diff mbox series

[net-next,v6,07/16] net_tstamp: Add TIMESTAMPING SOFTWARE and HARDWARE mask

Message ID 20231019-feature_ptp_netnext-v6-7-71affc27b0e5@bootlin.com
State New
Headers
Series net: Make timestamping selectable |

Commit Message

Köry Maincent Oct. 19, 2023, 2:29 p.m. UTC 
  Timestamping software or hardware flags are often used as a group,
therefore adding these masks will easier future use.

I did not use SOF_TIMESTAMPING_SYS_HARDWARE flag as it is deprecated and
not use at all.

Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
---
 include/uapi/linux/net_tstamp.h | 8 ++++++++
 1 file changed, 8 insertions(+)

Comments

Willem de Bruijn Oct. 19, 2023, 2:48 p.m. UTC | #1 
On Thu, Oct 19, 2023 at 10:29 AM Kory Maincent
<kory.maincent@bootlin.com> wrote:
>
> Timestamping software or hardware flags are often used as a group,
> therefore adding these masks will easier future use.

This assumes that device support for timestamping is often symmetric:
a device supports both rx and tx, or neither.

All devices support software receive timestamping, as that timestamp
is taken in the core network stack. But to support transmit timestamps
drivers have to call sbk_tstamp_tx in their ndo_start_xmit.

For hardware timestamping it may be more common to support both or
neither, not sure.

> I did not use SOF_TIMESTAMPING_SYS_HARDWARE flag as it is deprecated and
> not use at all.
>
> Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
> ---
>  include/uapi/linux/net_tstamp.h | 8 ++++++++
>  1 file changed, 8 insertions(+)
>
> diff --git a/include/uapi/linux/net_tstamp.h b/include/uapi/linux/net_tstamp.h
> index a2c66b3d7f0f..df8091998c8d 100644
> --- a/include/uapi/linux/net_tstamp.h
> +++ b/include/uapi/linux/net_tstamp.h
> @@ -48,6 +48,14 @@ enum {
>                                          SOF_TIMESTAMPING_TX_SCHED | \
>                                          SOF_TIMESTAMPING_TX_ACK)
>
> +#define SOF_TIMESTAMPING_SOFTWARE_MASK (SOF_TIMESTAMPING_RX_SOFTWARE | \
> +                                        SOF_TIMESTAMPING_TX_SOFTWARE | \
> +                                        SOF_TIMESTAMPING_SOFTWARE)
> +
> +#define SOF_TIMESTAMPING_HARDWARE_MASK (SOF_TIMESTAMPING_RX_HARDWARE | \
> +                                        SOF_TIMESTAMPING_TX_HARDWARE | \
> +                                        SOF_TIMESTAMPING_RAW_HARDWARE)
> +
>  /**
>   * struct so_timestamping - SO_TIMESTAMPING parameter
>   *
>
> --
> 2.25.1
>
Köry Maincent Oct. 24, 2023, 1:40 p.m. UTC | #2 
On Thu, 19 Oct 2023 10:48:04 -0400
Willem de Bruijn <willemdebruijn.kernel@gmail.com> wrote:

> On Thu, Oct 19, 2023 at 10:29 AM Kory Maincent
> <kory.maincent@bootlin.com> wrote:
> >
> > Timestamping software or hardware flags are often used as a group,
> > therefore adding these masks will easier future use.  
> 
> This assumes that device support for timestamping is often symmetric:
> a device supports both rx and tx, or neither.
>
> All devices support software receive timestamping, as that timestamp
> is taken in the core network stack. But to support transmit timestamps
> drivers have to call sbk_tstamp_tx in their ndo_start_xmit.

Yes, and in that software only case they often call ethtool_op_get_ts_info to
fill the timestamp info.

There is several drivers that support hardware and software timestamp, these
mask could be useful for these. In case of asymmetric support we can still use
the SOF_TIMESTAMPING_RX/TX_SOFTWARE flags.

I forgot to specify, in the commit message but this patch is mainly to ease
the next patch of this series to deal with software/hardware time stamping.
Maybe you prefer to have this squash into next patch as had suggested Florian
in last version.

Köry
Willem de Bruijn Oct. 24, 2023, 2:21 p.m. UTC | #3 
On Tue, Oct 24, 2023 at 9:40 AM Köry Maincent <kory.maincent@bootlin.com> wrote:
>
> On Thu, 19 Oct 2023 10:48:04 -0400
> Willem de Bruijn <willemdebruijn.kernel@gmail.com> wrote:
>
> > On Thu, Oct 19, 2023 at 10:29 AM Kory Maincent
> > <kory.maincent@bootlin.com> wrote:
> > >
> > > Timestamping software or hardware flags are often used as a group,
> > > therefore adding these masks will easier future use.
> >
> > This assumes that device support for timestamping is often symmetric:
> > a device supports both rx and tx, or neither.
> >
> > All devices support software receive timestamping, as that timestamp
> > is taken in the core network stack. But to support transmit timestamps
> > drivers have to call sbk_tstamp_tx in their ndo_start_xmit.
>
> Yes, and in that software only case they often call ethtool_op_get_ts_info to
> fill the timestamp info.
>
> There is several drivers that support hardware and software timestamp, these
> mask could be useful for these. In case of asymmetric support we can still use
> the SOF_TIMESTAMPING_RX/TX_SOFTWARE flags.
>
> I forgot to specify, in the commit message but this patch is mainly to ease
> the next patch of this series to deal with software/hardware time stamping.
> Maybe you prefer to have this squash into next patch as had suggested Florian
> in last version.

Thanks. I did not see such use in net/ or in my experience of using
setsockopt SO_TIMESTAMPING.

But I had not considered net device get_ts_info. It would be useful there.

Slight preference from me to keep this a separate patch, as it applies
not only to the use case in your follow-on patch.
diff mbox series

Patch

diff --git a/include/uapi/linux/net_tstamp.h b/include/uapi/linux/net_tstamp.h
index a2c66b3d7f0f..df8091998c8d 100644
--- a/include/uapi/linux/net_tstamp.h
+++ b/include/uapi/linux/net_tstamp.h
@@ -48,6 +48,14 @@  enum {
 					 SOF_TIMESTAMPING_TX_SCHED | \
 					 SOF_TIMESTAMPING_TX_ACK)
 
+#define SOF_TIMESTAMPING_SOFTWARE_MASK	(SOF_TIMESTAMPING_RX_SOFTWARE | \
+					 SOF_TIMESTAMPING_TX_SOFTWARE | \
+					 SOF_TIMESTAMPING_SOFTWARE)
+
+#define SOF_TIMESTAMPING_HARDWARE_MASK	(SOF_TIMESTAMPING_RX_HARDWARE | \
+					 SOF_TIMESTAMPING_TX_HARDWARE | \
+					 SOF_TIMESTAMPING_RAW_HARDWARE)
+
 /**
  * struct so_timestamping - SO_TIMESTAMPING parameter
  *