* [PATCH net-next v2 1/3] mctp: Return new key from mctp_alloc_local_tag
2021-10-29 3:01 [PATCH net-next v2 0/3] MCTP flow support Jeremy Kerr
@ 2021-10-29 3:01 ` Jeremy Kerr
2021-10-29 3:01 ` [PATCH net-next v2 2/3] mctp: Add flow extension to skb Jeremy Kerr
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Jeremy Kerr @ 2021-10-29 3:01 UTC (permalink / raw)
To: netdev; +Cc: David S. Miller, Jakub Kicinski, Matt Johnston, Andrew Jeffery
In a future change, we will want the key available for future use after
allocating a new tag. This change returns the key from
mctp_alloc_local_tag, rather than just key->tag.
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
---
net/mctp/route.c | 28 ++++++++++++++++------------
1 file changed, 16 insertions(+), 12 deletions(-)
diff --git a/net/mctp/route.c b/net/mctp/route.c
index c23ab3547ee5..82cc10a2fb0c 100644
--- a/net/mctp/route.c
+++ b/net/mctp/route.c
@@ -532,14 +532,14 @@ static void mctp_reserve_tag(struct net *net, struct mctp_sk_key *key,
/* Allocate a locally-owned tag value for (saddr, daddr), and reserve
* it for the socket msk
*/
-static int mctp_alloc_local_tag(struct mctp_sock *msk,
- mctp_eid_t saddr, mctp_eid_t daddr, u8 *tagp)
+static struct mctp_sk_key *mctp_alloc_local_tag(struct mctp_sock *msk,
+ mctp_eid_t saddr,
+ mctp_eid_t daddr, u8 *tagp)
{
struct net *net = sock_net(&msk->sk);
struct netns_mctp *mns = &net->mctp;
struct mctp_sk_key *key, *tmp;
unsigned long flags;
- int rc = -EAGAIN;
u8 tagbits;
/* for NULL destination EIDs, we may get a response from any peer */
@@ -549,7 +549,7 @@ static int mctp_alloc_local_tag(struct mctp_sock *msk,
/* be optimistic, alloc now */
key = mctp_key_alloc(msk, saddr, daddr, 0, GFP_KERNEL);
if (!key)
- return -ENOMEM;
+ return ERR_PTR(-ENOMEM);
/* 8 possible tag values */
tagbits = 0xff;
@@ -591,18 +591,16 @@ static int mctp_alloc_local_tag(struct mctp_sock *msk,
trace_mctp_key_acquire(key);
*tagp = key->tag;
- /* done with the key in this scope */
- mctp_key_unref(key);
- key = NULL;
- rc = 0;
}
spin_unlock_irqrestore(&mns->keys_lock, flags);
- if (!tagbits)
+ if (!tagbits) {
kfree(key);
+ return ERR_PTR(-EBUSY);
+ }
- return rc;
+ return key;
}
/* routing lookups */
@@ -740,6 +738,7 @@ int mctp_local_output(struct sock *sk, struct mctp_route *rt,
struct mctp_sock *msk = container_of(sk, struct mctp_sock, sk);
struct mctp_skb_cb *cb = mctp_cb(skb);
struct mctp_route tmp_rt;
+ struct mctp_sk_key *key;
struct net_device *dev;
struct mctp_hdr *hdr;
unsigned long flags;
@@ -799,11 +798,16 @@ int mctp_local_output(struct sock *sk, struct mctp_route *rt,
goto out_release;
if (req_tag & MCTP_HDR_FLAG_TO) {
- rc = mctp_alloc_local_tag(msk, saddr, daddr, &tag);
- if (rc)
+ key = mctp_alloc_local_tag(msk, saddr, daddr, &tag);
+ if (IS_ERR(key)) {
+ rc = PTR_ERR(key);
goto out_release;
+ }
+ /* done with the key in this scope */
+ mctp_key_unref(key);
tag |= MCTP_HDR_FLAG_TO;
} else {
+ key = NULL;
tag = req_tag;
}
--
2.33.0
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH net-next v2 2/3] mctp: Add flow extension to skb
2021-10-29 3:01 [PATCH net-next v2 0/3] MCTP flow support Jeremy Kerr
2021-10-29 3:01 ` [PATCH net-next v2 1/3] mctp: Return new key from mctp_alloc_local_tag Jeremy Kerr
@ 2021-10-29 3:01 ` Jeremy Kerr
2021-10-29 3:01 ` [PATCH net-next v2 3/3] mctp: Pass flow data & flow release events to drivers Jeremy Kerr
2021-10-29 12:40 ` [PATCH net-next v2 0/3] MCTP flow support patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: Jeremy Kerr @ 2021-10-29 3:01 UTC (permalink / raw)
To: netdev; +Cc: David S. Miller, Jakub Kicinski, Matt Johnston, Andrew Jeffery
This change adds a new skb extension for MCTP, to represent a
request/response flow.
The intention is to use this in a later change to allow i2c controllers
to correctly configure a multiplexer over a flow.
Since we have a cleanup function in the core path (if an extension is
present), we'll need to make CONFIG_MCTP a bool, rather than a tristate.
Includes a fix for a build warning with clang:
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
---
v2:
- #ifdef CONFIG_MCTP, rather than #if
- fix CONFIG_SKB_EXTENSIONS && !CONFIG_MCTP_FLOWS build
---
include/linux/skbuff.h | 3 +++
include/net/mctp.h | 7 +++++++
net/core/skbuff.c | 19 +++++++++++++++++++
net/mctp/Kconfig | 7 ++++++-
4 files changed, 35 insertions(+), 1 deletion(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index cb96f1e6460c..0bd6520329f6 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -4243,6 +4243,9 @@ enum skb_ext_id {
#endif
#if IS_ENABLED(CONFIG_MPTCP)
SKB_EXT_MPTCP,
+#endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+ SKB_EXT_MCTP,
#endif
SKB_EXT_NUM, /* must be last */
};
diff --git a/include/net/mctp.h b/include/net/mctp.h
index 23bec708f4c7..7a5ba801703c 100644
--- a/include/net/mctp.h
+++ b/include/net/mctp.h
@@ -189,6 +189,13 @@ static inline struct mctp_skb_cb *mctp_cb(struct sk_buff *skb)
return (void *)(skb->cb);
}
+/* If CONFIG_MCTP_FLOWS, we may add one of these as a SKB extension,
+ * indicating the flow to the device driver.
+ */
+struct mctp_flow {
+ struct mctp_sk_key *key;
+};
+
/* Route definition.
*
* These are held in the pernet->mctp.routes list, with RCU protection for
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 74601bbc56ac..b948fd1c6f75 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -70,6 +70,7 @@
#include <net/xfrm.h>
#include <net/mpls.h>
#include <net/mptcp.h>
+#include <net/mctp.h>
#include <net/page_pool.h>
#include <linux/uaccess.h>
@@ -4430,6 +4431,9 @@ static const u8 skb_ext_type_len[] = {
#if IS_ENABLED(CONFIG_MPTCP)
[SKB_EXT_MPTCP] = SKB_EXT_CHUNKSIZEOF(struct mptcp_ext),
#endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+ [SKB_EXT_MCTP] = SKB_EXT_CHUNKSIZEOF(struct mctp_flow),
+#endif
};
static __always_inline unsigned int skb_ext_total_length(void)
@@ -4446,6 +4450,9 @@ static __always_inline unsigned int skb_ext_total_length(void)
#endif
#if IS_ENABLED(CONFIG_MPTCP)
skb_ext_type_len[SKB_EXT_MPTCP] +
+#endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+ skb_ext_type_len[SKB_EXT_MCTP] +
#endif
0;
}
@@ -6519,6 +6526,14 @@ static void skb_ext_put_sp(struct sec_path *sp)
}
#endif
+#ifdef CONFIG_MCTP_FLOWS
+static void skb_ext_put_mctp(struct mctp_flow *flow)
+{
+ if (flow->key)
+ mctp_key_unref(flow->key);
+}
+#endif
+
void __skb_ext_del(struct sk_buff *skb, enum skb_ext_id id)
{
struct skb_ext *ext = skb->extensions;
@@ -6554,6 +6569,10 @@ void __skb_ext_put(struct skb_ext *ext)
if (__skb_ext_exist(ext, SKB_EXT_SEC_PATH))
skb_ext_put_sp(skb_ext_get_ptr(ext, SKB_EXT_SEC_PATH));
#endif
+#ifdef CONFIG_MCTP_FLOWS
+ if (__skb_ext_exist(ext, SKB_EXT_MCTP))
+ skb_ext_put_mctp(skb_ext_get_ptr(ext, SKB_EXT_MCTP));
+#endif
kmem_cache_free(skbuff_ext_cache, ext);
}
diff --git a/net/mctp/Kconfig b/net/mctp/Kconfig
index 868c92272cbd..3a5c0e70da77 100644
--- a/net/mctp/Kconfig
+++ b/net/mctp/Kconfig
@@ -1,7 +1,7 @@
menuconfig MCTP
depends on NET
- tristate "MCTP core protocol support"
+ bool "MCTP core protocol support"
help
Management Component Transport Protocol (MCTP) is an in-system
protocol for communicating between management controllers and
@@ -16,3 +16,8 @@ config MCTP_TEST
bool "MCTP core tests" if !KUNIT_ALL_TESTS
depends on MCTP=y && KUNIT=y
default KUNIT_ALL_TESTS
+
+config MCTP_FLOWS
+ bool
+ depends on MCTP
+ select SKB_EXTENSIONS
--
2.33.0
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH net-next v2 3/3] mctp: Pass flow data & flow release events to drivers
2021-10-29 3:01 [PATCH net-next v2 0/3] MCTP flow support Jeremy Kerr
2021-10-29 3:01 ` [PATCH net-next v2 1/3] mctp: Return new key from mctp_alloc_local_tag Jeremy Kerr
2021-10-29 3:01 ` [PATCH net-next v2 2/3] mctp: Add flow extension to skb Jeremy Kerr
@ 2021-10-29 3:01 ` Jeremy Kerr
2021-10-29 12:40 ` [PATCH net-next v2 0/3] MCTP flow support patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: Jeremy Kerr @ 2021-10-29 3:01 UTC (permalink / raw)
To: netdev; +Cc: David S. Miller, Jakub Kicinski, Matt Johnston, Andrew Jeffery
Now that we have an extension for MCTP data in skbs, populate the flow
when a key has been created for the packet, and add a device driver
operation to inform of flow destruction.
Includes a fix for a warning with test builds:
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
---
v2:
- forward-declare struct mctp_sk_key
---
include/net/mctp.h | 6 +++++
include/net/mctpdevice.h | 16 ++++++++++++
net/mctp/device.c | 51 +++++++++++++++++++++++++++++++++++++
net/mctp/route.c | 55 ++++++++++++++++++++++++++++++++++++++--
4 files changed, 126 insertions(+), 2 deletions(-)
diff --git a/include/net/mctp.h b/include/net/mctp.h
index 7a5ba801703c..7e35ec79b909 100644
--- a/include/net/mctp.h
+++ b/include/net/mctp.h
@@ -152,6 +152,12 @@ struct mctp_sk_key {
/* expiry timeout; valid (above) cleared on expiry */
unsigned long expiry;
+
+ /* free to use for device flow state tracking. Initialised to
+ * zero on initial key creation
+ */
+ unsigned long dev_flow_state;
+ struct mctp_dev *dev;
};
struct mctp_skb_cb {
diff --git a/include/net/mctpdevice.h b/include/net/mctpdevice.h
index 3a439463f055..5c0d04b5c12c 100644
--- a/include/net/mctpdevice.h
+++ b/include/net/mctpdevice.h
@@ -14,6 +14,8 @@
#include <linux/types.h>
#include <linux/refcount.h>
+struct mctp_sk_key;
+
struct mctp_dev {
struct net_device *dev;
@@ -21,6 +23,8 @@ struct mctp_dev {
unsigned int net;
+ const struct mctp_netdev_ops *ops;
+
/* Only modified under RTNL. Reads have addrs_lock held */
u8 *addrs;
size_t num_addrs;
@@ -29,12 +33,24 @@ struct mctp_dev {
struct rcu_head rcu;
};
+struct mctp_netdev_ops {
+ void (*release_flow)(struct mctp_dev *dev,
+ struct mctp_sk_key *key);
+};
+
#define MCTP_INITIAL_DEFAULT_NET 1
struct mctp_dev *mctp_dev_get_rtnl(const struct net_device *dev);
struct mctp_dev *__mctp_dev_get(const struct net_device *dev);
+int mctp_register_netdev(struct net_device *dev,
+ const struct mctp_netdev_ops *ops);
+void mctp_unregister_netdev(struct net_device *dev);
+
void mctp_dev_hold(struct mctp_dev *mdev);
void mctp_dev_put(struct mctp_dev *mdev);
+void mctp_dev_set_key(struct mctp_dev *dev, struct mctp_sk_key *key);
+void mctp_dev_release_key(struct mctp_dev *dev, struct mctp_sk_key *key);
+
#endif /* __NET_MCTPDEVICE_H */
diff --git a/net/mctp/device.c b/net/mctp/device.c
index 3827d62f52c9..8799ee77e7b7 100644
--- a/net/mctp/device.c
+++ b/net/mctp/device.c
@@ -260,6 +260,24 @@ void mctp_dev_put(struct mctp_dev *mdev)
}
}
+void mctp_dev_release_key(struct mctp_dev *dev, struct mctp_sk_key *key)
+ __must_hold(&key->lock)
+{
+ if (!dev)
+ return;
+ if (dev->ops && dev->ops->release_flow)
+ dev->ops->release_flow(dev, key);
+ key->dev = NULL;
+ mctp_dev_put(dev);
+}
+
+void mctp_dev_set_key(struct mctp_dev *dev, struct mctp_sk_key *key)
+ __must_hold(&key->lock)
+{
+ mctp_dev_hold(dev);
+ key->dev = dev;
+}
+
static struct mctp_dev *mctp_add_dev(struct net_device *dev)
{
struct mctp_dev *mdev;
@@ -414,6 +432,39 @@ static int mctp_dev_notify(struct notifier_block *this, unsigned long event,
return NOTIFY_OK;
}
+static int mctp_register_netdevice(struct net_device *dev,
+ const struct mctp_netdev_ops *ops)
+{
+ struct mctp_dev *mdev;
+
+ mdev = mctp_add_dev(dev);
+ if (IS_ERR(mdev))
+ return PTR_ERR(mdev);
+
+ mdev->ops = ops;
+
+ return register_netdevice(dev);
+}
+
+int mctp_register_netdev(struct net_device *dev,
+ const struct mctp_netdev_ops *ops)
+{
+ int rc;
+
+ rtnl_lock();
+ rc = mctp_register_netdevice(dev, ops);
+ rtnl_unlock();
+
+ return rc;
+}
+EXPORT_SYMBOL_GPL(mctp_register_netdev);
+
+void mctp_unregister_netdev(struct net_device *dev)
+{
+ unregister_netdev(dev);
+}
+EXPORT_SYMBOL_GPL(mctp_unregister_netdev);
+
static struct rtnl_af_ops mctp_af_ops = {
.family = AF_MCTP,
.fill_link_af = mctp_fill_link_af,
diff --git a/net/mctp/route.c b/net/mctp/route.c
index 82cc10a2fb0c..46c44823edb7 100644
--- a/net/mctp/route.c
+++ b/net/mctp/route.c
@@ -29,6 +29,8 @@
static const unsigned int mctp_message_maxlen = 64 * 1024;
static const unsigned long mctp_key_lifetime = 6 * CONFIG_HZ;
+static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev);
+
/* route output callbacks */
static int mctp_route_discard(struct mctp_route *route, struct sk_buff *skb)
{
@@ -152,8 +154,19 @@ static struct mctp_sk_key *mctp_key_alloc(struct mctp_sock *msk,
void mctp_key_unref(struct mctp_sk_key *key)
{
- if (refcount_dec_and_test(&key->refs))
- kfree(key);
+ unsigned long flags;
+
+ if (!refcount_dec_and_test(&key->refs))
+ return;
+
+ /* even though no refs exist here, the lock allows us to stay
+ * consistent with the locking requirement of mctp_dev_release_key
+ */
+ spin_lock_irqsave(&key->lock, flags);
+ mctp_dev_release_key(key->dev, key);
+ spin_unlock_irqrestore(&key->lock, flags);
+
+ kfree(key);
}
static int mctp_key_add(struct mctp_sk_key *key, struct mctp_sock *msk)
@@ -204,6 +217,7 @@ static void __mctp_key_unlock_drop(struct mctp_sk_key *key, struct net *net,
key->reasm_head = NULL;
key->reasm_dead = true;
key->valid = false;
+ mctp_dev_release_key(key->dev, key);
spin_unlock_irqrestore(&key->lock, flags);
spin_lock_irqsave(&net->mctp.keys_lock, flags);
@@ -222,6 +236,40 @@ static void __mctp_key_unlock_drop(struct mctp_sk_key *key, struct net *net,
}
+#ifdef CONFIG_MCTP_FLOWS
+static void mctp_skb_set_flow(struct sk_buff *skb, struct mctp_sk_key *key)
+{
+ struct mctp_flow *flow;
+
+ flow = skb_ext_add(skb, SKB_EXT_MCTP);
+ if (!flow)
+ return;
+
+ refcount_inc(&key->refs);
+ flow->key = key;
+}
+
+static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev)
+{
+ struct mctp_sk_key *key;
+ struct mctp_flow *flow;
+
+ flow = skb_ext_find(skb, SKB_EXT_MCTP);
+ if (!flow)
+ return;
+
+ key = flow->key;
+
+ if (WARN_ON(key->dev && key->dev != dev))
+ return;
+
+ mctp_dev_set_key(dev, key);
+}
+#else
+static void mctp_skb_set_flow(struct sk_buff *skb, struct mctp_sk_key *key) {}
+static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev) {}
+#endif
+
static int mctp_frag_queue(struct mctp_sk_key *key, struct sk_buff *skb)
{
struct mctp_hdr *hdr = mctp_hdr(skb);
@@ -465,6 +513,8 @@ static int mctp_route_output(struct mctp_route *route, struct sk_buff *skb)
return -EHOSTUNREACH;
}
+ mctp_flow_prepare_output(skb, route->dev);
+
rc = dev_queue_xmit(skb);
if (rc)
rc = net_xmit_errno(rc);
@@ -803,6 +853,7 @@ int mctp_local_output(struct sock *sk, struct mctp_route *rt,
rc = PTR_ERR(key);
goto out_release;
}
+ mctp_skb_set_flow(skb, key);
/* done with the key in this scope */
mctp_key_unref(key);
tag |= MCTP_HDR_FLAG_TO;
--
2.33.0
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH net-next v2 0/3] MCTP flow support
2021-10-29 3:01 [PATCH net-next v2 0/3] MCTP flow support Jeremy Kerr
` (2 preceding siblings ...)
2021-10-29 3:01 ` [PATCH net-next v2 3/3] mctp: Pass flow data & flow release events to drivers Jeremy Kerr
@ 2021-10-29 12:40 ` patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: patchwork-bot+netdevbpf @ 2021-10-29 12:40 UTC (permalink / raw)
To: Jeremy Kerr; +Cc: netdev, davem, kuba, matt, andrew
Hello:
This series was applied to netdev/net-next.git (master)
by David S. Miller <davem@davemloft.net>:
On Fri, 29 Oct 2021 11:01:42 +0800 you wrote:
> For certain MCTP transport bindings, the binding driver will need to be
> aware of request/response pairing. For example, the i2c binding may need
> to set multiplexer configuration when expecting a response from a device
> behind a mux.
>
> This series implements a mechanism for informing the driver about these
> flows, so it can implement transport-specific behaviour when a flow is
> in progress (ie, a response is expected, and/or we time-out on that
> expectation). We use a skb extension to notify the driver about the
> presence of a flow, and a new dev->ops callback to notify about a flow's
> destruction.
>
> [...]
Here is the summary with links:
- [net-next,v2,1/3] mctp: Return new key from mctp_alloc_local_tag
https://git.kernel.org/netdev/net-next/c/212c10c3c658
- [net-next,v2,2/3] mctp: Add flow extension to skb
https://git.kernel.org/netdev/net-next/c/78476d315e19
- [net-next,v2,3/3] mctp: Pass flow data & flow release events to drivers
https://git.kernel.org/netdev/net-next/c/67737c457281
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] 5+ messages in thread