Skip to contentSkip to conversation

mlx5 fixes 2023-07-26

GIT PULL Applied Saeed Mahameed 2 participants 17 messages Jul 26, 2023 – Jul 28, 2023

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

From: Saeed Mahameed <[email protected]>

This series provides bug fixes to mlx5 driver. Please pull and let me know if there is any problem.

Thanks,
Saeed.

The following changes since commit fa29d467977d50400f6bb1374e942e7474fdf53c:

Merge branch 'tools-ynl-gen-fix-parse-multi-attr-enum-attribute' (2023-07-26 13:38:29 -0700)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-fixes-2023-07-26

for you to fetch changes up to 53d737dfd3d7b023fa9fa445ea3f3db0ac9da402:

net/mlx5: Unregister devlink params in case interface is down (2023-07-26 14:31:05 -0700)

----------------------------------------------------------------
mlx5-fixes-2023-07-26

----------------------------------------------------------------
Amir Tzin (1):
net/mlx5e: Fix crash moving to switchdev mode when ntuple offload is set

Chris Mi (2):
net/mlx5e: Don't hold encap tbl lock if there is no encap action net/mlx5: fs_chains: Fix ft prio if ignore_flow_level is not supported

Dragos Tatulea (2):
net/mlx5e: xsk: Fix invalid buffer access for legacy rq net/mlx5e: xsk: Fix crash on regular rq reactivation

Jianbo Liu (2):
net/mlx5e: Move representor neigh cleanup to profile cleanup_tx net/mlx5e: kTLS, Fix protection domain in use syndrome when devlink reload

Shay Drory (3):
net/mlx5: Honor user input for migratable port fn attr net/mlx5: DR, Fix peer domain namespace setting net/mlx5: Unregister devlink params in case interface is down

Vlad Buslov (1):
net/mlx5: Bridge, set debugfs access right to root-only

Yuanjun Gong (1):
net/mlx5e: fix return value check in mlx5e_ipsec_remove_trailer()

Zhengchao Shao (3):
net/mlx5e: fix double free in macsec_fs_tx_create_crypto_table_groups net/mlx5: DR, fix memory leak in mlx5dr_cmd_create_reformat_ctx net/mlx5: fix potential memory leak in mlx5e_init_rep_rx

 .../ethernet/mellanox/mlx5/core/en/tc_tun_encap.c  |  3 ---
 .../net/ethernet/mellanox/mlx5/core/en/xsk/rx.c    |  5 +++-
 .../mellanox/mlx5/core/en_accel/ipsec_rxtx.c       |  4 ++-
 .../ethernet/mellanox/mlx5/core/en_accel/ktls.c    |  8 ------
 .../ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c | 29 +++++++++++++++++++---
 .../mellanox/mlx5/core/en_accel/macsec_fs.c        |  1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c  | 10 ++++++++
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  | 29 ++++++++++++++++------
 drivers/net/ethernet/mellanox/mlx5/core/en_rep.c   | 20 +++++++--------
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    | 21 +++++++++++++---
 .../mellanox/mlx5/core/esw/bridge_debugfs.c        |  2 +-
 .../ethernet/mellanox/mlx5/core/eswitch_offloads.c | 17 ++++++-------
 drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c   |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.h   |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c  |  4 +--
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.h  |  2 +-
 .../ethernet/mellanox/mlx5/core/lib/fs_chains.c    |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/main.c     |  1 +
 .../mellanox/mlx5/core/steering/dr_action.c        |  2 +-
 .../ethernet/mellanox/mlx5/core/steering/dr_cmd.c  |  5 ++--
 .../mellanox/mlx5/core/steering/dr_domain.c        | 19 +++++++++-----
 .../mellanox/mlx5/core/steering/dr_ste_v0.c        |  7 +++---
 .../mellanox/mlx5/core/steering/dr_ste_v1.c        |  7 +++---
 .../mellanox/mlx5/core/steering/dr_types.h         |  2 +-
 .../ethernet/mellanox/mlx5/core/steering/fs_dr.c   |  4 +--
 .../ethernet/mellanox/mlx5/core/steering/mlx5dr.h  |  2 +-
 26 files changed, 137 insertions(+), 73 deletions(-)
#1 lore raw

PATCH 1/15 net/mlx5e: fix double free in macsec_fs_tx_create_crypto_table_groups

From: Zhengchao Shao <[email protected]>

In function macsec_fs_tx_create_crypto_table_groups(), when the ft->g memory is successfully allocated but the 'in' memory fails to be allocated, the memory pointed to by ft->g is released once. And in function macsec_fs_tx_create(), macsec_fs_tx_destroy() is called to release the memory pointed to by ft->g again. This will cause double free problem.

Fixes e467b283ffd5 ("net/mlx5e: Add MACsec TX steering rules")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c +1 -0
index 7fc901a6ec5f..414e28584881 100644
@@ -161,6 +161,7 @@ static int macsec_fs_tx_create_crypto_table_groups(struct mlx5e_flow_table *ft)
 
 	if (!in) {
 		kfree(ft->g);
+		ft->g = NULL;
 		return -ENOMEM;
 	}
 
Signature · 1 line
2.41.0
#3 lore raw

PATCH 2/15 net/mlx5: DR, fix memory leak in mlx5dr_cmd_create_reformat_ctx

In reply to Saeed Mahameed

From: Zhengchao Shao <[email protected]>

when mlx5_cmd_exec failed in mlx5dr_cmd_create_reformat_ctx, the memory pointed by 'in' is not released, which will cause memory leak. Move memory release after mlx5_cmd_exec.

Fixes 1d9186476e12 ("net/mlx5: DR, Add direct rule command utilities")
---
 drivers/net/ethernet/mellanox/mlx5/core/steering/dr_cmd.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_cmd.c +3 -2
index 7491911ebcb5..8c2a34a0d6be 100644
@@ -564,11 +564,12 @@ int mlx5dr_cmd_create_reformat_ctx(struct mlx5_core_dev *mdev,
 
 	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
 	if (err)
-		return err;
+		goto err_free_in;
 
 	*reformat_id = MLX5_GET(alloc_packet_reformat_context_out, out, packet_reformat_id);
-	kvfree(in);
 
+err_free_in:
+	kvfree(in);
 	return err;
 }
 
Signature · 1 line
2.41.0
#4 lore raw

PATCH 3/15 net/mlx5: fix potential memory leak in mlx5e_init_rep_rx

In reply to Saeed Mahameed

From: Zhengchao Shao <[email protected]>

The memory pointed to by the priv->rx_res pointer is not freed in the error path of mlx5e_init_rep_rx, which can lead to a memory leak. Fix by freeing the memory in the error path, thereby making the error path identical to mlx5e_cleanup_rep_rx().

Fixes af8bbf730068 ("net/mlx5e: Convert mlx5e_flow_steering member of mlx5e_priv to pointer")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +2 -1
index 152b62138450..0b265a3f9b76 100644
@@ -1012,7 +1012,7 @@ static int mlx5e_init_rep_rx(struct mlx5e_priv *priv)
 	err = mlx5e_open_drop_rq(priv, &priv->drop_rq);
 	if (err) {
 		mlx5_core_err(mdev, "open drop rq failed, %d\n", err);
-		return err;
+		goto err_rx_res_free;
 	}
 
 	err = mlx5e_rx_res_init(priv->rx_res, priv->mdev, 0,
@@ -1046,6 +1046,7 @@ static int mlx5e_init_rep_rx(struct mlx5e_priv *priv)
 	mlx5e_rx_res_destroy(priv->rx_res);
 err_close_drop_rq:
 	mlx5e_close_drop_rq(&priv->drop_rq);
+err_rx_res_free:
 	mlx5e_rx_res_free(priv->rx_res);
 	priv->rx_res = NULL;
 err_free_fs:
Signature · 1 line
2.41.0
#5 lore raw

PATCH 4/15 net/mlx5e: fix return value check in mlx5e_ipsec_remove_trailer()

In reply to Saeed Mahameed

From: Yuanjun Gong <[email protected]>

mlx5e_ipsec_remove_trailer() should return an error code if function pskb_trim() returns an unexpected value.

Fixes 2ac9cfe78223 ("net/mlx5e: IPSec, Add Innova IPSec offload TX data path")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c +3 -1
index eab5bc718771..8d995e304869 100644
@@ -58,7 +58,9 @@ static int mlx5e_ipsec_remove_trailer(struct sk_buff *skb, struct xfrm_state *x)
 
 	trailer_len = alen + plen + 2;
 
-	pskb_trim(skb, skb->len - trailer_len);
+	ret = pskb_trim(skb, skb->len - trailer_len);
+	if (unlikely(ret))
+		return ret;
 	if (skb->protocol == htons(ETH_P_IP)) {
 		ipv4hdr->tot_len = htons(ntohs(ipv4hdr->tot_len) - trailer_len);
 		ip_send_check(ipv4hdr);
Signature · 1 line
2.41.0
#6 lore raw

PATCH 5/15 net/mlx5: Honor user input for migratable port fn attr

In reply to Saeed Mahameed

From: Shay Drory <[email protected]>

Currently, whenever a user is setting migratable port fn attr, the driver is always turn migratable capability on.
Fix it by honor the user input

Fixes e5b9642a33be ("net/mlx5: E-Switch, Implement devlink port function cmds to control migratable")
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c +1 -1
index bdfe609cc9ec..93b2b94d41cd 100644
@@ -4196,7 +4196,7 @@ int mlx5_devlink_port_fn_migratable_set(struct devlink_port *port, bool enable,
 	}
 
 	hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
-	MLX5_SET(cmd_hca_cap_2, hca_caps, migratable, 1);
+	MLX5_SET(cmd_hca_cap_2, hca_caps, migratable, enable);
 
 	err = mlx5_vport_set_other_func_cap(esw->dev, hca_caps, vport->vport,
 					    MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE2);
Signature · 1 line
2.41.0
#7 lore raw

PATCH 6/15 net/mlx5e: Don't hold encap tbl lock if there is no encap action

In reply to Saeed Mahameed

From: Chris Mi <[email protected]>

The cited commit holds encap tbl lock unconditionally when setting up dests. But it may cause the following deadlock:

PID: 1063722 TASK: ffffa062ca5d0000 CPU: 13 COMMAND: "handler8"
#0 [ffffb14de05b7368] __schedule at ffffffffa1d5aa91
#1 [ffffb14de05b7410] schedule at ffffffffa1d5afdb
#2 [ffffb14de05b7430] schedule_preempt_disabled at ffffffffa1d5b528 #3 [ffffb14de05b7440] __mutex_lock at ffffffffa1d5d6cb
#4 [ffffb14de05b74e8] mutex_lock_nested at ffffffffa1d5ddeb
#5 [ffffb14de05b74f8] mlx5e_tc_tun_encap_dests_set at ffffffffc12f2096 [mlx5_core] #6 [ffffb14de05b7568] post_process_attr at ffffffffc12d9fc5 [mlx5_core] #7 [ffffb14de05b75a0] mlx5e_tc_add_fdb_flow at ffffffffc12de877 [mlx5_core] #8 [ffffb14de05b75f0] __mlx5e_add_fdb_flow at ffffffffc12e0eef [mlx5_core] #9 [ffffb14de05b7660] mlx5e_tc_add_flow at ffffffffc12e12f7 [mlx5_core]
#10 [ffffb14de05b76b8] mlx5e_configure_flower at ffffffffc12e1686 [mlx5_core] #11 [ffffb14de05b7720] mlx5e_rep_indr_offload at ffffffffc12e3817 [mlx5_core] #12 [ffffb14de05b7730] mlx5e_rep_indr_setup_tc_cb at ffffffffc12e388a [mlx5_core] #13 [ffffb14de05b7740] tc_setup_cb_add at ffffffffa1ab2ba8
#14 [ffffb14de05b77a0] fl_hw_replace_filter at ffffffffc0bdec2f [cls_flower] #15 [ffffb14de05b7868] fl_change at ffffffffc0be6caa [cls_flower]
#16 [ffffb14de05b7908] tc_new_tfilter at ffffffffa1ab71f0

[1031218.028143]  wait_for_completion+0x24/0x30
[1031218.028589]  mlx5e_update_route_decap_flows+0x9a/0x1e0 [mlx5_core]
[1031218.029256]  mlx5e_tc_fib_event_work+0x1ad/0x300 [mlx5_core]
[1031218.029885]  process_one_work+0x24e/0x510

Actually no need to hold encap tbl lock if there is no encap action. Fix it by checking if encap action exists or not before holding encap tbl lock.

Fixes 37c3b9fa7ccf ("net/mlx5e: Prevent encap offload when neigh update is running")
---
 .../mellanox/mlx5/core/en/tc_tun_encap.c      |  3 ---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 21 ++++++++++++++++---
 2 files changed, 18 insertions(+), 6 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c +0 -3
index f0c3464f037f..0c88cf47af01 100644
@@ -1030,9 +1030,6 @@ int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
 	int out_index;
 	int err = 0;
 
-	if (!mlx5e_is_eswitch_flow(flow))
-		return 0;
-
 	parse_attr = attr->parse_attr;
 	esw_attr = attr->esw_attr;
 	*vf_tun = false;
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +18 -3
index 8d0a3f69693e..92377632f9e0 100644
@@ -1725,6 +1725,19 @@ verify_attr_actions(u32 actions, struct netlink_ext_ack *extack)
 	return 0;
 }
 
+static bool
+has_encap_dests(struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	int out_index;
+
+	for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++)
+		if (esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP)
+			return true;
+
+	return false;
+}
+
 static int
 post_process_attr(struct mlx5e_tc_flow *flow,
 		  struct mlx5_flow_attr *attr,
@@ -1737,9 +1750,11 @@ post_process_attr(struct mlx5e_tc_flow *flow,
 	if (err)
 		goto err_out;
 
-	err = mlx5e_tc_tun_encap_dests_set(flow->priv, flow, attr, extack, &vf_tun);
-	if (err)
-		goto err_out;
+	if (mlx5e_is_eswitch_flow(flow) && has_encap_dests(attr)) {
+		err = mlx5e_tc_tun_encap_dests_set(flow->priv, flow, attr, extack, &vf_tun);
+		if (err)
+			goto err_out;
+	}
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
 		err = mlx5e_tc_attach_mod_hdr(flow->priv, flow, attr);
Signature · 1 line
2.41.0
#8 lore raw

PATCH 7/15 net/mlx5e: Fix crash moving to switchdev mode when ntuple offload is set

In reply to Saeed Mahameed

From: Amir Tzin <[email protected]>

Moving to switchdev mode with ntuple offload on causes the kernel to crash since fs->arfs is freed during nic profile cleanup flow.

Ntuple offload is not supported in switchdev mode and it is already unset by mlx5 fix feature ndo in switchdev mode. Verify fs->arfs is valid before disabling it.

trace:
[] RIP: 0010:_raw_spin_lock_bh+0x17/0x30
[] arfs_del_rules+0x44/0x1a0 [mlx5_core]
[] mlx5e_arfs_disable+0xe/0x20 [mlx5_core]
[] mlx5e_handle_feature+0x3d/0xb0 [mlx5_core] [] ? __rtnl_unlock+0x25/0x50
[] mlx5e_set_features+0xfe/0x160 [mlx5_core]
[] __netdev_update_features+0x278/0xa50
[] ? netdev_run_todo+0x5e/0x2a0
[] netdev_update_features+0x22/0x70
[] ? _cond_resched+0x15/0x30
[] mlx5e_attach_netdev+0x12a/0x1e0 [mlx5_core] [] mlx5e_netdev_attach_profile+0xa1/0xc0 [mlx5_core] [] mlx5e_netdev_change_profile+0x77/0xe0 [mlx5_core] [] mlx5e_vport_rep_load+0x1ed/0x290 [mlx5_core] [] mlx5_esw_offloads_rep_load+0x88/0xd0 [mlx5_core] [] esw_offloads_load_rep.part.38+0x31/0x50 [mlx5_core] [] esw_offloads_enable+0x6c5/0x710 [mlx5_core] [] mlx5_eswitch_enable_locked+0x1bb/0x290 [mlx5_core] [] mlx5_devlink_eswitch_mode_set+0x14f/0x320 [mlx5_core] [] devlink_nl_cmd_eswitch_set_doit+0x94/0x120 [] genl_family_rcv_msg_doit.isra.17+0x113/0x150 [] genl_family_rcv_msg+0xb7/0x170
[] ? devlink_nl_cmd_port_split_doit+0x100/0x100 [] genl_rcv_msg+0x47/0xa0
[] ? genl_family_rcv_msg+0x170/0x170
[] netlink_rcv_skb+0x4c/0x130
[] genl_rcv+0x24/0x40
[] netlink_unicast+0x19a/0x230
[] netlink_sendmsg+0x204/0x3d0
[] sock_sendmsg+0x50/0x60

Fixes 90b22b9bcd24 ("net/mlx5e: Disable Rx ntuple offload for uplink representor")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c +10 -0
index 933a7772a7a3..5aa51d74f8b4 100644
@@ -135,6 +135,16 @@ static void arfs_del_rules(struct mlx5e_flow_steering *fs);
 
 int mlx5e_arfs_disable(struct mlx5e_flow_steering *fs)
 {
+	/* Moving to switchdev mode, fs->arfs is freed by mlx5e_nic_profile
+	 * cleanup_rx callback and it is not recreated when
+	 * mlx5e_uplink_rep_profile is loaded as mlx5e_create_flow_steering()
+	 * is not called by the uplink_rep profile init_rx callback. Thus, if
+	 * ntuple is set, moving to switchdev flow will enter this function
+	 * with fs->arfs nullified.
+	 */
+	if (!mlx5e_fs_get_arfs(fs))
+		return 0;
+
 	arfs_del_rules(fs);
 
 	return arfs_disable(fs);
Signature · 1 line
2.41.0
#9 lore raw

PATCH 8/15 net/mlx5e: Move representor neigh cleanup to profile cleanup_tx

In reply to Saeed Mahameed

From: Jianbo Liu <[email protected]>

For IP tunnel encapsulation in ECMP (Equal-Cost Multipath) mode, as the flow is duplicated to the peer eswitch, the related neighbour information on the peer uplink representor is created as well.

In the cited commit, eswitch devcom unpair is moved to uplink unload API, specifically the profile->cleanup_tx. If there is a encap rule offloaded in ECMP mode, when one eswitch does unpair (because of unloading the driver, for instance), and the peer rule from the peer eswitch is going to be deleted, the use-after-free error is triggered while accessing neigh info, as it is already cleaned up in uplink's profile->disable, which is before its profile->cleanup_tx.

To fix this issue, move the neigh cleanup to profile's cleanup_tx callback, and after mlx5e_cleanup_uplink_rep_tx is called. The neigh init is moved to init_tx for symmeter.

[ 2453.376299] BUG: KASAN: slab-use-after-free in mlx5e_rep_neigh_entry_release+0x109/0x3a0 [mlx5_core]
[ 2453.379125] Read of size 4 at addr ffff888127af9008 by task modprobe/2496
[ 2453.381542] CPU: 7 PID: 2496 Comm: modprobe Tainted: G    B              6.4.0-rc7+ #15
[ 2453.383386] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
[ 2453.384335] Call Trace:
[ 2453.384625]  <TASK>
[ 2453.384891]  dump_stack_lvl+0x33/0x50
[ 2453.385285]  print_report+0xc2/0x610
[ 2453.385667]  ? __virt_addr_valid+0xb1/0x130
[ 2453.386091]  ? mlx5e_rep_neigh_entry_release+0x109/0x3a0 [mlx5_core]
[ 2453.386757]  kasan_report+0xae/0xe0
[ 2453.387123]  ? mlx5e_rep_neigh_entry_release+0x109/0x3a0 [mlx5_core]
[ 2453.387798]  mlx5e_rep_neigh_entry_release+0x109/0x3a0 [mlx5_core]
[ 2453.388465]  mlx5e_rep_encap_entry_detach+0xa6/0xe0 [mlx5_core]
[ 2453.389111]  mlx5e_encap_dealloc+0xa7/0x100 [mlx5_core]
[ 2453.389706]  mlx5e_tc_tun_encap_dests_unset+0x61/0xb0 [mlx5_core]
[ 2453.390361]  mlx5_free_flow_attr_actions+0x11e/0x340 [mlx5_core]
[ 2453.391015]  ? complete_all+0x43/0xd0
[ 2453.391398]  ? free_flow_post_acts+0x38/0x120 [mlx5_core]
[ 2453.392004]  mlx5e_tc_del_fdb_flow+0x4ae/0x690 [mlx5_core]
[ 2453.392618]  mlx5e_tc_del_fdb_peers_flow+0x308/0x370 [mlx5_core]
[ 2453.393276]  mlx5e_tc_clean_fdb_peer_flows+0xf5/0x140 [mlx5_core]
[ 2453.393925]  mlx5_esw_offloads_unpair+0x86/0x540 [mlx5_core]
[ 2453.394546]  ? mlx5_esw_offloads_set_ns_peer.isra.0+0x180/0x180 [mlx5_core]
[ 2453.395268]  ? down_write+0xaa/0x100
[ 2453.395652]  mlx5_esw_offloads_devcom_event+0x203/0x530 [mlx5_core]
[ 2453.396317]  mlx5_devcom_send_event+0xbb/0x190 [mlx5_core]
[ 2453.396917]  mlx5_esw_offloads_devcom_cleanup+0xb0/0xd0 [mlx5_core]
[ 2453.397582]  mlx5e_tc_esw_cleanup+0x42/0x120 [mlx5_core]
[ 2453.398182]  mlx5e_rep_tc_cleanup+0x15/0x30 [mlx5_core]
[ 2453.398768]  mlx5e_cleanup_rep_tx+0x6c/0x80 [mlx5_core]
[ 2453.399367]  mlx5e_detach_netdev+0xee/0x120 [mlx5_core]
[ 2453.399957]  mlx5e_netdev_change_profile+0x84/0x170 [mlx5_core]
[ 2453.400598]  mlx5e_vport_rep_unload+0xe0/0xf0 [mlx5_core]
[ 2453.403781]  mlx5_eswitch_unregister_vport_reps+0x15e/0x190 [mlx5_core]
[ 2453.404479]  ? mlx5_eswitch_register_vport_reps+0x200/0x200 [mlx5_core]
[ 2453.405170]  ? up_write+0x39/0x60
[ 2453.405529]  ? kernfs_remove_by_name_ns+0xb7/0xe0
[ 2453.405985]  auxiliary_bus_remove+0x2e/0x40
[ 2453.406405]  device_release_driver_internal+0x243/0x2d0
[ 2453.406900]  ? kobject_put+0x42/0x2d0
[ 2453.407284]  bus_remove_device+0x128/0x1d0
[ 2453.407687]  device_del+0x240/0x550
[ 2453.408053]  ? waiting_for_supplier_show+0xe0/0xe0
[ 2453.408511]  ? kobject_put+0xfa/0x2d0
[ 2453.408889]  ? __kmem_cache_free+0x14d/0x280
[ 2453.409310]  mlx5_rescan_drivers_locked.part.0+0xcd/0x2b0 [mlx5_core]
[ 2453.409973]  mlx5_unregister_device+0x40/0x50 [mlx5_core]
[ 2453.410561]  mlx5_uninit_one+0x3d/0x110 [mlx5_core]
[ 2453.411111]  remove_one+0x89/0x130 [mlx5_core]
[ 2453.411628]  pci_device_remove+0x59/0xf0
[ 2453.412026]  device_release_driver_internal+0x243/0x2d0
[ 2453.412511]  ? parse_option_str+0x14/0x90
[ 2453.412915]  driver_detach+0x7b/0xf0
[ 2453.413289]  bus_remove_driver+0xb5/0x160
[ 2453.413685]  pci_unregister_driver+0x3f/0xf0
[ 2453.414104]  mlx5_cleanup+0xc/0x20 [mlx5_core]
Fixes 2be5bd42a5bb ("net/mlx5: Handle pairing of E-switch via uplink un/load APIs")
---
 .../net/ethernet/mellanox/mlx5/core/en_rep.c    | 17 +++++++----------
 1 file changed, 7 insertions(+), 10 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +7 -10
index 0b265a3f9b76..99b3843396f3 100644
@@ -1160,6 +1160,10 @@ static int mlx5e_init_rep_tx(struct mlx5e_priv *priv)
 		return err;
 	}
 
+	err = mlx5e_rep_neigh_init(rpriv);
+	if (err)
+		goto err_neigh_init;
+
 	if (rpriv->rep->vport == MLX5_VPORT_UPLINK) {
 		err = mlx5e_init_uplink_rep_tx(rpriv);
 		if (err)
@@ -1176,6 +1180,8 @@ static int mlx5e_init_rep_tx(struct mlx5e_priv *priv)
 	if (rpriv->rep->vport == MLX5_VPORT_UPLINK)
 		mlx5e_cleanup_uplink_rep_tx(rpriv);
 err_init_tx:
+	mlx5e_rep_neigh_cleanup(rpriv);
+err_neigh_init:
 	mlx5e_destroy_tises(priv);
 	return err;
 }
@@ -1189,22 +1195,17 @@ static void mlx5e_cleanup_rep_tx(struct mlx5e_priv *priv)
 	if (rpriv->rep->vport == MLX5_VPORT_UPLINK)
 		mlx5e_cleanup_uplink_rep_tx(rpriv);
 
+	mlx5e_rep_neigh_cleanup(rpriv);
 	mlx5e_destroy_tises(priv);
 }
 
 static void mlx5e_rep_enable(struct mlx5e_priv *priv)
 {
-	struct mlx5e_rep_priv *rpriv = priv->ppriv;
-
 	mlx5e_set_netdev_mtu_boundaries(priv);
-	mlx5e_rep_neigh_init(rpriv);
 }
 
 static void mlx5e_rep_disable(struct mlx5e_priv *priv)
 {
-	struct mlx5e_rep_priv *rpriv = priv->ppriv;
-
-	mlx5e_rep_neigh_cleanup(rpriv);
 }
 
 static int mlx5e_update_rep_rx(struct mlx5e_priv *priv)
@@ -1254,7 +1255,6 @@ static int uplink_rep_async_event(struct notifier_block *nb, unsigned long event
 
 static void mlx5e_uplink_rep_enable(struct mlx5e_priv *priv)
 {
-	struct mlx5e_rep_priv *rpriv = priv->ppriv;
 	struct net_device *netdev = priv->netdev;
 	struct mlx5_core_dev *mdev = priv->mdev;
 	u16 max_mtu;
@@ -1276,7 +1276,6 @@ static void mlx5e_uplink_rep_enable(struct mlx5e_priv *priv)
 	mlx5_notifier_register(mdev, &priv->events_nb);
 	mlx5e_dcbnl_initialize(priv);
 	mlx5e_dcbnl_init_app(priv);
-	mlx5e_rep_neigh_init(rpriv);
 	mlx5e_rep_bridge_init(priv);
 
 	netdev->wanted_features |= NETIF_F_HW_TC;
@@ -1291,7 +1290,6 @@ static void mlx5e_uplink_rep_enable(struct mlx5e_priv *priv)
 
 static void mlx5e_uplink_rep_disable(struct mlx5e_priv *priv)
 {
-	struct mlx5e_rep_priv *rpriv = priv->ppriv;
 	struct mlx5_core_dev *mdev = priv->mdev;
 
 	rtnl_lock();
@@ -1301,7 +1299,6 @@ static void mlx5e_uplink_rep_disable(struct mlx5e_priv *priv)
 	rtnl_unlock();
 
 	mlx5e_rep_bridge_cleanup(priv);
-	mlx5e_rep_neigh_cleanup(rpriv);
 	mlx5e_dcbnl_delete_app(priv);
 	mlx5_notifier_unregister(mdev, &priv->events_nb);
 	mlx5e_rep_tc_disable(priv);
Signature · 1 line
2.41.0
#10 lore raw

PATCH 9/15 net/mlx5e: xsk: Fix invalid buffer access for legacy rq

In reply to Saeed Mahameed

From: Dragos Tatulea <[email protected]>

The below crash can be encountered when using xdpsock in rx mode for legacy rq: the buffer gets released in the XDP_REDIRECT path, and then once again in the driver. This fix sets the flag to avoid releasing on the driver side.

XSK handling of buffers for legacy rq was relying on the caller to set the skip release flag. But the referenced fix started using fragment counts for pages instead of the skip flag.

Crash log:

 general protection fault, probably for non-canonical address 0xffff8881217e3a: 0000 [#1] SMP
 CPU: 0 PID: 14 Comm: ksoftirqd/0 Not tainted 6.5.0-rc1+ #31
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
 RIP: 0010:bpf_prog_03b13f331978c78c+0xf/0x28
 Code:  ...
 RSP: 0018:ffff88810082fc98 EFLAGS: 00010246
 RAX: 0000000000000000 RBX: ffff888138404901 RCX: c0ffffc900027cbc
 RDX: ffffffffa000b514 RSI: 00ffff8881217e32 RDI: ffff888138404901
 RBP: ffff88810082fc98 R08: 0000000000091100 R09: 0000000000000006
 R10: 0000000000000800 R11: 0000000000000800 R12: ffffc9000027a000
 R13: ffff8881217e2dc0 R14: ffff8881217e2910 R15: ffff8881217e2f00
 FS:  0000000000000000(0000) GS:ffff88852c800000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000564cb2e2cde0 CR3: 000000010e603004 CR4: 0000000000370eb0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 Call Trace:
  <TASK>
  ? die_addr+0x32/0x80
  ? exc_general_protection+0x192/0x390
  ? asm_exc_general_protection+0x22/0x30
  ? 0xffffffffa000b514
  ? bpf_prog_03b13f331978c78c+0xf/0x28
  mlx5e_xdp_handle+0x48/0x670 [mlx5_core]
  ? dev_gro_receive+0x3b5/0x6e0
  mlx5e_xsk_skb_from_cqe_linear+0x6e/0x90 [mlx5_core]
  mlx5e_handle_rx_cqe+0x55/0x100 [mlx5_core]
  mlx5e_poll_rx_cq+0x87/0x6e0 [mlx5_core]
  mlx5e_napi_poll+0x45e/0x6b0 [mlx5_core]
  __napi_poll+0x25/0x1a0
  net_rx_action+0x28a/0x300
  __do_softirq+0xcd/0x279
  ? sort_range+0x20/0x20
  run_ksoftirqd+0x1a/0x20
  smpboot_thread_fn+0xa2/0x130
  kthread+0xc9/0xf0
  ? kthread_complete_and_exit+0x20/0x20
  ret_from_fork+0x1f/0x30
  </TASK>
 Modules linked in: mlx5_ib mlx5_core rpcrdma rdma_ucm ib_iser libiscsi scsi_transport_iscsi ib_umad rdma_cm ib_ipoib iw_cm ib_cm ib_uverbs ib_core xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink xt_addrtype iptable_nat nf_nat br_netfilter overlay zram zsmalloc fuse [last unloaded: mlx5_core]
 ---[ end trace 0000000000000000 ]---
Fixes 7abd955a58fb ("net/mlx5e: RX, Fix page_pool page fragment tracking for XDP")
---
 drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c +4 -1
index d97e6df66f45..b8dd74453655 100644
@@ -323,8 +323,11 @@ struct sk_buff *mlx5e_xsk_skb_from_cqe_linear(struct mlx5e_rq *rq,
 	net_prefetch(mxbuf->xdp.data);
 
 	prog = rcu_dereference(rq->xdp_prog);
-	if (likely(prog && mlx5e_xdp_handle(rq, prog, mxbuf)))
+	if (likely(prog && mlx5e_xdp_handle(rq, prog, mxbuf))) {
+		if (likely(__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)))
+			wi->flags |= BIT(MLX5E_WQE_FRAG_SKIP_RELEASE);
 		return NULL; /* page/packet was consumed by XDP */
+	}
 
 	/* XDP_PASS: copy the data from the UMEM to a new SKB. The frame reuse
 	 * will be handled by mlx5e_free_rx_wqe.
Signature · 1 line
2.41.0
#11 lore raw

PATCH 10/15 net/mlx5e: xsk: Fix crash on regular rq reactivation

In reply to Saeed Mahameed

From: Dragos Tatulea <[email protected]>

When the regular rq is reactivated after the XSK socket is closed it could be reading stale cqes which eventually corrupts the rq. This leads to no more traffic being received on the regular rq and a crash on the next close or deactivation of the rq.

Kal Cuttler Conely reported this issue as a crash on the release path when the xdpsock sample program is stopped (killed) and restarted in sequence while traffic is running.

This patch flushes all cqes when during the rq flush. The cqe flushing is done in the reset state of the rq. mlx5e_rq_to_ready code is moved into the flush function to allow for this.

---
 .../net/ethernet/mellanox/mlx5/core/en_main.c | 29 ++++++++++++++-----
 1 file changed, 21 insertions(+), 8 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_main.c +21 -8
index defb1efccb78..1c820119e438 100644
@@ -1036,7 +1036,23 @@ static int mlx5e_modify_rq_state(struct mlx5e_rq *rq, int curr_state, int next_s
 	return err;
 }
 
-static int mlx5e_rq_to_ready(struct mlx5e_rq *rq, int curr_state)
+static void mlx5e_flush_rq_cq(struct mlx5e_rq *rq)
+{
+	struct mlx5_cqwq *cqwq = &rq->cq.wq;
+	struct mlx5_cqe64 *cqe;
+
+	if (test_bit(MLX5E_RQ_STATE_MINI_CQE_ENHANCED, &rq->state)) {
+		while ((cqe = mlx5_cqwq_get_cqe_enahnced_comp(cqwq)))
+			mlx5_cqwq_pop(cqwq);
+	} else {
+		while ((cqe = mlx5_cqwq_get_cqe(cqwq)))
+			mlx5_cqwq_pop(cqwq);
+	}
+
+	mlx5_cqwq_update_db_record(cqwq);
+}
+
+int mlx5e_flush_rq(struct mlx5e_rq *rq, int curr_state)
 {
 	struct net_device *dev = rq->netdev;
 	int err;
@@ -1046,6 +1062,10 @@ static int mlx5e_rq_to_ready(struct mlx5e_rq *rq, int curr_state)
 		netdev_err(dev, "Failed to move rq 0x%x to reset\n", rq->rqn);
 		return err;
 	}
+
+	mlx5e_free_rx_descs(rq);
+	mlx5e_flush_rq_cq(rq);
+
 	err = mlx5e_modify_rq_state(rq, MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY);
 	if (err) {
 		netdev_err(dev, "Failed to move rq 0x%x to ready\n", rq->rqn);
@@ -1055,13 +1075,6 @@ static int mlx5e_rq_to_ready(struct mlx5e_rq *rq, int curr_state)
 	return 0;
 }
 
-int mlx5e_flush_rq(struct mlx5e_rq *rq, int curr_state)
-{
-	mlx5e_free_rx_descs(rq);
-
-	return mlx5e_rq_to_ready(rq, curr_state);
-}
-
 static int mlx5e_modify_rq_vsd(struct mlx5e_rq *rq, bool vsd)
 {
 	struct mlx5_core_dev *mdev = rq->mdev;
Signature · 1 line
2.41.0
#12 lore raw

PATCH 11/15 net/mlx5: Bridge, set debugfs access right to root-only

In reply to Saeed Mahameed

From: Vlad Buslov <[email protected]>

As suggested during code review set the access rights for bridge 'fdb' debugfs file to root-only.

---
 drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_debugfs.c +1 -1
index b6a45eff28f5..dbd7cbe6cbf3 100644
@@ -64,7 +64,7 @@ void mlx5_esw_bridge_debugfs_init(struct net_device *br_netdev, struct mlx5_esw_
 
 	bridge->debugfs_dir = debugfs_create_dir(br_netdev->name,
 						 bridge->br_offloads->debugfs_root);
-	debugfs_create_file("fdb", 0444, bridge->debugfs_dir, bridge,
+	debugfs_create_file("fdb", 0400, bridge->debugfs_dir, bridge,
 			    &mlx5_esw_bridge_debugfs_fops);
 }
 
Signature · 1 line
2.41.0
#13 lore raw

PATCH 12/15 net/mlx5e: kTLS, Fix protection domain in use syndrome when devlink reload

In reply to Saeed Mahameed

From: Jianbo Liu <[email protected]>

There are DEK objects cached in DEK pool after kTLS is used, and they are freed only in mlx5e_ktls_cleanup().

mlx5e_destroy_mdev_resources() is called in mlx5e_suspend() to free mdev resources, including protection domain (PD). However, PD is still referenced by the cached DEK objects in this case, because profile->cleanup() (and therefore mlx5e_ktls_cleanup()) is called after mlx5e_suspend() during devlink reload. So the following FW syndrome is generated:

mlx5_cmd_out_err:803:(pid 12948): DEALLOC_PD(0x801) op_mod(0x0) failed,

    status bad resource state(0x9), syndrome (0xef0c8a), err(-22)

To avoid this syndrome, move DEK pool destruction to mlx5e_ktls_cleanup_tx(), which is called by profile->cleanup_tx(). And move pool creation to mlx5e_ktls_init_tx() for symmetry.

Fixes f741db1a5171 ("net/mlx5e: kTLS, Improve connection rate by using fast update encryption key")
---
 .../mellanox/mlx5/core/en_accel/ktls.c        |  8 -----
 .../mellanox/mlx5/core/en_accel/ktls_tx.c     | 29 +++++++++++++++++--
 2 files changed, 26 insertions(+), 11 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c +0 -8
index cf704f106b7c..984fa04bd331 100644
@@ -188,7 +188,6 @@ static void mlx5e_tls_debugfs_init(struct mlx5e_tls *tls,
 
 int mlx5e_ktls_init(struct mlx5e_priv *priv)
 {
-	struct mlx5_crypto_dek_pool *dek_pool;
 	struct mlx5e_tls *tls;
 
 	if (!mlx5e_is_ktls_device(priv->mdev))
@@ -199,12 +198,6 @@ int mlx5e_ktls_init(struct mlx5e_priv *priv)
 		return -ENOMEM;
 	tls->mdev = priv->mdev;
 
-	dek_pool = mlx5_crypto_dek_pool_create(priv->mdev, MLX5_ACCEL_OBJ_TLS_KEY);
-	if (IS_ERR(dek_pool)) {
-		kfree(tls);
-		return PTR_ERR(dek_pool);
-	}
-	tls->dek_pool = dek_pool;
 	priv->tls = tls;
 
 	mlx5e_tls_debugfs_init(tls, priv->dfs_root);
@@ -222,7 +215,6 @@ void mlx5e_ktls_cleanup(struct mlx5e_priv *priv)
 	debugfs_remove_recursive(tls->debugfs.dfs);
 	tls->debugfs.dfs = NULL;
 
-	mlx5_crypto_dek_pool_destroy(tls->dek_pool);
 	kfree(priv->tls);
 	priv->tls = NULL;
 }
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c +26 -3
index efb2cf74ad6a..d61be26a4df1 100644
@@ -908,28 +908,51 @@ static void mlx5e_tls_tx_debugfs_init(struct mlx5e_tls *tls,
 
 int mlx5e_ktls_init_tx(struct mlx5e_priv *priv)
 {
+	struct mlx5_crypto_dek_pool *dek_pool;
 	struct mlx5e_tls *tls = priv->tls;
+	int err;
+
+	if (!mlx5e_is_ktls_device(priv->mdev))
+		return 0;
+
+	/* DEK pool could be used by either or both of TX and RX. But we have to
+	 * put the creation here to avoid syndrome when doing devlink reload.
+	 */
+	dek_pool = mlx5_crypto_dek_pool_create(priv->mdev, MLX5_ACCEL_OBJ_TLS_KEY);
+	if (IS_ERR(dek_pool))
+		return PTR_ERR(dek_pool);
+	tls->dek_pool = dek_pool;
 
 	if (!mlx5e_is_ktls_tx(priv->mdev))
 		return 0;
 
 	priv->tls->tx_pool = mlx5e_tls_tx_pool_init(priv->mdev, &priv->tls->sw_stats);
-	if (!priv->tls->tx_pool)
-		return -ENOMEM;
+	if (!priv->tls->tx_pool) {
+		err = -ENOMEM;
+		goto err_tx_pool_init;
+	}
 
 	mlx5e_tls_tx_debugfs_init(tls, tls->debugfs.dfs);
 
 	return 0;
+
+err_tx_pool_init:
+	mlx5_crypto_dek_pool_destroy(dek_pool);
+	return err;
 }
 
 void mlx5e_ktls_cleanup_tx(struct mlx5e_priv *priv)
 {
 	if (!mlx5e_is_ktls_tx(priv->mdev))
-		return;
+		goto dek_pool_destroy;
 
 	debugfs_remove_recursive(priv->tls->debugfs.dfs_tx);
 	priv->tls->debugfs.dfs_tx = NULL;
 
 	mlx5e_tls_tx_pool_cleanup(priv->tls->tx_pool);
 	priv->tls->tx_pool = NULL;
+
+dek_pool_destroy:
+	if (mlx5e_is_ktls_device(priv->mdev))
+		mlx5_crypto_dek_pool_destroy(priv->tls->dek_pool);
 }
Signature · 1 line
2.41.0
#14 lore raw

PATCH 13/15 net/mlx5: fs_chains: Fix ft prio if ignore_flow_level is not supported

In reply to Saeed Mahameed

From: Chris Mi <[email protected]>

The cited commit sets ft prio to fs_base_prio. But if ignore_flow_level it not supported, ft prio must be set based on tc filter prio. Otherwise, all the ft prio are the same on the same chain. It is invalid if ignore_flow_level is not supported.

Fix it by setting ft prio based on tc filter prio and setting fs_base_prio to 0 for fdb.

Fixes 8e80e5648092 ("net/mlx5: fs_chains: Refactor to detach chains from tc usage")
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 1 -
 drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c    | 2 +-
 2 files changed, 1 insertion(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c +0 -1
index 93b2b94d41cd..bc04abb31f9c 100644
@@ -1436,7 +1436,6 @@ esw_chains_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *miss_fdb)
 
 	esw_init_chains_offload_flags(esw, &attr.flags);
 	attr.ns = MLX5_FLOW_NAMESPACE_FDB;
-	attr.fs_base_prio = FDB_TC_OFFLOAD;
 	attr.max_grp_num = esw->params.large_group_num;
 	attr.default_ft = miss_fdb;
 	attr.mapping = esw->offloads.reg_c0_obj_pool;
drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c +1 -1
index db9df9798ffa..a80ecb672f33 100644
@@ -178,7 +178,7 @@ mlx5_chains_create_table(struct mlx5_fs_chains *chains,
 	if (!mlx5_chains_ignore_flow_level_supported(chains) ||
 	    (chain == 0 && prio == 1 && level == 0)) {
 		ft_attr.level = chains->fs_base_level;
-		ft_attr.prio = chains->fs_base_prio;
+		ft_attr.prio = chains->fs_base_prio + prio - 1;
 		ns = (chains->ns == MLX5_FLOW_NAMESPACE_FDB) ?
 			mlx5_get_fdb_sub_ns(chains->dev, chain) :
 			mlx5_get_flow_namespace(chains->dev, chains->ns);
Signature · 1 line
2.41.0
#15 lore raw

PATCH 14/15 net/mlx5: DR, Fix peer domain namespace setting

In reply to Saeed Mahameed

From: Shay Drory <[email protected]>

The offending patch is based on the assumption that for PFs, mlx5_get_dev_index() is the same as vhca_id. However, this assumption is wrong in case of DPU (ECPF).
Fix it by using vhca_id directly, and switch the array of peers to xarray.

Fixes 6d5b7321d8af ("net/mlx5: DR, handle more than one peer domain")
---
 .../mellanox/mlx5/core/eswitch_offloads.c     | 14 +++++++-------
 .../net/ethernet/mellanox/mlx5/core/fs_cmd.c  |  2 +-
 .../net/ethernet/mellanox/mlx5/core/fs_cmd.h  |  2 +-
 .../net/ethernet/mellanox/mlx5/core/fs_core.c |  4 ++--
 .../net/ethernet/mellanox/mlx5/core/fs_core.h |  2 +-
 .../mellanox/mlx5/core/steering/dr_action.c   |  2 +-
 .../mellanox/mlx5/core/steering/dr_domain.c   | 19 +++++++++++++------
 .../mellanox/mlx5/core/steering/dr_ste_v0.c   |  7 ++++---
 .../mellanox/mlx5/core/steering/dr_ste_v1.c   |  7 ++++---
 .../mellanox/mlx5/core/steering/dr_types.h    |  2 +-
 .../mellanox/mlx5/core/steering/fs_dr.c       |  4 ++--
 .../mellanox/mlx5/core/steering/mlx5dr.h      |  2 +-
 12 files changed, 38 insertions(+), 29 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c +7 -7
index bc04abb31f9c..e59380ee1ead 100644
@@ -2778,9 +2778,9 @@ static int mlx5_esw_offloads_set_ns_peer(struct mlx5_eswitch *esw,
 					 struct mlx5_eswitch *peer_esw,
 					 bool pair)
 {
-	u8 peer_idx = mlx5_get_dev_index(peer_esw->dev);
+	u16 peer_vhca_id = MLX5_CAP_GEN(peer_esw->dev, vhca_id);
+	u16 vhca_id = MLX5_CAP_GEN(esw->dev, vhca_id);
 	struct mlx5_flow_root_namespace *peer_ns;
-	u8 idx = mlx5_get_dev_index(esw->dev);
 	struct mlx5_flow_root_namespace *ns;
 	int err;
 
@@ -2788,18 +2788,18 @@ static int mlx5_esw_offloads_set_ns_peer(struct mlx5_eswitch *esw,
 	ns = esw->dev->priv.steering->fdb_root_ns;
 
 	if (pair) {
-		err = mlx5_flow_namespace_set_peer(ns, peer_ns, peer_idx);
+		err = mlx5_flow_namespace_set_peer(ns, peer_ns, peer_vhca_id);
 		if (err)
 			return err;
 
-		err = mlx5_flow_namespace_set_peer(peer_ns, ns, idx);
+		err = mlx5_flow_namespace_set_peer(peer_ns, ns, vhca_id);
 		if (err) {
-			mlx5_flow_namespace_set_peer(ns, NULL, peer_idx);
+			mlx5_flow_namespace_set_peer(ns, NULL, peer_vhca_id);
 			return err;
 		}
 	} else {
-		mlx5_flow_namespace_set_peer(ns, NULL, peer_idx);
-		mlx5_flow_namespace_set_peer(peer_ns, NULL, idx);
+		mlx5_flow_namespace_set_peer(ns, NULL, peer_vhca_id);
+		mlx5_flow_namespace_set_peer(peer_ns, NULL, vhca_id);
 	}
 
 	return 0;
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c +1 -1
index 91dcb0dcad10..aab7059bf6e9 100644
@@ -140,7 +140,7 @@ static void mlx5_cmd_stub_modify_header_dealloc(struct mlx5_flow_root_namespace
 
 static int mlx5_cmd_stub_set_peer(struct mlx5_flow_root_namespace *ns,
 				  struct mlx5_flow_root_namespace *peer_ns,
-				  u8 peer_idx)
+				  u16 peer_vhca_id)
 {
 	return 0;
 }
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.h +1 -1
index b6b9a5a20591..7790ae5531e1 100644
@@ -94,7 +94,7 @@ struct mlx5_flow_cmds {
 
 	int (*set_peer)(struct mlx5_flow_root_namespace *ns,
 			struct mlx5_flow_root_namespace *peer_ns,
-			u8 peer_idx);
+			u16 peer_vhca_id);
 
 	int (*create_ns)(struct mlx5_flow_root_namespace *ns);
 	int (*destroy_ns)(struct mlx5_flow_root_namespace *ns);
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c +2 -2
index 4ef04aa28771..b4eb27e7f28b 100644
@@ -3621,7 +3621,7 @@ void mlx5_destroy_match_definer(struct mlx5_core_dev *dev,
 
 int mlx5_flow_namespace_set_peer(struct mlx5_flow_root_namespace *ns,
 				 struct mlx5_flow_root_namespace *peer_ns,
-				 u8 peer_idx)
+				 u16 peer_vhca_id)
 {
 	if (peer_ns && ns->mode != peer_ns->mode) {
 		mlx5_core_err(ns->dev,
@@ -3629,7 +3629,7 @@ int mlx5_flow_namespace_set_peer(struct mlx5_flow_root_namespace *ns,
 		return -EINVAL;
 	}
 
-	return ns->cmds->set_peer(ns, peer_ns, peer_idx);
+	return ns->cmds->set_peer(ns, peer_ns, peer_vhca_id);
 }
 
 /* This function should be called only at init stage of the namespace.
drivers/net/ethernet/mellanox/mlx5/core/fs_core.h +1 -1
index 03e64c4c245d..4aed1768b85f 100644
@@ -303,7 +303,7 @@ const struct mlx5_flow_cmds *mlx5_fs_cmd_get_fw_cmds(void);
 
 int mlx5_flow_namespace_set_peer(struct mlx5_flow_root_namespace *ns,
 				 struct mlx5_flow_root_namespace *peer_ns,
-				 u8 peer_idx);
+				 u16 peer_vhca_id);
 
 int mlx5_flow_namespace_set_mode(struct mlx5_flow_namespace *ns,
 				 enum mlx5_flow_steering_mode mode);
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_action.c +1 -1
index e739ec6cdf90..54bb0866ed72 100644
@@ -2079,7 +2079,7 @@ mlx5dr_action_create_dest_vport(struct mlx5dr_domain *dmn,
 
 	peer_vport = vhca_id_valid && mlx5_core_is_pf(dmn->mdev) &&
 		(vhca_id != dmn->info.caps.gvmi);
-	vport_dmn = peer_vport ? dmn->peer_dmn[vhca_id] : dmn;
+	vport_dmn = peer_vport ? xa_load(&dmn->peer_dmn_xa, vhca_id) : dmn;
 	if (!vport_dmn) {
 		mlx5dr_dbg(dmn, "No peer vport domain for given vhca_id\n");
 		return NULL;
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_domain.c +13 -6
index 75dc85dc24ef..3d74109f8230 100644
@@ -475,6 +475,7 @@ mlx5dr_domain_create(struct mlx5_core_dev *mdev, enum mlx5dr_domain_type type)
 	mutex_init(&dmn->info.rx.mutex);
 	mutex_init(&dmn->info.tx.mutex);
 	xa_init(&dmn->definers_xa);
+	xa_init(&dmn->peer_dmn_xa);
 
 	if (dr_domain_caps_init(mdev, dmn)) {
 		mlx5dr_err(dmn, "Failed init domain, no caps\n");
@@ -507,6 +508,7 @@ mlx5dr_domain_create(struct mlx5_core_dev *mdev, enum mlx5dr_domain_type type)
 uninit_caps:
 	dr_domain_caps_uninit(dmn);
 def_xa_destroy:
+	xa_destroy(&dmn->peer_dmn_xa);
 	xa_destroy(&dmn->definers_xa);
 	kfree(dmn);
 	return NULL;
@@ -547,6 +549,7 @@ int mlx5dr_domain_destroy(struct mlx5dr_domain *dmn)
 	dr_domain_uninit_csum_recalc_fts(dmn);
 	dr_domain_uninit_resources(dmn);
 	dr_domain_caps_uninit(dmn);
+	xa_destroy(&dmn->peer_dmn_xa);
 	xa_destroy(&dmn->definers_xa);
 	mutex_destroy(&dmn->info.tx.mutex);
 	mutex_destroy(&dmn->info.rx.mutex);
@@ -556,17 +559,21 @@ int mlx5dr_domain_destroy(struct mlx5dr_domain *dmn)
 
 void mlx5dr_domain_set_peer(struct mlx5dr_domain *dmn,
 			    struct mlx5dr_domain *peer_dmn,
-			    u8 peer_idx)
+			    u16 peer_vhca_id)
 {
+	struct mlx5dr_domain *peer;
+
 	mlx5dr_domain_lock(dmn);
 
-	if (dmn->peer_dmn[peer_idx])
-		refcount_dec(&dmn->peer_dmn[peer_idx]->refcount);
+	peer = xa_load(&dmn->peer_dmn_xa, peer_vhca_id);
+	if (peer)
+		refcount_dec(&peer->refcount);
 
-	dmn->peer_dmn[peer_idx] = peer_dmn;
+	WARN_ON(xa_err(xa_store(&dmn->peer_dmn_xa, peer_vhca_id, peer_dmn, GFP_KERNEL)));
 
-	if (dmn->peer_dmn[peer_idx])
-		refcount_inc(&dmn->peer_dmn[peer_idx]->refcount);
+	peer = xa_load(&dmn->peer_dmn_xa, peer_vhca_id);
+	if (peer)
+		refcount_inc(&peer->refcount);
 
 	mlx5dr_domain_unlock(dmn);
 }
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v0.c +4 -3
index 69d7a8f3c402..f708b029425a 100644
@@ -1652,17 +1652,18 @@ dr_ste_v0_build_src_gvmi_qpn_tag(struct mlx5dr_match_param *value,
 	struct mlx5dr_domain *dmn = sb->dmn;
 	struct mlx5dr_domain *vport_dmn;
 	u8 *bit_mask = sb->bit_mask;
+	struct mlx5dr_domain *peer;
 	bool source_gvmi_set;
 
 	DR_STE_SET_TAG(src_gvmi_qp, tag, source_qp, misc, source_sqn);
 
 	if (sb->vhca_id_valid) {
+		peer = xa_load(&dmn->peer_dmn_xa, id);
 		/* Find port GVMI based on the eswitch_owner_vhca_id */
 		if (id == dmn->info.caps.gvmi)
 			vport_dmn = dmn;
-		else if (id < MLX5_MAX_PORTS && dmn->peer_dmn[id] &&
-			 (id == dmn->peer_dmn[id]->info.caps.gvmi))
-			vport_dmn = dmn->peer_dmn[id];
+		else if (peer && (id == peer->info.caps.gvmi))
+			vport_dmn = peer;
 		else
 			return -EINVAL;
 
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.c +4 -3
index f4ef0b22b991..dd856cde188d 100644
@@ -1984,16 +1984,17 @@ static int dr_ste_v1_build_src_gvmi_qpn_tag(struct mlx5dr_match_param *value,
 	struct mlx5dr_domain *dmn = sb->dmn;
 	struct mlx5dr_domain *vport_dmn;
 	u8 *bit_mask = sb->bit_mask;
+	struct mlx5dr_domain *peer;
 
 	DR_STE_SET_TAG(src_gvmi_qp_v1, tag, source_qp, misc, source_sqn);
 
 	if (sb->vhca_id_valid) {
+		peer = xa_load(&dmn->peer_dmn_xa, id);
 		/* Find port GVMI based on the eswitch_owner_vhca_id */
 		if (id == dmn->info.caps.gvmi)
 			vport_dmn = dmn;
-		else if (id < MLX5_MAX_PORTS && dmn->peer_dmn[id] &&
-			 (id == dmn->peer_dmn[id]->info.caps.gvmi))
-			vport_dmn = dmn->peer_dmn[id];
+		else if (peer && (id == peer->info.caps.gvmi))
+			vport_dmn = peer;
 		else
 			return -EINVAL;
 
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_types.h +1 -1
index 1622dbbe6b97..6c59de3e28f6 100644
@@ -935,7 +935,6 @@ struct mlx5dr_domain_info {
 };
 
 struct mlx5dr_domain {
-	struct mlx5dr_domain *peer_dmn[MLX5_MAX_PORTS];
 	struct mlx5_core_dev *mdev;
 	u32 pdn;
 	struct mlx5_uars_page *uar;
@@ -956,6 +955,7 @@ struct mlx5dr_domain {
 	struct list_head dbg_tbl_list;
 	struct mlx5dr_dbg_dump_info dump_info;
 	struct xarray definers_xa;
+	struct xarray peer_dmn_xa;
 	/* memory management statistics */
 	u32 num_buddies[DR_ICM_TYPE_MAX];
 };
drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c +2 -2
index 6aac5f006bf8..feb307fb3440 100644
@@ -781,14 +781,14 @@ static int mlx5_cmd_dr_update_fte(struct mlx5_flow_root_namespace *ns,
 
 static int mlx5_cmd_dr_set_peer(struct mlx5_flow_root_namespace *ns,
 				struct mlx5_flow_root_namespace *peer_ns,
-				u8 peer_idx)
+				u16 peer_vhca_id)
 {
 	struct mlx5dr_domain *peer_domain = NULL;
 
 	if (peer_ns)
 		peer_domain = peer_ns->fs_dr_domain.dr_domain;
 	mlx5dr_domain_set_peer(ns->fs_dr_domain.dr_domain,
-			       peer_domain, peer_idx);
+			       peer_domain, peer_vhca_id);
 	return 0;
 }
 
drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5dr.h +1 -1
index 24cbb33ecd6c..89fced86936f 100644
@@ -49,7 +49,7 @@ int mlx5dr_domain_sync(struct mlx5dr_domain *domain, u32 flags);
 
 void mlx5dr_domain_set_peer(struct mlx5dr_domain *dmn,
 			    struct mlx5dr_domain *peer_dmn,
-			    u8 peer_idx);
+			    u16 peer_vhca_id);
 
 struct mlx5dr_table *
 mlx5dr_table_create(struct mlx5dr_domain *domain, u32 level, u32 flags,
Signature · 1 line
2.41.0
#16 lore raw

PATCH 15/15 net/mlx5: Unregister devlink params in case interface is down

In reply to Saeed Mahameed

From: Shay Drory <[email protected]>

Currently, in case an interface is down, mlx5 driver doesn't unregister its devlink params, which leads to this WARN[1]. Fix it by unregistering devlink params in that case as well.

[1]

[  295.244769 ] WARNING: CPU: 15 PID: 1 at net/core/devlink.c:9042 devlink_free+0x174/0x1fc
[  295.488379 ] CPU: 15 PID: 1 Comm: shutdown Tainted: G S         OE 5.15.0-1017.19.3.g0677e61-bluefield #g0677e61
[  295.509330 ] Hardware name: https://www.mellanox.com BlueField SoC/BlueField SoC, BIOS 4.2.0.12761 Jun  6 2023
[  295.543096 ] pc : devlink_free+0x174/0x1fc
[  295.551104 ] lr : mlx5_devlink_free+0x18/0x2c [mlx5_core]
[  295.561816 ] sp : ffff80000809b850
[  295.711155 ] Call trace:
[  295.716030 ]  devlink_free+0x174/0x1fc
[  295.723346 ]  mlx5_devlink_free+0x18/0x2c [mlx5_core]
[  295.733351 ]  mlx5_sf_dev_remove+0x98/0xb0 [mlx5_core]
[  295.743534 ]  auxiliary_bus_remove+0x2c/0x50
[  295.751893 ]  __device_release_driver+0x19c/0x280
[  295.761120 ]  device_release_driver+0x34/0x50
[  295.769649 ]  bus_remove_device+0xdc/0x170
[  295.777656 ]  device_del+0x17c/0x3a4
[  295.784620 ]  mlx5_sf_dev_remove+0x28/0xf0 [mlx5_core]
[  295.794800 ]  mlx5_sf_dev_table_destroy+0x98/0x110 [mlx5_core]
[  295.806375 ]  mlx5_unload+0x34/0xd0 [mlx5_core]
[  295.815339 ]  mlx5_unload_one+0x70/0xe4 [mlx5_core]
[  295.824998 ]  shutdown+0xb0/0xd8 [mlx5_core]
[  295.833439 ]  pci_device_shutdown+0x3c/0xa0
[  295.841651 ]  device_shutdown+0x170/0x340
[  295.849486 ]  __do_sys_reboot+0x1f4/0x2a0
[  295.857322 ]  __arm64_sys_reboot+0x2c/0x40
[  295.865329 ]  invoke_syscall+0x78/0x100
[  295.872817 ]  el0_svc_common.constprop.0+0x54/0x184
[  295.882392 ]  do_el0_svc+0x30/0xac
[  295.889008 ]  el0_svc+0x48/0x160
[  295.895278 ]  el0t_64_sync_handler+0xa4/0x130
[  295.903807 ]  el0t_64_sync+0x1a4/0x1a8
[  295.911120 ] ---[ end trace 4f1d2381d00d9dce  ]---
Fixes fe578cbb2f05 ("net/mlx5: Move devlink registration before mlx5_load")
---
 drivers/net/ethernet/mellanox/mlx5/core/main.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/ethernet/mellanox/mlx5/core/main.c +1 -0
index 88dbea6631d5..f42abc2ea73c 100644
@@ -1506,6 +1506,7 @@ void mlx5_uninit_one(struct mlx5_core_dev *dev)
 	if (!test_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state)) {
 		mlx5_core_warn(dev, "%s: interface is down, NOP\n",
 			       __func__);
+		mlx5_devlink_params_unregister(priv_to_devlink(dev));
 		mlx5_cleanup_once(dev);
 		goto out;
 	}
Signature · 1 line
2.41.0
#2 lore raw

Re: [net 01/15] net/mlx5e: fix double free in macsec_fs_tx_create_crypto_table_groups

In reply to Saeed Mahameed
This series was applied to netdev/net.git (main) by Saeed Mahameed <[email protected]>:

Hello:

This series was applied to netdev/net.git (main)
by Saeed Mahameed <[email protected]>:

Quoted 9 lines · On Wed, 26 Jul 2023 14:31:52 -0700 you wrote:

From: Zhengchao Shao <[email protected]>

In function macsec_fs_tx_create_crypto_table_groups(), when the ft->g memory is successfully allocated but the 'in' memory fails to be allocated, the memory pointed to by ft->g is released once. And in function macsec_fs_tx_create(), macsec_fs_tx_destroy() is called to release the memory pointed to by ft->g again. This will cause double free problem.

[...]

Here is the summary with links:
- [net,01/15] net/mlx5e: fix double free in macsec_fs_tx_create_crypto_table_groups

    https://git.kernel.org/netdev/net/c/aeb660171b06
  - [net,02/15] net/mlx5: DR, fix memory leak in mlx5dr_cmd_create_reformat_ctx
    https://git.kernel.org/netdev/net/c/5dd77585dd9d
  - [net,03/15] net/mlx5: fix potential memory leak in mlx5e_init_rep_rx
    https://git.kernel.org/netdev/net/c/c6cf0b6097bf
  - [net,04/15] net/mlx5e: fix return value check in mlx5e_ipsec_remove_trailer()
    https://git.kernel.org/netdev/net/c/e5bcb7564d3b
  - [net,05/15] net/mlx5: Honor user input for migratable port fn attr
    https://git.kernel.org/netdev/net/c/0507f2c8be0d
  - [net,06/15] net/mlx5e: Don't hold encap tbl lock if there is no encap action
    https://git.kernel.org/netdev/net/c/93a331939d1d
  - [net,07/15] net/mlx5e: Fix crash moving to switchdev mode when ntuple offload is set
    https://git.kernel.org/netdev/net/c/3ec43c1b082a
  - [net,08/15] net/mlx5e: Move representor neigh cleanup to profile cleanup_tx
    https://git.kernel.org/netdev/net/c/d03b6e6f3182
  - [net,09/15] net/mlx5e: xsk: Fix invalid buffer access for legacy rq
    https://git.kernel.org/netdev/net/c/e0f52298fee4
  - [net,10/15] net/mlx5e: xsk: Fix crash on regular rq reactivation
    https://git.kernel.org/netdev/net/c/39646d9bcd1a
  - [net,11/15] net/mlx5: Bridge, set debugfs access right to root-only
    https://git.kernel.org/netdev/net/c/eb02b93aad95
  - [net,12/15] net/mlx5e: kTLS, Fix protection domain in use syndrome when devlink reload
    https://git.kernel.org/netdev/net/c/3e4cf1dd2ce4
  - [net,13/15] net/mlx5: fs_chains: Fix ft prio if ignore_flow_level is not supported
    https://git.kernel.org/netdev/net/c/61eab651f6e9
  - [net,14/15] net/mlx5: DR, Fix peer domain namespace setting
    https://git.kernel.org/netdev/net/c/62752c0bc67f
  - [net,15/15] net/mlx5: Unregister devlink params in case interface is down
    https://git.kernel.org/netdev/net/c/53d737dfd3d7

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html