From: Jose Abreu <joabreu@xxxxxxxxxxxx> Date: Jul/22/2019, 10:47:44 (UTC+00:00) > From: Jon Hunter <jonathanh@xxxxxxxxxx> > Date: Jul/22/2019, 10:37:18 (UTC+00:00) > > > > > On 22/07/2019 08:23, Jose Abreu wrote: > > > From: Jon Hunter <jonathanh@xxxxxxxxxx> > > > Date: Jul/19/2019, 14:35:52 (UTC+00:00) > > > > > >> > > >> On 19/07/2019 13:32, Jose Abreu wrote: > > >>> From: Jon Hunter <jonathanh@xxxxxxxxxx> > > >>> Date: Jul/19/2019, 13:30:10 (UTC+00:00) > > >>> > > >>>> I booted the board without using NFS and then started used dhclient to > > >>>> bring up the network interface and it appears to be working fine. I can > > >>>> even mount the NFS share fine. So it does appear to be particular to > > >>>> using NFS to mount the rootfs. > > >>> > > >>> Damn. Can you send me your .config ? > > >> > > >> Yes no problem. Attached. > > > > > > Can you compile your image without modules (i.e. all built-in) and let > > > me know if the error still happens ? > > > > I simply removed the /lib/modules directory from the NFS share and > > verified that I still see the same issue. So it is not loading the > > modules that is a problem. > > Well, I meant that loading modules can be an issue but that's not the > way to verify that. > > You need to have all modules built-in so that it proves that no module > will try to be loaded. > > Anyway, this is probably not the cause as you wouldn't even be able to > compile kernel if you need a symbol from a module with stmmac built-in. > Kconfig would complain about that. > > The other cause could be data corruption in the RX path. Are you able to > send me packet dump by running wireshark either in the transmitter side > (i.e. NFS server), or using some kind of switch ? > > --- > Thanks, > Jose Miguel Abreu Also, please add attached patch. You'll get a compiler warning, just disregard it. --- Thanks, Jose Miguel Abreu
Attachment:
0001-net-stmmac-Debug-print.patch
Description: 0001-net-stmmac-Debug-print.patch
