FDR10 is wrongly displayed as QDR

[0blu]

Hey,
unfortunately FDR10 connections are displayed as QDR in ibtool.

$ ibtool ibnetdiscover
[...]
[1]	"S-<GUID>"[5]	# "desc" lid 2 4xQDR
[...]

$ ibnetdiscover
[...]
[1]	"S-<GUID>"[5]	# "desc" lid 2 4xFDR10
[...]

[jgunthorpe]

FDR10 is a non-standard Mellanox extension, I never looked into how to support it. If you have hardware that can support it, perhaps you could send a patch?

[0blu]

Thanks for the quick response.
Unfortunately i hardly have time to look into the IBA_struct.py code.

If you could make a new branch and add these:

	IB_MLNX_EXT_PORT_STATE_CHG_ENABLE_F,
	IB_MLNX_EXT_PORT_LINK_SPEED_SUPPORTED_F,
	IB_MLNX_EXT_PORT_LINK_SPEED_ENABLED_F,
	IB_MLNX_EXT_PORT_LINK_SPEED_ACTIVE_F,
	IB_MLNX_EXT_PORT_LAST_F,

https://github.com/chu11/libibmad-snapshot/blob/master/include/infiniband/mad.h#L1070-L1074
into IBA_struct.py i could test it and verify the result on my fabric.
https://github.com/jgunthorpe/python-rdma/blob/master/rdma/IBA_struct.py

After this we can compare IB_MLNX_EXT_PORT_LINK_SPEED_ACTIVE_F & 0x8(?) if its really FDR10

[jgunthorpe]

The information in libibmad does not seem to be complete. What is the attribute ID of the vendor structure, what operation is used to read it? The layout in libibmad is also incomplete, lots of holes.

Adding a new struct is easy, if you know what is in it, see for instance:

https://github.com/jgunthorpe/python-rdma/blob/master/codegen/vend_mlx.xml

Can you find a publication where mellanox documents this MAD?

[0blu]

How did you generate/get the vend_mlx.xml?
The file is ~6 Years old, maybe its deprecated.

[jgunthorpe]

It is hand written from the specification of the wire format. I expect that specific mad is out of date, but something that looks basically like it is what is needed to add the new mellanox vendor mad, I just don't have enough information to do it.