The dr_domain_add_vport_cap() function generally returns NULL on error
but sometimes we want it to return ERR_PTR(-EBUSY) so the caller can
retry.  The problem here is that "ret" can be either -EBUSY or -ENOMEM
and if it's and -ENOMEM then the error pointer is propogated back and
eventually dereferenced in dr_ste_v0_build_src_gvmi_qpn_tag().

Fixes: 11a45def2e19 ("net/mlx5: DR, Add support for SF vports")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
---
v2: Fix a typo in the commit message.  "generally".

 .../net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c    | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
index 3d74109f8230..a379e8358f82 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
@@ -297,6 +297,8 @@ dr_domain_add_vport_cap(struct mlx5dr_domain *dmn, u16 vport)
 	if (ret) {
 		mlx5dr_dbg(dmn, "Couldn't insert new vport into xarray (%d)\n", ret);
 		kvfree(vport_caps);
+		if (ret != -EBUSY)
+			return NULL;
 		return ERR_PTR(ret);
 	}
 
-- 
2.45.2

On 03/12/2024 11:45, Dan Carpenter wrote:
> The dr_domain_add_vport_cap() function generally returns NULL on error
> but sometimes we want it to return ERR_PTR(-EBUSY) so the caller can
> retry.  The problem here is that "ret" can be either -EBUSY or -ENOMEM
> and if it's and -ENOMEM then the error pointer is propogated back and
> eventually dereferenced in dr_ste_v0_build_src_gvmi_qpn_tag().
> 
> Fixes: 11a45def2e19 ("net/mlx5: DR, Add support for SF vports")
> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> ---
> v2: Fix a typo in the commit message.  "generally".
> 
>   .../net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c    | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> index 3d74109f8230..a379e8358f82 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> @@ -297,6 +297,8 @@ dr_domain_add_vport_cap(struct mlx5dr_domain *dmn, u16 vport)
>   	if (ret) {
>   		mlx5dr_dbg(dmn, "Couldn't insert new vport into xarray (%d)\n", ret);
>   		kvfree(vport_caps);
> +		if (ret != -EBUSY)
> +			return NULL;
>   		return ERR_PTR(ret);
>   	}
>   

Thanks for your patch.

It would be clearer to the reader if you test against -EBUSY and return 
ERR_PTR(-EBUSY).
Otherwise, return NULL.

Or at least modify ERR_PTR(ret) to be ERR_PTR(-EBUSY).

Keeping "return ERR_PTR(ret);" does not make it clear that we have only 
2 possible return values here.

On Wed, Dec 04, 2024 at 01:42:14PM +0200, Tariq Toukan wrote:
> 
> 
> On 03/12/2024 11:45, Dan Carpenter wrote:
> > The dr_domain_add_vport_cap() function generally returns NULL on error
> > but sometimes we want it to return ERR_PTR(-EBUSY) so the caller can
> > retry.  The problem here is that "ret" can be either -EBUSY or -ENOMEM
> > and if it's and -ENOMEM then the error pointer is propogated back and
> > eventually dereferenced in dr_ste_v0_build_src_gvmi_qpn_tag().
> > 
> > Fixes: 11a45def2e19 ("net/mlx5: DR, Add support for SF vports")
> > Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
> > Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> > ---
> > v2: Fix a typo in the commit message.  "generally".
> > 
> >   .../net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c    | 2 ++
> >   1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > index 3d74109f8230..a379e8358f82 100644
> > --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > @@ -297,6 +297,8 @@ dr_domain_add_vport_cap(struct mlx5dr_domain *dmn, u16 vport)
> >   	if (ret) {
> >   		mlx5dr_dbg(dmn, "Couldn't insert new vport into xarray (%d)\n", ret);
> >   		kvfree(vport_caps);
> > +		if (ret != -EBUSY)
> > +			return NULL;
> >   		return ERR_PTR(ret);
> >   	}
> 
> Thanks for your patch.
> 
> It would be clearer to the reader if you test against -EBUSY and return
> ERR_PTR(-EBUSY).
> Otherwise, return NULL.
> 
> Or at least modify ERR_PTR(ret) to be ERR_PTR(-EBUSY).
> 
> Keeping "return ERR_PTR(ret);" does not make it clear that we have only 2
> possible return values here.

Sure, that's a good point.

regards,
dan carpenter