Hello everyone,I've been playing around with a Utilite Pro recently. It features a Freescale imx6q with SoC FEC as eth0 and a second ether Intel I211 interface on PCIE being mapped to eth1 amongst various other bits.
ATM, it's running debian jessie and kernel 3.16.1 with the recent three patch sets from Russell King at http://ftp.arm.linux.org.uk/cgit/linux-arm.git/log/?h=fec-testing; from the tag imx-drm-fixes-urgent to fec-testing, minus the "clear IPV6 header CS".
During normal operation, I have had the occasional FEC Watchdog Timeout (FWT). However, when eth0 und eth1 are bonded to balance-rr mode and two independent iperfs cliented by different machines with each 4 threads run free on the Utilite, it inevitably comes to FWTs. Raising packets_per_slave to four seems to help somewhat.
The bond is routed through two aggregated (trunk) ports on a Netgear JGS524E running to the iperf clients.
Drop me a line, if you need more information. TIA, John[1] Patch set A: http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/270173.html [2] Patch set B: http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/270373.html [3] Patch set C: http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/270393.html [4] Linux bonding documentation https://www.kernel.org/doc/Documentation/networking/bonding.txt [5] Utilite Pro product page http://utilite-computer.com/web/utilite-pro-specifications [6] Utilite Pro bonding configuration http://www.utilite-co mputer.com/forum/viewtopic.php?f=73&t=1980&sid=31eb3a1273da28916d665112f8af2350
Attachment:
fec-watchdog.log.gz
Description: application/gzip
Attachment:
config.gz
Description: application/gzip
Attachment:
imx6q-cm-fx6.dtb
Description: Binary data
