TDA4VEN-Q1: dp83822 RMII mode ping failed

KONG XIANGXU

Expert 2046 points

Part Number: TDA4VEN-Q1

Tool/software:

The product board phy chip we currently produce uses dp83822, and RMII can not be used to ping successfully. The details are as follows:

1.Generated eth0 device, use ethtool to verify link detected:yes.

2. Can not ping at 10M/100M, and REF_CLK is measured to be 50MHz using an oscilloscope.

3.Read the RCSR register with a status value of 0x65 and a Bit 2 RMII Underflow Status of 1, which means RX FIFO Underflow detected.

4.Use an oscilloscope to measure the signals of TXD and REF_CLK, and the time delay between them is about 8ns.

5.Picture under device tree configuration. RMII and the GPIO used are checked using the devmem2 command.

6.The phy driver uses the driver file that comes with the kernel. The path is: tda4ven_sdk_v1106_cms\tda4_linux\board-support\ti-linux-kernel-6.12.17+git-ti\drivers\net\phy\dp83822.c

7.The Ethernet related dmesg is shown in the following figure.

Could you please tell me why the ping failed?

thank!

10 months ago

0 KONG XIANGXU 10 months ago

Expert 2046 points

Additional notes：sdk version is 11.0

0 Tanmay Patil 10 months ago in reply to KONG XIANGXU

TI__Mastermind 43550 points

Hi,

Can you ensure the IP addressess being used are in the same subnet.

I can see that the IP address for eth0 is 192.168.195.5/24 while you seem to be pinging 192.168.1.32. Both of these are in different subnets.

Regards,
Tanmay

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

Sorry, the photo is not very accurate.It can be confirmed that the IP address is in the same subnets. The inconsistency between the IP address in the figure and the ping is caused by the restart of the development board. We have tried to modify the eth0 IP address to 192.168.1.32 to ping 192.168.1.90, but the result is failed.

Regards,

Xiangxu,Kong

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

Regards,

Xiangxu,Kong

0 Tanmay Patil 10 months ago in reply to KONG XIANGXU

TI__Mastermind 43550 points

Hi,

Got it. Can you please explain to me how are the clocks being on the board. Are you using the internally generated clocks or are you using an external clock to drive both MAC and Phy?

Please note that in case of RMII, you need to route the the CLKOUT pin from the SoC to phy externally if you are using the SoC clock.

The rest of the things look okay.

Regards,
Tanmay

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

Drive using an external clock. The schematic of the external clock is shown below.

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

1.When pinging the computer from the development board, we measured the TXD pin of the development board with an oscilloscope, and no signal was measured. Again, devmem2 command was used to confirm that the pins were correctly multiplexed.

2.Supplement:The defconfig file in the kernel has been configured, as shown in the following figure. The file path is tda4ven_sdk_v1106_cms\tda4_linux\board-supprot\ti-linux-kernel-6.12.17+git-ti\arch\arm64\configa\defconfig

Regards,

Xiangxu Kong

0 Tanmay Patil 10 months ago in reply to KONG XIANGXU

TI__Mastermind 43550 points

Hi,

Can you confirm the RMII_REF_CLK in your schematic is the one which you measured to be 50 MHz

Everything seems to be correct as well from your logs.

Can you share the output of "ethtool -S eth0" before and after you try the ping.

Regards,
Tanmay

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

Hi,

It can be determined that the RMII_REF_CLK measurement is 50 MHz. OK, next I will share the picture of me before and after the ping using the "ethtool -s eth0".

Before the ping:

After the ping:

Regards,

Xiangxu,Kong

0 Tanmay Patil 10 months ago in reply to KONG XIANGXU

TI__Mastermind 43550 points

Hi,

Is it possible to get the logs in text file, We can then compare the easily.

Regards,
Tanmay

0 KONG XIANGXU 10 months ago in reply to Tanmay Patil

Expert 2046 points

Sorry,we can not share in the form of text. In order to facilitate the comparison, I arranged the pictures in the left and right format, the left side is berfore the ping, and the right side is after the ping, hoping to help the comparison.

The above is all logs content.

Regards,

Xiangxu,Kong

0 KONG XIANGXU 10 months ago

Expert 2046 points

Hi，

If you want to check the sending and receiving of the dp83822 and RMII interfaces, what registers can you check or what method can you use to query them？

Regards,

Xiangxu,Kong

0 KONG XIANGXU 9 months ago in reply to Tanmay Patil

Expert 2046 points

Hi，

The net work problem has been successfully, and there is a new problem :our development board shares one MDIO with two RMII ethernet . Is this supported?

Regards,

Xiangxu, Kong

+1 Tanmay Patil 9 months ago in reply to KONG XIANGXU

TI__Mastermind 43550 points

Hi,

There is no hardware limitation for this. The MDIO bus can support upto 32 device on a bus.

But this will have to be managed by software. Which software component is controlling the two RMII phys?

Regards,
Tanmay

0 KONG XIANGXU 9 months ago in reply to Tanmay Patil

Expert 2046 points

Hi,

Thank you for your help! Now we can control by switching eth0 and eth1.

Regards,

Xiangxu, Kong

0 KONG XIANGXU 9 months ago in reply to Tanmay Patil

Expert 2046 points

Hi,

Thank you for your help! Now we can control by switching eth0 and eth1.

Regards,

Xiangxu, Kong

Processors

Processors forum

TDA4VEN-Q1: dp83822 RMII mode ping failed