Re: mv88e6171r and mv88e6161 switch not working properly after commit 0f3c66a3c7b4e8b9f654b3c998e9674376a51b0f

Hi,

On 23/04/21 7:57 pm, 曹煜 wrote:

Hi,
     I've confirmed that the mv88e6171r and mv88e6161 switch run into
MTU issue after that commit (from kernel 5.9.0 to kernel 5.12-rc):

Sorry to hear that.

commit 0f3c66a3c7b4e8b9f654b3c998e9674376a51b0f
Author: Chris Packham [off-list ref]
Date:   Fri Jul 24 11:21:20 2020 +1200

     net: dsa: mv88e6xxx: MV88E6097 does not support jumbo configuration

     The MV88E6097 chip does not support configuring jumbo frames. Prior to
     commit 5f4366660d65 only the 6352, 6351, 6165 and 6320 chips configured
     jumbo mode. The refactor accidentally added the function for the 6097.
     Remove the erroneous function pointer assignment.

Do you mean one of the other commits in that series? I think perhaps the 
88e6161 is missing from commit 1baf0fac10fb ("net: dsa: mv88e6xxx: Use 
chip-wide max frame size for MTU"). I was doing that mostly from the 
datasheets I had available so could have easily missed one.

After my modify:

remove
.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,

add
.set_max_frame_size = mv88e6185_g1_set_max_frame_size,

The issue is gone, so could you please commit a fix for these two chips?

I'm traveling right now but I should be able to take a look next week 
(assuming someone else doesn't beat me to it).

PS: there are some more info on this post:
https://forum.doozan.com/read.php?2,119449


Regards!

On Mon, Apr 26, 2021 at 09:25:42PM +0000, Chris Packham wrote:

Hi,

On 23/04/21 7:57 pm, 曹煜 wrote:

quoted

Hi,
     I've confirmed that the mv88e6171r and mv88e6161 switch run into
MTU issue after that commit (from kernel 5.9.0 to kernel 5.12-rc):

Sorry to hear that.

quoted

commit 0f3c66a3c7b4e8b9f654b3c998e9674376a51b0f
Author: Chris Packham [off-list ref]
Date:   Fri Jul 24 11:21:20 2020 +1200

     net: dsa: mv88e6xxx: MV88E6097 does not support jumbo configuration

     The MV88E6097 chip does not support configuring jumbo frames. Prior to
     commit 5f4366660d65 only the 6352, 6351, 6165 and 6320 chips configured
     jumbo mode. The refactor accidentally added the function for the 6097.
     Remove the erroneous function pointer assignment.

Do you mean one of the other commits in that series? I think perhaps the 
88e6161 is missing from commit 1baf0fac10fb ("net: dsa: mv88e6xxx: Use 
chip-wide max frame size for MTU"). I was doing that mostly from the 
datasheets I had available so could have easily missed one.

quoted

After my modify:

remove
.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,

add
.set_max_frame_size = mv88e6185_g1_set_max_frame_size,

The issue is gone, so could you please commit a fix for these two chips?

The datasheet i have for the 6161 shows that bits 13:12 control jumbo
mode. So at least the code fits the datasheet. Also, when describing
global 1 register 4, bit 10, it is reserved. However, the diagram at
the beginning of the global1 section does list bit 10 as being MAX
frame size.

So your testing suggests the data sheet which Chris and I have is
wrong. The change you suggest makes use of the older method of
controlling the MTU.

I will create a patch based on your suggestion.

  Andrew