zynq_z_turn_defconfig: broken networking

[Michal Simek]

On 08. 04. 20 9:12, Matwey V. Kornilov wrote:
> ср, 8 апр. 2020 г. в 09:56, Michal Simek <monstr at monstr.eu>:
>>
>> On 07. 04. 20 22:12, Joe Hershberger wrote:
>>> On Tue, Apr 7, 2020 at 1:34 PM Matwey V. Kornilov
>>> <matwey.kornilov at gmail.com> wrote:
>>>>
>>>> вт, 7 апр. 2020 г. в 12:18, Michal Simek <monstr at monstr.eu>:
>>>>>
>>>>> On 07. 04. 20 11:16, Matwey V. Kornilov wrote:
>>>>>> вт, 7 апр. 2020 г. в 12:01, Michal Simek <monstr at monstr.eu>:
>>>>>>>
>>>>>>> On 07. 04. 20 9:58, Matwey V. Kornilov wrote:
>>>>>>>> пн, 6 апр. 2020 г. в 12:41, Michal Simek <monstr at monstr.eu>:
>>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> On 06. 04. 20 10:14, Matwey V. Kornilov wrote:
>>>>>>>>>> 06.04.2020 11:12, Matwey V. Kornilov пишет:
>>>>>>>>>>> Hello,
>>>>>>>>>>>
>>>>>>>>>>> I am running u-boot 2020.04-rc4-00100-g74bf17db39 with
>>>>>>>>>>> zynq_z_turn_defconfig configuration on MYIR Z-Turn board.
>>>>>>>>>>> And I see the following when I am trying to run `dhcp' command.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>>> mdio_register: non unique device name 'eth0'
>>>>>>>>>>> ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>>> mdio_register: non unique device name 'eth0'
>>>>>>>>>>> ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>>> mdio_register: non unique device name 'eth0'
>>>>>>>>>>> ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>>> mdio_register: non unique device name 'eth0'
>>>>>>>>>>> No ethernet found.
>>>>>>>>>>> ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>>> mdio_register: non unique device name 'eth0'
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Full boot log is the following:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> U-Boot SPL 2020.04-rc4-00100-g74bf17db39 (Apr 06 2020 - 11:00:45 +0300)
>>>>>>>>>> mmc boot
>>>>>>>>>> Trying to boot from MMC1
>>>>>>>>>> spl_load_image_fat_os: error reading image system.dtb, err - -2
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> U-Boot 2020.04-rc4-00100-g74bf17db39 (Apr 06 2020 - 11:00:45 +0300)
>>>>>>>>>>
>>>>>>>>>> CPU:   Zynq 7z010
>>>>>>>>>> Silicon: v3.1
>>>>>>>>>> Model: Zynq Z-Turn MYIR Board
>>>>>>>>>> DRAM:  ECC disabled 1 GiB
>>>>>>>>>> MMC:   mmc at e0100000: 0
>>>>>>>>>> In:    serial at e0001000
>>>>>>>>>> Out:   serial at e0001000
>>>>>>>>>> Err:   serial at e0001000
>>>>>>>>>> Net:   ZYNQ GEM: e000b000, phyaddr 0, interface rgmii-id
>>>>>>>>>> Could not get PHY for eth0: addr 0
>>>>>>>>>> No ethernet found.
>>>>>>>>>>
>>>>>>>>>> Hit any key to stop autoboot:  0
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> u-boot is not able to find out your phy. Address is setup 0 which means
>>>>>>>>> u-boot should do it autodetection. I don't have this board here to tell
>>>>>>>>> you what can be wrong.
>>>>>>>>>
>>>>>>>>> Alex, Anton: Can you please comment it?
>>>>>>>>>
>>>>>>>>
>>>>>>>> Is there a chance that something is wrong with ps7_init_gpl.c ?
>>>>>>>> The Ethernet PHY doesn't work in Linux too with the similar error message.
>>>>>>>
>>>>>>> Did you validate any golden images that phy is fine?
>>>>>>
>>>>>> Yes, I did. There was microsd card supplied with the board (it has
>>>>>> U-Boot 2013.10-svn7 and 3.15.0-xilinx kernel). The Ethernet is working
>>>>>> with this image.
>>>>>>
>>>>>>>
>>>>>>> Anyway as I said I don't have this board here that's why guys who has it
>>>>>>> should debug it or patches welcome.
>>>>>>> Or 3rd option is if none wants to keep it alive I will remove support
>>>>>>> for it.
>>>>>>
>>>>>> Well, unfortunately, I have little knowledge how to debug this, but if
>>>>>> you have an ideas what could I check, please ask.
>>>>>
>>>>> This board is in mainline for a while. That's why you can go back to
>>>>> commit which introduced this board and try it. Then you can bisect it
>>>>> and see which commit breaks it.
>>>>
>>>> Thanks. I've found where Ethernet was broken:
>>>>
>>>> 1f60789602e0d5f5f9a8b507f25737c65b5d8daa is the first bad commit
>>>> commit 1f60789602e0d5f5f9a8b507f25737c65b5d8daa
>>>> Author: Priyanka Jain <priyanka.jain at nxp.com>
>>>> Date:   Tue Nov 5 04:05:11 2019 +0000
>>>>
>>>>     net/phy: Fix phy_connect() for phy addr 0
>>>>
>>>>     Fix 'mask' calculation in phy_connect() for phy addr '0'.
>>>>     'mask' is getting set to '0xffffffff' for phy addr '0'
>>>>     in phy_connect() whereas expected value is '0'.
>>>>
>>>>     Signed-off-by: Priyanka Jain <priyanka.jain at nxp.com>
>>>>     Reported-by: tetsu-aoki via github
>>>>     Acked-by: Joe Hershberger <joe.hershberger at ni.com>
>>>>
>>>
>>> So presumably that means the phy addr should be set to -1 instead of 0
>>> for autodetect, right? I believe this was seen on at least one other
>>> board [1].
>>>
>>> [1] - https://www.mail-archive.com/u-boot@lists.denx.de/msg351386.html
>>
>> Good. I think that would be the best to setup direct phy address used on
>> that board instead of trying to do autodetection but up2you.
> 
> As far as I understand, the register comes from dtb which basically
> comes from Linux kernel sources.
> In the board schematics
> http://www.myirtech.com/download/Zynq7000/Z-TURNBOARD_schematic.pdf
> I see "PHY address: 000" (page 9), but I cannot understand what did they mean.

when you run mdio list you should see detected phy address.

M


-- 
Michal Simek, Ing. (M.Eng), OpenPGP -> KeyID: FE3D1F91
w: www.monstr.eu p: +42-0-721842854
Maintainer of Linux kernel - Xilinx Microblaze
Maintainer of Linux kernel - Xilinx Zynq ARM and ZynqMP ARM64 SoCs
U-Boot custodian - Xilinx Microblaze/Zynq/ZynqMP/Versal SoCs