> -----Original Message----- > From: André Sadie [mailto:andsa...@gmail.com] > Sent: Tuesday, August 07, 2018 5:58 AM > To: linuxptp-users@lists.sourceforge.net > Subject: [Linuxptp-users] High Propagation Time > > Hello All, > I have been trying to use ptp4l with a gPTP profile between multiple devices > and > an Extreme x440-48t. It works fine with the devices containing I210 network > controllers but it fails on a device using an I219-v network controller. > > The propagation time between the I219-v and the Extreme is around 7.3μs and > therefore does not fall below the default gPTP profile propagation delay > threshold of 800ns. For reference, the propagation time of the I210 is in the > range > of 20ns. > > > If I increase the propagation delay threshold to 10μs on the I219-v and the > Extreme, the device is set AS Capable and syncing happens with a very > inconsistent accuracy of rms 50-150ns and spikes to ~2μs on occasion. Whereas > the I210 has a stable accuracy in the range of 5-10ns. There is no other > traffic on > the network. > > > To me, the propagation delay seems much higher than it should be on the I219-v > since it is directly connected to the switch with a very short cable. I think > the > problem is either with the hardware or drivers. >
It is different hardware with a different design. From what I recall, the i210 had a better design for handling frequency tuning. > > I updated the e1000e drivers for the I219-v to version 3.4.1.1-NAPI with much > difficulty since it seems to require an older kernel than I had and I had to > make > changes to the makefile to get it working. I am now using 4.13.0-45-generic > kernel. e1000e also requires an e1000 version newer than 7.6.15.5 which I was > not able to upgrade to 8.0.35, again due to issues with the makefile and > requiring > deprecated kernel headers. > Just fyi, even if the version number in the e1000e upstream driver is old, it's actually likely contains more fixes/newer work than what's on sourceforge. The version numbers in the upstream driver are manual, so they don't always get updated in a timely fashion. Especially in regards to PTP functionality, I'd be surprised if your upstream driver was older than the out-of-tree code. Thanks, Jake > > Admittedly this is most likely a driver issue but I would rather not be > downgrading > or fiddling with kernel stuff unless I have to. You shouldn't need to. > > So I am just wondering if anyone has experienced similar issues with the > I219-v or > the e1000/e1000e drivers? I've got some experience, though I wasn't the driver author for either i210 or the e1000e driver, so I don't have that much knowledge. From what I recall, the i219-v device has a different register design for doing the frequency tuning. It's possible that tuning is somewhat buggy or less efficient than what's available in the i210. It would be helpful to see what the frequency adjustments are being made to the i219 device.. > > Is the 7μs really as unreasonable as I think? > Any recommendations on how to proceed? > Is increasing the propagation delay threshold a suitable fix? > > Thank you, > Andre. > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
