| Message ID | 20231025032712.79026-1-dust.li@linux.alibaba.com (mailing list archive) |
|---|---|
| State | Awaiting Upstream |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] net/mlx5e: fix double free of encap_header | expand |
On 25.10.2023 05:27, Dust Li wrote: > When mlx5_packet_reformat_alloc() fails, the encap_header allocated in > mlx5e_tc_tun_create_header_ipv4{6} will be released within it. However, > e->encap_header is already set to the previously freed encap_header > before mlx5_packet_reformat_alloc(). As a result, the later > mlx5e_encap_put() will free e->encap_header again, causing a double free > issue. > > mlx5e_encap_put() > --> mlx5e_encap_dealloc() > --> kfree(e->encap_header) nit: I think it should mlx5e_encap_put_locked not mlx5e_encap_put to be precise. > > This happens when cmd: MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT fail. > > This patch fix it by not setting e->encap_header until > mlx5_packet_reformat_alloc() success. > > Fixes: d589e785baf5e("net/mlx5e: Allow concurrent creation of encap entries") > Reported-by: Cruz Zhao <cruzzhao@linux.alibaba.com> > Reported-by: Tianchen Ding <dtcccc@linux.alibaba.com> > Signed-off-by: Dust Li <dust.li@linux.alibaba.com> > --- Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> > drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c | 10 ++++------ > 1 file changed, 4 insertions(+), 6 deletions(-) > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c > index 00a04fdd756f..8bca696b6658 100644 > --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c > @@ -300,9 +300,6 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, > if (err) > goto destroy_neigh_entry; > > - e->encap_size = ipv4_encap_size; > - e->encap_header = encap_header; > - > if (!(nud_state & NUD_VALID)) { > neigh_event_send(attr.n, NULL); > /* the encap entry will be made valid on neigh update event > @@ -322,6 +319,8 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, > goto destroy_neigh_entry; > } > > + e->encap_size = ipv4_encap_size; > + e->encap_header = encap_header; > e->flags |= MLX5_ENCAP_ENTRY_VALID; > mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); > mlx5e_route_lookup_ipv4_put(&attr); > @@ -568,9 +567,6 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, > if (err) > goto destroy_neigh_entry; > > - e->encap_size = ipv6_encap_size; > - e->encap_header = encap_header; > - > if (!(nud_state & NUD_VALID)) { > neigh_event_send(attr.n, NULL); > /* the encap entry will be made valid on neigh update event > @@ -590,6 +586,8 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, > goto destroy_neigh_entry; > } > > + e->encap_size = ipv6_encap_size; > + e->encap_header = encap_header; > e->flags |= MLX5_ENCAP_ENTRY_VALID; > mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); > mlx5e_route_lookup_ipv6_put(&attr);
On 25 Oct 11:27, Dust Li wrote: >When mlx5_packet_reformat_alloc() fails, the encap_header allocated in >mlx5e_tc_tun_create_header_ipv4{6} will be released within it. However, >e->encap_header is already set to the previously freed encap_header >before mlx5_packet_reformat_alloc(). As a result, the later >mlx5e_encap_put() will free e->encap_header again, causing a double free >issue. > >mlx5e_encap_put() > --> mlx5e_encap_dealloc() > --> kfree(e->encap_header) > >This happens when cmd: MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT fail. > >This patch fix it by not setting e->encap_header until >mlx5_packet_reformat_alloc() success. > >Fixes: d589e785baf5e("net/mlx5e: Allow concurrent creation of encap entries") >Reported-by: Cruz Zhao <cruzzhao@linux.alibaba.com> >Reported-by: Tianchen Ding <dtcccc@linux.alibaba.com> >Signed-off-by: Dust Li <dust.li@linux.alibaba.com> Applied to net-mlx5
On Wed, Oct 25, 2023 at 11:06:25AM +0200, Wojciech Drewek wrote: > > >On 25.10.2023 05:27, Dust Li wrote: >> When mlx5_packet_reformat_alloc() fails, the encap_header allocated in >> mlx5e_tc_tun_create_header_ipv4{6} will be released within it. However, >> e->encap_header is already set to the previously freed encap_header >> before mlx5_packet_reformat_alloc(). As a result, the later >> mlx5e_encap_put() will free e->encap_header again, causing a double free >> issue. >> >> mlx5e_encap_put() >> --> mlx5e_encap_dealloc() >> --> kfree(e->encap_header) > >nit: I think it should mlx5e_encap_put_locked not mlx5e_encap_put to be precise. You are right. The original version is mlx5e_encap_put, I mistakenly used the name of the old version when doing git blame. Best regards, Dust > >> >> This happens when cmd: MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT fail. >> >> This patch fix it by not setting e->encap_header until >> mlx5_packet_reformat_alloc() success. >> >> Fixes: d589e785baf5e("net/mlx5e: Allow concurrent creation of encap entries") >> Reported-by: Cruz Zhao <cruzzhao@linux.alibaba.com> >> Reported-by: Tianchen Ding <dtcccc@linux.alibaba.com> >> Signed-off-by: Dust Li <dust.li@linux.alibaba.com> >> --- > >Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> > >> drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c | 10 ++++------ >> 1 file changed, 4 insertions(+), 6 deletions(-) >> >> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c >> index 00a04fdd756f..8bca696b6658 100644 >> --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c >> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c >> @@ -300,9 +300,6 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, >> if (err) >> goto destroy_neigh_entry; >> >> - e->encap_size = ipv4_encap_size; >> - e->encap_header = encap_header; >> - >> if (!(nud_state & NUD_VALID)) { >> neigh_event_send(attr.n, NULL); >> /* the encap entry will be made valid on neigh update event >> @@ -322,6 +319,8 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, >> goto destroy_neigh_entry; >> } >> >> + e->encap_size = ipv4_encap_size; >> + e->encap_header = encap_header; >> e->flags |= MLX5_ENCAP_ENTRY_VALID; >> mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); >> mlx5e_route_lookup_ipv4_put(&attr); >> @@ -568,9 +567,6 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, >> if (err) >> goto destroy_neigh_entry; >> >> - e->encap_size = ipv6_encap_size; >> - e->encap_header = encap_header; >> - >> if (!(nud_state & NUD_VALID)) { >> neigh_event_send(attr.n, NULL); >> /* the encap entry will be made valid on neigh update event >> @@ -590,6 +586,8 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, >> goto destroy_neigh_entry; >> } >> >> + e->encap_size = ipv6_encap_size; >> + e->encap_header = encap_header; >> e->flags |= MLX5_ENCAP_ENTRY_VALID; >> mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); >> mlx5e_route_lookup_ipv6_put(&attr);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c index 00a04fdd756f..8bca696b6658 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c @@ -300,9 +300,6 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, if (err) goto destroy_neigh_entry; - e->encap_size = ipv4_encap_size; - e->encap_header = encap_header; - if (!(nud_state & NUD_VALID)) { neigh_event_send(attr.n, NULL); /* the encap entry will be made valid on neigh update event @@ -322,6 +319,8 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv, goto destroy_neigh_entry; } + e->encap_size = ipv4_encap_size; + e->encap_header = encap_header; e->flags |= MLX5_ENCAP_ENTRY_VALID; mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); mlx5e_route_lookup_ipv4_put(&attr); @@ -568,9 +567,6 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, if (err) goto destroy_neigh_entry; - e->encap_size = ipv6_encap_size; - e->encap_header = encap_header; - if (!(nud_state & NUD_VALID)) { neigh_event_send(attr.n, NULL); /* the encap entry will be made valid on neigh update event @@ -590,6 +586,8 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv, goto destroy_neigh_entry; } + e->encap_size = ipv6_encap_size; + e->encap_header = encap_header; e->flags |= MLX5_ENCAP_ENTRY_VALID; mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev)); mlx5e_route_lookup_ipv6_put(&attr);
When mlx5_packet_reformat_alloc() fails, the encap_header allocated in mlx5e_tc_tun_create_header_ipv4{6} will be released within it. However, e->encap_header is already set to the previously freed encap_header before mlx5_packet_reformat_alloc(). As a result, the later mlx5e_encap_put() will free e->encap_header again, causing a double free issue. mlx5e_encap_put() --> mlx5e_encap_dealloc() --> kfree(e->encap_header) This happens when cmd: MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT fail. This patch fix it by not setting e->encap_header until mlx5_packet_reformat_alloc() success. Fixes: d589e785baf5e("net/mlx5e: Allow concurrent creation of encap entries") Reported-by: Cruz Zhao <cruzzhao@linux.alibaba.com> Reported-by: Tianchen Ding <dtcccc@linux.alibaba.com> Signed-off-by: Dust Li <dust.li@linux.alibaba.com> --- drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)