Re: [PATCH 1/5] net: timestamp: add helper returning skb's tx tstamp

[Willem de Bruijn <willemdebruijn.kernel@gmail.com>]

Pavel Begunkov wrote:
> On 6/1/25 14:52, Willem de Bruijn wrote:
> > Pavel Begunkov wrote:
> >> Add a helper function skb_get_tx_timestamp() that returns a tx timestamp
> >> associated with an skb from an queue queue.
> > 
> > Just curious: why a timestamp specific operation, rather than a
> > general error queue report?
> 
> Timestamps still need custom code, not like we can do a generic
> implementation just by copying sock_extended_err to user. And then
> it'll be a problem to fit it into completions, it's already tight
> after placing the timeval directly into cqe, there are only
> few bits left.

Ok understood.
 
> Either way, I guess it can be extended if there are more use cases,
> or might be better introducing and new command to cover that and
> share some of the handling.

Not a request from me, to be clear. Just wanted to understand the
design choice.

> ...>> diff --git a/net/socket.c b/net/socket.c
> >> index 9a0e720f0859..d1dc8ab28e46 100644
> >> --- a/net/socket.c
> >> +++ b/net/socket.c
> >> @@ -843,6 +843,55 @@ static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb,
> >>   		 sizeof(ts_pktinfo), &ts_pktinfo);
> >>   }
> >>   
> >> +bool skb_has_tx_timestamp(struct sk_buff *skb, struct sock *sk)
> > 
> > Here and elsewhere: consider const pointers where possible
> 
> will do
> 
> > 
> >> +{
> >> +	u32 tsflags = READ_ONCE(sk->sk_tsflags);
> >> +	struct sock_exterr_skb *serr = SKB_EXT_ERR(skb);
> >> +
> >> +	if (serr->ee.ee_errno != ENOMSG ||
> >> +	   serr->ee.ee_origin != SO_EE_ORIGIN_TIMESTAMPING)
> >> +		return false;
> >> +
> >> +	/* software time stamp available and wanted */
> >> +	if ((tsflags & SOF_TIMESTAMPING_SOFTWARE) && skb->tstamp)
> >> +		return true;
> >> +	/* hardware time stamps available and wanted */
> >> +	return (tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) &&
> >> +		skb_hwtstamps(skb)->hwtstamp;
> >> +}
> >> +
> >> +bool skb_get_tx_timestamp(struct sk_buff *skb, struct sock *sk,
> >> +			  struct timespec64 *ts)
> >> +{
> >> +	u32 tsflags = READ_ONCE(sk->sk_tsflags);
> >> +	bool false_tstamp = false;
> >> +	ktime_t hwtstamp;
> >> +	int if_index = 0;
> >> +
> >> +	if (sock_flag(sk, SOCK_RCVTSTAMP) && skb->tstamp == 0) {
> >> +		__net_timestamp(skb);
> >> +		false_tstamp = true;
> >> +	}
> > 
> > This is for SO_TIMESTAMP, not SO_TIMESTAMPING, and intended in the
> > receive path only, where net_enable_timestamp may be too late for
> > initial packets.
> 
> Got it, I'll drop that chunk if you think it's fine. Thanks
> for review
> 
> >> +	if ((tsflags & SOF_TIMESTAMPING_SOFTWARE) &&
> >> +	    ktime_to_timespec64_cond(skb->tstamp, ts))
> >> +		return true;
> >> +
> >> +	if (!(tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) ||
> >> +	    skb_is_swtx_tstamp(skb, false_tstamp))
> >> +		return false;
> >> +
> >> +	if (skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP_NETDEV)
> >> +		hwtstamp = get_timestamp(sk, skb, &if_index);
> >> +	else
> >> +		hwtstamp = skb_hwtstamps(skb)->hwtstamp;
> >> +
> >> +	if (tsflags & SOF_TIMESTAMPING_BIND_PHC)
> >> +		hwtstamp = ptp_convert_timestamp(&hwtstamp,
> >> +						READ_ONCE(sk->sk_bind_phc));
> >> +	return ktime_to_timespec64_cond(hwtstamp, ts);
> > 
> > This duplicates code in __sock_recv_timestamp. Perhaps worth a helper.
> 
> I couldn't find a good way for doing that. There are rx checks in
> every if, there is also pkt info handling nested. And
> scm_timestamping_internal has 3 timeouts , so
> __sock_recv_timestamp() would need to duplicate some checks to
> choose the right place for the timeout or so.

Ack, then let's leave as is. Thanks for taking a stab.