> -----Original Message----- > From: Richard Cochran <richardcoch...@gmail.com> > Sent: Wednesday, July 14, 2021 6:12 AM > To: Keller, Jacob E <jacob.e.kel...@intel.com> > Cc: Miroslav Lichvar <mlich...@redhat.com>; linuxptp- > de...@lists.sourceforge.net > Subject: Re: [Linuxptp-devel] [PATCH] Increase the default > tx_timestamp_timeout > to 5 > > On Wed, Jul 14, 2021 at 11:20:00AM +0000, Keller, Jacob E wrote: > > > I think for Tx the challenges are higher: the timestamp is taken > > after we've filled in the descriptor and sent the frame. The only > > place it could reasonably be stored again is the descriptor > > writeback (since we don't get completion messages). > > Right, the would be the place to do it. > > > If I remember correctly, the challenge here is that in a traditional > > ring model the writeback is completed much earlier than the > > timestamp so we potentially delay cleanup of other packets by > > waiting to insert the timestamp into the writeback. > > If *every* frame gets a time stamp, then their write-backs would all > be delayed by the same amount. Hence no clean up operations would be > "delayed". They would all take the same amount of time. > > The only cost would be in space to keep the data for the write-back > around until the time stamp becomes available. Paying the price of > the little extra memory is well worth it, as it simplifies the time > stamping logic and removes every class of problem related to time > stamp delivery. > > IOW, KISS! > > Thanks, > Richard
Yea, if you timestamp every frame regardless of whether kernel requested it or not. Makes sense to me. Thanks, Jake _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
