when ar8033 working at 1000M mode,lost lots of package,reach of 20%, even 30%

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

when ar8033 working at 1000M mode,lost lots of package,reach of 20%, even 30%

1,487 Views
sunshineyang
Contributor I

my board is mx6,the driver is ar8031,not ar8033,it is ok when ar8033 working at 10/100M.  When it connect to 1000M interface,it can Automatic negotiation 1000M/Full, but when i ping my host ,it will lost a lot of packages.I dont know why.

Wait for your reply!

Labels (1)
0 Kudos
3 Replies

862 Views
lantianyu
Contributor I

we meet the same problem,can anyone help us?

1.Device: Customized imx6q-sapresd(only few change)
CUP:IMX6Q(AC or AD)
PHY:AR8033(change from 8031 to 8033)
2.kernel:git clone https://github.com/Freescale/linux-fslc.git (branch:4.1-2.0.x-imx)
version:4.1.44


I use PC connection equipment directly,configure port mode for "1000M/FULL",and then send ping package to device,but some packages lossed(not all,about 70%)
I change mode for "100M/FULL" or "10M/Full",no packet loss;
There was no such problem in the kernel 3.0.35 version
Look forward to your reply,Thank you very much!

log:
1000M,there is some pkg lost
I have no name!@(none) /mnt/cfg$ ifconfig eth0 down
I have no name!@(none) /mnt/cfg$ ifconfig eth0 up
fec 2188000.ethernet eth0: Freescale FEC PHY driver [Generic PHY] (mii_bus:phy_addr=2188000.ethernet:07, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
I have no name!@(none) /mnt/cfg$ fec 2188000.ethernet eth0: Link is Up - 1Gbps/Full - flow control off
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

I have no name!@(none) /mnt/cfg$ ping 198.1.10.1
PING 198.1.10.1 (198.1.10.1): 56 data bytes
64 bytes from 198.1.10.1: seq=3 ttl=64 time=1.580 ms
64 bytes from 198.1.10.1: seq=9 ttl=64 time=1.712 ms
64 bytes from 198.1.10.1: seq=10 ttl=64 time=1.596 ms
64 bytes from 198.1.10.1: seq=13 ttl=64 time=2.258 ms
--- 198.1.10.1 ping statistics ---
15 packets transmitted, 4 packets received, 73% packet loss
round-trip min/avg/max = 1.580/1.786/2.258 ms
I have no name!@(none) /mnt/cfg$

100M,it is ok
I have no name!@(none) /mnt/cfg$ ifconfig eth0 down
I have no name!@(none) /mnt/cfg$ ifconfig eth0 up
fec 2188000.ethernet eth0: Freescale FEC PHY driver [Generic PHY] (mii_bus:phy_addr=2188000.ethernet:07, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
I have no name!@(none) /mnt/cfg$ fec 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

PING 198.1.10.1 (198.1.10.1): 56 data bytes
64 bytes from 198.1.10.1: seq=0 ttl=64 time=1.670 ms
64 bytes from 198.1.10.1: seq=1 ttl=64 time=1.506 ms
64 bytes from 198.1.10.1: seq=2 ttl=64 time=1.610 ms
64 bytes from 198.1.10.1: seq=3 ttl=64 time=1.585 ms
64 bytes from 198.1.10.1: seq=4 ttl=64 time=1.564 ms
64 bytes from 198.1.10.1: seq=5 ttl=64 time=1.676 ms
64 bytes from 198.1.10.1: seq=6 ttl=64 time=1.585 ms
64 bytes from 198.1.10.1: seq=7 ttl=64 time=1.520 ms
64 bytes from 198.1.10.1: seq=8 ttl=64 time=1.664 ms
--- 198.1.10.1 ping statistics ---
9 packets transmitted, 9 packets received, 0% packet loss
round-trip min/avg/max = 1.506/1.597/1.676 ms
I have no name!@(none) /mnt/cfg$ fec 2188000.ethernet eth0: Link is Down
fec 2188000.ethernet eth0: Link is Up - 1Gbps/Full - flow control off

0 Kudos

862 Views
foxmaze
Contributor II

We found the same issue, assuming you were running andriod 5.0 or above, please try on kernerl 3.0.35/andriod 4.3, it works.

Sounds that freescale developer didn't test too much before new sw release.

0 Kudos

862 Views
jimmychan
NXP TechSupport
NXP TechSupport

There is an errata about the 1Gb Ethernet limitation.

http://cache.nxp.com/files/32bit/doc/errata/IMX6DQCE.pdf?fpsp=1&WT_TYPE=Errata&WT_VENDOR=FREESCALE&W...

" ERR004512 ENET: 1 Gb Ethernet MAC (ENET) system limitation "

0 Kudos