2023-11-28 08:29:34

by Dinghao Liu

Subject: [PATCH] net/mlx5e: fix a potential double-free in fs_any_create_groups

When kcalloc() for ft->g succeeds but kvzalloc() for in fails,
fs_any_create_groups() will free ft->g. However, its caller
fs_any_create_table() will free ft->g again through calling
mlx5e_destroy_flow_table(), which will lead to a double-free.
Fix this by removing the kfree(ft->g) in fs_any_create_groups().

Fixes: 0f575c20bf06 ("net/mlx5e: Introduce Flow Steering ANY API")
Signed-off-by: Dinghao Liu <[email protected]>
---
drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c | 1 -
1 file changed, 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
index be83ad9db82a..b222d23bfb9a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
@@ -434,7 +434,6 @@ static int fs_any_create_groups(struct mlx5e_flow_table *ft)
ft->g = kcalloc(MLX5E_FS_UDP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
in = kvzalloc(inlen, GFP_KERNEL);
if (!in || !ft->g) {
- kfree(ft->g);
kvfree(in);
return -ENOMEM;
}
--
2.17.1

2023-11-28 08:58:28

by Tariq Toukan

Subject: Re: [PATCH] net/mlx5e: fix a potential double-free in fs_any_create_groups

On 28/11/2023 10:28, Dinghao Liu wrote:
> When kcalloc() for ft->g succeeds but kvzalloc() for in fails,
> fs_any_create_groups() will free ft->g. However, its caller
> fs_any_create_table() will free ft->g again through calling
> mlx5e_destroy_flow_table(), which will lead to a double-free.
> Fix this by removing the kfree(ft->g) in fs_any_create_groups().
>
> Fixes: 0f575c20bf06 ("net/mlx5e: Introduce Flow Steering ANY API")
> Signed-off-by: Dinghao Liu <[email protected]>
> ---
> drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> index be83ad9db82a..b222d23bfb9a 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> @@ -434,7 +434,6 @@ static int fs_any_create_groups(struct mlx5e_flow_table *ft)
> ft->g = kcalloc(MLX5E_FS_UDP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
> in = kvzalloc(inlen, GFP_KERNEL);
> if (!in || !ft->g) {
> - kfree(ft->g);
> kvfree(in);
> return -ENOMEM;
> }

Function fs_any_create_groups should not return failure without cleaning
itself up. This is not the right fix.
Freeing ft->g and setting it to NULL will do it, and will avoid the
double free.

2023-11-28 09:15:57

by Dinghao Liu

Subject: Re: [PATCH] net/mlx5e: fix a potential double-free in fs_any_create_groups

>
> On 28/11/2023 10:28, Dinghao Liu wrote:
> > When kcalloc() for ft->g succeeds but kvzalloc() for in fails,
> > fs_any_create_groups() will free ft->g. However, its caller
> > fs_any_create_table() will free ft->g again through calling
> > mlx5e_destroy_flow_table(), which will lead to a double-free.
> > Fix this by removing the kfree(ft->g) in fs_any_create_groups().
> >
> > Fixes: 0f575c20bf06 ("net/mlx5e: Introduce Flow Steering ANY API")
> > Signed-off-by: Dinghao Liu <[email protected]>
> > ---
> > drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c | 1 -
> > 1 file changed, 1 deletion(-)
> >
> > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> > index be83ad9db82a..b222d23bfb9a 100644
> > --- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
> > @@ -434,7 +434,6 @@ static int fs_any_create_groups(struct mlx5e_flow_table *ft)
> > ft->g = kcalloc(MLX5E_FS_UDP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
> > in = kvzalloc(inlen, GFP_KERNEL);
> > if (!in || !ft->g) {
> > - kfree(ft->g);
> > kvfree(in);
> > return -ENOMEM;
> > }
>
> Function fs_any_create_groups should not return failure without cleaning
> itself up. This is not the right fix.
> Freeing ft->g and setting it to NULL will do it, and will avoid the
> double free.

Thanks for your advice! I will resend a new patch soon.

Regards,
Dinghao