[dpdk-dev] [PATCH v3] net/mlx5: support device removal event

Adrien Mazarguil adrien.mazarguil at 6wind.com
Tue Sep 5 11:28:02 CEST 2017

Hi Matan,

On Mon, Sep 04, 2017 at 05:52:55PM +0000, Matan Azrad wrote:
> Hi Adrien,
> > -----Original Message-----
> > From: Adrien Mazarguil [mailto:adrien.mazarguil at 6wind.com]
> > Sent: Monday, September 4, 2017 6:33 PM
> > To: Matan Azrad <matan at mellanox.com>
> > Cc: Nélio Laranjeiro <nelio.laranjeiro at 6wind.com>; dev at dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH v3] net/mlx5: support device removal event
> > 
> > Hi Matan,
> > 
> > One comment I have is, while this patch adds support for RMV, it also silently
> > addresses a bug (see large comment you added to
> > priv_link_status_update()).
> > 
> > This should be split in two commits, with the fix part coming first and CC
> > stable at dpdk.org, and a second commit adding RMV support proper.
> > 
> Actually, the mlx4 bug was not appeared in the mlx5 previous code,
> Probably because the RMV interrupt was not implemented in mlx5 before this patch.

Good point, no RMV could occur before it is implemented, however a dedicated
commit for the fix itself (i.e. alarm callback not supposed to end up
calling ibv_get_async_event()) might better explain the logic behind these
changes. What I mean is, if there was no problem, you wouldn't need to make
priv_link_status_update() a separate function, right?

> The big comment just explains the link inconsistent issue and was added
> here since Nelio and I think the new function, priv_link_status_update(),
> justifies this comment for future review.  

I understand, this could also have been part of the commit log of the
dedicated commit.


Adrien Mazarguil

More information about the dev mailing list