FS#27 - BT Home Hub 5 Type A - Switch problem when enabling WAN port about openwrt HOT 8 CLOSED

jow-:

Please try to re-acquire DHCP on your client after the device fully booted and see if it works then.

from openwrt.

bill888:

I'm afraid forcing PC to release and re-acquire DHCP IP address does not resolve issue.

I have also configured a static IP address and DNS setting on client PC. Here are ping results shortly after the Hub has been switched on:

C:\Users\admin>ping 8.8.8.8 -t

Pinging 8.8.8.8 with 32 bytes of data:
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=9ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=8ms TTL=53
Reply from 8.8.8.8: bytes=32 time=3616ms TTL=53
Reply from 8.8.8.8: bytes=32 time=1517ms TTL=53
Reply from 8.8.8.8: bytes=32 time=641ms TTL=53
Request timed out.
Reply from 8.8.8.8: bytes=32 time=1114ms TTL=53

When the above pings start to slow and fail from client PC connected to LAN port of Hub, I am still seeing these 'tx ring full' messages on serial console:

[ 95.372341] eth1: transmit timed out, disable the dma channel irq
[ 95.381099] eth1: tx ring full
[ 95.381134] eth0: tx ring full
[ 105.953032] eth1: transmit timed out, disable the dma channel irq
[ 105.957949] eth1: tx ring full
[ 124.961063] eth0: transmit timed out, disable the dma channel irq
[ 124.965868] eth0: tx ring full
[ 162.758301] eth0: tx ring full
[ 162.823880] eth1: tx ring full
[ 164.023015] eth1: tx ring full
[ 166.845388] eth0: port 4 lost link
[ 166.849247] br-lan: port 1(eth0.1) entered disabled state
[ 169.845683] eth1: port 5 lost link

As another test, I have just installed r48886 image from OpenWRT BT Home Hub 5 Type A wiki to a second Hub. Using the same network configuration file changes, internet through WAN port (DHCP client mode) works perfectly.

from openwrt.

bill888:

Further to my earlier posts, I have gone back and tried some other sysupgrade tar files.

openwrt r49377 is OK.
LEDE r508 is OK.
LEDE r619 has same issue as witnessed with R742.

I can only conclude something is 'broken' after r508 and before r619.

from openwrt.

bill888:

Commit cc3bfdb "lantiq: enable SMP for XRX200"

(https://git.lede-project.org/?p=source.git;a=commit;h=cc3bfdb62f941dff3e2983591c78b6d39ca8d88a)

is responsible for the TX ring full WAN port issue. I am advised the ticket needs to be assigned to 'nbd' for attention.

Tnx to mkresin for providing test images for me to help narrow down and identify the commit.

from openwrt.

nbd:

Commit reverted in 2270fc5

from openwrt.

nbd:

I will try to find the time to rework the ethernet driver soon and will write a comment in this ticket when I have something to test

from openwrt.

medber:

By the way, enabling SMP doesn't seem to cause any issues for ARV7519RW22 which is a XRX200 too. I've tested it extensively for days under tens of gigabytes of wan traffic daily.

from openwrt.

nbd:

I think the issue is probably specific to devices that use both eth0 and eth1

from openwrt.