[dpdk-dev] [PATCH v2] net/mlx5: fix init on secondary process

[Adrien Mazarguil]

On Wed, Oct 19, 2016 at 10:31:48AM +0100, Bruce Richardson wrote:
> On Mon, Oct 17, 2016 at 04:18:59PM +0200, Adrien Mazarguil wrote:
> > On Mon, Oct 17, 2016 at 02:52:39PM +0100, Ferruh Yigit wrote:
> > > Hi Adrien,
> > > 
> > > On 10/17/2016 1:56 PM, Olivier Gournet wrote:
> > > > Fixes: 1d88ba171942 ("net/mlx5: refactor Tx data path")
> > > > Fixes: 21c8bb4928c9 ("net/mlx5: split Tx queue structure")
> > > > 
> > > > Signed-off-by: Olivier Gournet <ogournet at corp.free.fr>
> > > 
> > > According your comment on previous version of it, I think you have your
> > > Ack on this patch, but can you please confirm?
> > 
> > Yes it's fine, thanks.
> > 
> > Acked-by: Adrien Mazarguil <adrien.mazarguil at 6wind.com>
> > 
> > -- 
> While this patch is acked, I'd still like a bit of detail in the commit
> message describing what the problem is and how the patch fixes it. The
> Chuck Norris approach of trying to stare down the code until it tells
> me just isn't working for me today! :-)
> 
> Adrien or Olivier, if you can supply a brief description of what this
> patch is doing and why I'll add it to the commit log on apply.

*cough* this patch restores the original behavior of not causing a secondary
process to segfault during init.

Seriously, one needs to look at mlx5_secondary_data_setup() in both commits
to really understand what happened, my suggestion for a commit log:

The changes introduced by these commits made secondaries attempt to
reinitialize the TX queue structures of the primary instead of their own,
for which they also do not allocate enough memory, leading to crashes.

-- 
Adrien Mazarguil
6WIND