public inbox for linux-rdma@vger.kernel.org
 help / color / mirror / Atom feed
From: Saeed Mahameed <saeedm@nvidia.com>
To: Roi Dayan <roid@nvidia.com>,
	"dan.carpenter@oracle.com" <dan.carpenter@oracle.com>,
	Feras Daoud <ferasda@nvidia.com>
Cc: "linux-rdma@vger.kernel.org" <linux-rdma@vger.kernel.org>
Subject: Re: [bug report] net/mxl5e: Add change profile method
Date: Wed, 10 Feb 2021 06:12:58 +0000	[thread overview]
Message-ID: <82d9b9b7b0b063aaab358041906baf3ac48ec4a9.camel@nvidia.com> (raw)
In-Reply-To: <YBz2CSKUBlUCRxsZ@mwanda>

On Fri, 2021-02-05 at 15:45 +0300, Dan Carpenter wrote:
> Hello Saeed Mahameed and Feras Daoud,
> 

Hi Dan, thanks for the report, adding Roi the owner of this change.

> I'm not exactly sure if I should blame commit c4d7eb57687f:
> "net/mxl5e:
> Add change profile method" fomr Mar 22, 2020 or commit 182570b26223
> ("net/mlx5e: Gather common netdev init/cleanup functionality in one
> place") from Oct 2, 2018.
> 
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5658
> mlx5e_netdev_change_profile() warn: 'priv->htb.qos_sq_stats' double
> freed
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5658
> mlx5e_netdev_change_profile() warn: 'priv->scratchpad.cpumask' double
> freed
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5789 mlx5e_probe()
> warn: 'priv->htb.qos_sq_stats' double freed
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5789 mlx5e_probe()
> warn: 'priv->scratchpad.cpumask' double freed
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5802 mlx5e_remove()
> warn: 'priv->htb.qos_sq_stats' double freed
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c:5802 mlx5e_remove()
> warn: 'priv->scratchpad.cpumask' double freed
> drivers/net/ethernet/mellanox/mlx5/core/en_rep.c:1317
> mlx5e_vport_rep_unload() warn: 'priv->htb.qos_sq_stats' double freed
> drivers/net/ethernet/mellanox/mlx5/core/en_rep.c:1317
> mlx5e_vport_rep_unload() warn: 'priv->scratchpad.cpumask' double
> freed

If I may ask,
What is this report ? Coverity ? static checker ? or runtime checks ?

> 
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c
>   5639  int mlx5e_netdev_change_profile(struct mlx5e_priv *priv,
>   5640                                  const struct mlx5e_profile
> *new_profile, void *new_ppriv)
>   5641  {
>   5642          unsigned int new_max_nch = mlx5e_calc_max_nch(priv,
> new_profile);
>   5643          const struct mlx5e_profile *orig_profile = priv-
> >profile;
>   5644          void *orig_ppriv = priv->ppriv;
>   5645          int err, rollback_err;
>   5646  
>   5647          /* sanity */
>   5648          if (new_max_nch != priv->max_nch) {
>   5649                  netdev_warn(priv->netdev,
>   5650                              "%s: Replacing profile with
> different max channles\n",
>   5651                              __func__);
>   5652                  return -EINVAL;
>   5653          }
>   5654  
>   5655          /* cleanup old profile */
>   5656          mlx5e_detach_netdev(priv);
>   5657          priv->profile->cleanup(priv);
> 
> The problem is that mlx5i_pkey_cleanup() calls mlx5e_priv_cleanup().
> 
>   5658          mlx5e_priv_cleanup(priv);
>                 ^^^^^^^^^^^^^^^^^^^^^^^^
> And then it gets called again here.

impossible, mlx5i would never call mlx5e_netdev_change_profile, but
anyway i see the issue with the error flow if this function fail after
mlx5e_priv_cleanup, then we remove the driver, we will actually end up
with double free, since this function is not supposed to free priv-
>fields and then just abort.. 

this function must guarantee that the priv is still intact after it
returns.  We will handle. 

  reply	other threads:[~2021-02-10  6:20 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-05 12:45 [bug report] net/mxl5e: Add change profile method Dan Carpenter
2021-02-10  6:12 ` Saeed Mahameed [this message]
2021-02-10  7:02   ` Dan Carpenter

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=82d9b9b7b0b063aaab358041906baf3ac48ec4a9.camel@nvidia.com \
    --to=saeedm@nvidia.com \
    --cc=dan.carpenter@oracle.com \
    --cc=ferasda@nvidia.com \
    --cc=linux-rdma@vger.kernel.org \
    --cc=roid@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox