* [net 01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-26 4:30 ` patchwork-bot+netdevbpf
2023-05-25 3:48 ` [net 02/17] net/mlx5e: Prevent encap offload when neigh update is running Saeed Mahameed
` (15 subsequent siblings)
16 siblings, 1 reply; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Chris Mi
From: Chris Mi <cmi@nvidia.com>
Move set_encap_dests() and clean_encap_dests() to the tunnel encap
dedicated file. And rename them to mlx5e_tc_tun_encap_dests_set()
and mlx5e_tc_tun_encap_dests_unset().
No functional change in this patch. It is needed in the next patch.
Signed-off-by: Chris Mi <cmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../mellanox/mlx5/core/en/tc_tun_encap.c | 83 +++++++++++++++++
.../mellanox/mlx5/core/en/tc_tun_encap.h | 9 ++
.../net/ethernet/mellanox/mlx5/core/en_tc.c | 89 +------------------
3 files changed, 94 insertions(+), 87 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
index 6a052c6cfc15..d516227b8fe8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
@@ -1016,6 +1016,89 @@ int mlx5e_attach_decap(struct mlx5e_priv *priv,
return err;
}
+int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
+ struct mlx5e_tc_flow *flow,
+ struct mlx5_flow_attr *attr,
+ struct netlink_ext_ack *extack,
+ bool *vf_tun)
+{
+ struct mlx5e_tc_flow_parse_attr *parse_attr;
+ struct mlx5_esw_flow_attr *esw_attr;
+ struct net_device *encap_dev = NULL;
+ struct mlx5e_rep_priv *rpriv;
+ struct mlx5e_priv *out_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;
+
+ for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
+ struct net_device *out_dev;
+ int mirred_ifindex;
+
+ if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
+ continue;
+
+ mirred_ifindex = parse_attr->mirred_ifindex[out_index];
+ out_dev = dev_get_by_index(dev_net(priv->netdev), mirred_ifindex);
+ if (!out_dev) {
+ NL_SET_ERR_MSG_MOD(extack, "Requested mirred device not found");
+ err = -ENODEV;
+ goto out;
+ }
+ err = mlx5e_attach_encap(priv, flow, attr, out_dev, out_index,
+ extack, &encap_dev);
+ dev_put(out_dev);
+ if (err)
+ goto out;
+
+ if (esw_attr->dests[out_index].flags &
+ MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE &&
+ !esw_attr->dest_int_port)
+ *vf_tun = true;
+
+ out_priv = netdev_priv(encap_dev);
+ rpriv = out_priv->ppriv;
+ esw_attr->dests[out_index].rep = rpriv->rep;
+ esw_attr->dests[out_index].mdev = out_priv->mdev;
+ }
+
+ if (*vf_tun && esw_attr->out_count > 1) {
+ NL_SET_ERR_MSG_MOD(extack, "VF tunnel encap with mirroring is not supported");
+ err = -EOPNOTSUPP;
+ goto out;
+ }
+
+out:
+ return err;
+}
+
+void mlx5e_tc_tun_encap_dests_unset(struct mlx5e_priv *priv,
+ struct mlx5e_tc_flow *flow,
+ struct mlx5_flow_attr *attr)
+{
+ struct mlx5_esw_flow_attr *esw_attr;
+ int out_index;
+
+ if (!mlx5e_is_eswitch_flow(flow))
+ return;
+
+ esw_attr = attr->esw_attr;
+
+ for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
+ if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
+ continue;
+
+ mlx5e_detach_encap(flow->priv, flow, attr, out_index);
+ kfree(attr->parse_attr->tun_info[out_index]);
+ }
+}
+
static int cmp_route_info(struct mlx5e_route_key *a,
struct mlx5e_route_key *b)
{
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h
index 8ad273dde40e..5d7d67687cbc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h
@@ -30,6 +30,15 @@ int mlx5e_attach_decap_route(struct mlx5e_priv *priv,
void mlx5e_detach_decap_route(struct mlx5e_priv *priv,
struct mlx5e_tc_flow *flow);
+int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
+ struct mlx5e_tc_flow *flow,
+ struct mlx5_flow_attr *attr,
+ struct netlink_ext_ack *extack,
+ bool *vf_tun);
+void mlx5e_tc_tun_encap_dests_unset(struct mlx5e_priv *priv,
+ struct mlx5e_tc_flow *flow,
+ struct mlx5_flow_attr *attr);
+
struct ip_tunnel_info *mlx5e_dup_tun_info(const struct ip_tunnel_info *tun_info);
int mlx5e_tc_set_attr_rx_tun(struct mlx5e_tc_flow *flow,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index e95414ef1f04..8935156f8f4e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -1699,91 +1699,6 @@ int mlx5e_tc_query_route_vport(struct net_device *out_dev, struct net_device *ro
return mlx5_eswitch_vhca_id_to_vport(esw, vhca_id, vport);
}
-static int
-set_encap_dests(struct mlx5e_priv *priv,
- struct mlx5e_tc_flow *flow,
- struct mlx5_flow_attr *attr,
- struct netlink_ext_ack *extack,
- bool *vf_tun)
-{
- struct mlx5e_tc_flow_parse_attr *parse_attr;
- struct mlx5_esw_flow_attr *esw_attr;
- struct net_device *encap_dev = NULL;
- struct mlx5e_rep_priv *rpriv;
- struct mlx5e_priv *out_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;
-
- for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
- struct net_device *out_dev;
- int mirred_ifindex;
-
- if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
- continue;
-
- mirred_ifindex = parse_attr->mirred_ifindex[out_index];
- out_dev = dev_get_by_index(dev_net(priv->netdev), mirred_ifindex);
- if (!out_dev) {
- NL_SET_ERR_MSG_MOD(extack, "Requested mirred device not found");
- err = -ENODEV;
- goto out;
- }
- err = mlx5e_attach_encap(priv, flow, attr, out_dev, out_index,
- extack, &encap_dev);
- dev_put(out_dev);
- if (err)
- goto out;
-
- if (esw_attr->dests[out_index].flags &
- MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE &&
- !esw_attr->dest_int_port)
- *vf_tun = true;
-
- out_priv = netdev_priv(encap_dev);
- rpriv = out_priv->ppriv;
- esw_attr->dests[out_index].rep = rpriv->rep;
- esw_attr->dests[out_index].mdev = out_priv->mdev;
- }
-
- if (*vf_tun && esw_attr->out_count > 1) {
- NL_SET_ERR_MSG_MOD(extack, "VF tunnel encap with mirroring is not supported");
- err = -EOPNOTSUPP;
- goto out;
- }
-
-out:
- return err;
-}
-
-static void
-clean_encap_dests(struct mlx5e_priv *priv,
- struct mlx5e_tc_flow *flow,
- struct mlx5_flow_attr *attr)
-{
- struct mlx5_esw_flow_attr *esw_attr;
- int out_index;
-
- if (!mlx5e_is_eswitch_flow(flow))
- return;
-
- esw_attr = attr->esw_attr;
-
- for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
- if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
- continue;
-
- mlx5e_detach_encap(priv, flow, attr, out_index);
- kfree(attr->parse_attr->tun_info[out_index]);
- }
-}
-
static int
verify_attr_actions(u32 actions, struct netlink_ext_ack *extack)
{
@@ -1820,7 +1735,7 @@ post_process_attr(struct mlx5e_tc_flow *flow,
if (err)
goto err_out;
- err = set_encap_dests(flow->priv, flow, attr, extack, &vf_tun);
+ err = mlx5e_tc_tun_encap_dests_set(flow->priv, flow, attr, extack, &vf_tun);
if (err)
goto err_out;
@@ -4324,7 +4239,7 @@ mlx5_free_flow_attr_actions(struct mlx5e_tc_flow *flow, struct mlx5_flow_attr *a
if (attr->post_act_handle)
mlx5e_tc_post_act_del(get_post_action(flow->priv), attr->post_act_handle);
- clean_encap_dests(flow->priv, flow, attr);
+ mlx5e_tc_tun_encap_dests_unset(flow->priv, flow, attr);
if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT)
mlx5_fc_destroy(counter_dev, attr->counter);
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* Re: [net 01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file
2023-05-25 3:48 ` [net 01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file Saeed Mahameed
@ 2023-05-26 4:30 ` patchwork-bot+netdevbpf
0 siblings, 0 replies; 19+ messages in thread
From: patchwork-bot+netdevbpf @ 2023-05-26 4:30 UTC (permalink / raw)
To: Saeed Mahameed; +Cc: davem, kuba, pabeni, edumazet, saeedm, netdev, tariqt, cmi
Hello:
This series was applied to netdev/net.git (main)
by Saeed Mahameed <saeedm@nvidia.com>:
On Wed, 24 May 2023 20:48:31 -0700 you wrote:
> From: Chris Mi <cmi@nvidia.com>
>
> Move set_encap_dests() and clean_encap_dests() to the tunnel encap
> dedicated file. And rename them to mlx5e_tc_tun_encap_dests_set()
> and mlx5e_tc_tun_encap_dests_unset().
>
> No functional change in this patch. It is needed in the next patch.
>
> [...]
Here is the summary with links:
- [net,01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file
https://git.kernel.org/netdev/net/c/e2ab5aa11f19
- [net,02/17] net/mlx5e: Prevent encap offload when neigh update is running
https://git.kernel.org/netdev/net/c/37c3b9fa7ccf
- [net,03/17] net/mlx5e: Consider internal buffers size in port buffer calculations
https://git.kernel.org/netdev/net/c/81fe2be06291
- [net,04/17] net/mlx5e: Do not update SBCM when prio2buffer command is invalid
https://git.kernel.org/netdev/net/c/623efc4cbd61
- [net,05/17] net/mlx5: Drain health before unregistering devlink
https://git.kernel.org/netdev/net/c/824c8dc4a470
- [net,06/17] net/mlx5: SF, Drain health before removing device
https://git.kernel.org/netdev/net/c/b4646da0573f
- [net,07/17] net/mlx5: fw_tracer, Fix event handling
https://git.kernel.org/netdev/net/c/341a80de2468
- [net,08/17] net/mlx5e: Use query_special_contexts cmd only once per mdev
https://git.kernel.org/netdev/net/c/1db1f21caebb
- [net,09/17] net/mlx5: Fix post parse infra to only parse every action once
https://git.kernel.org/netdev/net/c/5d862ec631f3
- [net,10/17] net/mlx5e: Don't attach netdev profile while handling internal error
https://git.kernel.org/netdev/net/c/bdf274750fca
- [net,11/17] net/mlx5e: Move Ethernet driver debugfs to profile init callback
https://git.kernel.org/netdev/net/c/c4c24fc30cc4
- [net,12/17] net/mlx5: DR, Add missing mutex init/destroy in pattern manager
https://git.kernel.org/netdev/net/c/fe5c2d3aef93
- [net,13/17] net/mlx5: Fix check for allocation failure in comp_irqs_request_pci()
https://git.kernel.org/netdev/net/c/d5972f1000c1
- [net,14/17] Documentation: net/mlx5: Wrap vnic reporter devlink commands in code blocks
https://git.kernel.org/netdev/net/c/c3acc46c602f
- [net,15/17] Documentation: net/mlx5: Use bullet and definition lists for vnic counters description
https://git.kernel.org/netdev/net/c/565b8c60e90f
- [net,16/17] Documentation: net/mlx5: Add blank line separator before numbered lists
https://git.kernel.org/netdev/net/c/92059da65d84
- [net,17/17] Documentation: net/mlx5: Wrap notes in admonition blocks
https://git.kernel.org/netdev/net/c/bb72b94c659f
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 19+ messages in thread
* [net 02/17] net/mlx5e: Prevent encap offload when neigh update is running
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
2023-05-25 3:48 ` [net 01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 03/17] net/mlx5e: Consider internal buffers size in port buffer calculations Saeed Mahameed
` (14 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Chris Mi, Roi Dayan,
Vlad Buslov
From: Chris Mi <cmi@nvidia.com>
The cited commit adds a compeletion to remove dependency on rtnl
lock. But it causes a deadlock for multiple encapsulations:
crash> bt ffff8aece8a64000
PID: 1514557 TASK: ffff8aece8a64000 CPU: 3 COMMAND: "tc"
#0 [ffffa6d14183f368] __schedule at ffffffffb8ba7f45
#1 [ffffa6d14183f3f8] schedule at ffffffffb8ba8418
#2 [ffffa6d14183f418] schedule_preempt_disabled at ffffffffb8ba8898
#3 [ffffa6d14183f428] __mutex_lock at ffffffffb8baa7f8
#4 [ffffa6d14183f4d0] mutex_lock_nested at ffffffffb8baabeb
#5 [ffffa6d14183f4e0] mlx5e_attach_encap at ffffffffc0f48c17 [mlx5_core]
#6 [ffffa6d14183f628] mlx5e_tc_add_fdb_flow at ffffffffc0f39680 [mlx5_core]
#7 [ffffa6d14183f688] __mlx5e_add_fdb_flow at ffffffffc0f3b636 [mlx5_core]
#8 [ffffa6d14183f6f0] mlx5e_tc_add_flow at ffffffffc0f3bcdf [mlx5_core]
#9 [ffffa6d14183f728] mlx5e_configure_flower at ffffffffc0f3c1d1 [mlx5_core]
#10 [ffffa6d14183f790] mlx5e_rep_setup_tc_cls_flower at ffffffffc0f3d529 [mlx5_core]
#11 [ffffa6d14183f7a0] mlx5e_rep_setup_tc_cb at ffffffffc0f3d714 [mlx5_core]
#12 [ffffa6d14183f7b0] tc_setup_cb_add at ffffffffb8931bb8
#13 [ffffa6d14183f810] fl_hw_replace_filter at ffffffffc0dae901 [cls_flower]
#14 [ffffa6d14183f8d8] fl_change at ffffffffc0db5c57 [cls_flower]
#15 [ffffa6d14183f970] tc_new_tfilter at ffffffffb8936047
#16 [ffffa6d14183fac8] rtnetlink_rcv_msg at ffffffffb88c7c31
#17 [ffffa6d14183fb50] netlink_rcv_skb at ffffffffb8942853
#18 [ffffa6d14183fbc0] rtnetlink_rcv at ffffffffb88c1835
#19 [ffffa6d14183fbd0] netlink_unicast at ffffffffb8941f27
#20 [ffffa6d14183fc18] netlink_sendmsg at ffffffffb8942245
#21 [ffffa6d14183fc98] sock_sendmsg at ffffffffb887d482
#22 [ffffa6d14183fcb8] ____sys_sendmsg at ffffffffb887d81a
#23 [ffffa6d14183fd38] ___sys_sendmsg at ffffffffb88806e2
#24 [ffffa6d14183fe90] __sys_sendmsg at ffffffffb88807a2
#25 [ffffa6d14183ff28] __x64_sys_sendmsg at ffffffffb888080f
#26 [ffffa6d14183ff38] do_syscall_64 at ffffffffb8b9b6a8
#27 [ffffa6d14183ff50] entry_SYSCALL_64_after_hwframe at ffffffffb8c0007c
crash> bt 0xffff8aeb07544000
PID: 1110766 TASK: ffff8aeb07544000 CPU: 0 COMMAND: "kworker/u20:9"
#0 [ffffa6d14e6b7bd8] __schedule at ffffffffb8ba7f45
#1 [ffffa6d14e6b7c68] schedule at ffffffffb8ba8418
#2 [ffffa6d14e6b7c88] schedule_timeout at ffffffffb8baef88
#3 [ffffa6d14e6b7d10] wait_for_completion at ffffffffb8ba968b
#4 [ffffa6d14e6b7d60] mlx5e_take_all_encap_flows at ffffffffc0f47ec4 [mlx5_core]
#5 [ffffa6d14e6b7da0] mlx5e_rep_update_flows at ffffffffc0f3e734 [mlx5_core]
#6 [ffffa6d14e6b7df8] mlx5e_rep_neigh_update at ffffffffc0f400bb [mlx5_core]
#7 [ffffa6d14e6b7e50] process_one_work at ffffffffb80acc9c
#8 [ffffa6d14e6b7ed0] worker_thread at ffffffffb80ad012
#9 [ffffa6d14e6b7f10] kthread at ffffffffb80b615d
#10 [ffffa6d14e6b7f50] ret_from_fork at ffffffffb8001b2f
After the first encap is attached, flow will be added to encap
entry's flows list. If neigh update is running at this time, the
following encaps of the flow can't hold the encap_tbl_lock and
sleep. If neigh update thread is waiting for that flow's init_done,
deadlock happens.
Fix it by holding lock outside of the for loop. If neigh update is
running, prevent encap flows from offloading. Since the lock is held
outside of the for loop, concurrent creation of encap entries is not
allowed. So remove unnecessary wait_for_completion call for res_ready.
Fixes: 95435ad7999b ("net/mlx5e: Only access fully initialized flows in neigh update")
Signed-off-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../mellanox/mlx5/core/en/tc_tun_encap.c | 37 ++++++++++---------
1 file changed, 20 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
index d516227b8fe8..f0c3464f037f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
@@ -492,6 +492,19 @@ void mlx5e_encap_put(struct mlx5e_priv *priv, struct mlx5e_encap_entry *e)
mlx5e_encap_dealloc(priv, e);
}
+static void mlx5e_encap_put_locked(struct mlx5e_priv *priv, struct mlx5e_encap_entry *e)
+{
+ struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+ lockdep_assert_held(&esw->offloads.encap_tbl_lock);
+
+ if (!refcount_dec_and_test(&e->refcnt))
+ return;
+ list_del(&e->route_list);
+ hash_del_rcu(&e->encap_hlist);
+ mlx5e_encap_dealloc(priv, e);
+}
+
static void mlx5e_decap_put(struct mlx5e_priv *priv, struct mlx5e_decap_entry *d)
{
struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
@@ -816,6 +829,8 @@ int mlx5e_attach_encap(struct mlx5e_priv *priv,
uintptr_t hash_key;
int err = 0;
+ lockdep_assert_held(&esw->offloads.encap_tbl_lock);
+
parse_attr = attr->parse_attr;
tun_info = parse_attr->tun_info[out_index];
mpls_info = &parse_attr->mpls_info[out_index];
@@ -829,7 +844,6 @@ int mlx5e_attach_encap(struct mlx5e_priv *priv,
hash_key = hash_encap_info(&key);
- mutex_lock(&esw->offloads.encap_tbl_lock);
e = mlx5e_encap_get(priv, &key, hash_key);
/* must verify if encap is valid or not */
@@ -840,15 +854,6 @@ int mlx5e_attach_encap(struct mlx5e_priv *priv,
goto out_err;
}
- mutex_unlock(&esw->offloads.encap_tbl_lock);
- wait_for_completion(&e->res_ready);
-
- /* Protect against concurrent neigh update. */
- mutex_lock(&esw->offloads.encap_tbl_lock);
- if (e->compl_result < 0) {
- err = -EREMOTEIO;
- goto out_err;
- }
goto attach_flow;
}
@@ -877,15 +882,12 @@ int mlx5e_attach_encap(struct mlx5e_priv *priv,
INIT_LIST_HEAD(&e->flows);
hash_add_rcu(esw->offloads.encap_tbl, &e->encap_hlist, hash_key);
tbl_time_before = mlx5e_route_tbl_get_last_update(priv);
- mutex_unlock(&esw->offloads.encap_tbl_lock);
if (family == AF_INET)
err = mlx5e_tc_tun_create_header_ipv4(priv, mirred_dev, e);
else if (family == AF_INET6)
err = mlx5e_tc_tun_create_header_ipv6(priv, mirred_dev, e);
- /* Protect against concurrent neigh update. */
- mutex_lock(&esw->offloads.encap_tbl_lock);
complete_all(&e->res_ready);
if (err) {
e->compl_result = err;
@@ -920,18 +922,15 @@ int mlx5e_attach_encap(struct mlx5e_priv *priv,
} else {
flow_flag_set(flow, SLOW);
}
- mutex_unlock(&esw->offloads.encap_tbl_lock);
return err;
out_err:
- mutex_unlock(&esw->offloads.encap_tbl_lock);
if (e)
- mlx5e_encap_put(priv, e);
+ mlx5e_encap_put_locked(priv, e);
return err;
out_err_init:
- mutex_unlock(&esw->offloads.encap_tbl_lock);
kfree(tun_info);
kfree(e);
return err;
@@ -1027,6 +1026,7 @@ int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
struct net_device *encap_dev = NULL;
struct mlx5e_rep_priv *rpriv;
struct mlx5e_priv *out_priv;
+ struct mlx5_eswitch *esw;
int out_index;
int err = 0;
@@ -1037,6 +1037,8 @@ int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
esw_attr = attr->esw_attr;
*vf_tun = false;
+ esw = priv->mdev->priv.eswitch;
+ mutex_lock(&esw->offloads.encap_tbl_lock);
for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
struct net_device *out_dev;
int mirred_ifindex;
@@ -1075,6 +1077,7 @@ int mlx5e_tc_tun_encap_dests_set(struct mlx5e_priv *priv,
}
out:
+ mutex_unlock(&esw->offloads.encap_tbl_lock);
return err;
}
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 03/17] net/mlx5e: Consider internal buffers size in port buffer calculations
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
2023-05-25 3:48 ` [net 01/17] net/mlx5e: Extract remaining tunnel encap code to dedicated file Saeed Mahameed
2023-05-25 3:48 ` [net 02/17] net/mlx5e: Prevent encap offload when neigh update is running Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 04/17] net/mlx5e: Do not update SBCM when prio2buffer command is invalid Saeed Mahameed
` (13 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Maher Sanalla,
Moshe Shemesh
From: Maher Sanalla <msanalla@nvidia.com>
Currently, when a user triggers a change in port buffer headroom
(buffers 0-7), the driver checks that the requested headroom does
not exceed the total port buffer size. However, this check does not
take into account the internal buffers (buffers 8-9), which are also
part of the total port buffer. This can result in treating invalid port
buffer change requests as valid, causing unintended changes to the shared
buffer.
To address this, include the internal buffers size in the calculation of
available port buffer space which ensures that port buffer requests do not
exceed the correct limit.
Furthermore, remove internal buffers (8-9) size from the total_size
calculation as these buffers are reserved for internal use and are not
exposed to the user.
While at it, add verbosity to the debug prints in
mlx5e_port_query_buffer() function to ease future debugging.
Fixes: ecdf2dadee8e ("net/mlx5e: Receive buffer support for DCBX")
Signed-off-by: Maher Sanalla <msanalla@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../mellanox/mlx5/core/en/port_buffer.c | 42 ++++++++++++-------
.../mellanox/mlx5/core/en/port_buffer.h | 8 ++--
.../ethernet/mellanox/mlx5/core/en_dcbnl.c | 7 ++--
3 files changed, 36 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
index 7ac1ad9c46de..0d78527451bc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
@@ -51,7 +51,7 @@ int mlx5e_port_query_buffer(struct mlx5e_priv *priv,
if (err)
goto out;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
buffer = MLX5_ADDR_OF(pbmc_reg, out, buffer[i]);
port_buffer->buffer[i].lossy =
MLX5_GET(bufferx_reg, buffer, lossy);
@@ -73,14 +73,24 @@ int mlx5e_port_query_buffer(struct mlx5e_priv *priv,
port_buffer->buffer[i].lossy);
}
- port_buffer->headroom_size = total_used;
+ port_buffer->internal_buffers_size = 0;
+ for (i = MLX5E_MAX_NETWORK_BUFFER; i < MLX5E_TOTAL_BUFFERS; i++) {
+ buffer = MLX5_ADDR_OF(pbmc_reg, out, buffer[i]);
+ port_buffer->internal_buffers_size +=
+ MLX5_GET(bufferx_reg, buffer, size) * port_buff_cell_sz;
+ }
+
port_buffer->port_buffer_size =
MLX5_GET(pbmc_reg, out, port_buffer_size) * port_buff_cell_sz;
- port_buffer->spare_buffer_size =
- port_buffer->port_buffer_size - total_used;
-
- mlx5e_dbg(HW, priv, "total buffer size=%d, spare buffer size=%d\n",
- port_buffer->port_buffer_size,
+ port_buffer->headroom_size = total_used;
+ port_buffer->spare_buffer_size = port_buffer->port_buffer_size -
+ port_buffer->internal_buffers_size -
+ port_buffer->headroom_size;
+
+ mlx5e_dbg(HW, priv,
+ "total buffer size=%u, headroom buffer size=%u, internal buffers size=%u, spare buffer size=%u\n",
+ port_buffer->port_buffer_size, port_buffer->headroom_size,
+ port_buffer->internal_buffers_size,
port_buffer->spare_buffer_size);
out:
kfree(out);
@@ -206,11 +216,11 @@ static int port_update_pool_cfg(struct mlx5_core_dev *mdev,
if (!MLX5_CAP_GEN(mdev, sbcam_reg))
return 0;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++)
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++)
lossless_buff_count += ((port_buffer->buffer[i].size) &&
(!(port_buffer->buffer[i].lossy)));
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
p = select_sbcm_params(&port_buffer->buffer[i], lossless_buff_count);
err = mlx5e_port_set_sbcm(mdev, 0, i,
MLX5_INGRESS_DIR,
@@ -293,7 +303,7 @@ static int port_set_buffer(struct mlx5e_priv *priv,
if (err)
goto out;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
void *buffer = MLX5_ADDR_OF(pbmc_reg, in, buffer[i]);
u64 size = port_buffer->buffer[i].size;
u64 xoff = port_buffer->buffer[i].xoff;
@@ -351,7 +361,7 @@ static int update_xoff_threshold(struct mlx5e_port_buffer *port_buffer,
{
int i;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
if (port_buffer->buffer[i].lossy) {
port_buffer->buffer[i].xoff = 0;
port_buffer->buffer[i].xon = 0;
@@ -408,7 +418,7 @@ static int update_buffer_lossy(struct mlx5_core_dev *mdev,
int err;
int i;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
prio_count = 0;
lossy_count = 0;
@@ -515,7 +525,7 @@ int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
if (change & MLX5E_PORT_BUFFER_PRIO2BUFFER) {
update_prio2buffer = true;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++)
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++)
mlx5e_dbg(HW, priv, "%s: requested to map prio[%d] to buffer %d\n",
__func__, i, prio2buffer[i]);
@@ -530,7 +540,7 @@ int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
}
if (change & MLX5E_PORT_BUFFER_SIZE) {
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
mlx5e_dbg(HW, priv, "%s: buffer[%d]=%d\n", __func__, i, buffer_size[i]);
if (!port_buffer.buffer[i].lossy && !buffer_size[i]) {
mlx5e_dbg(HW, priv, "%s: lossless buffer[%d] size cannot be zero\n",
@@ -544,7 +554,9 @@ int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
mlx5e_dbg(HW, priv, "%s: total buffer requested=%d\n", __func__, total_used);
- if (total_used > port_buffer.port_buffer_size)
+ if (total_used > port_buffer.headroom_size &&
+ (total_used - port_buffer.headroom_size) >
+ port_buffer.spare_buffer_size)
return -EINVAL;
update_buffer = true;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h
index a6ef118de758..f4a19ffbb641 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h
@@ -35,7 +35,8 @@
#include "en.h"
#include "port.h"
-#define MLX5E_MAX_BUFFER 8
+#define MLX5E_MAX_NETWORK_BUFFER 8
+#define MLX5E_TOTAL_BUFFERS 10
#define MLX5E_DEFAULT_CABLE_LEN 7 /* 7 meters */
#define MLX5_BUFFER_SUPPORTED(mdev) (MLX5_CAP_GEN(mdev, pcam_reg) && \
@@ -60,8 +61,9 @@ struct mlx5e_bufferx_reg {
struct mlx5e_port_buffer {
u32 port_buffer_size;
u32 spare_buffer_size;
- u32 headroom_size;
- struct mlx5e_bufferx_reg buffer[MLX5E_MAX_BUFFER];
+ u32 headroom_size; /* Buffers 0-7 */
+ u32 internal_buffers_size; /* Buffers 8-9 */
+ struct mlx5e_bufferx_reg buffer[MLX5E_MAX_NETWORK_BUFFER];
};
int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
index 89de92d06483..ebee52a8361a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
@@ -926,9 +926,10 @@ static int mlx5e_dcbnl_getbuffer(struct net_device *dev,
if (err)
return err;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++)
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++)
dcb_buffer->buffer_size[i] = port_buffer.buffer[i].size;
- dcb_buffer->total_size = port_buffer.port_buffer_size;
+ dcb_buffer->total_size = port_buffer.port_buffer_size -
+ port_buffer.internal_buffers_size;
return 0;
}
@@ -970,7 +971,7 @@ static int mlx5e_dcbnl_setbuffer(struct net_device *dev,
if (err)
return err;
- for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+ for (i = 0; i < MLX5E_MAX_NETWORK_BUFFER; i++) {
if (port_buffer.buffer[i].size != dcb_buffer->buffer_size[i]) {
changed |= MLX5E_PORT_BUFFER_SIZE;
buffer_size = dcb_buffer->buffer_size;
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 04/17] net/mlx5e: Do not update SBCM when prio2buffer command is invalid
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (2 preceding siblings ...)
2023-05-25 3:48 ` [net 03/17] net/mlx5e: Consider internal buffers size in port buffer calculations Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 05/17] net/mlx5: Drain health before unregistering devlink Saeed Mahameed
` (12 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Maher Sanalla,
Moshe Shemesh
From: Maher Sanalla <msanalla@nvidia.com>
The shared buffer pools configuration which are stored in the SBCM
register are updated when the user changes the prio2buffer mapping.
However, in case the user desired prio2buffer change is invalid,
which can occur due to mapping a lossless priority to a not large enough
buffer, the SBCM update should not be performed, as the user command is
failed.
Thus, Perform the SBCM update only after xoff threshold calculation is
performed and the user prio2buffer mapping is validated.
Fixes: a440030d8946 ("net/mlx5e: Update shared buffer along with device buffer changes")
Signed-off-by: Maher Sanalla <msanalla@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
index 0d78527451bc..7e8e96cc5cd0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
@@ -442,11 +442,11 @@ static int update_buffer_lossy(struct mlx5_core_dev *mdev,
}
if (changed) {
- err = port_update_pool_cfg(mdev, port_buffer);
+ err = update_xoff_threshold(port_buffer, xoff, max_mtu, port_buff_cell_sz);
if (err)
return err;
- err = update_xoff_threshold(port_buffer, xoff, max_mtu, port_buff_cell_sz);
+ err = port_update_pool_cfg(mdev, port_buffer);
if (err)
return err;
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 05/17] net/mlx5: Drain health before unregistering devlink
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (3 preceding siblings ...)
2023-05-25 3:48 ` [net 04/17] net/mlx5e: Do not update SBCM when prio2buffer command is invalid Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 06/17] net/mlx5: SF, Drain health before removing device Saeed Mahameed
` (11 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Shay Drory, Moshe Shemesh
From: Shay Drory <shayd@nvidia.com>
mlx5 health mechanism is using devlink APIs, which are using devlink
notify APIs. After the cited patch, using devlink notify APIs after
devlink is unregistered triggers a WARN_ON().
Hence, drain health WQ before devlink is unregistered.
Fixes: cf530217408e ("devlink: Notify users when objects are accessible")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/main.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index a7eb65cd0bdd..2132a6510639 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -1802,15 +1802,16 @@ static void remove_one(struct pci_dev *pdev)
struct devlink *devlink = priv_to_devlink(dev);
set_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state);
- /* mlx5_drain_fw_reset() is using devlink APIs. Hence, we must drain
- * fw_reset before unregistering the devlink.
+ /* mlx5_drain_fw_reset() and mlx5_drain_health_wq() are using
+ * devlink notify APIs.
+ * Hence, we must drain them before unregistering the devlink.
*/
mlx5_drain_fw_reset(dev);
+ mlx5_drain_health_wq(dev);
devlink_unregister(devlink);
mlx5_sriov_disable(pdev);
mlx5_thermal_uninit(dev);
mlx5_crdump_disable(dev);
- mlx5_drain_health_wq(dev);
mlx5_uninit_one(dev);
mlx5_pci_close(dev);
mlx5_mdev_uninit(dev);
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 06/17] net/mlx5: SF, Drain health before removing device
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (4 preceding siblings ...)
2023-05-25 3:48 ` [net 05/17] net/mlx5: Drain health before unregistering devlink Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 07/17] net/mlx5: fw_tracer, Fix event handling Saeed Mahameed
` (10 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Shay Drory, Moshe Shemesh
From: Shay Drory <shayd@nvidia.com>
There is no point in recovery during device removal. Also, if health
work started need to wait for it to avoid races and NULL pointer
access.
Hence, drain health WQ before removing device.
Fixes: 1958fc2f0712 ("net/mlx5: SF, Add auxiliary device driver")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
index e2f26d0bc615..0692363cf80e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
@@ -63,6 +63,7 @@ static void mlx5_sf_dev_remove(struct auxiliary_device *adev)
struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
struct devlink *devlink = priv_to_devlink(sf_dev->mdev);
+ mlx5_drain_health_wq(sf_dev->mdev);
devlink_unregister(devlink);
mlx5_uninit_one(sf_dev->mdev);
iounmap(sf_dev->mdev->iseg);
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 07/17] net/mlx5: fw_tracer, Fix event handling
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (5 preceding siblings ...)
2023-05-25 3:48 ` [net 06/17] net/mlx5: SF, Drain health before removing device Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 08/17] net/mlx5e: Use query_special_contexts cmd only once per mdev Saeed Mahameed
` (9 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Shay Drory, Moshe Shemesh
From: Shay Drory <shayd@nvidia.com>
mlx5 driver needs to parse traces with event_id inside the range of
first_string_trace and num_string_trace. However, mlx5 is parsing all
events with event_id >= first_string_trace.
Fix it by checking for the correct range.
Fixes: c71ad41ccb0c ("net/mlx5: FW tracer, events handling")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
index f40497823e65..7c0f2adbea00 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
@@ -490,7 +490,7 @@ static void poll_trace(struct mlx5_fw_tracer *tracer,
(u64)timestamp_low;
break;
default:
- if (tracer_event->event_id >= tracer->str_db.first_string_trace ||
+ if (tracer_event->event_id >= tracer->str_db.first_string_trace &&
tracer_event->event_id <= tracer->str_db.first_string_trace +
tracer->str_db.num_string_trace) {
tracer_event->type = TRACER_EVENT_TYPE_STRING;
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 08/17] net/mlx5e: Use query_special_contexts cmd only once per mdev
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (6 preceding siblings ...)
2023-05-25 3:48 ` [net 07/17] net/mlx5: fw_tracer, Fix event handling Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 09/17] net/mlx5: Fix post parse infra to only parse every action once Saeed Mahameed
` (8 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Dragos Tatulea, Or Har-Toov
From: Dragos Tatulea <dtatulea@nvidia.com>
Don't query the firmware so many times (num rqs * num wqes * wqe frags)
because it slows down linearly the interface creation time when the
product is larger. Do it only once per mdev and store the result in
mlx5e_param.
Due to helper function being called from different files, move it to
an appropriate location. Rename the function with a proper prefix and
add a small cleanup.
This fix applies only for legacy rq.
Fixes: 1b1e4868836a ("net/mlx5e: Use query_special_contexts for mkeys")
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Or Har-Toov <ohartoov@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 1 +
.../net/ethernet/mellanox/mlx5/core/en_main.c | 24 +++----------------
drivers/net/ethernet/mellanox/mlx5/core/mr.c | 21 ++++++++++++++++
include/linux/mlx5/driver.h | 1 +
4 files changed, 26 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index b8987a404d75..8e999f238194 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -327,6 +327,7 @@ struct mlx5e_params {
unsigned int sw_mtu;
int hard_mtu;
bool ptp_rx;
+ __be32 terminate_lkey_be;
};
static inline u8 mlx5e_get_dcb_num_tc(struct mlx5e_params *params)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 2944691f06ad..0235adcbc609 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -727,26 +727,6 @@ static void mlx5e_rq_free_shampo(struct mlx5e_rq *rq)
mlx5e_rq_shampo_hd_free(rq);
}
-static __be32 mlx5e_get_terminate_scatter_list_mkey(struct mlx5_core_dev *dev)
-{
- u32 out[MLX5_ST_SZ_DW(query_special_contexts_out)] = {};
- u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {};
- int res;
-
- if (!MLX5_CAP_GEN(dev, terminate_scatter_list_mkey))
- return MLX5_TERMINATE_SCATTER_LIST_LKEY;
-
- MLX5_SET(query_special_contexts_in, in, opcode,
- MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS);
- res = mlx5_cmd_exec_inout(dev, query_special_contexts, in, out);
- if (res)
- return MLX5_TERMINATE_SCATTER_LIST_LKEY;
-
- res = MLX5_GET(query_special_contexts_out, out,
- terminate_scatter_list_mkey);
- return cpu_to_be32(res);
-}
-
static int mlx5e_alloc_rq(struct mlx5e_params *params,
struct mlx5e_xsk_param *xsk,
struct mlx5e_rq_param *rqp,
@@ -908,7 +888,7 @@ static int mlx5e_alloc_rq(struct mlx5e_params *params,
/* check if num_frags is not a pow of two */
if (rq->wqe.info.num_frags < (1 << rq->wqe.info.log_num_frags)) {
wqe->data[f].byte_count = 0;
- wqe->data[f].lkey = mlx5e_get_terminate_scatter_list_mkey(mdev);
+ wqe->data[f].lkey = params->terminate_lkey_be;
wqe->data[f].addr = 0;
}
}
@@ -5007,6 +4987,8 @@ void mlx5e_build_nic_params(struct mlx5e_priv *priv, struct mlx5e_xsk *xsk, u16
/* RQ */
mlx5e_build_rq_params(mdev, params);
+ params->terminate_lkey_be = mlx5_core_get_terminate_scatter_list_mkey(mdev);
+
params->packet_merge.timeout = mlx5e_choose_lro_timeout(mdev, MLX5E_DEFAULT_LRO_TIMEOUT);
/* CQ moderation params */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mr.c b/drivers/net/ethernet/mellanox/mlx5/core/mr.c
index 9d735c343a3b..678f0be81375 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mr.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mr.c
@@ -32,6 +32,7 @@
#include <linux/kernel.h>
#include <linux/mlx5/driver.h>
+#include <linux/mlx5/qp.h>
#include "mlx5_core.h"
int mlx5_core_create_mkey(struct mlx5_core_dev *dev, u32 *mkey, u32 *in,
@@ -122,3 +123,23 @@ int mlx5_core_destroy_psv(struct mlx5_core_dev *dev, int psv_num)
return mlx5_cmd_exec_in(dev, destroy_psv, in);
}
EXPORT_SYMBOL(mlx5_core_destroy_psv);
+
+__be32 mlx5_core_get_terminate_scatter_list_mkey(struct mlx5_core_dev *dev)
+{
+ u32 out[MLX5_ST_SZ_DW(query_special_contexts_out)] = {};
+ u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {};
+ u32 mkey;
+
+ if (!MLX5_CAP_GEN(dev, terminate_scatter_list_mkey))
+ return MLX5_TERMINATE_SCATTER_LIST_LKEY;
+
+ MLX5_SET(query_special_contexts_in, in, opcode,
+ MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS);
+ if (mlx5_cmd_exec_inout(dev, query_special_contexts, in, out))
+ return MLX5_TERMINATE_SCATTER_LIST_LKEY;
+
+ mkey = MLX5_GET(query_special_contexts_out, out,
+ terminate_scatter_list_mkey);
+ return cpu_to_be32(mkey);
+}
+EXPORT_SYMBOL(mlx5_core_get_terminate_scatter_list_mkey);
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index a4c4f737f9c1..94d2be5848ae 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -1093,6 +1093,7 @@ void mlx5_cmdif_debugfs_cleanup(struct mlx5_core_dev *dev);
int mlx5_core_create_psv(struct mlx5_core_dev *dev, u32 pdn,
int npsvs, u32 *sig_index);
int mlx5_core_destroy_psv(struct mlx5_core_dev *dev, int psv_num);
+__be32 mlx5_core_get_terminate_scatter_list_mkey(struct mlx5_core_dev *dev);
void mlx5_core_put_rsc(struct mlx5_core_rsc_common *common);
int mlx5_query_odp_caps(struct mlx5_core_dev *dev,
struct mlx5_odp_caps *odp_caps);
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 09/17] net/mlx5: Fix post parse infra to only parse every action once
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (7 preceding siblings ...)
2023-05-25 3:48 ` [net 08/17] net/mlx5e: Use query_special_contexts cmd only once per mdev Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 10/17] net/mlx5e: Don't attach netdev profile while handling internal error Saeed Mahameed
` (7 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Vlad Buslov, Roi Dayan
From: Vlad Buslov <vladbu@nvidia.com>
Caller of mlx5e_tc_act_post_parse() needs it to parse only the subset of
actions starting after previous split and ending at the current action.
However, that range is not provided as arguments and
mlx5e_tc_act_post_parse() uses generic flow_action_for_each() that iterates
over all flow actions. Not only this is redundant, it also causes a bug
when mlx5e_tc_act->post_parse() callback is not idempotent since it will be
called for every split. For example, ct action tc_act_post_parse_ct()
callback obtains a reference to mlx5_ct_ft instance and calling it several
times during parsing stage will cause reference counter imbalance.
Fix the issue by providing a proper action range of the current split
subset to mlx5e_tc_act_post_parse() and only calling
mlx5e_tc_act->post_parse() for actions inside the subset range.
Fixes: 8300f225268b ("net/mlx5e: Create new flow attr for multi table actions")
Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c | 7 ++++++-
drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 8 +++++---
3 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c
index fc923a99b6a4..0380a04c3691 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c
@@ -84,7 +84,7 @@ mlx5e_tc_act_init_parse_state(struct mlx5e_tc_act_parse_state *parse_state,
int
mlx5e_tc_act_post_parse(struct mlx5e_tc_act_parse_state *parse_state,
- struct flow_action *flow_action,
+ struct flow_action *flow_action, int from, int to,
struct mlx5_flow_attr *attr,
enum mlx5_flow_namespace_type ns_type)
{
@@ -96,6 +96,11 @@ mlx5e_tc_act_post_parse(struct mlx5e_tc_act_parse_state *parse_state,
priv = parse_state->flow->priv;
flow_action_for_each(i, act, flow_action) {
+ if (i < from)
+ continue;
+ else if (i > to)
+ break;
+
tc_act = mlx5e_tc_act_get(act->id, ns_type);
if (!tc_act || !tc_act->post_parse)
continue;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
index 0e6e1872ac62..d6c12d0ea55b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
@@ -112,7 +112,7 @@ mlx5e_tc_act_init_parse_state(struct mlx5e_tc_act_parse_state *parse_state,
int
mlx5e_tc_act_post_parse(struct mlx5e_tc_act_parse_state *parse_state,
- struct flow_action *flow_action,
+ struct flow_action *flow_action, int from, int to,
struct mlx5_flow_attr *attr,
enum mlx5_flow_namespace_type ns_type);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 8935156f8f4e..8a5a8703f0a3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -3859,8 +3859,8 @@ parse_tc_actions(struct mlx5e_tc_act_parse_state *parse_state,
struct mlx5_flow_attr *prev_attr;
struct flow_action_entry *act;
struct mlx5e_tc_act *tc_act;
+ int err, i, i_split = 0;
bool is_missable;
- int err, i;
ns_type = mlx5e_get_flow_namespace(flow);
list_add(&attr->list, &flow->attrs);
@@ -3901,7 +3901,8 @@ parse_tc_actions(struct mlx5e_tc_act_parse_state *parse_state,
i < flow_action->num_entries - 1)) {
is_missable = tc_act->is_missable ? tc_act->is_missable(act) : false;
- err = mlx5e_tc_act_post_parse(parse_state, flow_action, attr, ns_type);
+ err = mlx5e_tc_act_post_parse(parse_state, flow_action, i_split, i, attr,
+ ns_type);
if (err)
goto out_free_post_acts;
@@ -3911,6 +3912,7 @@ parse_tc_actions(struct mlx5e_tc_act_parse_state *parse_state,
goto out_free_post_acts;
}
+ i_split = i + 1;
list_add(&attr->list, &flow->attrs);
}
@@ -3925,7 +3927,7 @@ parse_tc_actions(struct mlx5e_tc_act_parse_state *parse_state,
}
}
- err = mlx5e_tc_act_post_parse(parse_state, flow_action, attr, ns_type);
+ err = mlx5e_tc_act_post_parse(parse_state, flow_action, i_split, i, attr, ns_type);
if (err)
goto out_free_post_acts;
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 10/17] net/mlx5e: Don't attach netdev profile while handling internal error
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (8 preceding siblings ...)
2023-05-25 3:48 ` [net 09/17] net/mlx5: Fix post parse infra to only parse every action once Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 11/17] net/mlx5e: Move Ethernet driver debugfs to profile init callback Saeed Mahameed
` (6 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Dmytro Linkin
From: Dmytro Linkin <dlinkin@nvidia.com>
As part of switchdev mode disablement, driver changes port netdevice
profile from uplink to nic. If this process is triggered by health
recovery flow (PCI reset, for ex.) profile attach would fail because all
fw commands aborted when internal error flag is set. As a result, nic
netdevice profile is not attached and driver fails to rollback to uplink
profile, which leave driver in broken state and cause crash later.
To handle broken state do netdevice profile initialization only instead
of full attachment and release mdev resources on driver suspend as
expected. Actual netdevice attachment is done during driver load.
Fixes: c4d7eb57687f ("net/mxl5e: Add change profile method")
Signed-off-by: Dmytro Linkin <dlinkin@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/en_main.c | 35 ++++++++++++++++---
1 file changed, 31 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 0235adcbc609..a07bbe9a61be 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5833,8 +5833,8 @@ void mlx5e_detach_netdev(struct mlx5e_priv *priv)
}
static int
-mlx5e_netdev_attach_profile(struct net_device *netdev, struct mlx5_core_dev *mdev,
- const struct mlx5e_profile *new_profile, void *new_ppriv)
+mlx5e_netdev_init_profile(struct net_device *netdev, struct mlx5_core_dev *mdev,
+ const struct mlx5e_profile *new_profile, void *new_ppriv)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
int err;
@@ -5850,6 +5850,25 @@ mlx5e_netdev_attach_profile(struct net_device *netdev, struct mlx5_core_dev *mde
err = new_profile->init(priv->mdev, priv->netdev);
if (err)
goto priv_cleanup;
+
+ return 0;
+
+priv_cleanup:
+ mlx5e_priv_cleanup(priv);
+ return err;
+}
+
+static int
+mlx5e_netdev_attach_profile(struct net_device *netdev, struct mlx5_core_dev *mdev,
+ const struct mlx5e_profile *new_profile, void *new_ppriv)
+{
+ struct mlx5e_priv *priv = netdev_priv(netdev);
+ int err;
+
+ err = mlx5e_netdev_init_profile(netdev, mdev, new_profile, new_ppriv);
+ if (err)
+ return err;
+
err = mlx5e_attach_netdev(priv);
if (err)
goto profile_cleanup;
@@ -5857,7 +5876,6 @@ mlx5e_netdev_attach_profile(struct net_device *netdev, struct mlx5_core_dev *mde
profile_cleanup:
new_profile->cleanup(priv);
-priv_cleanup:
mlx5e_priv_cleanup(priv);
return err;
}
@@ -5876,6 +5894,12 @@ int mlx5e_netdev_change_profile(struct mlx5e_priv *priv,
priv->profile->cleanup(priv);
mlx5e_priv_cleanup(priv);
+ if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) {
+ mlx5e_netdev_init_profile(netdev, mdev, new_profile, new_ppriv);
+ set_bit(MLX5E_STATE_DESTROYING, &priv->state);
+ return -EIO;
+ }
+
err = mlx5e_netdev_attach_profile(netdev, mdev, new_profile, new_ppriv);
if (err) { /* roll back to original profile */
netdev_warn(netdev, "%s: new profile init failed, %d\n", __func__, err);
@@ -5937,8 +5961,11 @@ static int mlx5e_suspend(struct auxiliary_device *adev, pm_message_t state)
struct net_device *netdev = priv->netdev;
struct mlx5_core_dev *mdev = priv->mdev;
- if (!netif_device_present(netdev))
+ if (!netif_device_present(netdev)) {
+ if (test_bit(MLX5E_STATE_DESTROYING, &priv->state))
+ mlx5e_destroy_mdev_resources(mdev);
return -ENODEV;
+ }
mlx5e_detach_netdev(priv);
mlx5e_destroy_mdev_resources(mdev);
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 11/17] net/mlx5e: Move Ethernet driver debugfs to profile init callback
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (9 preceding siblings ...)
2023-05-25 3:48 ` [net 10/17] net/mlx5e: Don't attach netdev profile while handling internal error Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 12/17] net/mlx5: DR, Add missing mutex init/destroy in pattern manager Saeed Mahameed
` (5 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Jianbo Liu
From: Jianbo Liu <jianbol@nvidia.com>
As priv->dfs_root is cleared, and therefore missed, when change
eswitch mode, move the creation of the root debugfs to the init
callback of mlx5e_nic_profile and mlx5e_uplink_rep_profile, and
the destruction to the cleanup callback for symmeter.
Fixes: 288eca60cc31 ("net/mlx5e: Add Ethernet driver debugfs")
Signed-off-by: Jianbo Liu <jianbol@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 10 +++++-----
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 6 ++++++
2 files changed, 11 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index a07bbe9a61be..a7c526ee5024 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5261,12 +5261,16 @@ static int mlx5e_nic_init(struct mlx5_core_dev *mdev,
mlx5e_timestamp_init(priv);
+ priv->dfs_root = debugfs_create_dir("nic",
+ mlx5_debugfs_get_dev_root(mdev));
+
fs = mlx5e_fs_init(priv->profile, mdev,
!test_bit(MLX5E_STATE_DESTROYING, &priv->state),
priv->dfs_root);
if (!fs) {
err = -ENOMEM;
mlx5_core_err(mdev, "FS initialization failed, %d\n", err);
+ debugfs_remove_recursive(priv->dfs_root);
return err;
}
priv->fs = fs;
@@ -5287,6 +5291,7 @@ static void mlx5e_nic_cleanup(struct mlx5e_priv *priv)
mlx5e_health_destroy_reporters(priv);
mlx5e_ktls_cleanup(priv);
mlx5e_fs_cleanup(priv->fs);
+ debugfs_remove_recursive(priv->dfs_root);
priv->fs = NULL;
}
@@ -6011,9 +6016,6 @@ static int mlx5e_probe(struct auxiliary_device *adev,
priv->profile = profile;
priv->ppriv = NULL;
- priv->dfs_root = debugfs_create_dir("nic",
- mlx5_debugfs_get_dev_root(priv->mdev));
-
err = profile->init(mdev, netdev);
if (err) {
mlx5_core_err(mdev, "mlx5e_nic_profile init failed, %d\n", err);
@@ -6042,7 +6044,6 @@ static int mlx5e_probe(struct auxiliary_device *adev,
err_profile_cleanup:
profile->cleanup(priv);
err_destroy_netdev:
- debugfs_remove_recursive(priv->dfs_root);
mlx5e_destroy_netdev(priv);
err_devlink_port_unregister:
mlx5e_devlink_port_unregister(mlx5e_dev);
@@ -6062,7 +6063,6 @@ static void mlx5e_remove(struct auxiliary_device *adev)
unregister_netdev(priv->netdev);
mlx5e_suspend(adev, state);
priv->profile->cleanup(priv);
- debugfs_remove_recursive(priv->dfs_root);
mlx5e_destroy_netdev(priv);
mlx5e_devlink_port_unregister(mlx5e_dev);
mlx5e_destroy_devlink(mlx5e_dev);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 1fc386eccaf8..3e7041bd5705 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -30,6 +30,7 @@
* SOFTWARE.
*/
+#include <linux/debugfs.h>
#include <linux/mlx5/fs.h>
#include <net/switchdev.h>
#include <net/pkt_cls.h>
@@ -812,11 +813,15 @@ static int mlx5e_init_ul_rep(struct mlx5_core_dev *mdev,
{
struct mlx5e_priv *priv = netdev_priv(netdev);
+ priv->dfs_root = debugfs_create_dir("nic",
+ mlx5_debugfs_get_dev_root(mdev));
+
priv->fs = mlx5e_fs_init(priv->profile, mdev,
!test_bit(MLX5E_STATE_DESTROYING, &priv->state),
priv->dfs_root);
if (!priv->fs) {
netdev_err(priv->netdev, "FS allocation failed\n");
+ debugfs_remove_recursive(priv->dfs_root);
return -ENOMEM;
}
@@ -829,6 +834,7 @@ static int mlx5e_init_ul_rep(struct mlx5_core_dev *mdev,
static void mlx5e_cleanup_rep(struct mlx5e_priv *priv)
{
mlx5e_fs_cleanup(priv->fs);
+ debugfs_remove_recursive(priv->dfs_root);
priv->fs = NULL;
}
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 12/17] net/mlx5: DR, Add missing mutex init/destroy in pattern manager
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (10 preceding siblings ...)
2023-05-25 3:48 ` [net 11/17] net/mlx5e: Move Ethernet driver debugfs to profile init callback Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 13/17] net/mlx5: Fix check for allocation failure in comp_irqs_request_pci() Saeed Mahameed
` (4 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Yevgeny Kliteynik
From: Yevgeny Kliteynik <kliteyn@nvidia.com>
Add missing mutex init/destroy as caught by the lock's debug warning:
DEBUG_LOCKS_WARN_ON(lock->magic != lock)
Fixes: da5d0027d666 ("net/mlx5: DR, Add cache for modify header pattern")
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ptrn.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ptrn.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ptrn.c
index 13e06a6a6b22..d6947fe13d56 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ptrn.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ptrn.c
@@ -213,6 +213,8 @@ struct mlx5dr_ptrn_mgr *mlx5dr_ptrn_mgr_create(struct mlx5dr_domain *dmn)
}
INIT_LIST_HEAD(&mgr->ptrn_list);
+ mutex_init(&mgr->modify_hdr_mutex);
+
return mgr;
free_mgr:
@@ -237,5 +239,6 @@ void mlx5dr_ptrn_mgr_destroy(struct mlx5dr_ptrn_mgr *mgr)
}
mlx5dr_icm_pool_destroy(mgr->ptrn_icm_pool);
+ mutex_destroy(&mgr->modify_hdr_mutex);
kfree(mgr);
}
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 13/17] net/mlx5: Fix check for allocation failure in comp_irqs_request_pci()
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (11 preceding siblings ...)
2023-05-25 3:48 ` [net 12/17] net/mlx5: DR, Add missing mutex init/destroy in pattern manager Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 14/17] Documentation: net/mlx5: Wrap vnic reporter devlink commands in code blocks Saeed Mahameed
` (3 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Dan Carpenter,
kernel test robot, Simon Horman, Leon Romanovsky
From: Dan Carpenter <dan.carpenter@linaro.org>
This function accidentally dereferences "cpus" instead of returning
directly.
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/r/202305200354.KV3jU94w-lkp@intel.com/
Fixes: b48a0f72bc3e ("net/mlx5: Refactor completion irq request/release code")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/eq.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
index fe698c79616c..3db4866d7880 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -824,7 +824,7 @@ static int comp_irqs_request_pci(struct mlx5_core_dev *dev)
ncomp_eqs = table->num_comp_eqs;
cpus = kcalloc(ncomp_eqs, sizeof(*cpus), GFP_KERNEL);
if (!cpus)
- ret = -ENOMEM;
+ return -ENOMEM;
i = 0;
rcu_read_lock();
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 14/17] Documentation: net/mlx5: Wrap vnic reporter devlink commands in code blocks
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (12 preceding siblings ...)
2023-05-25 3:48 ` [net 13/17] net/mlx5: Fix check for allocation failure in comp_irqs_request_pci() Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 15/17] Documentation: net/mlx5: Use bullet and definition lists for vnic counters description Saeed Mahameed
` (2 subsequent siblings)
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Bagas Sanjaya,
Leon Romanovsky, Simon Horman
From: Bagas Sanjaya <bagasdotme@gmail.com>
Sphinx reports htmldocs warnings:
Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst:287: WARNING: Unexpected indentation.
Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst:288: WARNING: Block quote ends without a blank line; unexpected unindent.
Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst:290: WARNING: Unexpected indentation.
Fix above warnings by wrapping diagnostic devlink commands in "vnic
reporter" section in code blocks to be consistent with other devlink
command snippets.
Fixes: b0bc615df488ab ("net/mlx5: Add vnic devlink health reporter to PFs/VFs")
Fixes: cf14af140a5ad0 ("net/mlx5e: Add vnic devlink health reporter to representors")
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../device_drivers/ethernet/mellanox/mlx5/devlink.rst | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
index 3a7a714cc08f..0f0598caea14 100644
--- a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
+++ b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
@@ -283,10 +283,14 @@ nic_receive_steering_discard: number of packets that completed RX flow
steering but were discarded due to a mismatch in flow table.
User commands examples:
-- Diagnose PF/VF vnic counters
+
+- Diagnose PF/VF vnic counters::
+
$ devlink health diagnose pci/0000:82:00.1 reporter vnic
+
- Diagnose representor vnic counters (performed by supplying devlink port of the
- representor, which can be obtained via devlink port command)
+ representor, which can be obtained via devlink port command)::
+
$ devlink health diagnose pci/0000:82:00.1/65537 reporter vnic
NOTE: This command can run over all interfaces such as PF/VF and representor ports.
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 15/17] Documentation: net/mlx5: Use bullet and definition lists for vnic counters description
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (13 preceding siblings ...)
2023-05-25 3:48 ` [net 14/17] Documentation: net/mlx5: Wrap vnic reporter devlink commands in code blocks Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 16/17] Documentation: net/mlx5: Add blank line separator before numbered lists Saeed Mahameed
2023-05-25 3:48 ` [net 17/17] Documentation: net/mlx5: Wrap notes in admonition blocks Saeed Mahameed
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Bagas Sanjaya,
Leon Romanovsky, Simon Horman, Daniel Machon
From: Bagas Sanjaya <bagasdotme@gmail.com>
"vnic reporter" section contains unformatted description for vnic
counters, which is rendered as one long paragraph instead of list.
Use bullet and definition lists to match other lists.
Fixes: b0bc615df488ab ("net/mlx5: Add vnic devlink health reporter to PFs/VFs")
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Daniel Machon <daniel.machon@microchip.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../ethernet/mellanox/mlx5/devlink.rst | 36 ++++++++++---------
1 file changed, 20 insertions(+), 16 deletions(-)
diff --git a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
index 0f0598caea14..00687425d8b7 100644
--- a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
+++ b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
@@ -265,22 +265,26 @@ It is responsible for querying the vnic diagnostic counters from fw and displayi
them in realtime.
Description of the vnic counters:
-total_q_under_processor_handle: number of queues in an error state due to
-an async error or errored command.
-send_queue_priority_update_flow: number of QP/SQ priority/SL update
-events.
-cq_overrun: number of times CQ entered an error state due to an
-overflow.
-async_eq_overrun: number of times an EQ mapped to async events was
-overrun.
-comp_eq_overrun: number of times an EQ mapped to completion events was
-overrun.
-quota_exceeded_command: number of commands issued and failed due to quota
-exceeded.
-invalid_command: number of commands issued and failed dues to any reason
-other than quota exceeded.
-nic_receive_steering_discard: number of packets that completed RX flow
-steering but were discarded due to a mismatch in flow table.
+
+- total_q_under_processor_handle
+ number of queues in an error state due to
+ an async error or errored command.
+- send_queue_priority_update_flow
+ number of QP/SQ priority/SL update events.
+- cq_overrun
+ number of times CQ entered an error state due to an overflow.
+- async_eq_overrun
+ number of times an EQ mapped to async events was overrun.
+ comp_eq_overrun number of times an EQ mapped to completion events was
+ overrun.
+- quota_exceeded_command
+ number of commands issued and failed due to quota exceeded.
+- invalid_command
+ number of commands issued and failed dues to any reason other than quota
+ exceeded.
+- nic_receive_steering_discard
+ number of packets that completed RX flow
+ steering but were discarded due to a mismatch in flow table.
User commands examples:
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 16/17] Documentation: net/mlx5: Add blank line separator before numbered lists
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (14 preceding siblings ...)
2023-05-25 3:48 ` [net 15/17] Documentation: net/mlx5: Use bullet and definition lists for vnic counters description Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
2023-05-25 3:48 ` [net 17/17] Documentation: net/mlx5: Wrap notes in admonition blocks Saeed Mahameed
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Bagas Sanjaya,
Leon Romanovsky, Simon Horman
From: Bagas Sanjaya <bagasdotme@gmail.com>
The doc forgets to add separator before numbered lists, which causes the
lists to be appended to previous paragraph inline instead.
Add the missing separator.
Fixes: f2d51e579359b7 ("net/mlx5: Separate mlx5 driver documentation into multiple pages")
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../device_drivers/ethernet/mellanox/mlx5/devlink.rst | 2 ++
1 file changed, 2 insertions(+)
diff --git a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
index 00687425d8b7..f962c0975d84 100644
--- a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
+++ b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
@@ -40,6 +40,7 @@ flow_steering_mode: Device flow steering mode
---------------------------------------------
The flow steering mode parameter controls the flow steering mode of the driver.
Two modes are supported:
+
1. 'dmfs' - Device managed flow steering.
2. 'smfs' - Software/Driver managed flow steering.
@@ -99,6 +100,7 @@ between representors and stacked devices.
By default metadata is enabled on the supported devices in E-switch.
Metadata is applicable only for E-switch in switchdev mode and
users may disable it when NONE of the below use cases will be in use:
+
1. HCA is in Dual/multi-port RoCE mode.
2. VF/SF representor bonding (Usually used for Live migration)
3. Stacked devices
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread* [net 17/17] Documentation: net/mlx5: Wrap notes in admonition blocks
2023-05-25 3:48 [pull request][net 00/17] mlx5 fixes 2023-05-24 Saeed Mahameed
` (15 preceding siblings ...)
2023-05-25 3:48 ` [net 16/17] Documentation: net/mlx5: Add blank line separator before numbered lists Saeed Mahameed
@ 2023-05-25 3:48 ` Saeed Mahameed
16 siblings, 0 replies; 19+ messages in thread
From: Saeed Mahameed @ 2023-05-25 3:48 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: Saeed Mahameed, netdev, Tariq Toukan, Bagas Sanjaya,
Leon Romanovsky
From: Bagas Sanjaya <bagasdotme@gmail.com>
Wrap note paragraphs in note:: directive as it better fit for the
purpose of noting devlink commands.
Fixes: f2d51e579359b7 ("net/mlx5: Separate mlx5 driver documentation into multiple pages")
Fixes: cf14af140a5ad0 ("net/mlx5e: Add vnic devlink health reporter to representors")
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../ethernet/mellanox/mlx5/devlink.rst | 14 +++++++++-----
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
index f962c0975d84..3354ca3608ee 100644
--- a/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
+++ b/Documentation/networking/device_drivers/ethernet/mellanox/mlx5/devlink.rst
@@ -182,7 +182,8 @@ User commands examples:
$ devlink health diagnose pci/0000:82:00.0 reporter tx
-NOTE: This command has valid output only when interface is up, otherwise the command has empty output.
+.. note::
+ This command has valid output only when interface is up, otherwise the command has empty output.
- Show number of tx errors indicated, number of recover flows ended successfully,
is autorecover enabled and graceful period from last recover::
@@ -234,8 +235,9 @@ User commands examples:
$ devlink health dump show pci/0000:82:00.0 reporter fw
-NOTE: This command can run only on the PF which has fw tracer ownership,
-running it on other PF or any VF will return "Operation not permitted".
+.. note::
+ This command can run only on the PF which has fw tracer ownership,
+ running it on other PF or any VF will return "Operation not permitted".
fw fatal reporter
-----------------
@@ -258,7 +260,8 @@ User commands examples:
$ devlink health dump show pci/0000:82:00.1 reporter fw_fatal
-NOTE: This command can run only on PF.
+.. note::
+ This command can run only on PF.
vnic reporter
-------------
@@ -299,4 +302,5 @@ User commands examples:
$ devlink health diagnose pci/0000:82:00.1/65537 reporter vnic
-NOTE: This command can run over all interfaces such as PF/VF and representor ports.
+.. note::
+ This command can run over all interfaces such as PF/VF and representor ports.
--
2.40.1
^ permalink raw reply related [flat|nested] 19+ messages in thread