[dpdk-stable] [PATCH 00/33] mlx5 backports for 18.02.2 stable

Shahaf Shuler shahafs at mellanox.com
Wed May 9 13:18:17 CEST 2018


Wednesday, May 9, 2018 1:28 PM, Luca Boccassi:
> Subject: Re: [dpdk-stable] [PATCH 00/33] mlx5 backports for 18.02.2 stable
> 
> On Wed, 2018-05-09 at 07:51 +0000, Shahaf Shuler wrote:
> > Hi Luca,
> >
> > Tuesday, May 8, 2018 12:56 PM, Luca Boccassi:
> > > Subject: Re: [PATCH 00/33] mlx5 backports for 18.02.2 stable
> > >
> > > On Tue, 2018-05-08 at 10:49 +0100, Luca Boccassi wrote:
> > > > On Sun, 2018-05-06 at 09:05 +0300, Shahaf Shuler wrote:
> > > to
> > > > 29/33
> > >
> > > Also many of the patches do not apply. Have you checked out the
> > > dpdk-
> > > stable/18.02 branch? I've already backported many fixes, as
> > > mentioned in the email series sent to stable at dpdk.org
> >
> > Yes I see this now. Looks like there is some mess in the stable branch
> > for mlx5. Some of the patches you integrated breaks the compilation.
> > My suggestion is to revert the mlx5 patches you already applied
> > (starting from 18.02.1) , and apply the series I sent.
> >
> > If you agree I will send the remaining missing 30-33. Let me know.
> 
> Yes a DRV_LOG sneaked in one of the patches, didn't notice it, sorry about
> that. Nobody noticed either when I sent the backports to stable at d pdk.org
> last week for review.
> It's a trivial fix though, so I'd much rather fix it than redo the whole lot, see
> the diff at the bottom.

OK, if you prefer this way than ping me when you push the fix. So I can rebase and re-send the series. 

> 
> --
> Kind regards,
> Luca Boccassi
> 
> diff --git a/drivers/net/mlx5/mlx5_ethdev.c
> b/drivers/net/mlx5/mlx5_ethdev.c index 7284dfdee..d8286295e 100644
> --- a/drivers/net/mlx5/mlx5_ethdev.c
> +++ b/drivers/net/mlx5/mlx5_ethdev.c
> @@ -514,7 +514,7 @@ dev_configure(struct rte_eth_dev *dev)
>         if (use_app_rss_key &&
>             (dev->data->dev_conf.rx_adv_conf.rss_conf.rss_key_len !=
>              rss_hash_default_key_len)) {
> -               DRV_LOG(ERR, "port %u RSS key len must be %zu Bytes long",
> +               ERROR("port %u RSS key len must be %zu Bytes long",
>                         dev->data->port_id, rss_hash_default_key_len);
>                 return EINVAL;
>         }
> diff --git a/drivers/net/mlx5/mlx5_rss.c b/drivers/net/mlx5/mlx5_rss.c index
> c9db2a64f..27a579148 100644
> --- a/drivers/net/mlx5/mlx5_rss.c
> +++ b/drivers/net/mlx5/mlx5_rss.c
> @@ -51,11 +51,9 @@ mlx5_rss_hash_update(struct rte_eth_dev *dev,
>         }
>         if (rss_conf->rss_key && rss_conf->rss_key_len) {
>                 if (rss_conf->rss_key_len != rss_hash_default_key_len) {
> -                       DRV_LOG(ERR,
> -                               "port %u RSS key len must be %zu Bytes long",
> +                       ERROR("port %u RSS key len must be %zu Bytes
> + long",
>                                 dev->data->port_id, rss_hash_default_key_len);
> -                       rte_errno = EINVAL;
> -                       return -rte_errno;
> +                       return EINVAL;
>                 }
>                 priv->rss_conf.rss_key = rte_realloc(priv->rss_conf.rss_key,
>                                                      rss_conf->rss_key_len, 0);


More information about the stable mailing list