Skip to contentSkip to conversation

netdevsim: tc: serialize access to nsim_block_cb_list

PATCH In review Weiming Shi 3 participants 6 messages Jul 19 – Aug 3

Subsystems (7)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 netdevsim: tc: serialize access to nsim_block_cb_list

nsim_setup_tc() passes a single global nsim_block_cb_list, shared by every netdevsim device, to flow_block_cb_setup_simple(). That helper does list_add_tail()/list_del() on the list on block bind/unbind and walks it in flow_block_cb_is_busy(); it takes no lock of its own and relies on the caller for serialization.

The tc control path calls ndo_setup_tc(TC_SETUP_BLOCK) under rtnl, but the nf_tables hardware offload path reaches it under the per-netns nftables commit_mutex only. Two nft transactions committing offload chains from different netns thus run flow_block_cb_setup_simple() on the same list concurrently, corrupting it and freeing a flow_block_cb that the other CPU still walks:

list_del corruption. prev->next should be ffff88801f18ff00, but was ffff88801de82b00. kernel BUG at lib/list_debug.c:62!

 RIP: 0010:__list_del_entry_valid_or_report (lib/list_debug.c:62)
  flow_block_cb_setup_simple (net/core/flow_offload.c:369)
  nsim_setup_tc (drivers/net/netdevsim/tc.c)
  nft_block_offload_cmd (net/netfilter/nf_tables_offload.c:394)
  nft_flow_rule_offload_commit (net/netfilter/nf_tables_offload.c:585)
  nf_tables_commit (net/netfilter/nf_tables_api.c:10490)
  nfnetlink_rcv_batch (net/netfilter/nfnetlink.c:577)
  nfnetlink_rcv (net/netfilter/nfnetlink.c:649)
  netlink_unicast (net/netlink/af_netlink.c:1314)
  netlink_sendmsg (net/netlink/af_netlink.c:1889)
  __sys_sendto (net/socket.c:729)

With KASAN the same race is reported as a slab-use-after-free read of the freed flow_block_cb (kmalloc-192) in flow_block_cb_setup_simple(). The trace above is from a 6.12.y reproduction.

Serialize the list with a mutex. ndo_setup_tc(TC_SETUP_BLOCK) always runs in process context, so sleeping on the mutex is fine.

Fixes 955bcb6ea0df ("drivers: net: use flow block API")
Assisted-by Claude:claude-opus-4-8
---
 drivers/net/netdevsim/tc.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
drivers/net/netdevsim/tc.c +9 -4
index a415e02a6df1..30dd7f924b71 100644
@@ -72,11 +72,13 @@ static int nsim_setup_tc_ets(struct net_device *dev,
 }
 
 static LIST_HEAD(nsim_block_cb_list);
+static DEFINE_MUTEX(nsim_block_cb_lock);
 
 int
 nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
 {
 	struct netdevsim *ns = netdev_priv(dev);
+	int err;
 
 	switch (type) {
 	case TC_SETUP_QDISC_TAPRIO:
@@ -84,10 +86,13 @@ nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
 	case TC_SETUP_QDISC_ETS:
 		return nsim_setup_tc_ets(dev, type_data);
 	case TC_SETUP_BLOCK:
-		return flow_block_cb_setup_simple(type_data,
-						  &nsim_block_cb_list,
-						  nsim_setup_tc_block_cb,
-						  ns, ns, true);
+		mutex_lock(&nsim_block_cb_lock);
+		err = flow_block_cb_setup_simple(type_data,
+						 &nsim_block_cb_list,
+						 nsim_setup_tc_block_cb,
+						 ns, ns, true);
+		mutex_unlock(&nsim_block_cb_lock);
+		return err;
 	case TC_SETUP_FT:
 		return 0;
 	default:
Signature · 1 line
2.43.0
reviewer
#1 lore raw

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
Full review at: https://sashiko.dev/#/patchset/[email protected]?part=1
---
--- Patch [1]: [PATCH net] netdevsim: tc: serialize access to nsim_block_cb_list --- commit 27d18d348023b46d852b44b24e6f4ca81ab39bf3
Author: Weiming Shi <[email protected]>

netdevsim: tc: serialize access to nsim_block_cb_list

This commit attempts to fix a list corruption and slab-use-after-free race condition between the nftables hardware offload path and the TC control path. It adds a local mutex to netdevsim to serialize access to the global nsim_block_cb_list during flow_block_cb_setup_simple().

Quoted 18 lines
drivers/net/netdevsim/tc.c +2 -0
index a415e02a6df1e..30dd7f924b716 100644
@@ -72,11 +72,13 @@ static int nsim_setup_tc_ets(struct net_device *dev,
 }
 
 static LIST_HEAD(nsim_block_cb_list);
+static DEFINE_MUTEX(nsim_block_cb_lock);
 
 int
 nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
 {
 	struct netdevsim *ns = netdev_priv(dev);
+	int err;
 
 	switch (type) {
 	case TC_SETUP_QDISC_TAPRIO:

[ ... ]

drivers/net/netdevsim/tc.c nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data) · line 89

Quoted 9 lines
drivers/net/netdevsim/tc.c +1 -4
@@ -84,10 +86,13 @@ nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
 	case TC_SETUP_QDISC_ETS:
 		return nsim_setup_tc_ets(dev, type_data);
 	case TC_SETUP_BLOCK:
-		return flow_block_cb_setup_simple(type_data,
-						  &nsim_block_cb_list,
-						  nsim_setup_tc_block_cb,
-						  ns, ns, true);
+		mutex_lock(&nsim_block_cb_lock);

Simon says: I'm wondering if you could take a look at the issue flagged
below as it seems to undermine the fix proposed by this patch.

[Severity: High]
This isn't a bug introduced by this patch, but will the new nsim_block_cb_lock actually fully serialize access to the list?

If tcf_block_bind() fails in the core TC stack, its error unroll path directly executes list_del(&block_cb->driver_list) while holding only rtnl_lock.

If this error unroll happens concurrently with an nftables transaction that holds the commit_mutex and the newly introduced nsim_block_cb_lock, the core TC stack will modify nsim_block_cb_list completely bypassing the driver's local lock.

Could this still lead to list corruption and slab-use-after-free bugs when the concurrent nftables transaction walks the mutating list?

drivers/net/netdevsim/tc.c nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data) · line 90

Quoted 1 line
drivers/net/netdevsim/tc.c +1 -0
+		err = flow_block_cb_setup_simple(type_data,

Simon says: I think the problem flagged below can be considered
in the context of possible follow-up.

[Severity: High]
This is a pre-existing issue, but does adding a local lock in netdevsim mask the underlying systemic race?

flow_block_cb_setup_simple() modifies driver-provided lists directly without internal locking. Since the nftables hardware offload path runs without rtnl_lock, unprivileged users can configure nftables concurrently and trigger list corruption.

While this patch addresses the issue for netdevsim locally, won't this same vulnerability persist across other hardware drivers (e.g., cxgb4, prestera, i40e) that use TC_SETUP_BLOCK with statically allocated global lists?

Quoted 8 lines
drivers/net/netdevsim/tc.c +5 -0
+						 &nsim_block_cb_list,
+						 nsim_setup_tc_block_cb,
+						 ns, ns, true);
+		mutex_unlock(&nsim_block_cb_lock);
+		return err;
 	case TC_SETUP_FT:
 		return 0;
 	default:
#2 lore raw
Quoted 75 lines · Simon Horman <[email protected]> 于2026年7月24日周五 22:23写道:

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
Full review at: https://sashiko.dev/#/patchset/[email protected]?part=1
---
--- Patch [1]: [PATCH net] netdevsim: tc: serialize access to nsim_block_cb_list --- commit 27d18d348023b46d852b44b24e6f4ca81ab39bf3
Author: Weiming Shi <[email protected]>

netdevsim: tc: serialize access to nsim_block_cb_list

This commit attempts to fix a list corruption and slab-use-after-free race condition between the nftables hardware offload path and the TC control path. It adds a local mutex to netdevsim to serialize access to the global nsim_block_cb_list during flow_block_cb_setup_simple().

Quoted 18 lines
drivers/net/netdevsim/tc.c +2 -0
index a415e02a6df1e..30dd7f924b716 100644
@@ -72,11 +72,13 @@ static int nsim_setup_tc_ets(struct net_device *dev,
 }
 
 static LIST_HEAD(nsim_block_cb_list);
+static DEFINE_MUTEX(nsim_block_cb_lock);
 
 int
 nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
 {
      struct netdevsim *ns = netdev_priv(dev);
+     int err;
 
      switch (type) {
      case TC_SETUP_QDISC_TAPRIO:

[ ... ]

drivers/net/netdevsim/tc.c nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data) · line 89

Quoted 9 lines
drivers/net/netdevsim/tc.c +1 -4
@@ -84,10 +86,13 @@ nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data)
      case TC_SETUP_QDISC_ETS:
              return nsim_setup_tc_ets(dev, type_data);
      case TC_SETUP_BLOCK:
-             return flow_block_cb_setup_simple(type_data,
-                                               &nsim_block_cb_list,
-                                               nsim_setup_tc_block_cb,
-                                               ns, ns, true);
+             mutex_lock(&nsim_block_cb_lock);

Simon says: I'm wondering if you could take a look at the issue flagged
below as it seems to undermine the fix proposed by this patch.

[Severity: High]
This isn't a bug introduced by this patch, but will the new nsim_block_cb_lock actually fully serialize access to the list?

If tcf_block_bind() fails in the core TC stack, its error unroll path directly executes list_del(&block_cb->driver_list) while holding only rtnl_lock.

If this error unroll happens concurrently with an nftables transaction that holds the commit_mutex and the newly introduced nsim_block_cb_lock, the core TC stack will modify nsim_block_cb_list completely bypassing the driver's local lock.

Could this still lead to list corruption and slab-use-after-free bugs when the concurrent nftables transaction walks the mutating list?

drivers/net/netdevsim/tc.c nsim_setup_tc(struct net_device *dev, enum tc_setup_type type, void *type_data) · line 90

Quoted 1 line
drivers/net/netdevsim/tc.c +1 -0
+             err = flow_block_cb_setup_simple(type_data,

Simon says: I think the problem flagged below can be considered
in the context of possible follow-up.

[Severity: High]
This is a pre-existing issue, but does adding a local lock in netdevsim mask the underlying systemic race?

Quoted 17 lines

flow_block_cb_setup_simple() modifies driver-provided lists directly without internal locking. Since the nftables hardware offload path runs without rtnl_lock, unprivileged users can configure nftables concurrently and trigger list corruption.

While this patch addresses the issue for netdevsim locally, won't this same vulnerability persist across other hardware drivers (e.g., cxgb4, prestera, i40e) that use TC_SETUP_BLOCK with statically allocated global lists?

Quoted 8 lines
drivers/net/netdevsim/tc.c +5 -0
+                                              &nsim_block_cb_list,
+                                              nsim_setup_tc_block_cb,
+                                              ns, ns, true);
+             mutex_unlock(&nsim_block_cb_lock);
+             return err;
      case TC_SETUP_FT:
              return 0;
      default:

Hi Simon,

Thanks for the feedback. That makes sense, and I agree with the concern. I'll work on a v2 patch to address this.

Thanks,
Weiming

#3 lore raw

PATCH v2 1/1 net: flow_offload: serialize driver callback lists

In reply to Weiming Shi

Drivers keep flow_block_cb objects on callback lists which are often shared globally. TC setup normally runs under RTNL, but nftables invokes ndo_setup_tc() while holding only its per-netns commit mutex. Concurrent transactions in different netns can therefore corrupt a shared driver callback list.

The same list can also be changed after ndo_setup_tc() returns. If filter replay fails while binding a TC block, tcf_block_bind() removes and frees callbacks from its error path without synchronizing against an nftables transaction walking the driver list.

Add a flow-offload callback mutex and hold it across direct and indirect setup, callback ownership transfer and rollback. Move indirect-unregister list removal into the flow-offload core, and keep indirect-device registration transactional on setup failure. When TC must reject an indirect bind after callback installation, unwind only the callbacks created by that bind so existing callbacks remain intact.

Initialize the auxiliary callback list nodes so common error paths can safely detach them. ENETC also inspects its callback list outside the setup callback, so protect those accesses with the same mutex.

This fixes the following netdevsim failure:

list_del corruption. next->prev should be ffff88800b3e30c0, but was ffff88800b539e40.

  WARNING: CPU: 0 PID: 79 at lib/list_debug.c:65
  RIP: __list_del_entry_valid_or_report
  flow_block_cb_setup_simple
  nft_block_offload_cmd
  nft_flow_rule_offload_commit
  nf_tables_commit
  nfnetlink_rcv_batch
Fixes c9626a2cbdb2 ("netfilter: nf_tables: add hardware offload support")
Assisted-by Claude:claude-opus-4-8

---

v2:
- Move serialization to common flow-offload paths, including rollback and indirect cleanup.
- Preserve existing callbacks when an indirect bind fails.

v1: https://lore.kernel.org/netdev/[email protected]/
review: https://lore.kernel.org/netdev/[email protected]/

 .../net/ethernet/freescale/enetc/enetc_qos.c  | 27 ++++--
 include/net/flow_offload.h                    |  6 +-
 net/core/flow_offload.c                       | 77 +++++++++++++++--
 net/netfilter/nf_flow_table_offload.c         | 28 +++---
 net/netfilter/nf_tables_offload.c             | 18 +++-
 net/sched/act_api.c                           |  4 +-
 net/sched/cls_api.c                           | 86 ++++++++++++++++---
 7 files changed, 201 insertions(+), 45 deletions(-)
drivers/net/ethernet/freescale/enetc/enetc_qos.c +19 -8
index 7b17bca24f26..c994f878cacd 100644
@@ -1601,30 +1601,41 @@ int enetc_set_psfp(struct net_device *ndev, bool en)
 
 int enetc_psfp_init(struct enetc_ndev_priv *priv)
 {
+	int err = 0;
+
+	flow_block_cb_lock();
 	if (epsfp.psfp_sfi_bitmap)
-		return 0;
+		goto out;
 
 	epsfp.psfp_sfi_bitmap = bitmap_zalloc(priv->psfp_cap.max_psfp_filter,
 					      GFP_KERNEL);
-	if (!epsfp.psfp_sfi_bitmap)
-		return -ENOMEM;
+	if (!epsfp.psfp_sfi_bitmap) {
+		err = -ENOMEM;
+		goto out;
+	}
 
 	spin_lock_init(&epsfp.psfp_lock);
 
 	if (list_empty(&enetc_block_cb_list))
 		epsfp.dev_bitmap = 0;
 
-	return 0;
+out:
+	flow_block_cb_unlock();
+	return err;
 }
 
 int enetc_psfp_clean(struct enetc_ndev_priv *priv)
 {
-	if (!list_empty(&enetc_block_cb_list))
-		return -EBUSY;
+	int err = -EBUSY;
 
-	clean_psfp_all();
+	flow_block_cb_lock();
+	if (list_empty(&enetc_block_cb_list)) {
+		clean_psfp_all();
+		err = 0;
+	}
+	flow_block_cb_unlock();
 
-	return 0;
+	return err;
 }
 
 int enetc_setup_tc_psfp(struct net_device *ndev, void *type_data)
include/net/flow_offload.h +5 -1
index 70a02ee14308..96a46ba2769d 100644
@@ -647,6 +647,8 @@ struct flow_block_cb *flow_block_cb_lookup(struct flow_block *block,
 void *flow_block_cb_priv(struct flow_block_cb *block_cb);
 void flow_block_cb_incref(struct flow_block_cb *block_cb);
 unsigned int flow_block_cb_decref(struct flow_block_cb *block_cb);
+void flow_block_cb_lock(void);
+void flow_block_cb_unlock(void);
 
 static inline void flow_block_cb_add(struct flow_block_cb *block_cb,
 				     struct flow_block_offload *offload)
@@ -741,7 +743,9 @@ void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv,
 int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch,
 				enum tc_setup_type type, void *data,
 				struct flow_block_offload *bo,
-				void (*cleanup)(struct flow_block_cb *block_cb));
+				void (*cleanup)(struct flow_block_cb *block_cb),
+				bool *indr_dev_added);
+void flow_indr_dev_setup_abort(void *data);
 bool flow_indr_dev_exists(void);
 
 #endif /* _NET_FLOW_OFFLOAD_H */
net/core/flow_offload.c +70 -7

174 lines not on this page · raw message

net/netfilter/nf_flow_table_offload.c +16 -12
index 801a3dd9ceea..2612978b27b4 100644
@@ -1186,7 +1186,6 @@ static int nf_flow_table_block_setup(struct nf_flowtable *flowtable,
 	struct flow_block_cb *block_cb, *next;
 	int err = 0;
 
-	down_write(&flowtable->flow_block_lock);
 	switch (cmd) {
 	case FLOW_BLOCK_BIND:
 		list_splice(&bo->cb_list, &flowtable->flow_block.cb_list);
@@ -1201,8 +1200,6 @@ static int nf_flow_table_block_setup(struct nf_flowtable *flowtable,
 		WARN_ON_ONCE(1);
 		err = -EOPNOTSUPP;
 	}
-	up_write(&flowtable->flow_block_lock);
-
 	return err;
 }
 
@@ -1230,7 +1227,6 @@ static void nf_flow_table_indr_cleanup(struct flow_block_cb *block_cb)
 	nf_flow_table_gc_cleanup(flowtable, dev);
 	down_write(&flowtable->flow_block_lock);
 	list_del(&block_cb->list);
-	list_del(&block_cb->driver_list);
 	flow_block_cb_free(block_cb);
 	up_write(&flowtable->flow_block_lock);
 }
@@ -1239,13 +1235,15 @@ static int nf_flow_table_indr_offload_cmd(struct flow_block_offload *bo,
 					  struct nf_flowtable *flowtable,
 					  struct net_device *dev,
 					  enum flow_block_command cmd,
-					  struct netlink_ext_ack *extack)
+					  struct netlink_ext_ack *extack,
+					  bool *indr_dev_added)
 {
 	nf_flow_table_block_offload_init(bo, dev_net(dev), cmd, flowtable,
 					 extack);
 
 	return flow_indr_dev_setup_offload(dev, NULL, TC_SETUP_FT, flowtable, bo,
-					   nf_flow_table_indr_cleanup);
+					   nf_flow_table_indr_cleanup,
+					   indr_dev_added);
 }
 
 static int nf_flow_table_offload_cmd(struct flow_block_offload *bo,
@@ -1258,9 +1256,7 @@ static int nf_flow_table_offload_cmd(struct flow_block_offload *bo,
 
 	nf_flow_table_block_offload_init(bo, dev_net(dev), cmd, flowtable,
 					 extack);
-	down_write(&flowtable->flow_block_lock);
 	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_FT, bo);
-	up_write(&flowtable->flow_block_lock);
 	if (err < 0)
 		return err;
 
@@ -1273,21 +1269,29 @@ int nf_flow_table_offload_setup(struct nf_flowtable *flowtable,
 {
 	struct netlink_ext_ack extack = {};
 	struct flow_block_offload bo;
+	bool indr_dev_added = false;
 	int err;
 
 	if (!nf_flowtable_hw_offload(flowtable))
 		return nf_flow_offload_xdp_setup(flowtable, dev, cmd);
 
+	flow_block_cb_lock();
+	down_write(&flowtable->flow_block_lock);
 	if (dev->netdev_ops->ndo_setup_tc)
 		err = nf_flow_table_offload_cmd(&bo, flowtable, dev, cmd,
 						&extack);
 	else
 		err = nf_flow_table_indr_offload_cmd(&bo, flowtable, dev, cmd,
-						     &extack);
-	if (err < 0)
-		return err;
+						     &extack,
+						     &indr_dev_added);
+	if (err >= 0)
+		err = nf_flow_table_block_setup(flowtable, &bo, cmd);
+	if (err < 0 && indr_dev_added)
+		flow_indr_dev_setup_abort(flowtable);
+	up_write(&flowtable->flow_block_lock);
+	flow_block_cb_unlock();
 
-	return nf_flow_table_block_setup(flowtable, &bo, cmd);
+	return err;
 }
 EXPORT_SYMBOL_GPL(nf_flow_table_offload_setup);
 
net/netfilter/nf_tables_offload.c +14 -4
index 8998a24651ff..7189e8c4e5a0 100644
@@ -414,7 +414,6 @@ static void nft_indr_block_cleanup(struct flow_block_cb *block_cb)
 				    basechain, &extack);
 	nft_net = nft_pernet(net);
 	mutex_lock(&nft_net->commit_mutex);
-	list_del(&block_cb->driver_list);
 	list_move(&block_cb->list, &bo.cb_list);
 	nft_flow_offload_unbind(&bo, basechain);
 	mutex_unlock(&nft_net->commit_mutex);
@@ -426,19 +425,28 @@ static int nft_indr_block_offload_cmd(struct nft_base_chain *basechain,
 {
 	struct netlink_ext_ack extack = {};
 	struct flow_block_offload bo;
+	bool indr_dev_added = false;
 	int err;
 
 	nft_flow_block_offload_init(&bo, dev_net(dev), cmd, basechain, &extack);
 
 	err = flow_indr_dev_setup_offload(dev, NULL, TC_SETUP_BLOCK, basechain, &bo,
-					  nft_indr_block_cleanup);
-	if (err < 0)
+					  nft_indr_block_cleanup,
+					  &indr_dev_added);
+	if (err < 0) {
+		if (indr_dev_added)
+			flow_indr_dev_setup_abort(basechain);
 		return err;
+	}
 
 	if (list_empty(&bo.cb_list))
 		return -EOPNOTSUPP;
 
-	return nft_block_setup(basechain, &bo, cmd);
+	err = nft_block_setup(basechain, &bo, cmd);
+	if (err < 0 && indr_dev_added)
+		flow_indr_dev_setup_abort(basechain);
+
+	return err;
 }
 
 static int nft_chain_offload_cmd(struct nft_base_chain *basechain,
@@ -447,10 +455,12 @@ static int nft_chain_offload_cmd(struct nft_base_chain *basechain,
 {
 	int err;
 
+	flow_block_cb_lock();
 	if (dev->netdev_ops->ndo_setup_tc)
 		err = nft_block_offload_cmd(basechain, dev, cmd);
 	else
 		err = nft_indr_block_offload_cmd(basechain, dev, cmd);
+	flow_block_cb_unlock();
 
 	return err;
 }
net/sched/act_api.c +3 -1
index f141634df214..317b026f0463 100644
@@ -205,8 +205,10 @@ static int tcf_action_offload_cmd_ex(struct flow_offload_action *fl_act,
 {
 	int err;
 
+	flow_block_cb_lock();
 	err = flow_indr_dev_setup_offload(NULL, NULL, TC_SETUP_ACT,
-					  fl_act, NULL, NULL);
+					  fl_act, NULL, NULL, NULL);
+	flow_block_cb_unlock();
 	if (err < 0)
 		return err;
 
net/sched/cls_api.c +74 -12

160 lines not on this page · raw message

Signature · 1 line
2.55.0
reviewer
#4 lore raw

Re: [PATCH net v2] net: flow_offload: serialize driver callback lists

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
Full review at: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/[email protected]?part=1
---
--- Patch [1]: [PATCH net v2] net: flow_offload: serialize driver callback lists ---
commit 93c807b64d24dcddc15896bc3604b3d8341855e1
Author: Weiming Shi <[email protected]>

net: flow_offload: serialize driver callback lists

This adds a global flow-offload callback mutex that is held across direct and indirect ndo_setup_tc() setup, callback ownership transfer and rollback, moves indirect-unregister list removal into the flow-offload core, and makes indirect-device registration transactional on setup failure.

Quoted 13 lines
drivers/net/ethernet/freescale/enetc/enetc_qos.c +3 -1
index 7b17bca24f26..c994f878cacd 100644
@@ -1601,30 +1601,41 @@ int enetc_set_psfp(struct net_device *ndev, bool en)
 
 int enetc_psfp_init(struct enetc_ndev_priv *priv)
 {
+	int err = 0;
+
+	flow_block_cb_lock();
 	if (epsfp.psfp_sfi_bitmap)
-		return 0;

drivers/net/ethernet/freescale/enetc/enetc_qos.c int enetc_set_psfp(struct net_device *ndev, bool en) · line 1615

Quoted 10 lines
drivers/net/ethernet/freescale/enetc/enetc_qos.c +5 -2
+		goto out;
 
 	epsfp.psfp_sfi_bitmap = bitmap_zalloc(priv->psfp_cap.max_psfp_filter,
 					      GFP_KERNEL);
-	if (!epsfp.psfp_sfi_bitmap)
-		return -ENOMEM;
+	if (!epsfp.psfp_sfi_bitmap) {
+		err = -ENOMEM;
+		goto out;
+	}

[Severity: High]
This isn't a bug introduced by this patch, but since this hunk rewrites exactly the bitmap allocation site, is the psfp_sfi_bitmap lifecycle asymmetric here?

The UNBIND path in enetc_setup_tc_psfp() frees the bitmap:

	case FLOW_BLOCK_UNBIND:
		...
		clear_bit(port, &epsfp.dev_bitmap);
		if (!epsfp.dev_bitmap)
			clean_psfp_all();

and clean_psfp_all() -> clean_psfp_sfi_bitmap() does bitmap_free() and sets epsfp.psfp_sfi_bitmap = NULL.

The BIND path only does set_bit(port, &epsfp.dev_bitmap), it never reallocates. enetc_psfp_init() is reached only from the inline enetc_psfp_enable() in enetc.h, called from enetc_pf_set_features() on a NETIF_F_HW_TC change and once at probe, and it early-returns when the pointer is non-NULL.

So after "tc qdisc del dev eth0 clsact" followed by "tc qdisc add dev eth0 clsact" (features unchanged, so tc_can_offload() still passes), does the first flower rule with a PSFP action hit a NULL dereference in enetc_get_free_index()?

	index = find_first_zero_bit(epsfp.psfp_sfi_bitmap, max_size);

There is no NULL guard on any of the psfp_sfi_bitmap users.

Quoted 19 lines
drivers/net/ethernet/freescale/enetc/enetc_qos.c +4 -4
 	spin_lock_init(&epsfp.psfp_lock);
 
 	if (list_empty(&enetc_block_cb_list))
 		epsfp.dev_bitmap = 0;
 
-	return 0;
+out:
+	flow_block_cb_unlock();
+	return err;
 }
 
 int enetc_psfp_clean(struct enetc_ndev_priv *priv)
 {
-	if (!list_empty(&enetc_block_cb_list))
-		return -EBUSY;
+	int err = -EBUSY;
 
-	clean_psfp_all();

drivers/net/ethernet/freescale/enetc/enetc_qos.c int enetc_set_psfp(struct net_device *ndev, bool en) · line 1639

Quoted 10 lines
drivers/net/ethernet/freescale/enetc/enetc_qos.c +7 -1
+	flow_block_cb_lock();
+	if (list_empty(&enetc_block_cb_list)) {
+		clean_psfp_all();
+		err = 0;
+	}
+	flow_block_cb_unlock();
 
-	return 0;
+	return err;
 }

[Severity: Low]
Is exporting the core mutex for drivers to take directly the intended layering here? The driver now runs bitmap_zalloc() and clean_psfp_all() while holding a net/core mutex, which inverts the usual "core calls driver" direction.

The new contract - a driver must hold flow_block_cb_mutex whenever it touches its own driver_block_list - is expressed only as lockdep_assert_held() inside flow_block_cb_is_busy() and flow_block_cb_setup_simple(), and is not documented anywhere in include/net/flow_offload.h. Only ENETC was converted, so other drivers that inspect their callback list outside ndo_setup_tc() remain unprotected.

Would it be preferable to keep the mutex private to net/core/flow_offload.c and give drivers a purpose-built helper instead?

Quoted 14 lines
include/net/flow_offload.h +2 -0
index 70a02ee14308..96a46ba2769d 100644
@@ -647,6 +647,8 @@ struct flow_block_cb *flow_block_cb_lookup(struct flow_block *block,
 void *flow_block_cb_priv(struct flow_block_cb *block_cb);
 void flow_block_cb_incref(struct flow_block_cb *block_cb);
 unsigned int flow_block_cb_decref(struct flow_block_cb *block_cb);
+void flow_block_cb_lock(void);
+void flow_block_cb_unlock(void);
 
 static inline void flow_block_cb_add(struct flow_block_cb *block_cb,
 				     struct flow_block_offload *offload)

include/net/flow_offload.h void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv, · line 748

Quoted 7 lines
include/net/flow_offload.h +3 -1
@@ -741,7 +743,9 @@ void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv,
 int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch,
 				enum tc_setup_type type, void *data,
 				struct flow_block_offload *bo,
-				void (*cleanup)(struct flow_block_cb *block_cb));
+				void (*cleanup)(struct flow_block_cb *block_cb),
+				bool *indr_dev_added);
+void flow_indr_dev_setup_abort(void *data);

[Severity: Low]
Could this be split into a series? As it stands one patch with Fixes: +
Cc: stable carries at least five independent changes:

- the new global mutex plus exported lock/unlock API
- INIT_LIST_HEAD() of the block_cb auxiliary list nodes
- moving the indirect-cleanup list_del(&block_cb->driver_list) into the core
- transactional indirect-device registration, changing the prototype of the exported flow_indr_dev_setup_offload() and adding the exported flow_indr_dev_setup_abort()
- a brand new TC rollback path for failed indirect binds

plus the unrelated ENETC change.

The single Fixes: c9626a2cbdb2 ("netfilter: nf_tables: add hardware offload support") does not cover the TC-side changes (the moved list_del() in tc_block_indr_cleanup() and the new tcf_block_indr_bind_rollback()), which are not introduced by the nftables offload commit.

Quoted 7 lines
include/net/flow_offload.h +0 -0
 bool flow_indr_dev_exists(void);
 
 #endif /* _NET_FLOW_OFFLOAD_H */
net/core/flow_offload.c +0 -0
index 5071d7fe6ce2..0b4d2a2cbaf8 100644

[ ... ]

net/core/flow_offload.c unsigned int flow_block_cb_decref(struct flow_block_cb *block_cb) · line 328

Quoted 11 lines
net/core/flow_offload.c +7 -0
@@ -317,11 +319,27 @@ unsigned int flow_block_cb_decref(struct flow_block_cb *block_cb)
 }
 EXPORT_SYMBOL(flow_block_cb_decref);
 
+static DEFINE_MUTEX(flow_block_cb_mutex);
+
+void flow_block_cb_lock(void)
+{
+	mutex_lock(&flow_block_cb_mutex);
+}
+EXPORT_SYMBOL(flow_block_cb_lock);

[Severity: Medium]
Should the scalability impact of a single system-wide mutex be discussed in the commit message?

This mutex becomes the outermost lock on every flow-offload configuration path (TC block bind/unbind, nftables chain offload, flowtable offload, TC action offload, indirect device register/unregister, and even flow_indr_dev_exists()), and it is held across the driver's ndo_setup_tc().

The two sides acquire it with different outer locks:

nft_chain_offload_cmd() (no RTNL)

    flow_block_cb_lock()
        dev->netdev_ops->ndo_setup_tc()

tcf_block_offload_bind() (under RTNL)

    down_write(&block->cb_lock)
    flow_block_cb_lock()

So an nftables commit in one netns that stalls inside a driver or firmware ndo_setup_tc() while holding flow_block_cb_mutex will stall any RTNL holder waiting for that mutex, i.e. netdev configuration on completely unrelated devices.

Were cheaper alternatives considered, such as a lock supplied per driver callback list, taking RTNL in the nft offload path, or scoping the lock per netdev?

Quoted 6 lines
net/core/flow_offload.c +6 -0
+
+void flow_block_cb_unlock(void)
+{
+	mutex_unlock(&flow_block_cb_mutex);
+}
+EXPORT_SYMBOL(flow_block_cb_unlock);

[ ... ]

Quoted 12 lines
net/core/flow_offload.c +1 -0
@@ -486,6 +512,7 @@ void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv,
 	struct flow_indr_dev *this, *next, *indr_dev = NULL;
 	LIST_HEAD(cleanup_list);
 
+	flow_block_cb_lock();
 	mutex_lock(&flow_indr_block_lock);
 	list_for_each_entry_safe(this, next, &flow_block_indr_dev_list, list) {
 		if (this->cb == cb &&
@@ -499,11 +526,13 @@ void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv,
 
 	if (!indr_dev) {
 		mutex_unlock(&flow_indr_block_lock);

net/core/flow_offload.c void flow_indr_dev_unregister(flow_indr_block_bind_cb_t *cb, void *cb_priv, · line 538

Quoted 10 lines
net/core/flow_offload.c +2 -0
+		flow_block_cb_unlock();
 		return;
 	}
 
 	__flow_block_indr_cleanup(release, cb_priv, &cleanup_list);
 	mutex_unlock(&flow_indr_block_lock);
+	flow_block_cb_unlock();
 
 	tcf_action_reoffload_cb(cb, cb_priv, false);
 	flow_block_indr_notify(&cleanup_list);

[Severity: High]
This isn't a bug introduced by this patch, but does the indirect cleanup path still leave the very corruption window the patch aims to close?

flow_block_cb_unlock() runs before flow_block_indr_notify(), and flow_block_indr_notify() calls this->indr.cleanup(this), i.e. tc_block_indr_cleanup() in net/sched/cls_api.c:

	rtnl_lock();
	down_write(&block->cb_lock);
	list_move(&block_cb->list, &bo.cb_list);
	tcf_block_unbind(block, &bo);

so that deletion from block->flow_block.cb_list happens with only RTNL + cb_lock held.

Meanwhile the register side splices into the same list head with a disjoint lock set:

net/core/flow_offload.c:existing_qdiscs_register() {
	...
	cb(cur->dev, cur->sch, cb_priv, cur->type, &bo, cur->data, cur->cleanup);
	list_splice(&bo.cb_list, cur->cb_list);
	...
}

That runs from flow_indr_dev_register() holding flow_block_cb_mutex + flow_indr_block_lock, but neither RTNL nor block->cb_lock (cb_list here is bo->cb_list_head, i.e. &block->flow_block.cb_list, recorded by indir_dev_add()).

Can two indirect-offload drivers (or repeated load/unload of one) then run splice and list_del concurrently on one live tcf_block cb_list, and produce the same "list_del corruption ... next->prev should be" splat? Note tc_block_indr_cleanup() itself takes RTNL, so it cannot simply be pulled under the new mutex.

Quoted 24 lines
net/core/flow_offload.c +9 -4
@@ -603,18 +632,31 @@ static int indir_dev_remove(void *data)
 int flow_indr_dev_setup_offload(struct net_device *dev,	struct Qdisc *sch,
 				enum tc_setup_type type, void *data,
 				struct flow_block_offload *bo,
-				void (*cleanup)(struct flow_block_cb *block_cb))
+				void (*cleanup)(struct flow_block_cb *block_cb),
+				bool *indr_dev_added)
 {
 	struct flow_indr_dev *this;
 	u32 count = 0;
 	int err;
 
+	lockdep_assert_held(&flow_block_cb_mutex);
+
+	if (indr_dev_added)
+		*indr_dev_added = false;
+
 	mutex_lock(&flow_indr_block_lock);
 	if (bo) {
-		if (bo->command == FLOW_BLOCK_BIND)
-			indir_dev_add(data, dev, sch, type, cleanup, bo);
-		else if (bo->command == FLOW_BLOCK_UNBIND)
+		if (bo->command == FLOW_BLOCK_BIND) {
+			err = indir_dev_add(data, dev, sch, type, cleanup, bo);

net/core/flow_offload.c static int indir_dev_remove(void *data) · line 660

Quoted 10 lines
net/core/flow_offload.c +8 -0
+			if (!err) {
+				if (indr_dev_added)
+					*indr_dev_added = true;
+			} else if (err != -EEXIST) {
+				goto out_unlock;
+			}
+		} else if (bo->command == FLOW_BLOCK_UNBIND) {
 			indir_dev_remove(data);
+		}
 	}

[Severity: Medium]
Does this turn a previously silent degradation into a user-visible failure?

Before this change the return value of indir_dev_add() was discarded, so a failed bookkeeping allocation did not affect the outcome. Now an -ENOMEM from indir_dev_add() aborts the whole setup before any indirect driver callback runs, and cls_api propagates it:

net/sched/cls_api.c:tcf_block_offload_cmd() {
	...
	if (err < 0)
		return err;
	cb_count = list_count_nodes(&bo.cb_list);
	err = tcf_block_setup(block, &bo);
	if (err)
		return err;
	...
}
net/sched/cls_api.c:tcf_block_offload_bind() {
	...
	if (err == -EOPNOTSUPP)
		goto no_offload_dev_inc;
	if (err)
		goto err_unlock;
	...
}

Only -EOPNOTSUPP still reaches the software fallback, so "tc qdisc add" on an indirect device, nft base-chain offload and flowtable offload can now fail to user space where they previously succeeded with block->nooffloaddevcnt++. The same applies to a non-EOPNOTSUPP error from tp->ops->reoffload during filter replay, which tcf_block_setup() now forwards.

The commit message only says "keep indirect-device registration transactional on setup failure". Could it state this new failure mode explicitly, given the Cc: stable tag?

Quoted 15 lines
net/core/flow_offload.c +4 -1
 	list_for_each_entry(this, &flow_block_indr_dev_list, list) {
@@ -623,14 +665,35 @@ int flow_indr_dev_setup_offload(struct net_device *dev,	struct Qdisc *sch,
 			count++;
 	}
 
+	err = (bo && list_empty(&bo->cb_list)) ? -EOPNOTSUPP : count;
+
+out_unlock:
 	mutex_unlock(&flow_indr_block_lock);
 
-	return (bo && list_empty(&bo->cb_list)) ? -EOPNOTSUPP : count;
+	return err;
 }
 EXPORT_SYMBOL(flow_indr_dev_setup_offload);

net/core/flow_offload.c int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch, · line 685

Quoted 10 lines
net/core/flow_offload.c +9 -0
 
+void flow_indr_dev_setup_abort(void *data)
+{
+	lockdep_assert_held(&flow_block_cb_mutex);
+
+	mutex_lock(&flow_indr_block_lock);
+	WARN_ON_ONCE(indir_dev_remove(data));
+	mutex_unlock(&flow_indr_block_lock);
+}
+EXPORT_SYMBOL(flow_indr_dev_setup_abort);

[Severity: Low]
Would it be better for the core to undo its own bookkeeping rather than export partial state plus an abort call?

indir_dev_add()/indir_dev_remove() are internal to net/core/flow_offload.c, yet callers now receive bool *indr_dev_added and must remember to call flow_indr_dev_setup_abort() on every later failure. The same snippet is duplicated in nf_tables_offload.c (twice), nf_flow_table_offload.c and cls_api.c (twice), with three different failure semantics for one core operation: nft and flowtable only unwind the indr_dev bookkeeping, cls_api additionally unwinds installed callbacks.

Some of the new arms are also dead, since nft_flow_offload_bind() only does list_splice() and returns 0, and nf_flow_table_block_setup() cannot fail for BIND.

If a future caller forgets the abort, does the stale struct flow_indir_dev_info left on flow_indir_dev_list later drive a bogus replay through existing_qdiscs_register()?

Quoted 5 lines
net/core/flow_offload.c +2 -1
+
 bool flow_indr_dev_exists(void)
 {
-	return !list_empty(&flow_block_indr_dev_list);
+	bool exists;

net/core/flow_offload.c int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch, · line 699

Quoted 10 lines
net/core/flow_offload.c +8 -0
+
+	flow_block_cb_lock();
+	mutex_lock(&flow_indr_block_lock);
+	exists = !list_empty(&flow_block_indr_dev_list);
+	mutex_unlock(&flow_indr_block_lock);
+	flow_block_cb_unlock();
+
+	return exists;
 }
 EXPORT_SYMBOL(flow_indr_dev_exists);

[Severity: Medium]
Is the global mutex needed for this predicate? flow_indr_dev_exists() is a cheap existence test used by nft_chain_offload_support() in net/netfilter/nf_tables_offload.c, and taking flow_block_cb_mutex here imposes the new outermost-lock ordering constraint on all of its callers.

Quoted 4 lines
net/netfilter/nf_flow_table_offload.c +0 -0
index 801a3dd9ceea..2612978b27b4 100644

[ ... ]

Quoted 38 lines
net/netfilter/nf_flow_table_offload.c +12 -4
@@ -1273,21 +1269,29 @@ int nf_flow_table_offload_setup(struct nf_flowtable *flowtable,
 {
 	struct netlink_ext_ack extack = {};
 	struct flow_block_offload bo;
+	bool indr_dev_added = false;
 	int err;
 
 	if (!nf_flowtable_hw_offload(flowtable))
 		return nf_flow_offload_xdp_setup(flowtable, dev, cmd);
 
+	flow_block_cb_lock();
+	down_write(&flowtable->flow_block_lock);
 	if (dev->netdev_ops->ndo_setup_tc)
 		err = nf_flow_table_offload_cmd(&bo, flowtable, dev, cmd,
 						&extack);
 	else
 		err = nf_flow_table_indr_offload_cmd(&bo, flowtable, dev, cmd,
-						     &extack);
-	if (err < 0)
-		return err;
+						     &extack,
+						     &indr_dev_added);
+	if (err >= 0)
+		err = nf_flow_table_block_setup(flowtable, &bo, cmd);
+	if (err < 0 && indr_dev_added)
+		flow_indr_dev_setup_abort(flowtable);
+	up_write(&flowtable->flow_block_lock);
+	flow_block_cb_unlock();
 
-	return nf_flow_table_block_setup(flowtable, &bo, cmd);
+	return err;
 }
 EXPORT_SYMBOL_GPL(nf_flow_table_offload_setup);
 
net/netfilter/nf_tables_offload.c +0 -0
index 8998a24651ff..7189e8c4e5a0 100644

[ ... ]

Quoted 12 lines
net/netfilter/nf_tables_offload.c +1 -2
@@ -426,19 +425,28 @@ static int nft_indr_block_offload_cmd(struct nft_base_chain *basechain,
 {
 	struct netlink_ext_ack extack = {};
 	struct flow_block_offload bo;
+	bool indr_dev_added = false;
 	int err;
 
 	nft_flow_block_offload_init(&bo, dev_net(dev), cmd, basechain, &extack);
 
 	err = flow_indr_dev_setup_offload(dev, NULL, TC_SETUP_BLOCK, basechain, &bo,
-					  nft_indr_block_cleanup);
-	if (err < 0)

net/netfilter/nf_tables_offload.c static int nft_indr_block_offload_cmd(struct nft_base_chain *basechain, · line 443

Quoted 10 lines
net/netfilter/nf_tables_offload.c +6 -0
+					  nft_indr_block_cleanup,
+					  &indr_dev_added);
+	if (err < 0) {
+		if (indr_dev_added)
+			flow_indr_dev_setup_abort(basechain);
 		return err;
+	}
 
 	if (list_empty(&bo.cb_list))
 		return -EOPNOTSUPP;

[Severity: Low]
Should this exit also call flow_indr_dev_setup_abort()? It is the only error return of the restructured function that skips the new cleanup.

The branch is dead today, since flow_indr_dev_setup_offload() already folds the empty-cb_list case into -EOPNOTSUPP:

net/core/flow_offload.c:flow_indr_dev_setup_offload() {
	...
	err = (bo && list_empty(&bo->cb_list)) ? -EOPNOTSUPP : count;
	...
}

with bo non-NULL here, and its only other negative return is -ENOMEM via goto out_unlock. Should it become reachable, would this exit leave the flow_indir_dev_info created by indir_dev_add() on flow_indir_dev_list with data pointing at a base chain the failing nft transaction is about to destroy?

Quoted 8 lines
net/netfilter/nf_tables_offload.c +5 -1
-	return nft_block_setup(basechain, &bo, cmd);
+	err = nft_block_setup(basechain, &bo, cmd);
+	if (err < 0 && indr_dev_added)
+		flow_indr_dev_setup_abort(basechain);
+
+	return err;
 }

[ ... ]

Quoted 4 lines
net/sched/cls_api.c +0 -0
index fee4524adc98..1c230e33e1a5 100644

[ ... ]

Quoted 44 lines
net/sched/cls_api.c +33 -3
@@ -845,21 +847,66 @@ static int tcf_block_offload_cmd(struct tcf_block *block,
 		return tcf_block_setup(block, &bo);
 	}
 
-	flow_indr_dev_setup_offload(dev, sch, TC_SETUP_BLOCK, block, &bo,
-				    tc_block_indr_cleanup);
-	tcf_block_setup(block, &bo);
+	err = flow_indr_dev_setup_offload(dev, sch, TC_SETUP_BLOCK, block, &bo,
+					  tc_block_indr_cleanup,
+					  indr_dev_added);
+	if (err < 0)
+		return err;
+	cb_count = list_count_nodes(&bo.cb_list);
+	err = tcf_block_setup(block, &bo);
+	if (err)
+		return err;
+	if (indr_cb_count)
+		*indr_cb_count = cb_count;
+	if (indr_unlocked_driver_cb)
+		*indr_unlocked_driver_cb = bo.unlocked_driver_cb;
 
 	return -EOPNOTSUPP;
 }
 
+static void
+tcf_block_indr_bind_rollback(struct tcf_block *block, struct net_device *dev,
+			     struct Qdisc *sch, struct tcf_block_ext_info *ei,
+			     unsigned int cb_count, bool unlocked_driver_cb,
+			     bool remove_indr_dev)
+{
+	struct flow_block_offload bo = {};
+	struct flow_block_cb *block_cb;
+
+	tcf_block_offload_init(&bo, dev, sch, FLOW_BLOCK_UNBIND,
+			       ei->binder_type, &block->flow_block,
+			       tcf_block_shared(block), NULL);
+	bo.unlocked_driver_cb = unlocked_driver_cb;
+
+	if (remove_indr_dev)
+		flow_indr_dev_setup_abort(block);
+
+	/* tcf_block_bind() splices this bind's callbacks at the list head. */
+	while (cb_count--) {
+		if (WARN_ON_ONCE(list_empty(&block->flow_block.cb_list)))

net/sched/cls_api.c static int tcf_block_offload_cmd(struct tcf_block *block, · line 896

Quoted 10 lines
net/sched/cls_api.c +10 -0
+			break;
+
+		block_cb = list_first_entry(&block->flow_block.cb_list,
+					    struct flow_block_cb, list);
+		list_del_init(&block_cb->driver_list);
+		list_del_init(&block_cb->indr.list);
+		list_move(&block_cb->list, &bo.cb_list);
+	}
+	tcf_block_unbind(block, &bo);
+}

[Severity: Low]
Is identifying "this bind's callbacks" by popping cb_count entries off the head of block->flow_block.cb_list robust enough?

The count is taken from bo.cb_list before tcf_block_setup(), and the recovery relies on tcf_block_bind() using list_splice() (head insertion) and on nothing else inserting in between. The WARN_ON_ONCE(list_empty(...)) inside the pop loop suggests the invariant isn't structurally guaranteed.

If the splice direction in tcf_block_bind() ever changes, or another writer inserts into that list, would this unbind and free callbacks owned by a different binder?

Separately, tcf_block_offload_cmd() still returns -EOPNOTSUPP as the success sentinel for the indirect path while carrying three new out-parameters, so a caller passing NULL loses the rollback ability silently. Could the bind's callbacks be tracked explicitly instead, for example on a local list?

[Severity: Low]
Which lock protects block_cb->indr.list here? This helper does list_del_init(&block_cb->indr.list) on the file-global flow_block_indr_list while holding only block->cb_lock and the new flow_block_cb_mutex.

Every mutator of that list in net/core/flow_offload.c takes flow_indr_block_lock, e.g.

net/core/flow_offload.c:__flow_block_indr_cleanup() {
	...
		list_del_init(&this->driver_list);
		list_move(&this->indr.list, cleanup_list);
	...
}

and flow_indr_dev_setup_abort() itself takes flow_indr_block_lock for the equivalent bookkeeping. Today the lock sets do intersect on flow_block_cb_mutex, so there is no race, but the rule becomes implicit. Could a comment plus lockdep asserts (including on the exported flow_indr_block_cb_alloc(), which has none) be added?

Quoted 16 lines
net/sched/cls_api.c +5 -0
+
 static int tcf_block_offload_bind(struct tcf_block *block, struct Qdisc *q,
 				  struct tcf_block_ext_info *ei,
 				  struct netlink_ext_ack *extack)
 {
 	struct net_device *dev = q->dev_queue->dev;
+	bool indr_unlocked_driver_cb = false;
+	bool indr_dev_added = false;
+	unsigned int indr_cb_count = 0;
 	int err;
 
 	down_write(&block->cb_lock);
+	flow_block_cb_lock();
 
 	/* If tc offload feature is disabled and the block we try to bind
 	 * to already has some offloaded filters, forbid to bind.

[ ... ]

Quoted 16 lines
net/sched/cls_api.c +9 -0
@@ -888,6 +936,15 @@ static int tcf_block_offload_bind(struct tcf_block *block, struct Qdisc *q,
 	err = 0;
 	block->nooffloaddevcnt++;
 err_unlock:
+	if (err && indr_cb_count)
+		tcf_block_indr_bind_rollback(block, dev, q, ei,
+					     indr_cb_count,
+					     indr_unlocked_driver_cb,
+					     indr_dev_added);
+	else if (err && indr_dev_added)
+		flow_indr_dev_setup_abort(block);
+out_unlock:
+	flow_block_cb_unlock();
 	up_write(&block->cb_lock);
 	return err;
 }

[ ... ]

net/sched/cls_api.c static int tcf_block_bind(struct tcf_block *block, · line 1714

Quoted 7 lines
net/sched/cls_api.c +2 -1
@@ -1649,7 +1710,8 @@ static int tcf_block_bind(struct tcf_block *block,
 
 err_unroll:
 	list_for_each_entry_safe(block_cb, next, &bo->cb_list, list) {
-		list_del(&block_cb->driver_list);
+		list_del_init(&block_cb->driver_list);
+		list_del_init(&block_cb->indr.list);

[Severity: Low]
Same question as above for this site: the err_unroll path now mutates the global flow_block_indr_list without flow_indr_block_lock, and tcf_block_bind()/tcf_block_unbind() only assert block->cb_lock.

Quoted 3 lines
net/sched/cls_api.c +0 -0
 		if (i-- > 0) {
 			list_del(&block_cb->list);
 			tcf_block_playback_offloads(block, block_cb->cb,
maintainer
#5 lore raw

Re: [PATCH net v2] net: flow_offload: serialize driver callback lists

In reply to Weiming Shi
Quoted 6 lines · On Sun, 26 Jul 2026 05:03:15 -0700 Weiming Shi wrote:

Add a flow-offload callback mutex and hold it across direct and indirect setup, callback ownership transfer and rollback. Move indirect-unregister list removal into the flow-offload core, and keep indirect-device registration transactional on setup failure. When TC must reject an indirect bind after callback installation, unwind only the callbacks created by that bind so existing callbacks remain intact.

Isn't the unbind / rollback stuff mostly orthogonal to the locking? Looks like the patch should be split.