[Intel-wired-lan] [PATCH net-next v4 1/7] igb: clear out tstamp after sending the packet

Eric Dumazet eric.dumazet at gmail.com
Thu Jun 20 21:56:28 UTC 2019



On 6/20/19 1:32 PM, Patel, Vedang wrote:
> 

> Ahh.. that’s clearly a false statement. Skb->tstamp is cleared so
> that it is not interpreted as a software timestamp when trying to
> send the Hardware TX timestamp to the userspace. I will rephrase the
> commit message in the next version.
> 
> Some more details: The problem occurs when using the txtime-assist
> mode of taprio with packets which also request the hardware transmit
> timestamp (e.g. PTP packets). Whenever txtime-assist mode is set,
> taprio will assign a hardware transmit timestamp to all the packets
> (in skb->tstamp). PTP packets will also request the hardware transmit
> timestamp be sent to the userspace after packet is transmitted.
> 
> Whenever a new timestamp is detected by the driver (this work is done
> in igb_ptp_tx_work() which calls igb_ptp_tx_hwtstamps() in
> igb_ptp.c[1]), it will queue the timestamp in the ERR_QUEUE for the
> userspace to read. When the userspace is ready, it will issue a
> recvmsg() call to collect this timestamp. The problem is in this
> recvmsg() call. If the skb->tstamp is not cleared out, it will be
> interpreted as a software timestamp and the hardware tx timestamp
> will not be successfully sent to the userspace. Look at
> skb_is_swtx_tstamp() and the callee function __sock_recv_timestamp()
> in net/socket.c for more details.


That amount of details in the changelog would be really nice ;)




More information about the Intel-wired-lan mailing list