OpenWrt Forum Archive

Topic: transport problem:switch doesn't work,PHY-ID read incorrect

The content of this topic has been archived on 26 Apr 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

I have problem with openwrt transport, the target platfrom is ar9344+ar8327.
I'm trying bring up the switch ar8327,while it doesn't work:
    mdio driver  scan the mdio-bus to find the device, and it find 2 device, phy-id is 0x12800000,0x00000760. but my ar8327 driver indicate the phy-id should be 0x004d****, as a result, ar8327 driver can't bind to device ar8327, and it doesn't work.
  you see, the main problem is phy-id read incorrect, but the mido config is default from bootloader(it is work on the original firmware), it should be right.
  i have no idea with this problem,  can you give me some hint?
  thanks very much!

read this patch
DB120 have ar9344+ar8327.

Welcome to the crazy world of the WD My Net N750 (I guess that's the device you're working on).
By now I pretty much know the AR9344 and AR8327N registers by heart, scanned the bootloader code and reversed the original network driver.

I have yet to find where they mess up the ethernet so badly.

Heffer wrote:

Welcome to the crazy world of the WD My Net N750 (I guess that's the device you're working on).
By now I pretty much know the AR9344 and AR8327N registers by heart, scanned the bootloader code and reversed the original network driver.

I have yet to find where they mess up the ethernet so badly.

really? I build the db120 version, but it still can't work: buildin switch can be detected(phy id read as 0x4d****), while the ar8327 can't be detected(phy id read as 0x1280000 or 0x760).
what's wrong with it?

If we're both talking about the WD My Net N750 I can't exactly tell you what's wrong. My current guess is that the MDIO bus is too fast for the switch so it doesn't correctly communicate.

Please read this thread for all the information we currently have on the problem: https://lists.openwrt.org/pipermail/ope … 22002.html

Heffer wrote:

If we're both talking about the WD My Net N750 I can't exactly tell you what's wrong. My current guess is that the MDIO bus is too fast for the switch so it doesn't correctly communicate.

Please read this thread for all the information we currently have on the problem: https://lists.openwrt.org/pipermail/ope … 22002.html

thank you, I finally solved it.
bootloader powerdown all the phy port, and phy CANNOT be READ in the powerdown mode (while it CAN be written).
I just poweron all phy port, and it works.

How did you solve it then? By going into the bootloader and then issuing boot or did you solve it from within OpenWRT?
Can you post a patch?

Heffer wrote:

How did you solve it then? By going into the bootloader and then issuing boot or did you solve it from within OpenWRT?
Can you post a patch?

I'm using a private bootloader, and I didn't change the bootloader.
I just poweron all phy port before scan mii bus, it's in the openwrt.
um... infact i'm new to linux, i'm not familiar with "post patch", and time is late(i'm in china), I will try to post patch tomorrow.

Thanks for the tip! I got it working too.

Now I'm working on preparing integration of this device into trunk.

The discussion might have continued from here.