RE: [net-next 07/12] ixgbe: Enable timesync clock-out feature for PPS support on X540

-----Original Message-----
From: Richard Cochran [mailto:richardcochran@gmail.com]
Sent: Thursday, May 10, 2012 10:41 PM
To: Keller, Jacob E
Cc: Kirsher, Jeffrey T; davem@davemloft.net; netdev@vger.kernel.org;
gospo@redhat.com; sassmann@redhat.com
Subject: Re: [net-next 07/12] ixgbe: Enable timesync clock-out feature for PPS
support on X540

On Thu, May 10, 2012 at 10:08:44PM +0000, Keller, Jacob E wrote:

quoted

Oops stupid mail program sent that on accident. Anyways: I think you
might be right, Richard. We don't read those timestamp values unless
the stat err bit for timestamps is set on the descriptor. But I am not
sure what happens when the tjmestamped packet is dropped off the end
of the ring. What would you propose here? How can we detect if this
timestamp doesn't match the packet? I can look into using the extra
hardware features for matching timestamps. That might be a more useful, in

that it would help prevent this case.

[ Talking about the Rx time stamping locking from other patch... ]

The IGB provides some PTP event packet identification fields (seqNum,
etc) just for the purpose of matching time stamps to packets. Some of the
other PHC drivers (ixp4xx, dp83640) have code that does the matching.

Ixgbe has the sequence number also. I'll take a look at the PHC drivers
that match already, and see how difficult it would be to perform this check.

- Jake

HTH,
Richard