[RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[Shmuel Hazan]

From bd7fb292219e6b64a650a6c79d09822dbef662d3 Mon Sep 17 00:00:00 2001

In some configurations, a single TAI can be
responsible for multiple mvpp2 interfaces.
However, the mvpp2 driver will call mvpp22_tai_stop
and mvpp22_tai_start per interface RX timestamp
disable/enable.

As a result, disabling timestamping for one
interface would stop the worker and corrupt
the other interface's RX timestamps.

This commit solves the issue by introducing a
simpler ref count for each TAI instance.

Fixes: ce3497e2072e ("net: mvpp2: ptp: add support for receive
timestamping")
Signed-off-by: Shmuel Hazan <shmuel.h@siklu.com>
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
index 95862aff49f1..1b57573dd866 100644
--- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
+++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
@@ -61,6 +61,7 @@ struct mvpp2_tai {
 	u64 period;		// nanosecond period in 32.32 fixed
point
 	/* This timestamp is updated every two seconds */
 	struct timespec64 stamp;
+	u16 poll_worker_refcount;
 };
 
 static void mvpp2_tai_modify(void __iomem *reg, u32 mask, u32 set)
@@ -372,6 +373,10 @@ void mvpp22_tai_start(struct mvpp2_tai *tai)
 {
 	long delay;
 
+	tai->poll_worker_refcount++;
+	if (tai->poll_worker_refcount > 1)
+		return;
+
 	delay = mvpp22_tai_aux_work(&tai->caps);
 
 	ptp_schedule_worker(tai->ptp_clock, delay);
@@ -379,6 +384,9 @@ void mvpp22_tai_start(struct mvpp2_tai *tai)
 
 void mvpp22_tai_stop(struct mvpp2_tai *tai)
 {
+	tai->poll_worker_refcount--;
+	if (tai->poll_worker_refcount)
+		return;
 	ptp_cancel_worker_sync(tai->ptp_clock);
 }
 
-- 
2.40.0

Re: [RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[Andrew Lunn]

> diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> index 95862aff49f1..1b57573dd866 100644
> --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> @@ -61,6 +61,7 @@ struct mvpp2_tai {
>  	u64 period;		// nanosecond period in 32.32 fixed
> point
>  	/* This timestamp is updated every two seconds */
>  	struct timespec64 stamp;
> +	u16 poll_worker_refcount;

What lock is protecting this? It would be nice to comment in the
commit message why it is safe to use a simple u16.

       Andrew

Re: [RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[Shmuel Hazan]

On Sun, 2023-04-16 at 16:52 +0200, Andrew Lunn wrote:
> 
> > diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > index 95862aff49f1..1b57573dd866 100644
> > --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > @@ -61,6 +61,7 @@ struct mvpp2_tai {
> >       u64 period;             // nanosecond period in 32.32 fixed
> > point
> >       /* This timestamp is updated every two seconds */
> >       struct timespec64 stamp;
> > +     u16 poll_worker_refcount;
> 
> What lock is protecting this? It would be nice to comment in the
> commit message why it is safe to use a simple u16.

Hi Andrew, 

thanks for your response. In theory, the only code path
to these functions (mvpp22_tai_start and mvpp22_tai_stop)
is ioctl (mvpp2_ioctl -> mvpp2_set_ts_config) which should lock
rtnl. However, 
It would probably be a good idea to also lock mvpp2_tai->lock too.

As for the integer size, it will be increased once per ethernet device
on that CP, which currently has only maximum of 3 ethernet ports. I
gave it u16 just to be futureproof. However, I can change it to a more
appropriate type if it looks more appropriate to you. 

> 
>        Andrew

Re: [RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[Andrew Lunn]

On Sun, Apr 16, 2023 at 03:25:55PM +0000, Shmuel Hazan wrote:
> On Sun, 2023-04-16 at 16:52 +0200, Andrew Lunn wrote:
> > 
> > > diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > index 95862aff49f1..1b57573dd866 100644
> > > --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > @@ -61,6 +61,7 @@ struct mvpp2_tai {
> > >       u64 period;             // nanosecond period in 32.32 fixed
> > > point
> > >       /* This timestamp is updated every two seconds */
> > >       struct timespec64 stamp;
> > > +     u16 poll_worker_refcount;
> > 
> > What lock is protecting this? It would be nice to comment in the
> > commit message why it is safe to use a simple u16.
> 
> Hi Andrew, 
> 
> thanks for your response. In theory, the only code path
> to these functions (mvpp22_tai_start and mvpp22_tai_stop)
> is ioctl (mvpp2_ioctl -> mvpp2_set_ts_config) which should lock
> rtnl. However, 
> It would probably be a good idea to also lock mvpp2_tai->lock too.

I cannot comment on what locks should be used, i don't know the code.

Which is why as a reviewer, i just want some indication you have
thought about locking, and you think it is safe, given that there are
not obvious locks in the code.

	Andrew

Re: [RFC PATCH] net: mvpp2: tai: add refcount for ptp worker

[Russell King (Oracle)]

On Sun, Apr 16, 2023 at 05:40:12PM +0200, Andrew Lunn wrote:
> On Sun, Apr 16, 2023 at 03:25:55PM +0000, Shmuel Hazan wrote:
> > On Sun, 2023-04-16 at 16:52 +0200, Andrew Lunn wrote:
> > > 
> > > > diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > > b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > > index 95862aff49f1..1b57573dd866 100644
> > > > --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > > +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_tai.c
> > > > @@ -61,6 +61,7 @@ struct mvpp2_tai {
> > > >       u64 period;             // nanosecond period in 32.32 fixed
> > > > point
> > > >       /* This timestamp is updated every two seconds */
> > > >       struct timespec64 stamp;
> > > > +     u16 poll_worker_refcount;
> > > 
> > > What lock is protecting this? It would be nice to comment in the
> > > commit message why it is safe to use a simple u16.
> > 
> > Hi Andrew, 
> > 
> > thanks for your response. In theory, the only code path
> > to these functions (mvpp22_tai_start and mvpp22_tai_stop)
> > is ioctl (mvpp2_ioctl -> mvpp2_set_ts_config) which should lock
> > rtnl. However, 
> > It would probably be a good idea to also lock mvpp2_tai->lock too.
> 
> I cannot comment on what locks should be used, i don't know the code.
> 
> Which is why as a reviewer, i just want some indication you have
> thought about locking, and you think it is safe, given that there are
> not obvious locks in the code.

... and probably is a good idea to place a comment in the code stating
what the locking for it is.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!