RE: [PATCH] IB/mlx5: avoid binding a new mpi unit to the same devices repeatedly

[Parav Pandit]

Hi Qing,


> -----Original Message-----
> From: Qing Huang [mailto:qing.huang@xxxxxxxxxx]
> Sent: Monday, July 23, 2018 10:36 AM
> To: Daniel Jurgens <danielj@xxxxxxxxxxxx>; Or Gerlitz
> <gerlitz.or@xxxxxxxxx>; Parav Pandit <parav@xxxxxxxxxxxx>
> Cc: Linux Kernel <linux-kernel@xxxxxxxxxxxxxxx>; RDMA mailing list <linux-
> rdma@xxxxxxxxxxxxxxx>; Jason Gunthorpe <jgg@xxxxxxxx>; Doug Ledford
> <dledford@xxxxxxxxxx>; Leon Romanovsky <leon@xxxxxxxxxx>;
> gerald.gibson@xxxxxxxxxx
> Subject: Re: [PATCH] IB/mlx5: avoid binding a new mpi unit to the same
> devices repeatedly
> 
> 
> 
> On 7/15/2018 12:48 PM, Daniel Jurgens wrote:
> > On 7/14/2018 10:57 AM, Or Gerlitz wrote:
> >> On Sat, Jul 14, 2018 at 2:50 AM, Qing Huang <qing.huang@xxxxxxxxxx>
> wrote:
> >>> When a CX5 device is configured in dual-port RoCE mode, after
> >>> creating many VFs against port 1, creating the same number of VFs
> >>> against port 2 will flood kernel/syslog with something like
> >>> "mlx5_*:mlx5_ib_bind_slave_port:4266:(pid 5269): port 2 already
> >>> affiliated."
> >>>
> >>> So basically, when traversing mlx5_ib_dev_list,
> >>> mlx5_ib_add_slave_port() shouldn't repeatedly attempt to bind the
> >>> new mpi data unit to every device on the list until it finds an unbound
> device.
> >> Daniel,
> >>
> >> What is mpi data unit?
> > It's a structure to keep track affiliated port info in dual port RoCE mode,
> mpi meaning multi-port info. Parav can review this it my absence, otherwise I
> can take a closer look when I return to the office.
> Hi Daniel/Parav,
> 
> Have you got a chance to review this patch? Thanks!
Didn't have chance yet.
Will do this week.