Error during burning CC image

Hello,

I’m working with ConnectX6-DX to activate PCC (programmable congestion control).

I’ve done the prequirements according the PCC manual ver1.0
1. using RHEL 7.9
2. matching versions of firmware image and PCC image to 22.31.10
3. mlxconfig to use PCC

But during burning CC image, the error occurs as follows:

tkhan@nxc-node0:~/mlnx_cc_user_app$ sudo flint -d /dev/mst/mt4125_pciconf0 -i cc_image -cc beta burn
FSMST_INITIALIZE -
OK
Writing CONGESTION_CONTROL component-
OK
-E- Burning CC image failed: MCC error, FW error

Can you give me any advice about this MCC and FW error?

Best regards,
Taekyoung

hi,
could you try to first set ROCE_CC_LEGACY_DCQCN to 0 and USER_PROGRAMMABLE_CC to 1.

the commands are:

mlxconfig -d /dev/mst/mt4125_pciconf0 set ROCE_CC_LEGACY_DCQCN=0
mlxfwreset -d /dev/mst/mt4125_pciconf0 reset
mlxconfig -d /dev/mst/mt4125_pciconf0 set USER_PROGRAMMABLE_CC=1
mlxfwreset -d /dev/mst/mt4125_pciconf0 reset

Then burn the image again.

If still can’t. try with the latest firmware. The firmware can be downloaded from:
https://network.nvidia.com/support/firmware/connectx6dx/

Regards,
Levei

Hello,

First, thank you for your attention in this issue.

I’ve actually done those prequirements (1. OS
setting (RHEL7), 2. matching submajor version and minor version of PCC and FW version, 3. mlxconfig to enable/disable USER_PROGRAMMABLE_CC /ROCE_CC_LEGACY_DCQCN )

And we strongly guess that the cause of the MCC error in this burning process is the security attributes which is on secure-fw state.

To avoid this, we tried to sign the cc_image by referring to the security firmware update in MFT4.17, but the same MCC error is repeated.
(Secure Firmware Update - MFT v4.17.0 - NVIDIA Networking Docs)

Is there a way to change these security attributes or sign cc_image?

Above all, I wonder if it is fundamentally possible to use PCC on a device in a secure-fw state.

Any help would be appreciated.

Best regards,
Taekyoung

Hi,

Sign the image should no need anymore.
The sign action is removed from the newer version.
https://docs.nvidia.com/networking/display/MFTv422/Secure+Firmware+Update

I think we need to investigate the issue more.
I see you’re using ConnnectX-6 Dx. Is it still under warranty?
If yes, I suggest sending the SN and the issue to our support: networking-support@nvidia.com
So we can involve more resources to analyze the issue.

Regards,
Levei