msg->req = devm_kzalloc(dev, req_size, GFP_KERNEL);
+ if (!msg->req)
+ goto err_free_msg;
+
+ req_head = msg->req;
+
+ msg->resp = devm_kzalloc(dev, resp_cap, GFP_KERNEL);
+ if (!msg->resp)
+ goto err_free_msg_req;
+
+ msg->viortc = viortc;
+ msg->req_size = req_size;
+ msg->resp_cap = resp_cap;
+
+ refcount_set(&msg->refcnt, 2);
+
+ req_head->msg_type = virtio_cpu_to_le(msg_type, req_head->msg_type);
+
+ return msg;
+
+err_free_msg_req:
+ devm_kfree(dev, msg->req);
+
+err_free_msg:
+ devm_kfree(dev, msg);
+
+ return NULL;
+}
+
+/**
+ * viortc_msg_release() - Decrement message refcnt, potentially free message.
+ * @msg: message requested by driver
+ *
+ * Context: Any context.
+ */
+static void viortc_msg_release(struct viortc_msg *msg)
+{
+ if (refcount_dec_and_test(&msg->refcnt)) {
+ struct device *dev = &msg->viortc->vdev->dev;
+
+ devm_kfree(dev, msg->req);
+ devm_kfree(dev, msg->resp);
+ devm_kfree(dev, msg);
+ }
+}
+
+/**
+ * viortc_cb() - callback for readq and controlq
+ * @vq: virtqueue with device response
+ *
+ * Signals completion for each received message.
+ *
+ * Context: virtqueue callback, typically interrupt. Takes and releases vq lock.
+ */
+static void viortc_cb(struct virtqueue *vq)
+{
+ struct viortc_dev *viortc = vq->vdev->priv;
+ spinlock_t *lock = &viortc->vqs[vq->index].lock;
+ unsigned long flags;
+ struct viortc_msg *msg;
+ unsigned int len;
+ bool cb_enabled = true;
+
+ for (;;) {
+ spin_lock_irqsave(lock, flags);
+
+ if (cb_enabled) {
+ virtqueue_disable_cb(vq);
+ cb_enabled = false;
+ }
+
+ msg = virtqueue_get_buf(vq, &len);
+ if (!msg) {
+ if (virtqueue_enable_cb(vq)) {
+ spin_unlock_irqrestore(lock, flags);
+ return;
+ }
+ cb_enabled = true;
+ }
+
+ spin_unlock_irqrestore(lock, flags);
+
+ if (msg) {
+ msg->resp_actual_size = len;
+
+ /*
+ * completion waiter must see our msg metadata, but
+ * complete() does not guarantee a memory barrier
+ */
+ smp_wmb();
+
+ complete(&msg->responded);
+ viortc_msg_release(msg);
+ }
+ }
+}
+
+/**
+ * viortc_get_resp_errno() - converts virtio_rtc errnos to system errnos
+ * @resp_head: message response header
+ *
+ * Return: negative system errno, or 0
+ */
+static int viortc_get_resp_errno(struct virtio_rtc_resp_head *resp_head)
+{
+ switch (virtio_le_to_cpu(resp_head->status)) {
+ case VIRTIO_RTC_S_OK:
+ return 0;
+ case VIRTIO_RTC_S_UNSUPP:
+ return -EOPNOTSUPP;
+ case VIRTIO_RTC_S_INVAL:
+ return -EINVAL;
+ case VIRTIO_RTC_S_NODEV:
+ return -ENODEV;
+ case VIRTIO_RTC_S_DEVERR:
+ default:
+ return -EIO;
+ }
+}
+
+/**
+ * viortc_msg_xfer() - send message request, wait until message response
+ * @vq: virtqueue
+ * @msg: message with driver request
+ * @timeout_jiffies: message response timeout, 0 for no timeout
+ *
+ * Context: Process context. Takes and releases vq.lock. May sleep.
+ */
+static int viortc_msg_xfer(struct viortc_vq *vq, struct viortc_msg *msg,
+ unsigned long timeout_jiffies)
+{
+ int ret;
+ unsigned long flags;
+ struct scatterlist out_sg[1];
+ struct scatterlist in_sg[1];
+ struct scatterlist *sgs[2] = { out_sg, in_sg };
+ bool notify;
+
+ sg_init_one(out_sg, msg->req, msg->req_size);
+ sg_init_one(in_sg, msg->resp, msg->resp_cap);
+
+ spin_lock_irqsave(&vq->lock, flags);
+
+ ret = virtqueue_add_sgs(vq->vq, sgs, 1, 1, msg, GFP_ATOMIC);
+ if (ret) {
+ spin_unlock_irqrestore(&vq->lock, flags);
+ /*
+ * Release in place of the response callback, which will never
+ * come.
+ */
+ viortc_msg_release(msg);
+ return ret;
+ }
+
+ notify = virtqueue_kick_prepare(vq->vq);
+
+ spin_unlock_irqrestore(&vq->lock, flags);
+
+ if (notify)
+ virtqueue_notify(vq->vq);
+
+ if (timeout_jiffies) {
+ long timeout_ret;
+
+ timeout_ret = wait_for_completion_interruptible_timeout(
+ &msg->responded, timeout_jiffies);
+
+ if (!timeout_ret)
+ return -ETIMEDOUT;
+ else if (timeout_ret < 0)
+ return (int)timeout_ret;
+ } else {
+ ret = wait_for_completion_interruptible(&msg->responded);
+ if (ret)
+ return ret;
+ }
+
+ /*
+ * Ensure we can read message metadata written in the virtqueue
+ * callback.
+ */
+ smp_rmb();
+
+ /*
+ * There is not yet a case where returning a short message would make
+ * sense, so consider any deviation an error.
+ */
+ if (msg->resp_actual_size != msg->resp_cap)
+ return -EINVAL;
+
+ return viortc_get_resp_errno(msg->resp);
+}
+
+/*
+ * common message handle macros for messages of different types
+ */
+
+/**
+ * VIORTC_DECLARE_MSG_HDL_ONSTACK() - declare message handle on stack
+ * @hdl: message handle name
+ * @msg_suf_lowerc: message type suffix in lowercase
+ * @msg_suf_upperc: message type suffix in uppercase
+ */
+#define VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, msg_suf_lowerc, msg_suf_upperc) \
+ struct { \
+ struct viortc_msg *msg; \
+ struct virtio_rtc_req_##msg_suf_lowerc *req; \
+ struct virtio_rtc_resp_##msg_suf_lowerc *resp; \
+ unsigned int req_size; \
+ unsigned int resp_cap; \
+ u16 msg_type; \
+ } hdl = { \
+ NULL, \
+ NULL, \
+ NULL, \
+ sizeof(struct virtio_rtc_req_##msg_suf_lowerc), \
+ sizeof(struct virtio_rtc_resp_##msg_suf_lowerc), \
+ VIRTIO_RTC_M_##msg_suf_upperc, \
+ }
+
+/**
+ * VIORTC_MSG() - extract message from message handle
+ *
+ * Return: struct viortc_msg
+ */
+#define VIORTC_MSG(hdl) ((hdl).msg)
+
+/**
+ * VIORTC_MSG_INIT() - initialize message handle
+ * @hdl: message handle
+ * @viortc: device data (struct viortc_dev *)
+ *
+ * Context: Process context.
+ * Return: 0 on success, -ENOMEM otherwise.
+ */
+#define VIORTC_MSG_INIT(hdl, viortc) \
+ ({ \
+ typeof(hdl) *_hdl = &(hdl); \
+ \
+ _hdl->msg = viortc_msg_init((viortc), _hdl->msg_type, \
+ _hdl->req_size, _hdl->resp_cap); \
+ if (_hdl->msg) { \
+ _hdl->req = _hdl->msg->req; \
+ _hdl->resp = _hdl->msg->resp; \
+ } \
+ _hdl->msg ? 0 : -ENOMEM; \
+ })
+
+/**
+ * VIORTC_MSG_WRITE() - write a request message field
+ * @hdl: message handle
+ * @dest_member: request message field name
+ * @src_ptr: pointer to data of compatible type
+ *
+ * Writes the field in little-endian format.
+ */
+#define VIORTC_MSG_WRITE(hdl, dest_member, src_ptr) \
+ do { \
+ typeof(hdl) _hdl = (hdl); \
+ typeof(src_ptr) _src_ptr = (src_ptr); \
+ \
+ /* Sanity check: must match the member's type */ \
+ typecheck(typeof(_hdl.req->dest_member), *_src_ptr); \
+ \
+ _hdl.req->dest_member = \
+ virtio_cpu_to_le(*_src_ptr, _hdl.req->dest_member); \
+ } while (0)
+
+/**
+ * VIORTC_MSG_READ() - read from a response message field
+ * @hdl: message handle
+ * @src_member: response message field name
+ * @dest_ptr: pointer to data of compatible type
+ *
+ * Converts from little-endian format and writes to dest_ptr.
+ */
+#define VIORTC_MSG_READ(hdl, src_member, dest_ptr) \
+ do { \
+ typeof(dest_ptr) _dest_ptr = (dest_ptr); \
+ \
+ /* Sanity check: must match the member's type */ \
+ typecheck(typeof((hdl).resp->src_member), *_dest_ptr); \
+ \
+ *_dest_ptr = virtio_le_to_cpu((hdl).resp->src_member); \
+ } while (0)
+
+/*
+ * readq messages
+ */
+
+/** timeout for clock readings, where timeouts are considered non-fatal */
+#define VIORTC_MSG_READ_TIMEOUT (msecs_to_jiffies(60 * 1000))
+
+/**
+ * viortc_read() - VIRTIO_RTC_M_READ message wrapper
+ * @viortc: device data
+ * @vio_clk_id: virtio_rtc clock id
+ * @reading: clock reading [ns]
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_read(struct viortc_dev *viortc, u64 vio_clk_id, u64 *reading)
+{
+ int ret;
+ VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, read, READ);
+
+ ret = VIORTC_MSG_INIT(hdl, viortc);
+ if (ret)
+ return ret;
+
+ VIORTC_MSG_WRITE(hdl, clock_id, &vio_clk_id);
+
+ ret = viortc_msg_xfer(&viortc->vqs[VIORTC_READQ], VIORTC_MSG(hdl),
+ VIORTC_MSG_READ_TIMEOUT);
+ if (ret) {
+ dev_dbg(&viortc->vdev->dev, "%s: xfer returned %d\n", __func__,
+ ret);
+ goto out_release;
+ }
+
+ VIORTC_MSG_READ(hdl, clock_reading, reading);
+
+out_release:
+ viortc_msg_release(VIORTC_MSG(hdl));
+
+ return ret;
+}
+
+/**
+ * viortc_read_cross() - VIRTIO_RTC_M_READ_CROSS message wrapper
+ * @viortc: device data
+ * @vio_clk_id: virtio_rtc clock id
+ * @hw_counter: virtio_rtc HW counter type
+ * @reading: clock reading [ns]
+ * @cycles: HW counter cycles during clock reading
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_read_cross(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
+ u64 *reading, u64 *cycles)
+{
+ int ret;
+ VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, read_cross, READ_CROSS);
+
+ ret = VIORTC_MSG_INIT(hdl, viortc);
+ if (ret)
+ return ret;
+
+ VIORTC_MSG_WRITE(hdl, clock_id, &vio_clk_id);
+ VIORTC_MSG_WRITE(hdl, hw_counter, &hw_counter);
+
+ ret = viortc_msg_xfer(&viortc->vqs[VIORTC_READQ], VIORTC_MSG(hdl),
+ VIORTC_MSG_READ_TIMEOUT);
+ if (ret) {
+ dev_dbg(&viortc->vdev->dev, "%s: xfer returned %d\n", __func__,
+ ret);
+ goto out_release;
+ }
+
+ VIORTC_MSG_READ(hdl, clock_reading, reading);
+ VIORTC_MSG_READ(hdl, counter_cycles, cycles);
+
+out_release:
+ viortc_msg_release(VIORTC_MSG(hdl));
+
+ return ret;
+}
+
+/*
+ * controlq messages
+ */
+
+/**
+ * viortc_cfg() - VIRTIO_RTC_M_CFG message wrapper
+ * @viortc: device data
+ * @num_clocks: # of virtio_rtc clocks
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_cfg(struct viortc_dev *viortc, u16 *num_clocks)
+{
+ int ret;
+ VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, cfg, CFG);
+
+ ret = VIORTC_MSG_INIT(hdl, viortc);
+ if (ret)
+ return ret;
+
+ ret = viortc_msg_xfer(&viortc->vqs[VIORTC_CONTROLQ], VIORTC_MSG(hdl),
+ 0);
+ if (ret) {
+ dev_dbg(&viortc->vdev->dev, "%s: xfer returned %d\n", __func__,
+ ret);
+ goto out_release;
+ }
+
+ VIORTC_MSG_READ(hdl, num_clocks, num_clocks);
+
+out_release:
+ viortc_msg_release(VIORTC_MSG(hdl));
+
+ return ret;
+}
+
+/**
+ * viortc_clock_cap() - VIRTIO_RTC_M_CLOCK_CAP message wrapper
+ * @viortc: device data
+ * @vio_clk_id: virtio_rtc clock id
+ * @type: virtio_rtc clock type
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_clock_cap(struct viortc_dev *viortc, u64 vio_clk_id,
+ u16 *type)
+{
+ int ret;
+ VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, clock_cap, CLOCK_CAP);
+
+ ret = VIORTC_MSG_INIT(hdl, viortc);
+ if (ret)
+ return ret;
+
+ VIORTC_MSG_WRITE(hdl, clock_id, &vio_clk_id);
+
+ ret = viortc_msg_xfer(&viortc->vqs[VIORTC_CONTROLQ], VIORTC_MSG(hdl),
+ 0);
+ if (ret) {
+ dev_dbg(&viortc->vdev->dev, "%s: xfer returned %d\n", __func__,
+ ret);
+ goto out_release;
+ }
+
+ VIORTC_MSG_READ(hdl, type, type);
+
+out_release:
+ viortc_msg_release(VIORTC_MSG(hdl));
+
+ return ret;
+}
+
+/**
+ * viortc_cross_cap() - VIRTIO_RTC_M_CROSS_CAP message wrapper
+ * @viortc: device data
+ * @vio_clk_id: virtio_rtc clock id
+ * @hw_counter: virtio_rtc HW counter type
+ * @supported: xtstamping is supported for the vio_clk_id/hw_counter pair
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_cross_cap(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
+ bool *supported)
+{
+ int ret;
+ VIORTC_DECLARE_MSG_HDL_ONSTACK(hdl, cross_cap, CROSS_CAP);
+ u8 flags;
+
+ ret = VIORTC_MSG_INIT(hdl, viortc);
+ if (ret)
+ return ret;
+
+ VIORTC_MSG_WRITE(hdl, clock_id, &vio_clk_id);
+ VIORTC_MSG_WRITE(hdl, hw_counter, &hw_counter);
+
+ ret = viortc_msg_xfer(&viortc->vqs[VIORTC_CONTROLQ], VIORTC_MSG(hdl),
+ 0);
+ if (ret) {
+ dev_dbg(&viortc->vdev->dev, "%s: xfer returned %d\n", __func__,
+ ret);
+ goto out_release;
+ }
+
+ VIORTC_MSG_READ(hdl, flags, &flags);
+ *supported = !!(flags & BIT(VIRTIO_RTC_FLAG_CROSS_CAP));
+
+out_release:
+ viortc_msg_release(VIORTC_MSG(hdl));
+
+ return ret;
+}
+
+/*
+ * init, deinit
+ */
+
+/**
+ * viortc_clocks_init() - init local representations of virtio_rtc clocks
+ * @viortc: device data
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_clocks_init(struct viortc_dev *viortc)
+{
+ int ret;
+ u16 num_clocks;
+
+ ret = viortc_cfg(viortc, &num_clocks);
+ if (ret)
+ return ret;
+
+ if (num_clocks < 1) {
+ dev_err(&viortc->vdev->dev, "device reported 0 clocks\n");
+ return -ENODEV;
+ }
+
+ viortc->num_clocks = num_clocks;
+
+ /* In the future, PTP clocks will be initialized here. */
+ (void)viortc_clock_cap;
+
+ return 0;
+}
+
+/**
+ * viortc_init_vqs() - init virtqueues
+ * @viortc: device data
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ *
+ * Init virtqueues, and their abstractions.
+ */
+static int viortc_init_vqs(struct viortc_dev *viortc)
+{
+ int ret;
+ struct virtio_device *vdev = viortc->vdev;
+ const char *names[VIORTC_MAX_NR_QUEUES];
+ vq_callback_t *callbacks[VIORTC_MAX_NR_QUEUES];
+ struct virtqueue *vqs[VIORTC_MAX_NR_QUEUES];
+ int nr_queues;
+
+ names[VIORTC_READQ] = "readq";
+ callbacks[VIORTC_READQ] = viortc_cb;
+
+ names[VIORTC_CONTROLQ] = "controlq";
+ callbacks[VIORTC_CONTROLQ] = viortc_cb;
+
+ nr_queues = 2;
+
+ ret = virtio_find_vqs(vdev, nr_queues, vqs, callbacks, names, NULL);
+ if (ret)
+ return ret;
+
+ viortc->vqs[VIORTC_READQ].vq = vqs[VIORTC_READQ];
+ spin_lock_init(&viortc->vqs[VIORTC_READQ].lock);
+
+ viortc->vqs[VIORTC_CONTROLQ].vq = vqs[VIORTC_CONTROLQ];
+ spin_lock_init(&viortc->vqs[VIORTC_CONTROLQ].lock);
+
+ return 0;
+}
+
+/**
+ * viortc_probe() - probe a virtio_rtc virtio device
+ * @vdev: virtio device
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_probe(struct virtio_device *vdev)
+{
+ struct viortc_dev *viortc;
+ int ret;
+
+ viortc = devm_kzalloc(&vdev->dev, sizeof(*viortc), GFP_KERNEL);
+ if (!viortc)
+ return -ENOMEM;
+
+ vdev->priv = viortc;
+ viortc->vdev = vdev;
+
+ ret = viortc_init_vqs(viortc);
+ if (ret)
+ return ret;
+
+ virtio_device_ready(vdev);
+
+ /* Ready vdev for use by frontend devices initialized next. */
+ smp_wmb();
+
+ ret = viortc_clocks_init(viortc);
+ if (ret)
+ goto err_reset_vdev;
+
+ return 0;
+
+err_reset_vdev:
+ virtio_reset_device(vdev);
+ vdev->config->del_vqs(vdev);
+
+ return ret;
+}
+
+/**
+ * viortc_remove() - remove a virtio_rtc virtio device
+ * @vdev: virtio device
+ */
+static void viortc_remove(struct virtio_device *vdev)
+{
+ /* In the future, PTP clocks will be deinitialized here. */
+
+ virtio_reset_device(vdev);
+ vdev->config->del_vqs(vdev);
+}
+
+static unsigned int features[] = {
+ VIRTIO_RTC_F_READ_CROSS,
+};
+
+static struct virtio_device_id id_table[] = {
+ { VIRTIO_ID_CLOCK, VIRTIO_DEV_ANY_ID },
+ { 0 },
+};
+MODULE_DEVICE_TABLE(virtio, id_table);
+
+static struct virtio_driver virtio_rtc_drv = {
+ .driver.name = KBUILD_MODNAME,
+ .driver.owner = THIS_MODULE,
+ .feature_table = features,
+ .feature_table_size = ARRAY_SIZE(features),
+ .id_table = id_table,
+ .probe = viortc_probe,
+ .remove = viortc_remove,
+};
+
+module_virtio_driver(virtio_rtc_drv);
+
+MODULE_DESCRIPTION("Virtio RTC driver");
+MODULE_AUTHOR("OpenSynergy GmbH");
+MODULE_LICENSE("GPL");
diff --git a/drivers/virtio/virtio_rtc_internal.h b/drivers/virtio/virtio_rtc_internal.h
new file mode 100644
index 000000000000..c2b5387f506f
--- /dev/null
+++ b/drivers/virtio/virtio_rtc_internal.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * virtio_rtc internal interfaces
+ *
+ * Copyright (C) 2022-2023 OpenSynergy GmbH
+ */
+
+#ifndef _VIRTIO_RTC_INTERNAL_H_
+#define _VIRTIO_RTC_INTERNAL_H_
+
+#include <linux/types.h>
+
+/* driver core IFs */
+
+struct viortc_dev;
+
+int viortc_read(struct viortc_dev *viortc, u64 vio_clk_id, u64 *reading);
+int viortc_read_cross(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
+ u64 *reading, u64 *cycles);
+int viortc_cross_cap(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
+ bool *supported);
+
+#endif /* _VIRTIO_RTC_INTERNAL_H_ */
diff --git a/include/uapi/linux/virtio_rtc.h b/include/uapi/linux/virtio_rtc.h
new file mode 100644
index 000000000000..0926b3d58254
--- /dev/null
+++ b/include/uapi/linux/virtio_rtc.h
@@ -0,0 +1,159 @@
+/* SPDX-License-Identifier: ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) */
+/*
+ * Copyright (C) 2022-2023 OpenSynergy GmbH
+ */
+
+#ifndef _LINUX_VIRTIO_RTC_H
+#define _LINUX_VIRTIO_RTC_H
+
+#include <linux/types.h>
+
+/* Device-specific features */
+
+#define VIRTIO_RTC_F_READ_CROSS 0
+
+/* readq message types */
+
+#define VIRTIO_RTC_M_READ 0x0001
+#define VIRTIO_RTC_M_READ_CROSS 0x0002
+
+/* controlq message types */
+
+#define VIRTIO_RTC_M_CFG 0x1000
+#define VIRTIO_RTC_M_CLOCK_CAP 0x1001
+#define VIRTIO_RTC_M_CROSS_CAP 0x1002
+
+/* Message headers */
+
+/** common request header */
+struct virtio_rtc_req_head {
+ __le16 msg_type;
+ __u8 reserved[2];
+};
+
+/** common response header */
+struct virtio_rtc_resp_head {
+#define VIRTIO_RTC_S_OK 0
+#define VIRTIO_RTC_S_UNSUPP 1
+#define VIRTIO_RTC_S_NODEV 2
+#define VIRTIO_RTC_S_INVAL 3
+#define VIRTIO_RTC_S_DEVERR 4
+ __u8 status;
+ __u8 reserved[3];
+};
+
+/* readq messages */
+
+/* VIRTIO_RTC_M_READ message */
+
+struct virtio_rtc_req_read {
+ struct virtio_rtc_req_head head;
+ __u8 reserved[4];
+ __le64 clock_id;
+};
+
+struct virtio_rtc_resp_read {
+ struct virtio_rtc_resp_head head;
+ __u8 reserved[4];
+ __le64 clock_reading;
+};
+
+/* VIRTIO_RTC_M_READ_CROSS message */
+
+struct virtio_rtc_req_read_cross {
+ struct virtio_rtc_req_head head;
+/** Arm Generic Timer Virtual Count */
+#define VIRTIO_RTC_COUNTER_ARM_VIRT 0
+/** Arm Generic Timer Physical Count */
+#define VIRTIO_RTC_COUNTER_ARM_PHYS 1
+/** x86 Time Stamp Counter */
+#define VIRTIO_RTC_COUNTER_X86_TSC 2
+ __le16 hw_counter;
+ __u8 reserved[2];
+ __le64 clock_id;
+};
+
+struct virtio_rtc_resp_read_cross {
+ struct virtio_rtc_resp_head head;
+ __u8 reserved[4];
+ __le64 clock_reading;
+ __le64 counter_cycles;
+};
+
+/** Union of request types for readq */
+union virtio_rtc_req_readq {
+ struct virtio_rtc_req_read read;
+ struct virtio_rtc_req_read_cross read_cross;
+};
+
+/** Union of response types for readq */
+union virtio_rtc_resp_readq {
+ struct virtio_rtc_resp_read read;
+ struct virtio_rtc_resp_read_cross read_cross;
+};
+
+/* controlq messages */
+
+/* VIRTIO_RTC_M_CFG message */
+
+struct virtio_rtc_req_cfg {
+ struct virtio_rtc_req_head head;
+ /* no request params */
+ __u8 reserved[4];
+};
+
+struct virtio_rtc_resp_cfg {
+ struct virtio_rtc_resp_head head;
+ /** # of clocks -> clock ids < num_clocks are valid */
+ __le16 num_clocks;
+ __u8 reserved[10];
+};
+
+/* VIRTIO_RTC_M_CLOCK_CAP message */
+
+struct virtio_rtc_req_clock_cap {
+ struct virtio_rtc_req_head head;
+ __u8 reserved[4];
+ __le64 clock_id;
+};
+
+struct virtio_rtc_resp_clock_cap {
+ struct virtio_rtc_resp_head head;
+#define VIRTIO_RTC_CLOCK_UTC 0
+#define VIRTIO_RTC_CLOCK_TAI 1
+#define VIRTIO_RTC_CLOCK_MONO 2
+ __le16 type;
+ __u8 reserved[10];
+};
+
+/* VIRTIO_RTC_M_CROSS_CAP message */
+
+struct virtio_rtc_req_cross_cap {
+ struct virtio_rtc_req_head head;
+ __le16 hw_counter;
+ __u8 reserved[2];
+ __le64 clock_id;
+};
+
+struct virtio_rtc_resp_cross_cap {
+ struct virtio_rtc_resp_head head;
+#define VIRTIO_RTC_FLAG_CROSS_CAP 0
+ __u8 flags;
+ __u8 reserved[11];
+};
+
+/** Union of request types for controlq */
+union virtio_rtc_req_controlq {
+ struct virtio_rtc_req_cfg cfg;
+ struct virtio_rtc_req_clock_cap clock_cap;
+ struct virtio_rtc_req_cross_cap cross_cap;
+};
+
+/** Union of response types for controlq */
+union virtio_rtc_resp_controlq {
+ struct virtio_rtc_resp_cfg cfg;
+ struct virtio_rtc_resp_clock_cap clock_cap;
+ struct virtio_rtc_resp_cross_cap cross_cap;
+};
+
+#endif /* _LINUX_VIRTIO_RTC_H */
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* [RFC PATCH 6/7] virtio_rtc: Add PTP clocks
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
` (4 preceding siblings ...)
2023-06-30 17:10 ` [RFC PATCH 5/7] virtio_rtc: Add module and driver core Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
2023-06-30 17:10 ` [RFC PATCH 7/7] virtio_rtc: Add Arm Generic Timer cross-timestamping Peter Hilber
6 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: virtualization, virtio-dev
Cc: Peter Hilber, linux-kernel, Michael S. Tsirkin, Jason Wang,
Xuan Zhuo, Richard Cochran, netdev
Expose the virtio_rtc clocks as PTP clocks to userspace, similar to
ptp_kvm. virtio_rtc can expose multiple clocks, e.g. a UTC clock and a
monotonic clock. Userspace should distinguish different clocks through the
name assigned by the driver. A udev rule such as the following can be used
to get a symlink /dev/ptp_virtio to the UTC clock:
SUBSYSTEM=="ptp", ATTR{clock_name}=="Virtio PTP UTC", SYMLINK += "ptp_virtio"
The preferred PTP clock reading method is ioctl PTP_SYS_OFFSET_PRECISE2,
through the ptp_clock_info.getcrosststamp() op. For now,
PTP_SYS_OFFSET_PRECISE2 will return -EOPNOTSUPP through a weak function.
PTP_SYS_OFFSET_PRECISE2 requires cross-timestamping support for specific
clocksources, which will be added in the following. If the clocksource
specific code is enabled, check that the Virtio RTC device supports the
respective HW counter before obtaining an actual cross-timestamp from the
Virtio device.
The Virtio RTC device response time may be higher than the timekeeper
seqcount increment interval. Therefore, obtain the cross-timestamp before
calling get_device_system_crosststamp().
As a fallback, support the ioctl PTP_SYS_OFFSET_EXTENDED2 for all
platforms.
Assume that concurrency issues during PTP clock removal are avoided by the
posix_clock framework.
Kconfig recursive dependencies prevent virtio_rtc from implicitly enabling
PTP_1588_CLOCK, therefore just warn the user if PTP_1588_CLOCK is not
available. Since virtio_rtc should in the future also expose clocks as RTC
class devices, it should make sense to not have VIRTIO_RTC depend on
PTP_1588_CLOCK.
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
---
drivers/virtio/Kconfig | 16 ++
drivers/virtio/Makefile | 1 +
drivers/virtio/virtio_rtc_driver.c | 111 +++++++-
drivers/virtio/virtio_rtc_internal.h | 62 +++++
drivers/virtio/virtio_rtc_ptp.c | 384 +++++++++++++++++++++++++++
5 files changed, 571 insertions(+), 3 deletions(-)
create mode 100644 drivers/virtio/virtio_rtc_ptp.c
diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index e3dbf16fa977..7369ecd7dd01 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -187,4 +187,20 @@ config VIRTIO_RTC
If unsure, say M.
+comment "WARNING: The Virtio RTC driver is useless without VIRTIO_RTC_PTP."
+ depends on VIRTIO_RTC && !VIRTIO_RTC_PTP
+
+comment "Enable PTP_1588_CLOCK in order to enable VIRTIO_RTC_PTP."
+ depends on VIRTIO_RTC && PTP_1588_CLOCK=n
+
+config VIRTIO_RTC_PTP
+ bool "Virtio RTC PTP clocks"
+ default y
+ depends on VIRTIO_RTC && PTP_1588_CLOCK
+ help
+ This exposes any Virtio RTC clocks as PTP Hardware Clocks (PHCs) to
+ userspace.
+
+ If unsure, say Y.
+
endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index f760414ed6ab..4d48cbcae6bb 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -14,3 +14,4 @@ obj-$(CONFIG_VIRTIO_MEM) += virtio_mem.o
obj-$(CONFIG_VIRTIO_DMA_SHARED_BUFFER) += virtio_dma_buf.o
obj-$(CONFIG_VIRTIO_RTC) += virtio_rtc.o
virtio_rtc-y := virtio_rtc_driver.o
+virtio_rtc-$(CONFIG_VIRTIO_RTC_PTP) += virtio_rtc_ptp.o
diff --git a/drivers/virtio/virtio_rtc_driver.c b/drivers/virtio/virtio_rtc_driver.c
index 424500d2c4f7..3c11fa95b9a7 100644
--- a/drivers/virtio/virtio_rtc_driver.c
+++ b/drivers/virtio/virtio_rtc_driver.c
@@ -36,11 +36,16 @@ struct viortc_vq {
* struct viortc_dev - virtio_rtc device data
* @vdev: virtio device
* @vqs: virtqueues
+ * @clocks_to_unregister: Clock references, which are only used during device
+ * removal.
+ * For other uses, there would be a race between device
+ * creation and setting the pointers here.
* @num_clocks: # of virtio_rtc clocks
*/
struct viortc_dev {
struct virtio_device *vdev;
struct viortc_vq vqs[VIORTC_MAX_NR_QUEUES];
+ struct viortc_ptp_clock **clocks_to_unregister;
u16 num_clocks;
};
@@ -588,6 +593,89 @@ int viortc_cross_cap(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
* init, deinit
*/
+/**
+ * viortc_init_clock() - init local representation of virtio_rtc clock (PHC)
+ * @viortc: device data
+ * @vio_clk_id: virtio_rtc clock id
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_init_clock(struct viortc_dev *viortc, u64 vio_clk_id)
+{
+ int ret;
+ u16 clock_type;
+ char ptp_clock_name[PTP_CLOCK_NAME_LEN];
+ const char *type_name;
+ /* fit prefix + u16 in decimal */
+ char type_name_buf[5 + 5 + 1];
+ bool has_xtstamp_feature;
+ struct viortc_ptp_clock *vio_ptp;
+ struct virtio_device *vdev = viortc->vdev;
+
+ ret = viortc_clock_cap(viortc, vio_clk_id, &clock_type);
+ if (ret)
+ return ret;
+
+ switch (clock_type) {
+ case VIRTIO_RTC_CLOCK_UTC:
+ type_name = "UTC";
+ break;
+ case VIRTIO_RTC_CLOCK_TAI:
+ type_name = "TAI";
+ break;
+ case VIRTIO_RTC_CLOCK_MONO:
+ type_name = "monotonic";
+ break;
+ default:
+ snprintf(type_name_buf, sizeof(type_name_buf), "type %hu",
+ clock_type);
+ type_name = type_name_buf;
+ }
+
+ snprintf(ptp_clock_name, PTP_CLOCK_NAME_LEN, "Virtio PTP %s",
+ type_name);
+
+ has_xtstamp_feature = virtio_has_feature(vdev, VIRTIO_RTC_F_READ_CROSS);
+
+ vio_ptp = viortc_ptp_register(viortc, &vdev->dev, vio_clk_id,
+ ptp_clock_name, has_xtstamp_feature);
+ if (IS_ERR(vio_ptp)) {
+ dev_err(&vdev->dev, "failed to register PTP clock '%s'\n",
+ ptp_clock_name);
+ return PTR_ERR(vio_ptp);
+ }
+
+ viortc->clocks_to_unregister[vio_clk_id] = vio_ptp;
+
+ if (!vio_ptp)
+ dev_warn(&vdev->dev, "clock %llu is not exposed to userspace\n",
+ vio_clk_id);
+
+ return 0;
+}
+
+/**
+ * viortc_clocks_exit() - unregister PHCs
+ * @viortc: device data
+ */
+static void viortc_clocks_exit(struct viortc_dev *viortc)
+{
+ unsigned int i;
+ struct viortc_ptp_clock *vio_ptp;
+
+ for (i = 0; i < viortc->num_clocks; i++) {
+ vio_ptp = viortc->clocks_to_unregister[i];
+
+ if (!vio_ptp)
+ continue;
+
+ viortc->clocks_to_unregister[i] = NULL;
+
+ WARN_ON(viortc_ptp_unregister(vio_ptp, &viortc->vdev->dev));
+ }
+}
+
/**
* viortc_clocks_init() - init local representations of virtio_rtc clocks
* @viortc: device data
@@ -599,6 +687,7 @@ static int viortc_clocks_init(struct viortc_dev *viortc)
{
int ret;
u16 num_clocks;
+ unsigned int i;
ret = viortc_cfg(viortc, &num_clocks);
if (ret)
@@ -611,10 +700,24 @@ static int viortc_clocks_init(struct viortc_dev *viortc)
viortc->num_clocks = num_clocks;
- /* In the future, PTP clocks will be initialized here. */
- (void)viortc_clock_cap;
+ viortc->clocks_to_unregister =
+ devm_kcalloc(&viortc->vdev->dev, num_clocks,
+ sizeof(*viortc->clocks_to_unregister), GFP_KERNEL);
+ if (!viortc->clocks_to_unregister)
+ return -ENOMEM;
+
+ for (i = 0; i < num_clocks; i++) {
+ ret = viortc_init_clock(viortc, i);
+ if (ret)
+ goto err_free_clocks;
+ }
return 0;
+
+err_free_clocks:
+ viortc_clocks_exit(viortc);
+
+ return ret;
}
/**
@@ -703,7 +806,9 @@ static int viortc_probe(struct virtio_device *vdev)
*/
static void viortc_remove(struct virtio_device *vdev)
{
- /* In the future, PTP clocks will be deinitialized here. */
+ struct viortc_dev *viortc = vdev->priv;
+
+ viortc_clocks_exit(viortc);
virtio_reset_device(vdev);
vdev->config->del_vqs(vdev);
diff --git a/drivers/virtio/virtio_rtc_internal.h b/drivers/virtio/virtio_rtc_internal.h
index c2b5387f506f..d8bd008cb8f6 100644
--- a/drivers/virtio/virtio_rtc_internal.h
+++ b/drivers/virtio/virtio_rtc_internal.h
@@ -9,6 +9,7 @@
#define _VIRTIO_RTC_INTERNAL_H_
#include <linux/types.h>
+#include <linux/ptp_clock_kernel.h>
/* driver core IFs */
@@ -20,4 +21,65 @@ int viortc_read_cross(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
int viortc_cross_cap(struct viortc_dev *viortc, u64 vio_clk_id, u16 hw_counter,
bool *supported);
+/* PTP IFs */
+
+struct viortc_ptp_clock;
+
+#if IS_ENABLED(CONFIG_VIRTIO_RTC_PTP)
+
+struct viortc_ptp_clock *viortc_ptp_register(struct viortc_dev *viortc,
+ struct device *parent_dev,
+ u64 vio_clk_id,
+ const char *ptp_clock_name,
+ bool try_enable_xtstamp);
+int viortc_ptp_unregister(struct viortc_ptp_clock *vio_ptp,
+ struct device *parent_dev);
+
+#else
+
+static inline struct viortc_ptp_clock *
+viortc_ptp_register(struct viortc_dev *viortc, struct device *parent_dev,
+ u64 vio_clk_id, const char *ptp_clock_name,
+ bool try_enable_xtstamp)
+{
+ return NULL;
+}
+
+int viortc_ptp_unregister(struct viortc_ptp_clock *vio_ptp,
+ struct device *parent_dev)
+{
+ return -ENODEV;
+}
+
+#endif
+
+/* HW counter IFs */
+
+/**
+ * Maximum # of HW counters which the driver can support - can be increased.
+ */
+#define VIORTC_CAP_HW_COUNTERS 4
+
+/**
+ * viortc_hw_get_counters() - get HW counters present
+ * @hw_counters: virtio_rtc HW counters
+ * @num_hw_counters: number of HW counters
+ *
+ * num_hw_counters must not exceed VIORTC_CAP_HW_COUNTERS.
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_hw_get_counters(const u16 **hw_counters, int *num_hw_counters);
+
+/**
+ * viortc_hw_xtstamp_params() - get HW-specific xtstamp params
+ * @hw_counter: virtio_rtc HW counter type
+ * @cs: clocksource corresponding to hw_counter
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_hw_xtstamp_params(u16 *hw_counter, struct clocksource **cs);
+
#endif /* _VIRTIO_RTC_INTERNAL_H_ */
diff --git a/drivers/virtio/virtio_rtc_ptp.c b/drivers/virtio/virtio_rtc_ptp.c
new file mode 100644
index 000000000000..e52205a1caa9
--- /dev/null
+++ b/drivers/virtio/virtio_rtc_ptp.c
@@ -0,0 +1,384 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Expose virtio_rtc clocks as PTP clocks.
+ *
+ * Copyright (C) 2022-2023 OpenSynergy GmbH
+ *
+ * Derived from ptp_kvm_common.c, virtual PTP 1588 clock for use with KVM
+ * guests.
+ *
+ * Copyright (C) 2017 Red Hat Inc.
+ */
+
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/ptp_clock_kernel.h>
+
+#include <uapi/linux/virtio_rtc.h>
+
+#include "virtio_rtc_internal.h"
+
+/**
+ * struct viortc_ptp_clock - PTP clock abstraction
+ * @vio_clk_id: virtio_rtc clock id
+ * @ptp_clock: PTP clock handle
+ * @viortc: virtio_rtc device data
+ * @ptp_info: PTP clock description
+ * @num_hw_counters: actual # of hw_counters
+ * @hw_counters: HW clocks which are supported for xtstamping
+ */
+struct viortc_ptp_clock {
+ u64 vio_clk_id;
+ struct ptp_clock *ptp_clock;
+ struct viortc_dev *viortc;
+ struct ptp_clock_info ptp_info;
+ u32 num_hw_counters;
+ u16 hw_counters[VIORTC_CAP_HW_COUNTERS];
+};
+
+/**
+ * struct viortc_ptp_cross_ctx - context for get_device_system_crosststamp()
+ * @device_time: device clock reading
+ * @system_counterval: HW counter value at device_time
+ *
+ * Provides the already obtained crosststamp to get_device_system_crosststamp().
+ */
+struct viortc_ptp_cross_ctx {
+ ktime_t device_time;
+ struct system_counterval_t system_counterval;
+};
+
+/* Weak functions in case get_device_system_crosststamp() is not supported */
+
+int __weak viortc_hw_get_counters(const u16 **hw_counters, int *num_hw_counters)
+{
+ *hw_counters = NULL;
+ *num_hw_counters = 0;
+ return 0;
+}
+
+int __weak viortc_hw_xtstamp_params(u16 *hw_counter, struct clocksource **cs)
+{
+ return -EOPNOTSUPP;
+}
+
+/**
+ * viortc_ptp_get_time_fn() - callback for get_device_system_crosststamp()
+ * @device_time: device clock reading
+ * @system_counterval: HW counter value at device_time
+ * @ctx: context with already obtained crosststamp
+ *
+ * Return: zero (success).
+ */
+static int viortc_ptp_get_time_fn(ktime_t *device_time,
+ struct system_counterval_t *system_counterval,
+ void *ctx)
+{
+ struct viortc_ptp_cross_ctx *vio_ctx = ctx;
+
+ *device_time = vio_ctx->device_time;
+ *system_counterval = vio_ctx->system_counterval;
+
+ return 0;
+}
+
+/**
+ * viortc_ptp_check_hw_counter_supported() - look up if xtstamp supported
+ * @vio_ptp: virtio_rtc PTP clock
+ * @hw_counter: virtio_rtc HW counter type
+ *
+ * Return: Zero if xtstamp is supported for hw_counter, negative error code
+ * otherwise.
+ */
+static int
+viortc_ptp_check_hw_counter_supported(struct viortc_ptp_clock *vio_ptp,
+ u16 hw_counter)
+{
+ u32 i;
+
+ for (i = 0; i < vio_ptp->num_hw_counters; i++) {
+ if (vio_ptp->hw_counters[i] == hw_counter)
+ return 0;
+ }
+
+ return -EOPNOTSUPP;
+}
+
+/**
+ * viortc_ptp_do_xtstamp() - get HW-specific crosststamp from device
+ * @vio_ptp: virtio_rtc PTP clock
+ * @ctx: context for get_device_system_crosststamp()
+ *
+ * Gets HW-specific crosststamp params and reads crosststamp from device.
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_ptp_do_xtstamp(struct viortc_ptp_clock *vio_ptp,
+ struct viortc_ptp_cross_ctx *ctx)
+{
+ u16 hw_counter;
+ u64 ns;
+ u64 max_ns;
+ int ret;
+
+ ret = viortc_hw_xtstamp_params(&hw_counter, &ctx->system_counterval.cs);
+ if (ret)
+ return ret;
+
+ ret = viortc_ptp_check_hw_counter_supported(vio_ptp, hw_counter);
+ if (ret)
+ return ret;
+
+ ret = viortc_read_cross(vio_ptp->viortc, vio_ptp->vio_clk_id,
+ hw_counter, &ns,
+ &ctx->system_counterval.cycles);
+ if (ret)
+ return ret;
+
+ max_ns = (u64)ktime_to_ns(KTIME_MAX);
+ if (ns > max_ns)
+ return -EINVAL;
+
+ ctx->device_time = ns_to_ktime(ns);
+
+ return 0;
+}
+
+/*
+ * PTP clock operations
+ */
+
+/**
+ * viortc_ptp_getcrosststamp() - PTP clock getcrosststamp op
+ * @vio_ptp: virtio_rtc PTP clock
+ * @xtstamp: crosststamp
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_ptp_getcrosststamp(struct ptp_clock_info *ptp,
+ struct system_device_crosststamp *xtstamp)
+{
+ struct viortc_ptp_clock *vio_ptp =
+ container_of(ptp, struct viortc_ptp_clock, ptp_info);
+ int ret;
+ struct system_time_snapshot history_begin;
+ struct viortc_ptp_cross_ctx ctx;
+
+ ktime_get_snapshot(&history_begin);
+
+ /*
+ * Getting the timestamp can take many milliseconds with a slow Virtio
+ * device. This is too long for viortc_ptp_get_time_fn() passed to
+ * get_device_system_crosststamp(), which has to usually return before
+ * the timekeeper seqcount increases (every tick or so).
+ *
+ * So, get the actual cross-timestamp first.
+ */
+ ret = viortc_ptp_do_xtstamp(vio_ptp, &ctx);
+ if (ret)
+ return ret;
+
+ ret = get_device_system_crosststamp(viortc_ptp_get_time_fn, &ctx,
+ &history_begin, xtstamp);
+ if (ret) {
+ pr_debug("%s: get_device_system_crosststamp() returned %d\n",
+ __func__, ret);
+ }
+
+ return ret;
+}
+
+/** viortc_ptp_adjfine() - unsupported PTP clock adjfine op */
+static int viortc_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm)
+{
+ return -EOPNOTSUPP;
+}
+
+/** viortc_ptp_adjtime() - unsupported PTP clock adjtime op */
+static int viortc_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
+{
+ return -EOPNOTSUPP;
+}
+
+/** viortc_ptp_settime64() - unsupported PTP clock settime64 op */
+static int viortc_ptp_settime64(struct ptp_clock_info *ptp,
+ const struct timespec64 *ts)
+{
+ return -EOPNOTSUPP;
+}
+
+/**
+ * viortc_ptp_gettimex64() - PTP clock gettimex64 op
+ *
+ * Context: Process context.
+ */
+static int viortc_ptp_gettimex64(struct ptp_clock_info *ptp,
+ struct timespec64 *ts,
+ struct ptp_system_timestamp *sts)
+{
+ struct viortc_ptp_clock *vio_ptp =
+ container_of(ptp, struct viortc_ptp_clock, ptp_info);
+ u64 ns;
+ int ret;
+
+ ptp_read_system_prets(sts);
+ ret = viortc_read(vio_ptp->viortc, vio_ptp->vio_clk_id, &ns);
+ ptp_read_system_postts(sts);
+
+ if (ret)
+ return ret;
+
+ if (ns > (u64)S64_MAX)
+ return -EINVAL;
+
+ *ts = ns_to_timespec64((s64)ns);
+
+ return 0;
+}
+
+/** viortc_ptp_enable() - unsupported PTP clock enable op */
+static int viortc_ptp_enable(struct ptp_clock_info *ptp,
+ struct ptp_clock_request *rq, int on)
+{
+ return -EOPNOTSUPP;
+}
+
+/**
+ * viortc_ptp_info_template - ptp_clock_info template
+ *
+ * The .name member will be set for individual virtio_rtc PTP clocks.
+ */
+static const struct ptp_clock_info viortc_ptp_info_template = {
+ .owner = THIS_MODULE,
+ /* .name is set according to clock type */
+ .adjfine = viortc_ptp_adjfine,
+ .adjtime = viortc_ptp_adjtime,
+ .gettimex64 = viortc_ptp_gettimex64,
+ .settime64 = viortc_ptp_settime64,
+ .enable = viortc_ptp_enable,
+ .getcrosststamp = viortc_ptp_getcrosststamp,
+};
+
+/**
+ * viortc_ptp_unregister() - PTP clock unregistering wrapper
+ * @vio_ptp: virtio_rtc PTP clock
+ * @parent_dev: parent device of PTP clock
+ *
+ * Return: Zero on success, negative error code otherwise.
+ */
+int viortc_ptp_unregister(struct viortc_ptp_clock *vio_ptp,
+ struct device *parent_dev)
+{
+ int ret = ptp_clock_unregister(vio_ptp->ptp_clock);
+
+ if (!ret)
+ devm_kfree(parent_dev, vio_ptp);
+
+ return ret;
+}
+
+/**
+ * viortc_ptp_get_cross_cap() - get xtstamp support info from device
+ * @viortc: virtio_rtc device data
+ * @vio_ptp: virtio_rtc PTP clock abstraction
+ *
+ * Context: Process context.
+ * Return: Zero on success, negative error code otherwise.
+ */
+static int viortc_ptp_get_cross_cap(struct viortc_dev *viortc,
+ struct viortc_ptp_clock *vio_ptp)
+{
+ int ret;
+ const u16 *hw_counters_driver;
+ u32 num_hw_counters_driver;
+ u32 i;
+ u32 num_hw_counters = 0;
+
+ ret = viortc_hw_get_counters(&hw_counters_driver,
+ &num_hw_counters_driver);
+ if (ret)
+ return ret;
+
+ if (num_hw_counters_driver > VIORTC_CAP_HW_COUNTERS) {
+ pr_err("%s: HW counter capacity exceeded\n", __func__);
+ return -ENOMEM;
+ }
+
+ for (i = 0; i < num_hw_counters_driver; i++) {
+ u16 hw_counter = hw_counters_driver[i];
+ bool xtstamp_supported;
+
+ ret = viortc_cross_cap(viortc, vio_ptp->vio_clk_id, hw_counter,
+ &xtstamp_supported);
+ if (ret)
+ return ret;
+
+ if (xtstamp_supported)
+ vio_ptp->hw_counters[num_hw_counters++] = hw_counter;
+ }
+
+ vio_ptp->num_hw_counters = num_hw_counters;
+
+ return 0;
+}
+
+/**
+ * viortc_ptp_register() - prepare and register PTP clock
+ * @viortc: virtio_rtc device data
+ * @parent_dev: parent device for PTP clock
+ * @vio_clk_id: id of virtio_rtc clock which backs PTP clock
+ * @ptp_clock_name: PTP clock name
+ * @try_enable_xtstamp: enable xtstamp op, if available
+ *
+ * Context: Process context.
+ * Return: Pointer on success, ERR_PTR() otherwise; NULL if PTP clock support
+ * not available.
+ */
+struct viortc_ptp_clock *viortc_ptp_register(struct viortc_dev *viortc,
+ struct device *parent_dev,
+ u64 vio_clk_id,
+ const char *ptp_clock_name,
+ bool try_enable_xtstamp)
+{
+ struct viortc_ptp_clock *vio_ptp;
+ struct ptp_clock *ptp_clock;
+ ssize_t len;
+ int ret;
+
+ vio_ptp = devm_kzalloc(parent_dev, sizeof(*vio_ptp), GFP_KERNEL);
+ if (!vio_ptp)
+ return ERR_PTR(-ENOMEM);
+
+ vio_ptp->viortc = viortc;
+ vio_ptp->vio_clk_id = vio_clk_id;
+ vio_ptp->ptp_info = viortc_ptp_info_template;
+ len = strscpy(vio_ptp->ptp_info.name, ptp_clock_name,
+ sizeof(vio_ptp->ptp_info.name));
+ if (len < 0) {
+ ret = len;
+ goto err_free_dev;
+ }
+
+ if (try_enable_xtstamp) {
+ ret = viortc_ptp_get_cross_cap(viortc, vio_ptp);
+ if (ret)
+ goto err_free_dev;
+ }
+
+ ptp_clock = ptp_clock_register(&vio_ptp->ptp_info, parent_dev);
+ if (IS_ERR(ptp_clock))
+ goto err_on_register;
+
+ vio_ptp->ptp_clock = ptp_clock;
+
+ return vio_ptp;
+
+err_on_register:
+ ret = PTR_ERR(ptp_clock);
+
+err_free_dev:
+ devm_kfree(parent_dev, vio_ptp);
+ return ERR_PTR(ret);
+}
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread* [RFC PATCH 7/7] virtio_rtc: Add Arm Generic Timer cross-timestamping
2023-06-30 17:10 [RFC PATCH 0/7] Add virtio_rtc module and related changes Peter Hilber
` (5 preceding siblings ...)
2023-06-30 17:10 ` [RFC PATCH 6/7] virtio_rtc: Add PTP clocks Peter Hilber
@ 2023-06-30 17:10 ` Peter Hilber
6 siblings, 0 replies; 18+ messages in thread
From: Peter Hilber @ 2023-06-30 17:10 UTC (permalink / raw)
To: virtualization, virtio-dev
Cc: Peter Hilber, linux-kernel, Michael S. Tsirkin, Jason Wang,
Xuan Zhuo
Add PTP_SYS_OFFSET_PRECISE2 support on platforms using the Arm Generic
Timer, by forwarding the clocksource information from arm_arch_timer.
Support only the CP15 counter interfaces, since the memory-mapped
interfaces are not supported by the Virtio RTC draft spec [1].
[1] https://lists.oasis-open.org/archives/virtio-comment/202306/msg00592.html
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
---
drivers/virtio/Kconfig | 13 ++++++++++
drivers/virtio/Makefile | 1 +
drivers/virtio/virtio_rtc_arm.c | 44 +++++++++++++++++++++++++++++++++
3 files changed, 58 insertions(+)
create mode 100644 drivers/virtio/virtio_rtc_arm.c
diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 7369ecd7dd01..ed3f541032a0 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -203,4 +203,17 @@ config VIRTIO_RTC_PTP
If unsure, say Y.
+config VIRTIO_RTC_ARM
+ bool "Virtio RTC cross-timestamping using Arm Generic Timer"
+ default y
+ depends on VIRTIO_RTC_PTP && ARM_ARCH_TIMER
+ help
+ This enables Virtio RTC cross-timestamping using the Arm Generic Timer.
+ It only has an effect if the Virtio RTC device also supports this. The
+ cross-timestamp is available through the PTP clock driver precise
+ cross-timestamp ioctl (PTP_SYS_OFFSET_PRECISE2 or
+ PTP_SYS_OFFSET_PRECISE).
+
+ If unsure, say Y.
+
endif # VIRTIO_MENU
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 4d48cbcae6bb..781dff9f8822 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -15,3 +15,4 @@ obj-$(CONFIG_VIRTIO_DMA_SHARED_BUFFER) += virtio_dma_buf.o
obj-$(CONFIG_VIRTIO_RTC) += virtio_rtc.o
virtio_rtc-y := virtio_rtc_driver.o
virtio_rtc-$(CONFIG_VIRTIO_RTC_PTP) += virtio_rtc_ptp.o
+virtio_rtc-$(CONFIG_VIRTIO_RTC_ARM) += virtio_rtc_arm.o
diff --git a/drivers/virtio/virtio_rtc_arm.c b/drivers/virtio/virtio_rtc_arm.c
new file mode 100644
index 000000000000..2367f054081c
--- /dev/null
+++ b/drivers/virtio/virtio_rtc_arm.c
@@ -0,0 +1,44 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Provides cross-timestamp params for Arm.
+ *
+ * Copyright (C) 2022-2023 OpenSynergy GmbH
+ */
+
+#include <clocksource/arm_arch_timer.h>
+#include <linux/err.h>
+
+#include <uapi/linux/virtio_rtc.h>
+
+#include "virtio_rtc_internal.h"
+
+static const u16 viortc_hw_counters[] = { VIRTIO_RTC_COUNTER_ARM_VIRT,
+ VIRTIO_RTC_COUNTER_ARM_PHYS };
+
+/* see header for doc */
+int viortc_hw_get_counters(const u16 **hw_counters, int *num_hw_counters)
+{
+ *hw_counters = viortc_hw_counters;
+ *num_hw_counters = ARRAY_SIZE(viortc_hw_counters);
+
+ return 0;
+}
+
+/* see header for doc */
+int viortc_hw_xtstamp_params(u16 *hw_counter, struct clocksource **cs)
+{
+ *cs = arch_timer_get_cs();
+
+ switch (arch_timer_counter_get_type()) {
+ case ARCH_COUNTER_CP15_VIRT:
+ *hw_counter = VIRTIO_RTC_COUNTER_ARM_VIRT;
+ break;
+ case ARCH_COUNTER_CP15_PHYS:
+ *hw_counter = VIRTIO_RTC_COUNTER_ARM_PHYS;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
--
2.39.2
^ permalink raw reply related [flat|nested] 18+ messages in thread