All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order
@ 2021-08-17 17:05 Eli Cohen
  2021-08-17 17:05 ` [PATCHv2 net-next 1/2] net/core: Remove unused field from struct flow_indr_dev Eli Cohen
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Eli Cohen @ 2021-08-17 17:05 UTC (permalink / raw)
  To: davem, kuba; +Cc: netdev, elic

The first patch is just a cleanup of the code.
The second patch is fixing the dependency in ingress qdisc creation
relative to offloading driver registration to filter configurations.

v1 -> v2:
Fix warning - variable set but not used

Eli Cohen (2):
  net/core: Remove unused field from struct flow_indr_dev
  net: Fix offloading indirect devices dependency on qdisc order
    creation

 include/net/flow_offload.h            |  1 +
 net/core/flow_offload.c               | 90 ++++++++++++++++++++++++++-
 net/netfilter/nf_flow_table_offload.c |  1 +
 net/netfilter/nf_tables_offload.c     |  1 +
 net/sched/cls_api.c                   |  1 +
 5 files changed, 92 insertions(+), 2 deletions(-)

-- 
2.32.0


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCHv2 net-next 1/2] net/core: Remove unused field from struct flow_indr_dev
  2021-08-17 17:05 [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order Eli Cohen
@ 2021-08-17 17:05 ` Eli Cohen
  2021-08-17 17:05 ` [PATCHv2 net-next 2/2] net: Fix offloading indirect devices dependency on qdisc order creation Eli Cohen
  2021-08-19 12:30 ` [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order patchwork-bot+netdevbpf
  2 siblings, 0 replies; 4+ messages in thread
From: Eli Cohen @ 2021-08-17 17:05 UTC (permalink / raw)
  To: davem, kuba; +Cc: netdev, elic, Jiri Pirko

rcu field is not used. Remove it.

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Eli Cohen <elic@nvidia.com>
---
 net/core/flow_offload.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/net/core/flow_offload.c b/net/core/flow_offload.c
index 715b67f6c62f..1da83997e86a 100644
--- a/net/core/flow_offload.c
+++ b/net/core/flow_offload.c
@@ -327,7 +327,6 @@ struct flow_indr_dev {
 	flow_indr_block_bind_cb_t	*cb;
 	void				*cb_priv;
 	refcount_t			refcnt;
-	struct rcu_head			rcu;
 };
 
 static struct flow_indr_dev *flow_indr_dev_alloc(flow_indr_block_bind_cb_t *cb,
-- 
2.32.0


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* [PATCHv2 net-next 2/2] net: Fix offloading indirect devices dependency on qdisc order creation
  2021-08-17 17:05 [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order Eli Cohen
  2021-08-17 17:05 ` [PATCHv2 net-next 1/2] net/core: Remove unused field from struct flow_indr_dev Eli Cohen
@ 2021-08-17 17:05 ` Eli Cohen
  2021-08-19 12:30 ` [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order patchwork-bot+netdevbpf
  2 siblings, 0 replies; 4+ messages in thread
From: Eli Cohen @ 2021-08-17 17:05 UTC (permalink / raw)
  To: davem, kuba; +Cc: netdev, elic, Jiri Pirko

Currently, when creating an ingress qdisc on an indirect device before
the driver registered for callbacks, the driver will not have a chance
to register its filter configuration callbacks.

To fix that, modify the code such that it keeps track of all the ingress
qdiscs that call flow_indr_dev_setup_offload(). When a driver calls
flow_indr_dev_register(),  go through the list of tracked ingress qdiscs
and call the driver callback entry point so as to give it a chance to
register its callback.

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Eli Cohen <elic@nvidia.com>
---
v1 -> v2:
Fix warning - variable set but not used

 include/net/flow_offload.h            |  1 +
 net/core/flow_offload.c               | 89 ++++++++++++++++++++++++++-
 net/netfilter/nf_flow_table_offload.c |  1 +
 net/netfilter/nf_tables_offload.c     |  1 +
 net/sched/cls_api.c                   |  1 +
 5 files changed, 92 insertions(+), 1 deletion(-)

diff --git a/include/net/flow_offload.h b/include/net/flow_offload.h
index f3c2841566a0..5aa27acdb0b3 100644
--- a/include/net/flow_offload.h
+++ b/include/net/flow_offload.h
@@ -453,6 +453,7 @@ struct flow_block_offload {
 	struct list_head *driver_block_list;
 	struct netlink_ext_ack *extack;
 	struct Qdisc *sch;
+	struct list_head *cb_list_head;
 };
 
 enum tc_setup_type;
diff --git a/net/core/flow_offload.c b/net/core/flow_offload.c
index 1da83997e86a..6beaea13564a 100644
--- a/net/core/flow_offload.c
+++ b/net/core/flow_offload.c
@@ -321,6 +321,7 @@ EXPORT_SYMBOL(flow_block_cb_setup_simple);
 static DEFINE_MUTEX(flow_indr_block_lock);
 static LIST_HEAD(flow_block_indr_list);
 static LIST_HEAD(flow_block_indr_dev_list);
+static LIST_HEAD(flow_indir_dev_list);
 
 struct flow_indr_dev {
 	struct list_head		list;
@@ -345,6 +346,33 @@ static struct flow_indr_dev *flow_indr_dev_alloc(flow_indr_block_bind_cb_t *cb,
 	return indr_dev;
 }
 
+struct flow_indir_dev_info {
+	void *data;
+	struct net_device *dev;
+	struct Qdisc *sch;
+	enum tc_setup_type type;
+	void (*cleanup)(struct flow_block_cb *block_cb);
+	struct list_head list;
+	enum flow_block_command command;
+	enum flow_block_binder_type binder_type;
+	struct list_head *cb_list;
+};
+
+static void existing_qdiscs_register(flow_indr_block_bind_cb_t *cb, void *cb_priv)
+{
+	struct flow_block_offload bo;
+	struct flow_indir_dev_info *cur;
+
+	list_for_each_entry(cur, &flow_indir_dev_list, list) {
+		memset(&bo, 0, sizeof(bo));
+		bo.command = cur->command;
+		bo.binder_type = cur->binder_type;
+		INIT_LIST_HEAD(&bo.cb_list);
+		cb(cur->dev, cur->sch, cb_priv, cur->type, &bo, cur->data, cur->cleanup);
+		list_splice(&bo.cb_list, cur->cb_list);
+	}
+}
+
 int flow_indr_dev_register(flow_indr_block_bind_cb_t *cb, void *cb_priv)
 {
 	struct flow_indr_dev *indr_dev;
@@ -366,6 +394,7 @@ int flow_indr_dev_register(flow_indr_block_bind_cb_t *cb, void *cb_priv)
 	}
 
 	list_add(&indr_dev->list, &flow_block_indr_dev_list);
+	existing_qdiscs_register(cb, cb_priv);
 	mutex_unlock(&flow_indr_block_lock);
 
 	return 0;
@@ -462,7 +491,59 @@ struct flow_block_cb *flow_indr_block_cb_alloc(flow_setup_cb_t *cb,
 }
 EXPORT_SYMBOL(flow_indr_block_cb_alloc);
 
-int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch,
+static struct flow_indir_dev_info *find_indir_dev(void *data)
+{
+	struct flow_indir_dev_info *cur;
+
+	list_for_each_entry(cur, &flow_indir_dev_list, list) {
+		if (cur->data == data)
+			return cur;
+	}
+	return NULL;
+}
+
+static int indir_dev_add(void *data, struct net_device *dev, struct Qdisc *sch,
+			 enum tc_setup_type type, void (*cleanup)(struct flow_block_cb *block_cb),
+			 struct flow_block_offload *bo)
+{
+	struct flow_indir_dev_info *info;
+
+	info = find_indir_dev(data);
+	if (info)
+		return -EEXIST;
+
+	info = kzalloc(sizeof(*info), GFP_KERNEL);
+	if (!info)
+		return -ENOMEM;
+
+	info->data = data;
+	info->dev = dev;
+	info->sch = sch;
+	info->type = type;
+	info->cleanup = cleanup;
+	info->command = bo->command;
+	info->binder_type = bo->binder_type;
+	info->cb_list = bo->cb_list_head;
+
+	list_add(&info->list, &flow_indir_dev_list);
+	return 0;
+}
+
+static int indir_dev_remove(void *data)
+{
+	struct flow_indir_dev_info *info;
+
+	info = find_indir_dev(data);
+	if (!info)
+		return -ENOENT;
+
+	list_del(&info->list);
+
+	kfree(info);
+	return 0;
+}
+
+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))
@@ -470,6 +551,12 @@ int flow_indr_dev_setup_offload(struct net_device *dev, struct Qdisc *sch,
 	struct flow_indr_dev *this;
 
 	mutex_lock(&flow_indr_block_lock);
+
+	if (bo->command == FLOW_BLOCK_BIND)
+		indir_dev_add(data, dev, sch, type, cleanup, bo);
+	else if (bo->command == FLOW_BLOCK_UNBIND)
+		indir_dev_remove(data);
+
 	list_for_each_entry(this, &flow_block_indr_dev_list, list)
 		this->cb(dev, sch, this->cb_priv, type, bo, data, cleanup);
 
diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c
index f92006cec94c..cbd9f59098b7 100644
--- a/net/netfilter/nf_flow_table_offload.c
+++ b/net/netfilter/nf_flow_table_offload.c
@@ -1097,6 +1097,7 @@ static void nf_flow_table_block_offload_init(struct flow_block_offload *bo,
 	bo->command	= cmd;
 	bo->binder_type	= FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS;
 	bo->extack	= extack;
+	bo->cb_list_head = &flowtable->flow_block.cb_list;
 	INIT_LIST_HEAD(&bo->cb_list);
 }
 
diff --git a/net/netfilter/nf_tables_offload.c b/net/netfilter/nf_tables_offload.c
index b58d73a96523..9656c1646222 100644
--- a/net/netfilter/nf_tables_offload.c
+++ b/net/netfilter/nf_tables_offload.c
@@ -353,6 +353,7 @@ static void nft_flow_block_offload_init(struct flow_block_offload *bo,
 	bo->command	= cmd;
 	bo->binder_type	= FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS;
 	bo->extack	= extack;
+	bo->cb_list_head = &basechain->flow_block.cb_list;
 	INIT_LIST_HEAD(&bo->cb_list);
 }
 
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index e3e79e9bd706..9b276d14be4c 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -634,6 +634,7 @@ static void tcf_block_offload_init(struct flow_block_offload *bo,
 	bo->block_shared = shared;
 	bo->extack = extack;
 	bo->sch = sch;
+	bo->cb_list_head = &flow_block->cb_list;
 	INIT_LIST_HEAD(&bo->cb_list);
 }
 
-- 
2.32.0


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order
  2021-08-17 17:05 [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order Eli Cohen
  2021-08-17 17:05 ` [PATCHv2 net-next 1/2] net/core: Remove unused field from struct flow_indr_dev Eli Cohen
  2021-08-17 17:05 ` [PATCHv2 net-next 2/2] net: Fix offloading indirect devices dependency on qdisc order creation Eli Cohen
@ 2021-08-19 12:30 ` patchwork-bot+netdevbpf
  2 siblings, 0 replies; 4+ messages in thread
From: patchwork-bot+netdevbpf @ 2021-08-19 12:30 UTC (permalink / raw)
  To: Eli Cohen; +Cc: davem, kuba, netdev

Hello:

This series was applied to netdev/net-next.git (refs/heads/master):

On Tue, 17 Aug 2021 20:05:16 +0300 you wrote:
> The first patch is just a cleanup of the code.
> The second patch is fixing the dependency in ingress qdisc creation
> relative to offloading driver registration to filter configurations.
> 
> v1 -> v2:
> Fix warning - variable set but not used
> 
> [...]

Here is the summary with links:
  - [PATCHv2,net-next,1/2] net/core: Remove unused field from struct flow_indr_dev
    https://git.kernel.org/netdev/net-next/c/c1c5cb3aee05
  - [PATCHv2,net-next,2/2] net: Fix offloading indirect devices dependency on qdisc order creation
    https://git.kernel.org/netdev/net-next/c/74fc4f828769

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] 4+ messages in thread

end of thread, other threads:[~2021-08-19 12:30 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-08-17 17:05 [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order Eli Cohen
2021-08-17 17:05 ` [PATCHv2 net-next 1/2] net/core: Remove unused field from struct flow_indr_dev Eli Cohen
2021-08-17 17:05 ` [PATCHv2 net-next 2/2] net: Fix offloading indirect devices dependency on qdisc order creation Eli Cohen
2021-08-19 12:30 ` [PATCHv2 net-next 0/2] Indirect dev ingress qdisc creation order patchwork-bot+netdevbpf

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.