* [net-next v2 12/18] tipc: introduce group anycast messaging
From: Jon Maloy @ 2017-10-13 9:04 UTC (permalink / raw)
To: davem, netdev; +Cc: parthasarathy.bhuvaragan, ying.xue, tipc-discussion
In-Reply-To: <1507885474-11213-1-git-send-email-jon.maloy@ericsson.com>
In this commit, we make it possible to send connectionless unicast
messages to any member corresponding to the given member identity,
when there is more than one such member. The sender must use a
TIPC_ADDR_NAME address to achieve this effect.
We also perform load balancing between the destinations, i.e., we
primarily select one which has advertised sufficient send window
to not cause a block/EAGAIN delay, if any. This mechanism is
overlayed on the always present round-robin selection.
Anycast messages are subject to the same start synchronization
and flow control mechanism as group broadcast messages.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
---
net/tipc/group.c | 7 +++++
net/tipc/group.h | 3 ++
net/tipc/name_table.c | 41 +++++++++++++++++++++++++
net/tipc/name_table.h | 3 ++
net/tipc/socket.c | 84 +++++++++++++++++++++++++++++++++++++++++++++++++++
5 files changed, 138 insertions(+)
diff --git a/net/tipc/group.c b/net/tipc/group.c
index 18440be..16aaaa9 100644
--- a/net/tipc/group.c
+++ b/net/tipc/group.c
@@ -116,6 +116,13 @@ static bool tipc_group_is_receiver(struct tipc_member *m)
return m && m->state >= MBR_JOINED;
}
+u32 tipc_group_exclude(struct tipc_group *grp)
+{
+ if (!grp->loopback)
+ return grp->portid;
+ return 0;
+}
+
int tipc_group_size(struct tipc_group *grp)
{
return grp->member_cnt;
diff --git a/net/tipc/group.h b/net/tipc/group.h
index 8f77290..e432066 100644
--- a/net/tipc/group.h
+++ b/net/tipc/group.h
@@ -49,6 +49,7 @@ void tipc_group_add_member(struct tipc_group *grp, u32 node, u32 port);
struct tipc_nlist *tipc_group_dests(struct tipc_group *grp);
void tipc_group_self(struct tipc_group *grp, struct tipc_name_seq *seq,
int *scope);
+u32 tipc_group_exclude(struct tipc_group *grp);
void tipc_group_filter_msg(struct tipc_group *grp,
struct sk_buff_head *inputq,
struct sk_buff_head *xmitq);
@@ -68,5 +69,7 @@ void tipc_group_update_rcv_win(struct tipc_group *grp, int blks, u32 node,
u32 port, struct sk_buff_head *xmitq);
u16 tipc_group_bc_snd_nxt(struct tipc_group *grp);
void tipc_group_update_member(struct tipc_member *m, int len);
+struct tipc_member *tipc_group_find_sender(struct tipc_group *grp,
+ u32 node, u32 port);
int tipc_group_size(struct tipc_group *grp);
#endif
diff --git a/net/tipc/name_table.c b/net/tipc/name_table.c
index 114d72b..2856e19 100644
--- a/net/tipc/name_table.c
+++ b/net/tipc/name_table.c
@@ -597,6 +597,47 @@ u32 tipc_nametbl_translate(struct net *net, u32 type, u32 instance,
return ref;
}
+bool tipc_nametbl_lookup(struct net *net, u32 type, u32 instance, u32 domain,
+ struct list_head *dsts, int *dstcnt, u32 exclude,
+ bool all)
+{
+ u32 self = tipc_own_addr(net);
+ struct publication *publ;
+ struct name_info *info;
+ struct name_seq *seq;
+ struct sub_seq *sseq;
+
+ if (!tipc_in_scope(domain, self))
+ return false;
+
+ *dstcnt = 0;
+ rcu_read_lock();
+ seq = nametbl_find_seq(net, type);
+ if (unlikely(!seq))
+ goto exit;
+ spin_lock_bh(&seq->lock);
+ sseq = nameseq_find_subseq(seq, instance);
+ if (likely(sseq)) {
+ info = sseq->info;
+ list_for_each_entry(publ, &info->zone_list, zone_list) {
+ if (!tipc_in_scope(domain, publ->node))
+ continue;
+ if (publ->ref == exclude && publ->node == self)
+ continue;
+ tipc_dest_push(dsts, publ->node, publ->ref);
+ (*dstcnt)++;
+ if (all)
+ continue;
+ list_move_tail(&publ->zone_list, &info->zone_list);
+ break;
+ }
+ }
+ spin_unlock_bh(&seq->lock);
+exit:
+ rcu_read_unlock();
+ return !list_empty(dsts);
+}
+
int tipc_nametbl_mc_translate(struct net *net, u32 type, u32 lower, u32 upper,
u32 limit, struct list_head *dports)
{
diff --git a/net/tipc/name_table.h b/net/tipc/name_table.h
index 97646b1..71926e4 100644
--- a/net/tipc/name_table.h
+++ b/net/tipc/name_table.h
@@ -107,6 +107,9 @@ void tipc_nametbl_build_group(struct net *net, struct tipc_group *grp,
void tipc_nametbl_lookup_dst_nodes(struct net *net, u32 type, u32 lower,
u32 upper, u32 domain,
struct tipc_nlist *nodes);
+bool tipc_nametbl_lookup(struct net *net, u32 type, u32 instance, u32 domain,
+ struct list_head *dsts, int *dstcnt, u32 exclude,
+ bool all);
struct publication *tipc_nametbl_publish(struct net *net, u32 type, u32 lower,
u32 upper, u32 scope, u32 port_ref,
u32 key);
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index e71c8d2..66165e1 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -905,6 +905,88 @@ static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
}
/**
+ * tipc_send_group_anycast - send message to any member with given identity
+ * @sock: socket structure
+ * @m: message to send
+ * @dlen: total length of message data
+ * @timeout: timeout to wait for wakeup
+ *
+ * Called from function tipc_sendmsg(), which has done all sanity checks
+ * Returns the number of bytes sent on success, or errno
+ */
+static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
+ int dlen, long timeout)
+{
+ DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
+ struct sock *sk = sock->sk;
+ struct tipc_sock *tsk = tipc_sk(sk);
+ struct list_head *cong_links = &tsk->cong_links;
+ int blks = tsk_blocks(GROUP_H_SIZE + dlen);
+ struct tipc_group *grp = tsk->group;
+ struct tipc_member *first = NULL;
+ struct tipc_member *mbr = NULL;
+ struct net *net = sock_net(sk);
+ u32 node, port, exclude;
+ u32 type, inst, domain;
+ struct list_head dsts;
+ int lookups = 0;
+ int dstcnt, rc;
+ bool cong;
+
+ INIT_LIST_HEAD(&dsts);
+
+ type = dest->addr.name.name.type;
+ inst = dest->addr.name.name.instance;
+ domain = addr_domain(net, dest->scope);
+ exclude = tipc_group_exclude(grp);
+
+ while (++lookups < 4) {
+ first = NULL;
+
+ /* Look for a non-congested destination member, if any */
+ while (1) {
+ if (!tipc_nametbl_lookup(net, type, inst, domain, &dsts,
+ &dstcnt, exclude, false))
+ return -EHOSTUNREACH;
+ tipc_dest_pop(&dsts, &node, &port);
+ cong = tipc_group_cong(grp, node, port, blks, &mbr);
+ if (!cong)
+ break;
+ if (mbr == first)
+ break;
+ if (!first)
+ first = mbr;
+ }
+
+ /* Start over if destination was not in member list */
+ if (unlikely(!mbr))
+ continue;
+
+ if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
+ break;
+
+ /* Block or return if destination link or member is congested */
+ rc = tipc_wait_for_cond(sock, &timeout,
+ !tipc_dest_find(cong_links, node, 0) &&
+ !tipc_group_cong(grp, node, port,
+ blks, &mbr));
+ if (unlikely(rc))
+ return rc;
+
+ /* Send, unless destination disappeared while waiting */
+ if (likely(mbr))
+ break;
+ }
+
+ if (unlikely(lookups >= 4))
+ return -EHOSTUNREACH;
+
+ rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
+
+ return rc ? rc : dlen;
+}
+
+/**
* tipc_send_group_bcast - send message to all members in communication group
* @sk: socket structure
* @m: message to send
@@ -1127,6 +1209,8 @@ static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
if (grp) {
if (!dest)
return tipc_send_group_bcast(sock, m, dlen, timeout);
+ if (dest->addrtype == TIPC_ADDR_NAME)
+ return tipc_send_group_anycast(sock, m, dlen, timeout);
if (dest->addrtype == TIPC_ADDR_ID)
return tipc_send_group_unicast(sock, m, dlen, timeout);
return -EINVAL;
--
2.1.4
^ permalink raw reply related
* [net-next v2 13/18] tipc: introduce group multicast messaging
From: Jon Maloy @ 2017-10-13 9:04 UTC (permalink / raw)
To: davem, netdev; +Cc: parthasarathy.bhuvaragan, ying.xue, tipc-discussion
In-Reply-To: <1507885474-11213-1-git-send-email-jon.maloy@ericsson.com>
The previously introduced message transport to all group members is
based on the tipc multicast service, but is logically a broadcast
service within the group, and that is what we call it.
We now add functionality for sending messages to all group members
having a certain identity. Correspondingly, we call this feature 'group
multicast'. The service is using unicast when only one destination is
found, otherwise it will use the bearer broadcast service to transfer
the messages. In the latter case, the receiving members filter arriving
messages by looking at the intended destination instance. If there is
no match, the message will be dropped, while still being considered
received and read as seen by the flow control mechanism.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
---
net/tipc/group.c | 14 +++++++++++++-
net/tipc/msg.h | 11 +++++++++--
net/tipc/socket.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++----
3 files changed, 74 insertions(+), 7 deletions(-)
diff --git a/net/tipc/group.c b/net/tipc/group.c
index 16aaaa9..ffac2f3 100644
--- a/net/tipc/group.c
+++ b/net/tipc/group.c
@@ -413,10 +413,22 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
if (!tipc_group_is_receiver(m))
goto drop;
+ m->bc_rcv_nxt = msg_grp_bc_seqno(hdr) + 1;
+
+ /* Drop multicast here if not for this member */
+ if (mtyp == TIPC_GRP_MCAST_MSG) {
+ if (msg_nameinst(hdr) != grp->instance) {
+ m->bc_rcv_nxt = msg_grp_bc_seqno(hdr) + 1;
+ tipc_group_update_rcv_win(grp, msg_blocks(hdr),
+ node, port, xmitq);
+ kfree_skb(skb);
+ return;
+ }
+ }
+
TIPC_SKB_CB(skb)->orig_member = m->instance;
__skb_queue_tail(inputq, skb);
- m->bc_rcv_nxt = msg_grp_bc_seqno(hdr) + 1;
return;
drop:
kfree_skb(skb);
diff --git a/net/tipc/msg.h b/net/tipc/msg.h
index f5033f4..d6f9821 100644
--- a/net/tipc/msg.h
+++ b/net/tipc/msg.h
@@ -67,7 +67,8 @@ struct plist;
#define TIPC_DIRECT_MSG 3
#define TIPC_GRP_MEMBER_EVT 4
#define TIPC_GRP_BCAST_MSG 5
-#define TIPC_GRP_UCAST_MSG 6
+#define TIPC_GRP_MCAST_MSG 6
+#define TIPC_GRP_UCAST_MSG 7
/*
* Internal message users
@@ -195,6 +196,11 @@ static inline u32 msg_size(struct tipc_msg *m)
return msg_bits(m, 0, 0, 0x1ffff);
}
+static inline u32 msg_blocks(struct tipc_msg *m)
+{
+ return (msg_size(m) / 1024) + 1;
+}
+
static inline u32 msg_data_sz(struct tipc_msg *m)
{
return msg_size(m) - msg_hdr_sz(m);
@@ -279,7 +285,8 @@ static inline u32 msg_mcast(struct tipc_msg *m)
{
int mtyp = msg_type(m);
- return ((mtyp == TIPC_MCAST_MSG) || (mtyp == TIPC_GRP_BCAST_MSG));
+ return ((mtyp == TIPC_MCAST_MSG) || (mtyp == TIPC_GRP_BCAST_MSG) ||
+ (mtyp == TIPC_GRP_MCAST_MSG));
}
static inline u32 msg_connected(struct tipc_msg *m)
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 66165e1..8fdd969 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -999,6 +999,7 @@ static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
int dlen, long timeout)
{
+ DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct tipc_sock *tsk = tipc_sk(sk);
@@ -1021,11 +1022,16 @@ static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
return rc;
/* Complete message header */
- msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
+ if (dest) {
+ msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
+ msg_set_nameinst(hdr, dest->addr.name.name.instance);
+ } else {
+ msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
+ msg_set_nameinst(hdr, 0);
+ }
msg_set_hdr_sz(hdr, GROUP_H_SIZE);
msg_set_destport(hdr, 0);
msg_set_destnode(hdr, 0);
- msg_set_nameinst(hdr, 0);
msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(grp));
/* Build message as chain of buffers */
@@ -1046,6 +1052,48 @@ static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
}
/**
+ * tipc_send_group_mcast - send message to all members with given identity
+ * @sock: socket structure
+ * @m: message to send
+ * @dlen: total length of message data
+ * @timeout: timeout to wait for wakeup
+ *
+ * Called from function tipc_sendmsg(), which has done all sanity checks
+ * Returns the number of bytes sent on success, or errno
+ */
+static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
+ int dlen, long timeout)
+{
+ struct sock *sk = sock->sk;
+ DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
+ struct tipc_name_seq *seq = &dest->addr.nameseq;
+ struct tipc_sock *tsk = tipc_sk(sk);
+ struct tipc_group *grp = tsk->group;
+ struct net *net = sock_net(sk);
+ u32 domain, exclude, dstcnt;
+ struct list_head dsts;
+
+ INIT_LIST_HEAD(&dsts);
+
+ if (seq->lower != seq->upper)
+ return -ENOTSUPP;
+
+ domain = addr_domain(net, dest->scope);
+ exclude = tipc_group_exclude(grp);
+ if (!tipc_nametbl_lookup(net, seq->type, seq->lower, domain,
+ &dsts, &dstcnt, exclude, true))
+ return -EHOSTUNREACH;
+
+ if (dstcnt == 1) {
+ tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref);
+ return tipc_send_group_unicast(sock, m, dlen, timeout);
+ }
+
+ tipc_dest_list_purge(&dsts);
+ return tipc_send_group_bcast(sock, m, dlen, timeout);
+}
+
+/**
* tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
* @arrvq: queue with arriving messages, to be cloned after destination lookup
* @inputq: queue with cloned messages, delivered to socket after dest lookup
@@ -1213,6 +1261,8 @@ static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
return tipc_send_group_anycast(sock, m, dlen, timeout);
if (dest->addrtype == TIPC_ADDR_ID)
return tipc_send_group_unicast(sock, m, dlen, timeout);
+ if (dest->addrtype == TIPC_ADDR_MCAST)
+ return tipc_send_group_mcast(sock, m, dlen, timeout);
return -EINVAL;
}
@@ -2022,8 +2072,6 @@ static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
if (unlikely(!msg_isdata(hdr)))
tipc_sk_proto_rcv(sk, &inputq, xmitq);
- else if (unlikely(msg_type(hdr) > TIPC_GRP_UCAST_MSG))
- return kfree_skb(skb);
if (unlikely(grp))
tipc_group_filter_msg(grp, &inputq, xmitq);
--
2.1.4
^ permalink raw reply related
* [net-next v2 14/18] tipc: guarantee group unicast doesn't bypass group broadcast
From: Jon Maloy @ 2017-10-13 9:04 UTC (permalink / raw)
To: davem, netdev; +Cc: parthasarathy.bhuvaragan, ying.xue, tipc-discussion
In-Reply-To: <1507885474-11213-1-git-send-email-jon.maloy@ericsson.com>
Group unicast messages don't follow the same path as broadcast messages,
and there is a high risk that unicasts sent from a socket might bypass
previously sent broadcasts from the same socket.
We fix this by letting all unicast messages carry the sequence number of
the next sent broadcast from the same node, but without updating this
number at the receiver. This way, a receiver can check and if necessary
re-order such messages before they are added to the socket receive buffer.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
---
net/tipc/group.c | 87 +++++++++++++++++++++++++++++++++++++++++++++----------
net/tipc/socket.c | 2 ++
2 files changed, 74 insertions(+), 15 deletions(-)
diff --git a/net/tipc/group.c b/net/tipc/group.c
index ffac2f3..985e0ce 100644
--- a/net/tipc/group.c
+++ b/net/tipc/group.c
@@ -62,6 +62,7 @@ struct tipc_member {
struct list_head list;
struct list_head congested;
struct sk_buff *event_msg;
+ struct sk_buff_head deferredq;
struct tipc_group *group;
u32 node;
u32 port;
@@ -253,6 +254,7 @@ static struct tipc_member *tipc_group_create_member(struct tipc_group *grp,
return NULL;
INIT_LIST_HEAD(&m->list);
INIT_LIST_HEAD(&m->congested);
+ __skb_queue_head_init(&m->deferredq);
m->group = grp;
m->node = node;
m->port = port;
@@ -380,29 +382,54 @@ bool tipc_group_bc_cong(struct tipc_group *grp, int len)
return tipc_group_cong(grp, m->node, m->port, len, &m);
}
+/* tipc_group_sort_msg() - sort msg into queue by bcast sequence number
+ */
+static void tipc_group_sort_msg(struct sk_buff *skb, struct sk_buff_head *defq)
+{
+ struct tipc_msg *_hdr, *hdr = buf_msg(skb);
+ u16 bc_seqno = msg_grp_bc_seqno(hdr);
+ struct sk_buff *_skb, *tmp;
+ int mtyp = msg_type(hdr);
+
+ /* Bcast may be bypassed by unicast, - sort it in */
+ if (mtyp == TIPC_GRP_BCAST_MSG || mtyp == TIPC_GRP_MCAST_MSG) {
+ skb_queue_walk_safe(defq, _skb, tmp) {
+ _hdr = buf_msg(_skb);
+ if (!less(bc_seqno, msg_grp_bc_seqno(_hdr)))
+ continue;
+ __skb_queue_before(defq, _skb, skb);
+ return;
+ }
+ /* Bcast was not bypassed, - add to tail */
+ }
+ /* Unicasts are never bypassed, - always add to tail */
+ __skb_queue_tail(defq, skb);
+}
+
/* tipc_group_filter_msg() - determine if we should accept arriving message
*/
void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
struct sk_buff_head *xmitq)
{
struct sk_buff *skb = __skb_dequeue(inputq);
+ struct sk_buff_head *defq;
struct tipc_member *m;
struct tipc_msg *hdr;
+ bool deliver, update;
u32 node, port;
- int mtyp;
+ int mtyp, blks;
if (!skb)
return;
hdr = buf_msg(skb);
- mtyp = msg_type(hdr);
node = msg_orignode(hdr);
port = msg_origport(hdr);
if (!msg_in_group(hdr))
goto drop;
- if (mtyp == TIPC_GRP_MEMBER_EVT) {
+ if (msg_is_grp_evt(hdr)) {
if (!grp->events)
goto drop;
__skb_queue_tail(inputq, skb);
@@ -413,22 +440,52 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
if (!tipc_group_is_receiver(m))
goto drop;
- m->bc_rcv_nxt = msg_grp_bc_seqno(hdr) + 1;
+ if (less(msg_grp_bc_seqno(hdr), m->bc_rcv_nxt))
+ goto drop;
- /* Drop multicast here if not for this member */
- if (mtyp == TIPC_GRP_MCAST_MSG) {
- if (msg_nameinst(hdr) != grp->instance) {
- m->bc_rcv_nxt = msg_grp_bc_seqno(hdr) + 1;
- tipc_group_update_rcv_win(grp, msg_blocks(hdr),
- node, port, xmitq);
- kfree_skb(skb);
- return;
+ TIPC_SKB_CB(skb)->orig_member = m->instance;
+ defq = &m->deferredq;
+ tipc_group_sort_msg(skb, defq);
+
+ while ((skb = skb_peek(defq))) {
+ hdr = buf_msg(skb);
+ mtyp = msg_type(hdr);
+ deliver = true;
+ update = false;
+
+ if (more(msg_grp_bc_seqno(hdr), m->bc_rcv_nxt))
+ break;
+
+ /* Decide what to do with message */
+ switch (mtyp) {
+ case TIPC_GRP_MCAST_MSG:
+ if (msg_nameinst(hdr) != grp->instance) {
+ update = true;
+ deliver = false;
+ }
+ /* Fall thru */
+ case TIPC_GRP_BCAST_MSG:
+ m->bc_rcv_nxt++;
+ break;
+ case TIPC_GRP_UCAST_MSG:
+ break;
+ default:
+ break;
}
- }
- TIPC_SKB_CB(skb)->orig_member = m->instance;
- __skb_queue_tail(inputq, skb);
+ /* Execute decisions */
+ __skb_dequeue(defq);
+ if (deliver)
+ __skb_queue_tail(inputq, skb);
+ else
+ kfree_skb(skb);
+
+ if (!update)
+ continue;
+ blks = msg_blocks(hdr);
+ tipc_group_update_rcv_win(grp, blks, node, port, xmitq);
+ }
return;
drop:
kfree_skb(skb);
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 8fdd969..3276b7a 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -830,6 +830,7 @@ static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
struct msghdr *m, struct tipc_member *mb,
u32 dnode, u32 dport, int dlen)
{
+ u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
int blks = tsk_blocks(GROUP_H_SIZE + dlen);
struct tipc_msg *hdr = &tsk->phdr;
struct sk_buff_head pkts;
@@ -840,6 +841,7 @@ static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
msg_set_hdr_sz(hdr, GROUP_H_SIZE);
msg_set_destport(hdr, dport);
msg_set_destnode(hdr, dnode);
+ msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
/* Build message as chain of buffers */
skb_queue_head_init(&pkts);
--
2.1.4
^ permalink raw reply related
* [net-next v2 18/18] tipc: add multipoint-to-point flow control
From: Jon Maloy @ 2017-10-13 9:04 UTC (permalink / raw)
To: davem, netdev; +Cc: parthasarathy.bhuvaragan, ying.xue, tipc-discussion
In-Reply-To: <1507885474-11213-1-git-send-email-jon.maloy@ericsson.com>
We already have point-to-multipoint flow control within a group. But
we even need the opposite; -a scheme which can handle that potentially
hundreds of sources may try to send messages to the same destination
simultaneously without causing buffer overflow at the recipient. This
commit adds such a mechanism.
The algorithm works as follows:
- When a member detects a new, joining member, it initially set its
state to JOINED and advertises a minimum window to the new member.
This window is chosen so that the new member can send exactly one
maximum sized message, or several smaller ones, to the recipient
before it must stop and wait for an additional advertisement. This
minimum window ADV_IDLE is set to 65 1kB blocks.
- When a member receives the first data message from a JOINED member,
it changes the state of the latter to ACTIVE, and advertises a larger
window ADV_ACTIVE = 12 x ADV_IDLE blocks to the sender, so it can
continue sending with minimal disturbances to the data flow.
- The active members are kept in a dedicated linked list. Each time a
message is received from an active member, it will be moved to the
tail of that list. This way, we keep a record of which members have
been most (tail) and least (head) recently active.
- There is a maximum number (16) of permitted simultaneous active
senders per receiver. When this limit is reached, the receiver will
not advertise anything immediately to a new sender, but instead put
it in a PENDING state, and add it to a corresponding queue. At the
same time, it will pick the least recently active member, send it an
advertisement RECLAIM message, and set this member to state
RECLAIMING.
- The reclaimee member has to respond with a REMIT message, meaning that
it goes back to a send window of ADV_IDLE, and returns its unused
advertised blocks beyond that value to the reclaiming member.
- When the reclaiming member receives the REMIT message, it unlinks
the reclaimee from its active list, resets its state to JOINED, and
notes that it is now back at ADV_IDLE advertised blocks to that
member. If there are still unread data messages sent out by
reclaimee before the REMIT, the member goes into an intermediate
state REMITTED, where it stays until the said messages have been
consumed.
- The returned advertised blocks can now be re-advertised to the
pending member, which is now set to state ACTIVE and added to
the active member list.
- To be proactive, i.e., to minimize the risk that any member will
end up in the pending queue, we start reclaiming resources already
when the number of active members exceeds 3/4 of the permitted
maximum.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
---
net/tipc/group.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++++++---
net/tipc/msg.h | 12 ++++++
2 files changed, 136 insertions(+), 5 deletions(-)
diff --git a/net/tipc/group.c b/net/tipc/group.c
index 8f0eb5d..7821085 100644
--- a/net/tipc/group.c
+++ b/net/tipc/group.c
@@ -54,6 +54,10 @@ enum mbr_state {
MBR_JOINING,
MBR_PUBLISHED,
MBR_JOINED,
+ MBR_PENDING,
+ MBR_ACTIVE,
+ MBR_RECLAIMING,
+ MBR_REMITTED,
MBR_LEAVING
};
@@ -79,6 +83,9 @@ struct tipc_member {
struct tipc_group {
struct rb_root members;
struct list_head congested;
+ struct list_head pending;
+ struct list_head active;
+ struct list_head reclaiming;
struct tipc_nlist dests;
struct net *net;
int subid;
@@ -88,6 +95,8 @@ struct tipc_group {
u32 scope;
u32 portid;
u16 member_cnt;
+ u16 active_cnt;
+ u16 max_active;
u16 bc_snd_nxt;
u16 bc_ackers;
bool loopback;
@@ -97,12 +106,29 @@ struct tipc_group {
static void tipc_group_proto_xmit(struct tipc_group *grp, struct tipc_member *m,
int mtyp, struct sk_buff_head *xmitq);
+static void tipc_group_decr_active(struct tipc_group *grp,
+ struct tipc_member *m)
+{
+ if (m->state == MBR_ACTIVE || m->state == MBR_RECLAIMING)
+ grp->active_cnt--;
+}
+
static int tipc_group_rcvbuf_limit(struct tipc_group *grp)
{
+ int max_active, active_pool, idle_pool;
int mcnt = grp->member_cnt + 1;
+ /* Limit simultaneous reception from other members */
+ max_active = min(mcnt / 8, 64);
+ max_active = max(max_active, 16);
+ grp->max_active = max_active;
+
+ /* Reserve blocks for active and idle members */
+ active_pool = max_active * ADV_ACTIVE;
+ idle_pool = (mcnt - max_active) * ADV_IDLE;
+
/* Scale to bytes, considering worst-case truesize/msgsize ratio */
- return mcnt * ADV_ACTIVE * FLOWCTL_BLK_SZ * 4;
+ return (active_pool + idle_pool) * FLOWCTL_BLK_SZ * 4;
}
u16 tipc_group_bc_snd_nxt(struct tipc_group *grp)
@@ -143,6 +169,9 @@ struct tipc_group *tipc_group_create(struct net *net, u32 portid,
return NULL;
tipc_nlist_init(&grp->dests, tipc_own_addr(net));
INIT_LIST_HEAD(&grp->congested);
+ INIT_LIST_HEAD(&grp->active);
+ INIT_LIST_HEAD(&grp->pending);
+ INIT_LIST_HEAD(&grp->reclaiming);
grp->members = RB_ROOT;
grp->net = net;
grp->portid = portid;
@@ -286,6 +315,7 @@ static void tipc_group_delete_member(struct tipc_group *grp,
list_del_init(&m->list);
list_del_init(&m->congested);
+ tipc_group_decr_active(grp, m);
/* If last member on a node, remove node from dest list */
if (!tipc_group_find_node(grp, m->node))
@@ -378,6 +408,10 @@ bool tipc_group_cong(struct tipc_group *grp, u32 dnode, u32 dport,
return true;
if (state == MBR_JOINED && adv == ADV_IDLE)
return true;
+ if (state == MBR_ACTIVE && adv == ADV_ACTIVE)
+ return true;
+ if (state == MBR_PENDING && adv == ADV_IDLE)
+ return true;
skb_queue_head_init(&xmitq);
tipc_group_proto_xmit(grp, m, GRP_ADV_MSG, &xmitq);
tipc_node_distr_xmit(grp->net, &xmitq);
@@ -523,7 +557,11 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
void tipc_group_update_rcv_win(struct tipc_group *grp, int blks, u32 node,
u32 port, struct sk_buff_head *xmitq)
{
- struct tipc_member *m;
+ struct list_head *active = &grp->active;
+ int max_active = grp->max_active;
+ int reclaim_limit = max_active * 3 / 4;
+ int active_cnt = grp->active_cnt;
+ struct tipc_member *m, *rm;
m = tipc_group_find_member(grp, node, port);
if (!m)
@@ -533,9 +571,41 @@ void tipc_group_update_rcv_win(struct tipc_group *grp, int blks, u32 node,
switch (m->state) {
case MBR_JOINED:
- if (m->advertised <= (ADV_ACTIVE - ADV_UNIT))
+ /* Reclaim advertised space from least active member */
+ if (!list_empty(active) && active_cnt >= reclaim_limit) {
+ rm = list_first_entry(active, struct tipc_member, list);
+ rm->state = MBR_RECLAIMING;
+ list_move_tail(&rm->list, &grp->reclaiming);
+ tipc_group_proto_xmit(grp, rm, GRP_RECLAIM_MSG, xmitq);
+ }
+ /* If max active, become pending and wait for reclaimed space */
+ if (active_cnt >= max_active) {
+ m->state = MBR_PENDING;
+ list_add_tail(&m->list, &grp->pending);
+ break;
+ }
+ /* Otherwise become active */
+ m->state = MBR_ACTIVE;
+ list_add_tail(&m->list, &grp->active);
+ grp->active_cnt++;
+ /* Fall through */
+ case MBR_ACTIVE:
+ if (!list_is_last(&m->list, &grp->active))
+ list_move_tail(&m->list, &grp->active);
+ if (m->advertised > (ADV_ACTIVE * 3 / 4))
+ break;
+ tipc_group_proto_xmit(grp, m, GRP_ADV_MSG, xmitq);
+ break;
+ case MBR_REMITTED:
+ if (m->advertised > ADV_IDLE)
+ break;
+ m->state = MBR_JOINED;
+ if (m->advertised < ADV_IDLE) {
+ pr_warn_ratelimited("Rcv unexpected msg after REMIT\n");
tipc_group_proto_xmit(grp, m, GRP_ADV_MSG, xmitq);
+ }
break;
+ case MBR_RECLAIMING:
case MBR_DISCOVERED:
case MBR_JOINING:
case MBR_LEAVING:
@@ -557,8 +627,10 @@ static void tipc_group_proto_xmit(struct tipc_group *grp, struct tipc_member *m,
if (!skb)
return;
- if (m->state == MBR_JOINED)
+ if (m->state == MBR_ACTIVE)
adv = ADV_ACTIVE - m->advertised;
+ else if (m->state == MBR_JOINED || m->state == MBR_PENDING)
+ adv = ADV_IDLE - m->advertised;
hdr = buf_msg(skb);
@@ -573,6 +645,8 @@ static void tipc_group_proto_xmit(struct tipc_group *grp, struct tipc_member *m,
m->advertised += adv;
} else if (mtyp == GRP_ACK_MSG) {
msg_set_grp_bc_acked(hdr, m->bc_rcv_nxt);
+ } else if (mtyp == GRP_REMIT_MSG) {
+ msg_set_grp_remitted(hdr, m->window);
}
__skb_queue_tail(xmitq, skb);
}
@@ -583,8 +657,9 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *usr_wakeup,
{
u32 node = msg_orignode(hdr);
u32 port = msg_origport(hdr);
- struct tipc_member *m;
+ struct tipc_member *m, *pm;
struct tipc_msg *ehdr;
+ u16 remitted, in_flight;
if (!grp)
return;
@@ -626,6 +701,7 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *usr_wakeup,
/* Wait until WITHDRAW event is received */
if (m->state != MBR_LEAVING) {
+ tipc_group_decr_active(grp, m);
m->state = MBR_LEAVING;
return;
}
@@ -653,6 +729,48 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *usr_wakeup,
*usr_wakeup = true;
m->usr_pending = false;
return;
+ case GRP_RECLAIM_MSG:
+ if (!m)
+ return;
+ *usr_wakeup = m->usr_pending;
+ m->usr_pending = false;
+ tipc_group_proto_xmit(grp, m, GRP_REMIT_MSG, xmitq);
+ m->window = ADV_IDLE;
+ return;
+ case GRP_REMIT_MSG:
+ if (!m || m->state != MBR_RECLAIMING)
+ return;
+
+ list_del_init(&m->list);
+ grp->active_cnt--;
+ remitted = msg_grp_remitted(hdr);
+
+ /* Messages preceding the REMIT still in receive queue */
+ if (m->advertised > remitted) {
+ m->state = MBR_REMITTED;
+ in_flight = m->advertised - remitted;
+ }
+ /* All messages preceding the REMIT have been read */
+ if (m->advertised <= remitted) {
+ m->state = MBR_JOINED;
+ in_flight = 0;
+ }
+ /* ..and the REMIT overtaken by more messages => re-advertise */
+ if (m->advertised < remitted)
+ tipc_group_proto_xmit(grp, m, GRP_ADV_MSG, xmitq);
+
+ m->advertised = ADV_IDLE + in_flight;
+
+ /* Set oldest pending member to active and advertise */
+ if (list_empty(&grp->pending))
+ return;
+ pm = list_first_entry(&grp->pending, struct tipc_member, list);
+ pm->state = MBR_ACTIVE;
+ list_move_tail(&pm->list, &grp->active);
+ grp->active_cnt++;
+ if (pm->advertised <= (ADV_ACTIVE * 3 / 4))
+ tipc_group_proto_xmit(grp, pm, GRP_ADV_MSG, xmitq);
+ return;
default:
pr_warn("Received unknown GROUP_PROTO message\n");
}
@@ -735,6 +853,7 @@ void tipc_group_member_evt(struct tipc_group *grp,
/* Hold back event if more messages might be expected */
if (m->state != MBR_LEAVING && node_up) {
m->event_msg = skb;
+ tipc_group_decr_active(grp, m);
m->state = MBR_LEAVING;
} else {
if (node_up)
diff --git a/net/tipc/msg.h b/net/tipc/msg.h
index 52c6a2e..cedf811 100644
--- a/net/tipc/msg.h
+++ b/net/tipc/msg.h
@@ -548,6 +548,8 @@ static inline void msg_set_nameupper(struct tipc_msg *m, u32 n)
#define GRP_LEAVE_MSG 1
#define GRP_ADV_MSG 2
#define GRP_ACK_MSG 3
+#define GRP_RECLAIM_MSG 4
+#define GRP_REMIT_MSG 5
/*
* Word 1
@@ -850,6 +852,16 @@ static inline void msg_set_grp_bc_acked(struct tipc_msg *m, u16 n)
msg_set_bits(m, 9, 16, 0xffff, n);
}
+static inline u16 msg_grp_remitted(struct tipc_msg *m)
+{
+ return msg_bits(m, 9, 16, 0xffff);
+}
+
+static inline void msg_set_grp_remitted(struct tipc_msg *m, u16 n)
+{
+ msg_set_bits(m, 9, 16, 0xffff, n);
+}
+
/* Word 10
*/
static inline u16 msg_grp_evt(struct tipc_msg *m)
--
2.1.4
^ permalink raw reply related
* [net-next v2 15/18] tipc: guarantee that group broadcast doesn't bypass group unicast
From: Jon Maloy @ 2017-10-13 9:04 UTC (permalink / raw)
To: davem, netdev; +Cc: parthasarathy.bhuvaragan, ying.xue, tipc-discussion
In-Reply-To: <1507885474-11213-1-git-send-email-jon.maloy@ericsson.com>
We need a mechanism guaranteeing that group unicasts sent out from a
socket are not bypassed by later sent broadcasts from the same socket.
We do this as follows:
- Each time a unicast is sent, we set a the broadcast method for the
socket to "replicast" and "mandatory". This forces the first
subsequent broadcast message to follow the same network and data path
as the preceding unicast to a destination, hence preventing it from
overtaking the latter.
- In order to make the 'same data path' statement above true, we let
group unicasts pass through the multicast link input queue, instead
of as previously through the unicast link input queue.
- In the first broadcast following a unicast, we set a new header flag,
requiring all recipients to immediately acknowledge its reception.
- During the period before all the expected acknowledges are received,
the socket refuses to accept any more broadcast attempts, i.e., by
blocking or returning EAGAIN. This period should typically not be
longer than a few microseconds.
- When all acknowledges have been received, the sending socket will
open up for subsequent broadcasts, this time giving the link layer
freedom to itself select the best transmission method.
- The forced and/or abrupt transmission method changes described above
may lead to broadcasts arriving out of order to the recipients. We
remedy this by introducing code that checks and if necessary
re-orders such messages at the receiving end.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
---
net/tipc/group.c | 47 +++++++++++++++++++++++++++++++++++++++++------
net/tipc/group.h | 4 +---
net/tipc/link.c | 5 ++---
net/tipc/msg.h | 21 +++++++++++++++++++++
net/tipc/socket.c | 34 +++++++++++++++++++++++++++++-----
5 files changed, 94 insertions(+), 17 deletions(-)
diff --git a/net/tipc/group.c b/net/tipc/group.c
index 985e0ce..eab862e 100644
--- a/net/tipc/group.c
+++ b/net/tipc/group.c
@@ -71,6 +71,7 @@ struct tipc_member {
u16 advertised;
u16 window;
u16 bc_rcv_nxt;
+ u16 bc_acked;
bool usr_pending;
};
@@ -87,6 +88,7 @@ struct tipc_group {
u32 portid;
u16 member_cnt;
u16 bc_snd_nxt;
+ u16 bc_ackers;
bool loopback;
bool events;
};
@@ -258,6 +260,7 @@ static struct tipc_member *tipc_group_create_member(struct tipc_group *grp,
m->group = grp;
m->node = node;
m->port = port;
+ m->bc_acked = grp->bc_snd_nxt - 1;
grp->member_cnt++;
tipc_group_add_to_tree(grp, m);
tipc_nlist_add(&grp->dests, m->node);
@@ -275,6 +278,11 @@ static void tipc_group_delete_member(struct tipc_group *grp,
{
rb_erase(&m->tree_node, &grp->members);
grp->member_cnt--;
+
+ /* Check if we were waiting for replicast ack from this member */
+ if (grp->bc_ackers && less(m->bc_acked, grp->bc_snd_nxt - 1))
+ grp->bc_ackers--;
+
list_del_init(&m->list);
list_del_init(&m->congested);
@@ -325,16 +333,23 @@ void tipc_group_update_member(struct tipc_member *m, int len)
list_add_tail(&m->congested, &grp->congested);
}
-void tipc_group_update_bc_members(struct tipc_group *grp, int len)
+void tipc_group_update_bc_members(struct tipc_group *grp, int len, bool ack)
{
+ u16 prev = grp->bc_snd_nxt - 1;
struct tipc_member *m;
struct rb_node *n;
for (n = rb_first(&grp->members); n; n = rb_next(n)) {
m = container_of(n, struct tipc_member, tree_node);
- if (tipc_group_is_enabled(m))
+ if (tipc_group_is_enabled(m)) {
tipc_group_update_member(m, len);
+ m->bc_acked = prev;
+ }
}
+
+ /* Mark number of acknowledges to expect, if any */
+ if (ack)
+ grp->bc_ackers = grp->member_cnt;
grp->bc_snd_nxt++;
}
@@ -372,6 +387,10 @@ bool tipc_group_bc_cong(struct tipc_group *grp, int len)
{
struct tipc_member *m = NULL;
+ /* If prev bcast was replicast, reject until all receivers have acked */
+ if (grp->bc_ackers)
+ return true;
+
if (list_empty(&grp->congested))
return false;
@@ -391,7 +410,7 @@ static void tipc_group_sort_msg(struct sk_buff *skb, struct sk_buff_head *defq)
struct sk_buff *_skb, *tmp;
int mtyp = msg_type(hdr);
- /* Bcast may be bypassed by unicast, - sort it in */
+ /* Bcast may be bypassed by unicast or other bcast, - sort it in */
if (mtyp == TIPC_GRP_BCAST_MSG || mtyp == TIPC_GRP_MCAST_MSG) {
skb_queue_walk_safe(defq, _skb, tmp) {
_hdr = buf_msg(_skb);
@@ -412,10 +431,10 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
struct sk_buff_head *xmitq)
{
struct sk_buff *skb = __skb_dequeue(inputq);
+ bool ack, deliver, update;
struct sk_buff_head *defq;
struct tipc_member *m;
struct tipc_msg *hdr;
- bool deliver, update;
u32 node, port;
int mtyp, blks;
@@ -451,6 +470,7 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
hdr = buf_msg(skb);
mtyp = msg_type(hdr);
deliver = true;
+ ack = false;
update = false;
if (more(msg_grp_bc_seqno(hdr), m->bc_rcv_nxt))
@@ -466,6 +486,7 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
/* Fall thru */
case TIPC_GRP_BCAST_MSG:
m->bc_rcv_nxt++;
+ ack = msg_grp_bc_ack_req(hdr);
break;
case TIPC_GRP_UCAST_MSG:
break;
@@ -480,6 +501,9 @@ void tipc_group_filter_msg(struct tipc_group *grp, struct sk_buff_head *inputq,
else
kfree_skb(skb);
+ if (ack)
+ tipc_group_proto_xmit(grp, m, GRP_ACK_MSG, xmitq);
+
if (!update)
continue;
@@ -540,6 +564,8 @@ static void tipc_group_proto_xmit(struct tipc_group *grp, struct tipc_member *m,
} else if (mtyp == GRP_ADV_MSG) {
msg_set_adv_win(hdr, adv);
m->advertised += adv;
+ } else if (mtyp == GRP_ACK_MSG) {
+ msg_set_grp_bc_acked(hdr, m->bc_rcv_nxt);
}
__skb_queue_tail(xmitq, skb);
}
@@ -593,7 +619,7 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *usr_wakeup,
}
/* Otherwise deliver already received WITHDRAW event */
__skb_queue_tail(inputq, m->event_msg);
- *usr_wakeup = m->usr_pending;
+ *usr_wakeup = true;
tipc_group_delete_member(grp, m);
list_del_init(&m->congested);
return;
@@ -605,6 +631,15 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *usr_wakeup,
m->usr_pending = false;
list_del_init(&m->congested);
return;
+ case GRP_ACK_MSG:
+ if (!m)
+ return;
+ m->bc_acked = msg_grp_bc_acked(hdr);
+ if (--grp->bc_ackers)
+ break;
+ *usr_wakeup = true;
+ m->usr_pending = false;
+ return;
default:
pr_warn("Received unknown GROUP_PROTO message\n");
}
@@ -678,7 +713,7 @@ void tipc_group_member_evt(struct tipc_group *grp,
TIPC_SKB_CB(skb)->orig_member = m->instance;
- *usr_wakeup = m->usr_pending;
+ *usr_wakeup = true;
m->usr_pending = false;
/* Hold back event if more messages might be expected */
diff --git a/net/tipc/group.h b/net/tipc/group.h
index e432066..d525e1c 100644
--- a/net/tipc/group.h
+++ b/net/tipc/group.h
@@ -61,7 +61,7 @@ void tipc_group_proto_rcv(struct tipc_group *grp, bool *wakeup,
struct tipc_msg *hdr,
struct sk_buff_head *inputq,
struct sk_buff_head *xmitq);
-void tipc_group_update_bc_members(struct tipc_group *grp, int len);
+void tipc_group_update_bc_members(struct tipc_group *grp, int len, bool ack);
bool tipc_group_cong(struct tipc_group *grp, u32 dnode, u32 dport,
int len, struct tipc_member **m);
bool tipc_group_bc_cong(struct tipc_group *grp, int len);
@@ -69,7 +69,5 @@ void tipc_group_update_rcv_win(struct tipc_group *grp, int blks, u32 node,
u32 port, struct sk_buff_head *xmitq);
u16 tipc_group_bc_snd_nxt(struct tipc_group *grp);
void tipc_group_update_member(struct tipc_member *m, int len);
-struct tipc_member *tipc_group_find_sender(struct tipc_group *grp,
- u32 node, u32 port);
int tipc_group_size(struct tipc_group *grp);
#endif
diff --git a/net/tipc/link.c b/net/tipc/link.c
index bd25bff..70a2149 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -1046,13 +1046,12 @@ static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
case TIPC_MEDIUM_IMPORTANCE:
case TIPC_HIGH_IMPORTANCE:
case TIPC_CRITICAL_IMPORTANCE:
- if (unlikely(msg_mcast(hdr))) {
+ if (unlikely(msg_in_group(hdr) || msg_mcast(hdr))) {
skb_queue_tail(l->bc_rcvlink->inputq, skb);
return true;
}
- case CONN_MANAGER:
case GROUP_PROTOCOL:
- skb_queue_tail(inputq, skb);
+ case CONN_MANAGER:
return true;
case NAME_DISTRIBUTOR:
l->bc_rcvlink->state = LINK_ESTABLISHED;
diff --git a/net/tipc/msg.h b/net/tipc/msg.h
index d6f9821..52c6a2e 100644
--- a/net/tipc/msg.h
+++ b/net/tipc/msg.h
@@ -547,6 +547,7 @@ static inline void msg_set_nameupper(struct tipc_msg *m, u32 n)
#define GRP_JOIN_MSG 0
#define GRP_LEAVE_MSG 1
#define GRP_ADV_MSG 2
+#define GRP_ACK_MSG 3
/*
* Word 1
@@ -839,6 +840,16 @@ static inline void msg_set_grp_bc_syncpt(struct tipc_msg *m, u16 n)
msg_set_bits(m, 9, 16, 0xffff, n);
}
+static inline u16 msg_grp_bc_acked(struct tipc_msg *m)
+{
+ return msg_bits(m, 9, 16, 0xffff);
+}
+
+static inline void msg_set_grp_bc_acked(struct tipc_msg *m, u16 n)
+{
+ msg_set_bits(m, 9, 16, 0xffff, n);
+}
+
/* Word 10
*/
static inline u16 msg_grp_evt(struct tipc_msg *m)
@@ -851,6 +862,16 @@ static inline void msg_set_grp_evt(struct tipc_msg *m, int n)
msg_set_bits(m, 10, 0, 0x3, n);
}
+static inline u16 msg_grp_bc_ack_req(struct tipc_msg *m)
+{
+ return msg_bits(m, 10, 0, 0x1);
+}
+
+static inline void msg_set_grp_bc_ack_req(struct tipc_msg *m, bool n)
+{
+ msg_set_bits(m, 10, 0, 0x1, n);
+}
+
static inline u16 msg_grp_bc_seqno(struct tipc_msg *m)
{
return msg_bits(m, 10, 16, 0xffff);
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 3276b7a..b1f1c3c 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -831,6 +831,7 @@ static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
u32 dnode, u32 dport, int dlen)
{
u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
+ struct tipc_mc_method *method = &tsk->mc_method;
int blks = tsk_blocks(GROUP_H_SIZE + dlen);
struct tipc_msg *hdr = &tsk->phdr;
struct sk_buff_head pkts;
@@ -857,9 +858,12 @@ static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
tsk->cong_link_cnt++;
}
- /* Update send window and sequence number */
+ /* Update send window */
tipc_group_update_member(mb, blks);
+ /* A broadcast sent within next EXPIRE period must follow same path */
+ method->rcast = true;
+ method->mandatory = true;
return dlen;
}
@@ -1008,6 +1012,7 @@ static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
struct tipc_group *grp = tsk->group;
struct tipc_nlist *dsts = tipc_group_dests(grp);
struct tipc_mc_method *method = &tsk->mc_method;
+ bool ack = method->mandatory && method->rcast;
int blks = tsk_blocks(MCAST_H_SIZE + dlen);
struct tipc_msg *hdr = &tsk->phdr;
int mtu = tipc_bcast_get_mtu(net);
@@ -1036,6 +1041,9 @@ static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
msg_set_destnode(hdr, 0);
msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(grp));
+ /* Avoid getting stuck with repeated forced replicasts */
+ msg_set_grp_bc_ack_req(hdr, ack);
+
/* Build message as chain of buffers */
skb_queue_head_init(&pkts);
rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
@@ -1043,13 +1051,17 @@ static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
return rc;
/* Send message */
- rc = tipc_mcast_xmit(net, &pkts, method, dsts,
- &tsk->cong_link_cnt);
+ rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
if (unlikely(rc))
return rc;
/* Update broadcast sequence number and send windows */
- tipc_group_update_bc_members(tsk->group, blks);
+ tipc_group_update_bc_members(tsk->group, blks, ack);
+
+ /* Broadcast link is now free to choose method for next broadcast */
+ method->mandatory = false;
+ method->expires = jiffies;
+
return dlen;
}
@@ -1113,7 +1125,7 @@ void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
u32 portid, oport, onode;
struct list_head dports;
struct tipc_msg *msg;
- int hsz;
+ int user, mtyp, hsz;
__skb_queue_head_init(&tmpq);
INIT_LIST_HEAD(&dports);
@@ -1121,6 +1133,18 @@ void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
skb = tipc_skb_peek(arrvq, &inputq->lock);
for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
msg = buf_msg(skb);
+ user = msg_user(msg);
+ mtyp = msg_type(msg);
+ if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
+ spin_lock_bh(&inputq->lock);
+ if (skb_peek(arrvq) == skb) {
+ __skb_dequeue(arrvq);
+ __skb_queue_tail(inputq, skb);
+ }
+ refcount_dec(&skb->users);
+ spin_unlock_bh(&inputq->lock);
+ continue;
+ }
hsz = skb_headroom(skb) + msg_hdr_sz(msg);
oport = msg_origport(msg);
onode = msg_orignode(msg);
--
2.1.4
^ permalink raw reply related
* Re: [net-next V7 PATCH 3/5] bpf: cpumap xdp_buff to skb conversion and allocation
From: Jesper Dangaard Brouer @ 2017-10-13 9:13 UTC (permalink / raw)
To: Edward Cree
Cc: netdev, jakub.kicinski, Michael S. Tsirkin, pavel.odintsov,
Jason Wang, mchan, John Fastabend, peter.waskiewicz.jr, ast,
Daniel Borkmann, Alexei Starovoitov, Andy Gospodarek, brouer
In-Reply-To: <21efa4a5-1e2f-bbe9-c1d0-115843fb885b@solarflare.com>
On Thu, 12 Oct 2017 22:13:43 +0100
Edward Cree <ecree@solarflare.com> wrote:
> On 12/10/17 13:26, Jesper Dangaard Brouer wrote:
> > This patch makes cpumap functional, by adding SKB allocation and
> > invoking the network stack on the dequeuing CPU.
> >
> > For constructing the SKB on the remote CPU, the xdp_buff in converted
> > into a struct xdp_pkt, and it mapped into the top headroom of the
> > packet, to avoid allocating separate mem. For now, struct xdp_pkt is
> > just a cpumap internal data structure, with info carried between
> > enqueue to dequeue.
>
> <snip>
>
> > +struct sk_buff *cpu_map_build_skb(struct bpf_cpu_map_entry *rcpu,
> > + struct xdp_pkt *xdp_pkt)
> > +{
> > + unsigned int frame_size;
> > + void *pkt_data_start;
> > + struct sk_buff *skb;
> > +
> > + /* build_skb need to place skb_shared_info after SKB end, and
> > + * also want to know the memory "truesize". Thus, need to
> > + * know the memory frame size backing xdp_buff.
> > + *
> > + * XDP was designed to have PAGE_SIZE frames, but this
> > + * assumption is not longer true with ixgbe and i40e. It
> > + * would be preferred to set frame_size to 2048 or 4096
> > + * depending on the driver.
> > + * frame_size = 2048;
> > + * frame_len = frame_size - sizeof(*xdp_pkt);
> > + *
> > + * Instead, with info avail, skb_shared_info in placed after
> > + * packet len. This, unfortunately fakes the truesize.
> > + * Another disadvantage of this approach, the skb_shared_info
> > + * is not at a fixed memory location, with mixed length
> > + * packets, which is bad for cache-line hotness.
> > + */
> > + frame_size = SKB_DATA_ALIGN(xdp_pkt->len) + xdp_pkt->headroom +
> > + SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> > +
> > + pkt_data_start = xdp_pkt->data - xdp_pkt->headroom;
> > + skb = build_skb(pkt_data_start, frame_size);
> > + if (!skb)
> > + return NULL;
> > +
> > + skb_reserve(skb, xdp_pkt->headroom);
> > + __skb_put(skb, xdp_pkt->len);
> > + if (xdp_pkt->metasize)
> > + skb_metadata_set(skb, xdp_pkt->metasize);
> > +
> > + /* Essential SKB info: protocol and skb->dev */
> > + skb->protocol = eth_type_trans(skb, xdp_pkt->dev_rx);
> > +
> > + /* Optional SKB info, currently missing:
> > + * - HW checksum info (skb->ip_summed)
> > + * - HW RX hash (skb_set_hash)
> > + * - RX ring dev queue index (skb_record_rx_queue)
> > + */
> One possibility for dealing with these and related issues — also things
> like the proper way to free an xdp_buff if SKB creation fails, which
> might not be page_frag_free() for some drivers with unusual recycle ring
> implementations — is to have a new ndo for 'receiving' an xdp_pkt from a
> cpumap redirect.
> Since you're always receiving from the same driver that enqueued it, even
> the structure of the metadata stored in the top of the packet page
> doesn't have to be standardised; instead, each driver can put there just
> whatever happens to be needed for its ndo_xdp_rx routine. (Though there
> would probably be standard enqueue and dequeue functions that the
> 'common-case' drivers could use.)
> In some cases, the driver could even just leave in the page the packet
> prefix it got from the NIC, rather than reading it and then writing an
> interpreted version back, thus minimising the number of packet-page
> cachelines the 'bottom half' RX function has to touch (it would still
> need to write in anything it got from the RX event, of course).
> It shouldn't be much work as many driver RX routines are already
> structured this way — sfc, for instance, has a split into efx_rx_packet()
> and __efx_rx_packet(), as a software pipeline for prefetching.
This is all talking about future work. I'll be happy to discuss this
outside/after this patchset. I see nothing blocking these ideas.
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
^ permalink raw reply
* Re: [v2] mwifiex: Use put_unaligned_le32
From: Kalle Valo @ 2017-10-13 9:42 UTC (permalink / raw)
To: Himanshu Jha
Cc: amitkarwar, nishants, gbhat, huxm, linux-wireless, netdev,
linux-kernel, Himanshu Jha
In-Reply-To: <1507302279-3875-1-git-send-email-himanshujha199640@gmail.com>
Himanshu Jha <himanshujha199640@gmail.com> wrote:
> Use put_unaligned_le32 rather than using byte ordering function and
> memcpy which makes code clear.
> Also, add the header file where it is declared.
>
> Done using Coccinelle and semantic patch used is :
>
> @ rule1 @
> identifier tmp; expression ptr,x; type T;
> @@
>
> - tmp = cpu_to_le32(x);
>
> <+... when != tmp
> - memcpy(ptr, (T)&tmp, ...);
> + put_unaligned_le32(x,ptr);
> ...+>
>
> @ depends on rule1 @
> type j; identifier tmp;
> @@
>
> - j tmp;
> ...when != tmp
>
> Signed-off-by: Himanshu Jha <himanshujha199640@gmail.com>
Patch applied to wireless-drivers-next.git, thanks.
317049204cd3 mwifiex: Use put_unaligned_le32
--
https://patchwork.kernel.org/patch/9989655/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
^ permalink raw reply
* [iproute2] regression in ss output
From: Humberto Alves @ 2017-10-13 9:57 UTC (permalink / raw)
To: netdev@vger.kernel.org
Hi! With the last iproute2 release, ss command output does not
differentiate between any-address IPv4 sockets from the IPv6 ones.
I don't know if this is an expected behaviour, but the old output looks
more useful. Below I pasted the difference between the old behaviour and
the new one.
I am happy to send a patch if you want. Thanks
iproute2-4.12
-------------
$ ss --version
ss utility, iproute2-ss170705
$ ss -ntl
State Recv-Q Send-Q Local Address:Port Peer
Address:Port
LISTEN 0 100 *:587 *:*
LISTEN 0 100 *:110 *:*
LISTEN 0 100 *:143 *:*
LISTEN 0 128 *:80 *:*
LISTEN 0 128 *:2225 *:*
LISTEN 0 10 127.0.0.1:5010 *:*
LISTEN 0 128 *:20 *:*
LISTEN 0 128 127.0.0.1:20150 *:*
LISTEN 0 128 *:61719 *:*
LISTEN 0 100 *:25 *:*
LISTEN 0 3 *:1723 *:*
LISTEN 0 128 *:17500 *:*
LISTEN 0 128 127.0.0.1:17600 *:*
LISTEN 0 128 *:8000 *:*
LISTEN 0 100 *:993 *:*
LISTEN 0 128 127.0.0.1:17603 *:*
LISTEN 0 100 *:995 *:*
LISTEN 0 100 :::587 :::*
LISTEN 0 100 :::110 :::*
LISTEN 0 100 :::143 :::*
LISTEN 0 128 :::22000 :::*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::2225 :::*
LISTEN 0 128 :::20 :::*
LISTEN 0 128 ::1:20150 :::*
LISTEN 0 100 :::25 :::*
LISTEN 0 128 :::17500 :::*
LISTEN 0 128 :::8384 :::*
LISTEN 0 100 :::993 :::*
LISTEN 0 100 :::995 :::*
iproute2-4.13
-------------
$ ss --version
ss utility, iproute2-ss170905
$ ss -ntl
State Recv-Q Send-Q Local Address:Port Peer
Address:Port
LISTEN 0 100 *:587 *:*
LISTEN 0 100 *:110 *:*
LISTEN 0 100 *:143 *:*
LISTEN 0 128 *:80 *:*
LISTEN 0 128 *:2225 *:*
LISTEN 0 10 127.0.0.1:5010 *:*
LISTEN 0 128 *:20 *:*
LISTEN 0 128 127.0.0.1:20150 *:*
LISTEN 0 128 *:61719 *:*
LISTEN 0 100 *:25 *:*
LISTEN 0 3 *:1723 *:*
LISTEN 0 128 *:17500 *:*
LISTEN 0 128 127.0.0.1:17600 *:*
LISTEN 0 128 *:8000 *:*
LISTEN 0 100 *:993 *:*
LISTEN 0 128 127.0.0.1:17603 *:*
LISTEN 0 100 *:995 *:*
LISTEN 0 100 *:587 *:*
LISTEN 0 100 *:110 *:*
LISTEN 0 100 *:143 *:*
LISTEN 0 128 *:22000 *:*
LISTEN 0 128 *:80 *:*
LISTEN 0 128 *:2225 *:*
LISTEN 0 128 *:20 *:*
LISTEN 0 128 [::1]:20150 *:*
LISTEN 0 100 *:25 *:*
LISTEN 0 128 *:17500 *:*
LISTEN 0 128 *:8384 *:*
LISTEN 0 100 *:993 *:*
LISTEN 0 100 *:995 *:*
^ permalink raw reply
* Re: rtlwifi: Remove unused cur_rfstate variables
From: Kalle Valo @ 2017-10-13 9:58 UTC (permalink / raw)
To: Christos Gkekas
Cc: Larry Finger, Chaoming Li, linux-wireless, netdev, linux-kernel,
Christos Gkekas
In-Reply-To: <1507756515-15738-1-git-send-email-chris.gekas@gmail.com>
Christos Gkekas <chris.gekas@gmail.com> wrote:
> Clean up unused cur_rfstate variables in rtl8188ee, rtl8723ae, rtl8723be
> and rtl8821ae.
>
> Signed-off-by: Christos Gkekas <chris.gekas@gmail.com>
Patch applied to wireless-drivers-next.git, thanks.
76d7b12cbbe2 rtlwifi: Remove unused cur_rfstate variables
--
https://patchwork.kernel.org/patch/10000659/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
^ permalink raw reply
* [PATCH net-next v2 0/2] net: stmmac: Improvements for multi-queuing and for AVB
From: Jose Abreu @ 2017-10-13 9:58 UTC (permalink / raw)
To: netdev
Cc: Jose Abreu, David S. Miller, Joao Pinto, Giuseppe Cavallaro,
Alexandre Torgue
Hi,
Two improvements for stmmac: First one corrects the available fifo size per queue, second one corrects enabling of AVB queues. More info in commit log.
Best regards,
Jose Miguel Abreu
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Changes from v1:
- Fix typo in second patch
Jose Abreu (2):
net: stmmac: Use correct values in TQS/RQS fields
net: stmmac: Disable flow ctrl for RX AVB queues and really enable TX
AVB queues
drivers/net/ethernet/stmicro/stmmac/common.h | 5 +--
drivers/net/ethernet/stmicro/stmmac/dwmac4.h | 2 ++
drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c | 27 ++++++++++------
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 39 +++++++++++++++++++----
4 files changed, 56 insertions(+), 17 deletions(-)
--
1.9.1
^ permalink raw reply
* [PATCH net-next v2 1/2] net: stmmac: Use correct values in TQS/RQS fields
From: Jose Abreu @ 2017-10-13 9:58 UTC (permalink / raw)
To: netdev
Cc: Jose Abreu, David S. Miller, Joao Pinto, Giuseppe Cavallaro,
Alexandre Torgue
In-Reply-To: <cover.1507888312.git.joabreu@synopsys.com>
Currently we are using all the available fifo size in RQS and
TQS fields. This will not work correctly in multi-queues IP's
because total fifo size must be splitted to the enabled queues.
Correct this by computing the available fifo size per queue and
setting the right value in TQS and RQS fields.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
---
drivers/net/ethernet/stmicro/stmmac/common.h | 3 ++-
drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c | 15 +++++++++------
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 22 ++++++++++++++++++++--
3 files changed, 31 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index e82b4b7..c26c8a7 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -443,7 +443,8 @@ struct stmmac_dma_ops {
int rxfifosz);
void (*dma_rx_mode)(void __iomem *ioaddr, int mode, u32 channel,
int fifosz);
- void (*dma_tx_mode)(void __iomem *ioaddr, int mode, u32 channel);
+ void (*dma_tx_mode)(void __iomem *ioaddr, int mode, u32 channel,
+ int fifosz);
/* To track extra statistic (if supported) */
void (*dma_diagnostic_fr) (void *data, struct stmmac_extra_stats *x,
void __iomem *ioaddr);
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c b/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
index e84831e..898849b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
@@ -271,9 +271,10 @@ static void dwmac4_dma_rx_chan_op_mode(void __iomem *ioaddr, int mode,
}
static void dwmac4_dma_tx_chan_op_mode(void __iomem *ioaddr, int mode,
- u32 channel)
+ u32 channel, int fifosz)
{
u32 mtl_tx_op = readl(ioaddr + MTL_CHAN_TX_OP_MODE(channel));
+ unsigned int tqs = fifosz / 256 - 1;
if (mode == SF_DMA_MODE) {
pr_debug("GMAC: enable TX store and forward mode\n");
@@ -306,12 +307,14 @@ static void dwmac4_dma_tx_chan_op_mode(void __iomem *ioaddr, int mode,
* For an IP with DWC_EQOS_NUM_TXQ > 1, the fields TXQEN and TQS are R/W
* with reset values: TXQEN off, TQS 256 bytes.
*
- * Write the bits in both cases, since it will have no effect when RO.
- * For DWC_EQOS_NUM_TXQ > 1, the top bits in MTL_OP_MODE_TQS_MASK might
- * be RO, however, writing the whole TQS field will result in a value
- * equal to DWC_EQOS_TXFIFO_SIZE, just like for DWC_EQOS_NUM_TXQ == 1.
+ * TXQEN must be written for multi-channel operation and TQS must
+ * reflect the available fifo size per queue (total fifo size / number
+ * of enabled queues).
*/
- mtl_tx_op |= MTL_OP_MODE_TXQEN | MTL_OP_MODE_TQS_MASK;
+ mtl_tx_op |= MTL_OP_MODE_TXQEN;
+ mtl_tx_op &= ~MTL_OP_MODE_TQS_MASK;
+ mtl_tx_op |= tqs << MTL_OP_MODE_TQS_SHIFT;
+
writel(mtl_tx_op, ioaddr + MTL_CHAN_TX_OP_MODE(channel));
}
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index f41661a..edf245b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -1750,12 +1750,19 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
u32 rx_channels_count = priv->plat->rx_queues_to_use;
u32 tx_channels_count = priv->plat->tx_queues_to_use;
int rxfifosz = priv->plat->rx_fifo_size;
+ int txfifosz = priv->plat->tx_fifo_size;
u32 txmode = 0;
u32 rxmode = 0;
u32 chan = 0;
if (rxfifosz == 0)
rxfifosz = priv->dma_cap.rx_fifo_size;
+ if (txfifosz == 0)
+ txfifosz = priv->dma_cap.tx_fifo_size;
+
+ /* Adjust for real per queue fifo size */
+ rxfifosz /= rx_channels_count;
+ txfifosz /= tx_channels_count;
if (priv->plat->force_thresh_dma_mode) {
txmode = tc;
@@ -1783,7 +1790,8 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
rxfifosz);
for (chan = 0; chan < tx_channels_count; chan++)
- priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan);
+ priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan,
+ txfifosz);
} else {
priv->hw->dma->dma_mode(priv->ioaddr, txmode, rxmode,
rxfifosz);
@@ -1946,15 +1954,25 @@ static void stmmac_tx_err(struct stmmac_priv *priv, u32 chan)
static void stmmac_set_dma_operation_mode(struct stmmac_priv *priv, u32 txmode,
u32 rxmode, u32 chan)
{
+ u32 rx_channels_count = priv->plat->rx_queues_to_use;
+ u32 tx_channels_count = priv->plat->tx_queues_to_use;
int rxfifosz = priv->plat->rx_fifo_size;
+ int txfifosz = priv->plat->tx_fifo_size;
if (rxfifosz == 0)
rxfifosz = priv->dma_cap.rx_fifo_size;
+ if (txfifosz == 0)
+ txfifosz = priv->dma_cap.tx_fifo_size;
+
+ /* Adjust for real per queue fifo size */
+ rxfifosz /= rx_channels_count;
+ txfifosz /= tx_channels_count;
if (priv->synopsys_id >= DWMAC_CORE_4_00) {
priv->hw->dma->dma_rx_mode(priv->ioaddr, rxmode, chan,
rxfifosz);
- priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan);
+ priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan,
+ txfifosz);
} else {
priv->hw->dma->dma_mode(priv->ioaddr, txmode, rxmode,
rxfifosz);
--
1.9.1
^ permalink raw reply related
* [PATCH net-next v2 2/2] net: stmmac: Disable flow ctrl for RX AVB queues and really enable TX AVB queues
From: Jose Abreu @ 2017-10-13 9:58 UTC (permalink / raw)
To: netdev
Cc: Jose Abreu, David S. Miller, Joao Pinto, Giuseppe Cavallaro,
Alexandre Torgue
In-Reply-To: <cover.1507888312.git.joabreu@synopsys.com>
Flow control must be disabled for AVB enabled queues and TX
AVB queues must be enabled by setting BIT(2) of TXQEN.
Correct this by passing the queue mode to DMA callbacks
and by checking in these functions wether we are in AVB
performing the necessary adjustments.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
---
Changes from v1:
- Fix typo in dwmac4_dma_tx_chan_op_mode(): mtl_tx_op -> qmode
---
drivers/net/ethernet/stmicro/stmmac/common.h | 4 ++--
drivers/net/ethernet/stmicro/stmmac/dwmac4.h | 2 ++
drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c | 16 +++++++++++-----
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 21 +++++++++++++++------
4 files changed, 30 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index c26c8a7..e1e5ac0 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -442,9 +442,9 @@ struct stmmac_dma_ops {
void (*dma_mode)(void __iomem *ioaddr, int txmode, int rxmode,
int rxfifosz);
void (*dma_rx_mode)(void __iomem *ioaddr, int mode, u32 channel,
- int fifosz);
+ int fifosz, u8 qmode);
void (*dma_tx_mode)(void __iomem *ioaddr, int mode, u32 channel,
- int fifosz);
+ int fifosz, u8 qmode);
/* To track extra statistic (if supported) */
void (*dma_diagnostic_fr) (void *data, struct stmmac_extra_stats *x,
void __iomem *ioaddr);
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4.h b/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
index d74cedf..aeda3ab 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
@@ -225,6 +225,8 @@ enum power_event {
#define MTL_CHAN_RX_DEBUG(x) (MTL_CHANX_BASE_ADDR(x) + 0x38)
#define MTL_OP_MODE_RSF BIT(5)
+#define MTL_OP_MODE_TXQEN_MASK GENMASK(3, 2)
+#define MTL_OP_MODE_TXQEN_AV BIT(2)
#define MTL_OP_MODE_TXQEN BIT(3)
#define MTL_OP_MODE_TSF BIT(1)
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c b/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
index 898849b..c110f68 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4_dma.c
@@ -191,7 +191,7 @@ static void dwmac4_rx_watchdog(void __iomem *ioaddr, u32 riwt, u32 number_chan)
}
static void dwmac4_dma_rx_chan_op_mode(void __iomem *ioaddr, int mode,
- u32 channel, int fifosz)
+ u32 channel, int fifosz, u8 qmode)
{
unsigned int rqs = fifosz / 256 - 1;
u32 mtl_rx_op, mtl_rx_int;
@@ -218,8 +218,10 @@ static void dwmac4_dma_rx_chan_op_mode(void __iomem *ioaddr, int mode,
mtl_rx_op &= ~MTL_OP_MODE_RQS_MASK;
mtl_rx_op |= rqs << MTL_OP_MODE_RQS_SHIFT;
- /* enable flow control only if each channel gets 4 KiB or more FIFO */
- if (fifosz >= 4096) {
+ /* Enable flow control only if each channel gets 4 KiB or more FIFO and
+ * only if channel is not an AVB channel.
+ */
+ if ((fifosz >= 4096) && (qmode != MTL_QUEUE_AVB)) {
unsigned int rfd, rfa;
mtl_rx_op |= MTL_OP_MODE_EHFC;
@@ -271,7 +273,7 @@ static void dwmac4_dma_rx_chan_op_mode(void __iomem *ioaddr, int mode,
}
static void dwmac4_dma_tx_chan_op_mode(void __iomem *ioaddr, int mode,
- u32 channel, int fifosz)
+ u32 channel, int fifosz, u8 qmode)
{
u32 mtl_tx_op = readl(ioaddr + MTL_CHAN_TX_OP_MODE(channel));
unsigned int tqs = fifosz / 256 - 1;
@@ -311,7 +313,11 @@ static void dwmac4_dma_tx_chan_op_mode(void __iomem *ioaddr, int mode,
* reflect the available fifo size per queue (total fifo size / number
* of enabled queues).
*/
- mtl_tx_op |= MTL_OP_MODE_TXQEN;
+ mtl_tx_op &= ~MTL_OP_MODE_TXQEN_MASK;
+ if (qmode != MTL_QUEUE_AVB)
+ mtl_tx_op |= MTL_OP_MODE_TXQEN;
+ else
+ mtl_tx_op |= MTL_OP_MODE_TXQEN_AV;
mtl_tx_op &= ~MTL_OP_MODE_TQS_MASK;
mtl_tx_op |= tqs << MTL_OP_MODE_TQS_SHIFT;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index edf245b..0e1b0a3 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -1754,6 +1754,7 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
u32 txmode = 0;
u32 rxmode = 0;
u32 chan = 0;
+ u8 qmode = 0;
if (rxfifosz == 0)
rxfifosz = priv->dma_cap.rx_fifo_size;
@@ -1785,13 +1786,19 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
/* configure all channels */
if (priv->synopsys_id >= DWMAC_CORE_4_00) {
- for (chan = 0; chan < rx_channels_count; chan++)
+ for (chan = 0; chan < rx_channels_count; chan++) {
+ qmode = priv->plat->rx_queues_cfg[chan].mode_to_use;
+
priv->hw->dma->dma_rx_mode(priv->ioaddr, rxmode, chan,
- rxfifosz);
+ rxfifosz, qmode);
+ }
+
+ for (chan = 0; chan < tx_channels_count; chan++) {
+ qmode = priv->plat->tx_queues_cfg[chan].mode_to_use;
- for (chan = 0; chan < tx_channels_count; chan++)
priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan,
- txfifosz);
+ txfifosz, qmode);
+ }
} else {
priv->hw->dma->dma_mode(priv->ioaddr, txmode, rxmode,
rxfifosz);
@@ -1954,6 +1961,8 @@ static void stmmac_tx_err(struct stmmac_priv *priv, u32 chan)
static void stmmac_set_dma_operation_mode(struct stmmac_priv *priv, u32 txmode,
u32 rxmode, u32 chan)
{
+ u8 rxqmode = priv->plat->rx_queues_cfg[chan].mode_to_use;
+ u8 txqmode = priv->plat->tx_queues_cfg[chan].mode_to_use;
u32 rx_channels_count = priv->plat->rx_queues_to_use;
u32 tx_channels_count = priv->plat->tx_queues_to_use;
int rxfifosz = priv->plat->rx_fifo_size;
@@ -1970,9 +1979,9 @@ static void stmmac_set_dma_operation_mode(struct stmmac_priv *priv, u32 txmode,
if (priv->synopsys_id >= DWMAC_CORE_4_00) {
priv->hw->dma->dma_rx_mode(priv->ioaddr, rxmode, chan,
- rxfifosz);
+ rxfifosz, rxqmode);
priv->hw->dma->dma_tx_mode(priv->ioaddr, txmode, chan,
- txfifosz);
+ txfifosz, txqmode);
} else {
priv->hw->dma->dma_mode(priv->ioaddr, txmode, rxmode,
rxfifosz);
--
1.9.1
^ permalink raw reply related
* Re: rsi: fix integer overflow warning
From: Kalle Valo @ 2017-10-13 10:00 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Prameela Rani Garnepudi, Amitkumar Karwar, Arnd Bergmann,
Pavani Muthyala, Karun Eagalapati,
linux-wireless-u79uwXL29TY76Z2rM5mHXA,
netdev-u79uwXL29TY76Z2rM5mHXA,
linux-kernel-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <20171005120547.328687-1-arnd-r2nGTMty4D4@public.gmane.org>
Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org> wrote:
> gcc produces a harmless warning about a recently introduced
> signed integer overflow:
>
> drivers/net/wireless/rsi/rsi_91x_hal.c: In function 'rsi_prepare_mgmt_desc':
> include/uapi/linux/swab.h:13:15: error: integer overflow in expression [-Werror=overflow]
> (((__u16)(x) & (__u16)0x00ffU) << 8) | \
> ~~~~~~~~~~~~^~~~~~~~~~~~~~~~~
> include/uapi/linux/swab.h:104:2: note: in expansion of macro '___constant_swab16'
> ___constant_swab16(x) : \
> ^~~~~~~~~~~~~~~~~~
> include/uapi/linux/byteorder/big_endian.h:34:43: note: in expansion of macro '__swab16'
> #define __cpu_to_le16(x) ((__force __le16)__swab16((x)))
> ^~~~~~~~
> include/linux/byteorder/generic.h:89:21: note: in expansion of macro '__cpu_to_le16'
> #define cpu_to_le16 __cpu_to_le16
> ^~~~~~~~~~~~~
> drivers/net/wireless/rsi/rsi_91x_hal.c:136:3: note: in expansion of macro 'cpu_to_le16'
> cpu_to_le16((tx_params->vap_id << RSI_DESC_VAP_ID_OFST) &
> ^~~~~~~~~~~
>
> The problem is that the 'mask' value is a signed integer that gets
> turned into a negative number when truncated to 16 bits. Making it
> an unsigned constant avoids this.
>
> Fixes: eac4eed3224b ("rsi: tx and rx path enhancements for p2p mode")
> Signed-off-by: Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org>
Patch applied to wireless-drivers-next.git, thanks.
a39644b235c1 rsi: fix integer overflow warning
--
https://patchwork.kernel.org/patch/9986961/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
^ permalink raw reply
* linux-next: manual merge of the rdma tree with the FIXME tree
From: Mark Brown @ 2017-10-13 10:18 UTC (permalink / raw)
To: Doug Ledford, David S. Miller, Allen Pais, Kees Cook
Cc: netdev, Linux-Next Mailing List, Linux Kernel Mailing List
[-- Attachment #1: Type: text/plain, Size: 740 bytes --]
Hi Doug,
Today's linux-next merge of the rdma tree got a conflict in:
drivers/net/ethernet/mellanox/mlx4/catas.c
between commit:
d2a0012e7632a5 ("drivers: net: mlx4: use setup_timer() helper.")
from the net-next tree and commit:
55c0fcc3de4605 ("net/mlx4_core: Convert timers to use timer_setup()")
from the rdma tree.
I fixed it up by taking the second commit and can carry the fix as
necessary. This is now fixed as far as linux-next is concerned, but any
non trivial conflicts should be mentioned to your upstream maintainer
when your tree is submitted for merging. You may also want to consider
cooperating with the maintainer of the conflicting tree to minimise any
particularly complex conflicts.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply
* Re: [PATCHv4 iproute2 2/2] lib/libnetlink: update rtnl_talk to support malloc buff at run time
From: Phil Sutter @ 2017-10-13 10:31 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: Michal Kubecek, Hangbin Liu, netdev, Hangbin Liu
In-Reply-To: <20171012090706.04c00ab5@xeon-e3>
On Thu, Oct 12, 2017 at 09:07:06AM -0700, Stephen Hemminger wrote:
> On Wed, 11 Oct 2017 13:10:07 +0200
> Phil Sutter <phil@nwl.cc> wrote:
>
> > On Tue, Oct 10, 2017 at 09:47:43AM -0700, Stephen Hemminger wrote:
> > > On Tue, 10 Oct 2017 08:41:17 +0200
> > > Michal Kubecek <mkubecek@suse.cz> wrote:
> > >
> > > > On Mon, Oct 09, 2017 at 10:25:25PM +0200, Phil Sutter wrote:
> > > > > Hi Stephen,
> > > > >
> > > > > On Mon, Oct 02, 2017 at 10:37:08AM -0700, Stephen Hemminger wrote:
> > > > > > On Thu, 28 Sep 2017 21:33:46 +0800
> > > > > > Hangbin Liu <haliu@redhat.com> wrote:
> > > > > >
> > > > > > > From: Hangbin Liu <liuhangbin@gmail.com>
> > > > > > >
> > > > > > > This is an update for 460c03f3f3cc ("iplink: double the buffer size also in
> > > > > > > iplink_get()"). After update, we will not need to double the buffer size
> > > > > > > every time when VFs number increased.
> > > > > > >
> > > > > > > With call like rtnl_talk(&rth, &req.n, NULL, 0), we can simply remove the
> > > > > > > length parameter.
> > > > > > >
> > > > > > > With call like rtnl_talk(&rth, nlh, nlh, sizeof(req), I add a new variable
> > > > > > > answer to avoid overwrite data in nlh, because it may has more info after
> > > > > > > nlh. also this will avoid nlh buffer not enough issue.
> > > > > > >
> > > > > > > We need to free answer after using.
> > > > > > >
> > > > > > > Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
> > > > > > > Signed-off-by: Phil Sutter <phil@nwl.cc>
> > > > > > > ---
> > > > > >
> > > > > > Most of the uses of rtnl_talk() don't need to this peek and dynamic sizing.
> > > > > > Can only those places that need that be targeted?
> > > > >
> > > > > We could probably do that, by having a buffer on stack in __rtnl_talk()
> > > > > which will be used instead of the allocated one if 'answer' is NULL. Or
> > > > > maybe even introduce a dedicated API call for the dynamically allocated
> > > > > receive buffer. But I really doubt that's feasible: AFAICT, that stack
> > > > > buffer still needs to be reasonably sized since the reply might be
> > > > > larger than the request (reusing the request buffer would be the most
> > > > > simple way to tackle this), also there is support for extack which may
> > > > > bloat the response to arbitrary size. Hangbin has shown in his benchmark
> > > > > that the overhead of the second syscall is negligible, so why care about
> > > > > that and increase code complexity even further?
> > > > >
> > > > > Not saying it's not possible, but I just doubt it's worth the effort.
> > > >
> > > > Agreed. Current code is based on the assumption that we can estimate the
> > > > maximum reply length in advance and the reason for this series is that
> > > > this assumption turned out to be wrong. I'm afraid that if we replace
> > > > it by an assumption that we can estimate the maximum reply length for
> > > > most requests with only few exceptions, it's only matter of time for us
> > > > to be proven wrong again.
> > > >
> > > > Michal Kubecek
> > > >
> > >
> > > For query responses, yes the response may be large. But for the common cases of
> > > add address or add route, the response should just be ack or error.
> >
> > And with extack, error is comprised of the original request plus an
> > arbitrarily sized error message, so we can't just reuse the request
> > buffer and are back to "guessing" the right length again.
> >
> > To get an idea of what we're talking about, I wrote a simple benchmark
> > which adds 256 * 254 (= 65024) addresses to an interface, then removes
> > them again one by one and measured the time that takes for binaries with
> > and without Hangbin's patches:
> >
> > OP Vanilla Hangbin Delta
> > --------------------------------------------------------
> > add real 2m16.244s real 2m27.964s +11.72s (108.6%)
> > user 0m15.241s user 0m17.295s +2.054s (113.5%)
> > sys 1m40.229s sys 1m48.239s +8.01s (108.0%)
> >
> > remove real 1m44.950s real 1m47.044s +2.094s (102.0%)
> > user 0m13.899s user 0m14.723s +0.824s (105.9%)
> > sys 1m30.798s sys 1m31.938s +1.140s (101.3%)
> >
> > So the overhead of the second syscall and dynamic memory allocation is
> > less than 10% overall. Given the short time a single call to 'ip'
> > typically takes, I don't think the difference is noticeable even in
> > highly performance critical applications.
> >
> > Cheers, Phil
>
> For a better benchmark, I generated 4 Million routes
> then did:
> # ip ---batch routes.txt
Ah, batch mode. Nice trick!
> OP Vanilla Hangbin Delta
> -----------------------------------------------------
> add real 1:25.840 1:33.677 +9.13%
> user 10.690 6.078 -56.85%
> sys 1:00.920 1:13.109 +20.00%
>
> remove real 2:29.881 2:25.872 -2.67%
> user 12.862 7.942 -38.25%
> sys 44.127 44.633 +1.15%
>
>
> So the answer is addition is slower but deletion appears faster?
Yeah, that's funny. Hangbin's tests show the same in his 'ip link show'
test. I can imagine a performance improvement in some situations since
the patches eliminate that memcpy() of the reply buffer in
__rtnl_talk(), but neither 'route add' nor 'route del' trigger that code
path.
> If I rerun the Vanilla test, get about the same times.
>
> The slowdown won't impact me, but what about large scale users
> like Cumulus.
If they delete routes as often as they add them, things don't look too
bad at least. :)
Cheers, Phil
^ permalink raw reply
* [patch net-next 0/9] net: sched: remove some tp->q usage
From: Jiri Pirko @ 2017-10-13 12:00 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
From: Jiri Pirko <jiri@mellanox.com>
In order to prepare for block sharing, tcf_proto instances need to be
independent on particular qdisc instances. This patchset takes care of
removal of couple occurrences of tp->q usage.
Jiri Pirko (9):
net: sched: store Qdisc pointer in struct block
net: sched: store net pointer in block and introduce qdisc_net helper
net: sched: introduce tcf_block_q and tcf_block_dev helpers
net: sched: teach tcf_bind/unbind_filter to use block->q
net: sched: ematch: obtain net pointer from blocks
net: sched: cls_u32: use block instead of q in tc_u_common
net: sched: tcindex, fw, flow: use tcf_block_q helper to get struct
Qdisc
net: sched: use tcf_block_q helper to get q pointer for sch_tree_lock
net: sched: propagate q and parent from caller down to tcf_fill_node
include/net/pkt_cls.h | 52 +++++++++++++++++++++++++++++++----------
include/net/pkt_sched.h | 7 ++++++
include/net/sch_generic.h | 5 ++--
net/sched/cls_api.c | 59 ++++++++++++++++++++++++++++++-----------------
net/sched/cls_flow.c | 7 ++++--
net/sched/cls_fw.c | 5 +++-
net/sched/cls_tcindex.c | 5 +++-
net/sched/cls_u32.c | 12 ++++------
net/sched/ematch.c | 2 +-
net/sched/sch_api.c | 6 +++--
net/sched/sch_atm.c | 4 ++--
net/sched/sch_cbq.c | 2 +-
net/sched/sch_drr.c | 2 +-
net/sched/sch_dsmark.c | 2 +-
net/sched/sch_fq_codel.c | 2 +-
net/sched/sch_hfsc.c | 4 ++--
net/sched/sch_htb.c | 4 ++--
net/sched/sch_ingress.c | 6 ++---
net/sched/sch_multiq.c | 2 +-
net/sched/sch_prio.c | 2 +-
net/sched/sch_qfq.c | 2 +-
net/sched/sch_sfb.c | 2 +-
net/sched/sch_sfq.c | 2 +-
23 files changed, 127 insertions(+), 69 deletions(-)
--
2.9.5
^ permalink raw reply
* [patch net-next 1/9] net: sched: store Qdisc pointer in struct block
From: Jiri Pirko @ 2017-10-13 12:00 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Prepare for removal of tp->q and store Qdisc pointer in the block
structure.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/pkt_cls.h | 4 ++--
include/net/sch_generic.h | 1 +
net/sched/cls_api.c | 3 ++-
net/sched/sch_atm.c | 4 ++--
net/sched/sch_cbq.c | 2 +-
net/sched/sch_drr.c | 2 +-
net/sched/sch_dsmark.c | 2 +-
net/sched/sch_fq_codel.c | 2 +-
net/sched/sch_hfsc.c | 4 ++--
net/sched/sch_htb.c | 4 ++--
net/sched/sch_ingress.c | 6 +++---
net/sched/sch_multiq.c | 2 +-
net/sched/sch_prio.c | 2 +-
net/sched/sch_qfq.c | 2 +-
net/sched/sch_sfb.c | 2 +-
net/sched/sch_sfq.c | 2 +-
16 files changed, 23 insertions(+), 21 deletions(-)
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index f526374..772dfa8 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -22,7 +22,7 @@ struct tcf_chain *tcf_chain_get(struct tcf_block *block, u32 chain_index,
bool create);
void tcf_chain_put(struct tcf_chain *chain);
int tcf_block_get(struct tcf_block **p_block,
- struct tcf_proto __rcu **p_filter_chain);
+ struct tcf_proto __rcu **p_filter_chain, struct Qdisc *q);
void tcf_block_put(struct tcf_block *block);
int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct tcf_result *res, bool compat_mode);
@@ -30,7 +30,7 @@ int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
#else
static inline
int tcf_block_get(struct tcf_block **p_block,
- struct tcf_proto __rcu **p_filter_chain)
+ struct tcf_proto __rcu **p_filter_chain, struct Qdisc *q)
{
return 0;
}
diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index 684d8ed..df4032c 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -270,6 +270,7 @@ struct tcf_chain {
struct tcf_block {
struct list_head chain_list;
+ struct Qdisc *q;
};
static inline void qdisc_cb_private_validate(const struct sk_buff *skb, int sz)
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 2977b8a..f7d3f1f 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -241,7 +241,7 @@ tcf_chain_filter_chain_ptr_set(struct tcf_chain *chain,
}
int tcf_block_get(struct tcf_block **p_block,
- struct tcf_proto __rcu **p_filter_chain)
+ struct tcf_proto __rcu **p_filter_chain, struct Qdisc *q)
{
struct tcf_block *block = kzalloc(sizeof(*block), GFP_KERNEL);
struct tcf_chain *chain;
@@ -257,6 +257,7 @@ int tcf_block_get(struct tcf_block **p_block,
goto err_chain_create;
}
tcf_chain_filter_chain_ptr_set(chain, p_filter_chain);
+ block->q = q;
*p_block = block;
return 0;
diff --git a/net/sched/sch_atm.c b/net/sched/sch_atm.c
index c5fcdf1..2dbd249 100644
--- a/net/sched/sch_atm.c
+++ b/net/sched/sch_atm.c
@@ -281,7 +281,7 @@ static int atm_tc_change(struct Qdisc *sch, u32 classid, u32 parent,
goto err_out;
}
- error = tcf_block_get(&flow->block, &flow->filter_list);
+ error = tcf_block_get(&flow->block, &flow->filter_list, sch);
if (error) {
kfree(flow);
goto err_out;
@@ -546,7 +546,7 @@ static int atm_tc_init(struct Qdisc *sch, struct nlattr *opt)
p->link.q = &noop_qdisc;
pr_debug("atm_tc_init: link (%p) qdisc %p\n", &p->link, p->link.q);
- err = tcf_block_get(&p->link.block, &p->link.filter_list);
+ err = tcf_block_get(&p->link.block, &p->link.filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c
index dcef97f..c3b92d6 100644
--- a/net/sched/sch_cbq.c
+++ b/net/sched/sch_cbq.c
@@ -1566,7 +1566,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t
if (cl == NULL)
goto failure;
- err = tcf_block_get(&cl->block, &cl->filter_list);
+ err = tcf_block_get(&cl->block, &cl->filter_list, sch);
if (err) {
kfree(cl);
return err;
diff --git a/net/sched/sch_drr.c b/net/sched/sch_drr.c
index 2d0e8d4..753dc7a 100644
--- a/net/sched/sch_drr.c
+++ b/net/sched/sch_drr.c
@@ -412,7 +412,7 @@ static int drr_init_qdisc(struct Qdisc *sch, struct nlattr *opt)
struct drr_sched *q = qdisc_priv(sch);
int err;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
err = qdisc_class_hash_init(&q->clhash);
diff --git a/net/sched/sch_dsmark.c b/net/sched/sch_dsmark.c
index 2836c80..fb4fb71 100644
--- a/net/sched/sch_dsmark.c
+++ b/net/sched/sch_dsmark.c
@@ -344,7 +344,7 @@ static int dsmark_init(struct Qdisc *sch, struct nlattr *opt)
if (!opt)
goto errout;
- err = tcf_block_get(&p->block, &p->filter_list);
+ err = tcf_block_get(&p->block, &p->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_fq_codel.c b/net/sched/sch_fq_codel.c
index de3b57c..3c40ede 100644
--- a/net/sched/sch_fq_codel.c
+++ b/net/sched/sch_fq_codel.c
@@ -481,7 +481,7 @@ static int fq_codel_init(struct Qdisc *sch, struct nlattr *opt)
return err;
}
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c
index 3f88b75..a692184 100644
--- a/net/sched/sch_hfsc.c
+++ b/net/sched/sch_hfsc.c
@@ -1033,7 +1033,7 @@ hfsc_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
if (cl == NULL)
return -ENOBUFS;
- err = tcf_block_get(&cl->block, &cl->filter_list);
+ err = tcf_block_get(&cl->block, &cl->filter_list, sch);
if (err) {
kfree(cl);
return err;
@@ -1405,7 +1405,7 @@ hfsc_init_qdisc(struct Qdisc *sch, struct nlattr *opt)
return err;
q->eligible = RB_ROOT;
- err = tcf_block_get(&q->root.block, &q->root.filter_list);
+ err = tcf_block_get(&q->root.block, &q->root.filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index c6d7ae8..57be73c 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -1030,7 +1030,7 @@ static int htb_init(struct Qdisc *sch, struct nlattr *opt)
if (!opt)
return -EINVAL;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
@@ -1393,7 +1393,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
if (!cl)
goto failure;
- err = tcf_block_get(&cl->block, &cl->filter_list);
+ err = tcf_block_get(&cl->block, &cl->filter_list, sch);
if (err) {
kfree(cl);
goto failure;
diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c
index 44de4ee..9ccc1b8 100644
--- a/net/sched/sch_ingress.c
+++ b/net/sched/sch_ingress.c
@@ -59,7 +59,7 @@ static int ingress_init(struct Qdisc *sch, struct nlattr *opt)
struct net_device *dev = qdisc_dev(sch);
int err;
- err = tcf_block_get(&q->block, &dev->ingress_cl_list);
+ err = tcf_block_get(&q->block, &dev->ingress_cl_list, sch);
if (err)
return err;
@@ -153,11 +153,11 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt)
struct net_device *dev = qdisc_dev(sch);
int err;
- err = tcf_block_get(&q->ingress_block, &dev->ingress_cl_list);
+ err = tcf_block_get(&q->ingress_block, &dev->ingress_cl_list, sch);
if (err)
return err;
- err = tcf_block_get(&q->egress_block, &dev->egress_cl_list);
+ err = tcf_block_get(&q->egress_block, &dev->egress_cl_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_multiq.c b/net/sched/sch_multiq.c
index ff4fc3e..31e0a28 100644
--- a/net/sched/sch_multiq.c
+++ b/net/sched/sch_multiq.c
@@ -245,7 +245,7 @@ static int multiq_init(struct Qdisc *sch, struct nlattr *opt)
if (opt == NULL)
return -EINVAL;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index 2dd6c68..95fad34 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -212,7 +212,7 @@ static int prio_init(struct Qdisc *sch, struct nlattr *opt)
if (!opt)
return -EINVAL;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_qfq.c b/net/sched/sch_qfq.c
index 6ddfd49..8694c7b 100644
--- a/net/sched/sch_qfq.c
+++ b/net/sched/sch_qfq.c
@@ -1419,7 +1419,7 @@ static int qfq_init_qdisc(struct Qdisc *sch, struct nlattr *opt)
int i, j, err;
u32 max_cl_shift, maxbudg_shift, max_classes;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_sfb.c b/net/sched/sch_sfb.c
index cc39e17..487d375 100644
--- a/net/sched/sch_sfb.c
+++ b/net/sched/sch_sfb.c
@@ -553,7 +553,7 @@ static int sfb_init(struct Qdisc *sch, struct nlattr *opt)
struct sfb_sched_data *q = qdisc_priv(sch);
int err;
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 74ea863..123a53a 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -725,7 +725,7 @@ static int sfq_init(struct Qdisc *sch, struct nlattr *opt)
setup_deferrable_timer(&q->perturb_timer, sfq_perturbation,
(unsigned long)sch);
- err = tcf_block_get(&q->block, &q->filter_list);
+ err = tcf_block_get(&q->block, &q->filter_list, sch);
if (err)
return err;
--
2.9.5
^ permalink raw reply related
* [patch net-next 2/9] net: sched: store net pointer in block and introduce qdisc_net helper
From: Jiri Pirko @ 2017-10-13 12:00 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Store net pointer in the block structure. Along the way, introduce
qdisc_net helper which allows to easily obtain net pointer for
qdisc instance.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/pkt_sched.h | 7 +++++++
include/net/sch_generic.h | 1 +
net/sched/cls_api.c | 1 +
3 files changed, 9 insertions(+)
diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
index 259bc19..2d234af 100644
--- a/include/net/pkt_sched.h
+++ b/include/net/pkt_sched.h
@@ -4,7 +4,9 @@
#include <linux/jiffies.h>
#include <linux/ktime.h>
#include <linux/if_vlan.h>
+#include <linux/netdevice.h>
#include <net/sch_generic.h>
+#include <net/net_namespace.h>
#include <uapi/linux/pkt_sched.h>
#define DEFAULT_TX_QUEUE_LEN 1000
@@ -146,4 +148,9 @@ static inline bool is_classid_clsact_egress(u32 classid)
TC_H_MIN(classid) == TC_H_MIN(TC_H_MIN_EGRESS);
}
+static inline struct net *qdisc_net(struct Qdisc *q)
+{
+ return dev_net(q->dev_queue->dev);
+}
+
#endif
diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index df4032c..9b2cb91 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -270,6 +270,7 @@ struct tcf_chain {
struct tcf_block {
struct list_head chain_list;
+ struct net *net;
struct Qdisc *q;
};
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index f7d3f1f..856003c 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -257,6 +257,7 @@ int tcf_block_get(struct tcf_block **p_block,
goto err_chain_create;
}
tcf_chain_filter_chain_ptr_set(chain, p_filter_chain);
+ block->net = qdisc_net(q);
block->q = q;
*p_block = block;
return 0;
--
2.9.5
^ permalink raw reply related
* [patch net-next 3/9] net: sched: introduce tcf_block_q and tcf_block_dev helpers
From: Jiri Pirko @ 2017-10-13 12:00 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
These helpers allows to get a q and netdev pointers
for given block easily.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/pkt_cls.h | 21 +++++++++++++++++++++
1 file changed, 21 insertions(+)
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index 772dfa8..01ffda5 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -24,6 +24,17 @@ void tcf_chain_put(struct tcf_chain *chain);
int tcf_block_get(struct tcf_block **p_block,
struct tcf_proto __rcu **p_filter_chain, struct Qdisc *q);
void tcf_block_put(struct tcf_block *block);
+
+static inline struct Qdisc *tcf_block_q(struct tcf_block *block)
+{
+ return block->q;
+}
+
+static inline struct net_device *tcf_block_dev(struct tcf_block *block)
+{
+ return tcf_block_q(block)->dev_queue->dev;
+}
+
int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct tcf_result *res, bool compat_mode);
@@ -39,6 +50,16 @@ static inline void tcf_block_put(struct tcf_block *block)
{
}
+static inline struct Qdisc *tcf_block_q(struct tcf_block *block)
+{
+ return NULL;
+}
+
+static inline struct net_device *tcf_block_dev(struct tcf_block *block)
+{
+ return NULL;
+}
+
static inline int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct tcf_result *res, bool compat_mode)
{
--
2.9.5
^ permalink raw reply related
* [patch net-next 4/9] net: sched: teach tcf_bind/unbind_filter to use block->q
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Whenever the block->q is set, it can be used instead of tp->q as it
contains the same value. When it is not set, which can't happen now but
it might happen with the follow-up shared blocks introduction, the class
is not set in the result. That would lead to a class lookup instead
of direct class pointer use for classful qdiscs. However, it is not
planned to support classful qdisqs sharing filter blocks, so that may
never happen.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/pkt_cls.h | 27 +++++++++++++++++----------
1 file changed, 17 insertions(+), 10 deletions(-)
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index 01ffda5..bf6c853 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -74,36 +74,43 @@ __cls_set_class(unsigned long *clp, unsigned long cl)
}
static inline unsigned long
-cls_set_class(struct tcf_proto *tp, unsigned long *clp,
- unsigned long cl)
+cls_set_class(struct Qdisc *q, unsigned long *clp, unsigned long cl)
{
unsigned long old_cl;
-
- tcf_tree_lock(tp);
+
+ sch_tree_lock(q);
old_cl = __cls_set_class(clp, cl);
- tcf_tree_unlock(tp);
-
+ sch_tree_unlock(q);
return old_cl;
}
static inline void
tcf_bind_filter(struct tcf_proto *tp, struct tcf_result *r, unsigned long base)
{
+ struct Qdisc *q = tp->chain->block->q;
unsigned long cl;
- cl = tp->q->ops->cl_ops->bind_tcf(tp->q, base, r->classid);
- cl = cls_set_class(tp, &r->class, cl);
+ /* Check q as it is not set for shared blocks. In that case,
+ * setting class is not supported.
+ */
+ if (!q)
+ return;
+ cl = q->ops->cl_ops->bind_tcf(q, base, r->classid);
+ cl = cls_set_class(q, &r->class, cl);
if (cl)
- tp->q->ops->cl_ops->unbind_tcf(tp->q, cl);
+ q->ops->cl_ops->unbind_tcf(q, cl);
}
static inline void
tcf_unbind_filter(struct tcf_proto *tp, struct tcf_result *r)
{
+ struct Qdisc *q = tp->chain->block->q;
unsigned long cl;
+ if (!q)
+ return;
if ((cl = __cls_set_class(&r->class, 0)) != 0)
- tp->q->ops->cl_ops->unbind_tcf(tp->q, cl);
+ q->ops->cl_ops->unbind_tcf(q, cl);
}
struct tcf_exts {
--
2.9.5
^ permalink raw reply related
* [patch net-next 5/9] net: sched: ematch: obtain net pointer from blocks
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Instead of using tp->q, use block to get the net pointer.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
net/sched/ematch.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/sched/ematch.c b/net/sched/ematch.c
index 03b677b..1331a4c 100644
--- a/net/sched/ematch.c
+++ b/net/sched/ematch.c
@@ -178,7 +178,7 @@ static int tcf_em_validate(struct tcf_proto *tp,
struct tcf_ematch_hdr *em_hdr = nla_data(nla);
int data_len = nla_len(nla) - sizeof(*em_hdr);
void *data = (void *) em_hdr + sizeof(*em_hdr);
- struct net *net = dev_net(qdisc_dev(tp->q));
+ struct net *net = tp->chain->block->net;
if (!TCF_EM_REL_VALID(em_hdr->flags))
goto errout;
--
2.9.5
^ permalink raw reply related
* [patch net-next 6/9] net: sched: cls_u32: use block instead of q in tc_u_common
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
tc_u_common is now per-q. With blocks, it has to be converted to be
per-block.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
net/sched/cls_u32.c | 12 ++++--------
1 file changed, 4 insertions(+), 8 deletions(-)
diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c
index 094d224..b6d4606 100644
--- a/net/sched/cls_u32.c
+++ b/net/sched/cls_u32.c
@@ -93,7 +93,7 @@ struct tc_u_hnode {
struct tc_u_common {
struct tc_u_hnode __rcu *hlist;
- struct Qdisc *q;
+ struct tcf_block *block;
int refcnt;
struct idr handle_idr;
struct hlist_node hnode;
@@ -335,11 +335,7 @@ static struct hlist_head *tc_u_common_hash;
static unsigned int tc_u_hash(const struct tcf_proto *tp)
{
- struct net_device *dev = tp->q->dev_queue->dev;
- u32 qhandle = tp->q->handle;
- int ifindex = dev->ifindex;
-
- return hash_64((u64)ifindex << 32 | qhandle, U32_HASH_SHIFT);
+ return hash_64((u64) tp->chain->block, U32_HASH_SHIFT);
}
static struct tc_u_common *tc_u_common_find(const struct tcf_proto *tp)
@@ -349,7 +345,7 @@ static struct tc_u_common *tc_u_common_find(const struct tcf_proto *tp)
h = tc_u_hash(tp);
hlist_for_each_entry(tc, &tc_u_common_hash[h], hnode) {
- if (tc->q == tp->q)
+ if (tc->block == tp->chain->block)
return tc;
}
return NULL;
@@ -378,7 +374,7 @@ static int u32_init(struct tcf_proto *tp)
kfree(root_ht);
return -ENOBUFS;
}
- tp_c->q = tp->q;
+ tp_c->block = tp->chain->block;
INIT_HLIST_NODE(&tp_c->hnode);
idr_init(&tp_c->handle_idr);
--
2.9.5
^ permalink raw reply related
* [patch net-next 8/9] net: sched: use tcf_block_q helper to get q pointer for sch_tree_lock
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Use tcf_block_q helper to get q pointer to be used for direct call of
sch_tree_lock/unlock instead of tcf_tree_lock/unlock.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/sch_generic.h | 3 ---
net/sched/sch_api.c | 6 ++++--
2 files changed, 4 insertions(+), 5 deletions(-)
diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index 9b2cb91..0aea9e2 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -359,9 +359,6 @@ static inline void sch_tree_unlock(const struct Qdisc *q)
spin_unlock_bh(qdisc_root_sleeping_lock(q));
}
-#define tcf_tree_lock(tp) sch_tree_lock((tp)->q)
-#define tcf_tree_unlock(tp) sch_tree_unlock((tp)->q)
-
extern struct Qdisc noop_qdisc;
extern struct Qdisc_ops noop_qdisc_ops;
extern struct Qdisc_ops pfifo_fast_ops;
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index aa82116..a9ac912 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1661,9 +1661,11 @@ static int tcf_node_bind(struct tcf_proto *tp, void *n, struct tcf_walker *arg)
struct tcf_bind_args *a = (void *)arg;
if (tp->ops->bind_class) {
- tcf_tree_lock(tp);
+ struct Qdisc *q = tcf_block_q(tp->chain->block);
+
+ sch_tree_lock(q);
tp->ops->bind_class(n, a->classid, a->cl);
- tcf_tree_unlock(tp);
+ sch_tree_unlock(q);
}
return 0;
}
--
2.9.5
^ permalink raw reply related
* [patch net-next 7/9] net: sched: tcindex, fw, flow: use tcf_block_q helper to get struct Qdisc
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
Use helper to get q pointer per block.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
net/sched/cls_flow.c | 7 +++++--
net/sched/cls_fw.c | 5 ++++-
net/sched/cls_tcindex.c | 5 ++++-
3 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/net/sched/cls_flow.c b/net/sched/cls_flow.c
index 2a3a60e..f3be666 100644
--- a/net/sched/cls_flow.c
+++ b/net/sched/cls_flow.c
@@ -491,8 +491,11 @@ static int flow_change(struct net *net, struct sk_buff *in_skb,
perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
}
- if (TC_H_MAJ(baseclass) == 0)
- baseclass = TC_H_MAKE(tp->q->handle, baseclass);
+ if (TC_H_MAJ(baseclass) == 0) {
+ struct Qdisc *q = tcf_block_q(tp->chain->block);
+
+ baseclass = TC_H_MAKE(q->handle, baseclass);
+ }
if (TC_H_MIN(baseclass) == 0)
baseclass = TC_H_MAKE(baseclass, 1);
diff --git a/net/sched/cls_fw.c b/net/sched/cls_fw.c
index 941245a..aa1e1f3 100644
--- a/net/sched/cls_fw.c
+++ b/net/sched/cls_fw.c
@@ -28,6 +28,7 @@
#include <net/netlink.h>
#include <net/act_api.h>
#include <net/pkt_cls.h>
+#include <net/sch_generic.h>
#define HTSIZE 256
@@ -83,9 +84,11 @@ static int fw_classify(struct sk_buff *skb, const struct tcf_proto *tp,
}
}
} else {
+ struct Qdisc *q = tcf_block_q(tp->chain->block);
+
/* Old method: classify the packet using its skb mark. */
if (id && (TC_H_MAJ(id) == 0 ||
- !(TC_H_MAJ(id ^ tp->q->handle)))) {
+ !(TC_H_MAJ(id ^ q->handle)))) {
res->classid = id;
res->class = 0;
return 0;
diff --git a/net/sched/cls_tcindex.c b/net/sched/cls_tcindex.c
index 14a7e08..d732b54 100644
--- a/net/sched/cls_tcindex.c
+++ b/net/sched/cls_tcindex.c
@@ -13,6 +13,7 @@
#include <net/act_api.h>
#include <net/netlink.h>
#include <net/pkt_cls.h>
+#include <net/sch_generic.h>
/*
* Passing parameters to the root seems to be done more awkwardly than really
@@ -90,9 +91,11 @@ static int tcindex_classify(struct sk_buff *skb, const struct tcf_proto *tp,
f = tcindex_lookup(p, key);
if (!f) {
+ struct Qdisc *q = tcf_block_q(tp->chain->block);
+
if (!p->fall_through)
return -1;
- res->classid = TC_H_MAKE(TC_H_MAJ(tp->q->handle), key);
+ res->classid = TC_H_MAKE(TC_H_MAJ(q->handle), key);
res->class = 0;
pr_debug("alg 0x%x\n", res->classid);
return 0;
--
2.9.5
^ permalink raw reply related
* [patch net-next 9/9] net: sched: propagate q and parent from caller down to tcf_fill_node
From: Jiri Pirko @ 2017-10-13 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, jhs, xiyou.wangcong, mlxsw
In-Reply-To: <20171013120105.23358-1-jiri@resnulli.us>
From: Jiri Pirko <jiri@mellanox.com>
The callers have this info, they will pass it down to tcf_fill_node.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
net/sched/cls_api.c | 55 ++++++++++++++++++++++++++++++++++-------------------
1 file changed, 35 insertions(+), 20 deletions(-)
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 856003c..2e8e87f 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -420,8 +420,8 @@ static struct tcf_proto *tcf_chain_tp_find(struct tcf_chain *chain,
}
static int tcf_fill_node(struct net *net, struct sk_buff *skb,
- struct tcf_proto *tp, void *fh, u32 portid,
- u32 seq, u16 flags, int event)
+ struct tcf_proto *tp, struct Qdisc *q, u32 parent,
+ void *fh, u32 portid, u32 seq, u16 flags, int event)
{
struct tcmsg *tcm;
struct nlmsghdr *nlh;
@@ -434,8 +434,8 @@ static int tcf_fill_node(struct net *net, struct sk_buff *skb,
tcm->tcm_family = AF_UNSPEC;
tcm->tcm__pad1 = 0;
tcm->tcm__pad2 = 0;
- tcm->tcm_ifindex = qdisc_dev(tp->q)->ifindex;
- tcm->tcm_parent = tp->classid;
+ tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
+ tcm->tcm_parent = parent;
tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
if (nla_put_string(skb, TCA_KIND, tp->ops->kind))
goto nla_put_failure;
@@ -458,6 +458,7 @@ static int tcf_fill_node(struct net *net, struct sk_buff *skb,
static int tfilter_notify(struct net *net, struct sk_buff *oskb,
struct nlmsghdr *n, struct tcf_proto *tp,
+ struct Qdisc *q, u32 parent,
void *fh, int event, bool unicast)
{
struct sk_buff *skb;
@@ -467,7 +468,7 @@ static int tfilter_notify(struct net *net, struct sk_buff *oskb,
if (!skb)
return -ENOBUFS;
- if (tcf_fill_node(net, skb, tp, fh, portid, n->nlmsg_seq,
+ if (tcf_fill_node(net, skb, tp, q, parent, fh, portid, n->nlmsg_seq,
n->nlmsg_flags, event) <= 0) {
kfree_skb(skb);
return -EINVAL;
@@ -482,6 +483,7 @@ static int tfilter_notify(struct net *net, struct sk_buff *oskb,
static int tfilter_del_notify(struct net *net, struct sk_buff *oskb,
struct nlmsghdr *n, struct tcf_proto *tp,
+ struct Qdisc *q, u32 parent,
void *fh, bool unicast, bool *last)
{
struct sk_buff *skb;
@@ -492,7 +494,7 @@ static int tfilter_del_notify(struct net *net, struct sk_buff *oskb,
if (!skb)
return -ENOBUFS;
- if (tcf_fill_node(net, skb, tp, fh, portid, n->nlmsg_seq,
+ if (tcf_fill_node(net, skb, tp, q, parent, fh, portid, n->nlmsg_seq,
n->nlmsg_flags, RTM_DELTFILTER) <= 0) {
kfree_skb(skb);
return -EINVAL;
@@ -512,6 +514,7 @@ static int tfilter_del_notify(struct net *net, struct sk_buff *oskb,
}
static void tfilter_notify_chain(struct net *net, struct sk_buff *oskb,
+ struct Qdisc *q, u32 parent,
struct nlmsghdr *n,
struct tcf_chain *chain, int event)
{
@@ -519,7 +522,7 @@ static void tfilter_notify_chain(struct net *net, struct sk_buff *oskb,
for (tp = rtnl_dereference(chain->filter_chain);
tp; tp = rtnl_dereference(tp->next))
- tfilter_notify(net, oskb, n, tp, 0, event, false);
+ tfilter_notify(net, oskb, n, tp, q, parent, 0, event, false);
}
/* Add/change/delete/get a filter node */
@@ -638,7 +641,8 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n,
}
if (n->nlmsg_type == RTM_DELTFILTER && prio == 0) {
- tfilter_notify_chain(net, skb, n, chain, RTM_DELTFILTER);
+ tfilter_notify_chain(net, skb, q, parent, n,
+ chain, RTM_DELTFILTER);
tcf_chain_flush(chain);
err = 0;
goto errout;
@@ -685,7 +689,7 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n,
if (!fh) {
if (n->nlmsg_type == RTM_DELTFILTER && t->tcm_handle == 0) {
tcf_chain_tp_remove(chain, &chain_info, tp);
- tfilter_notify(net, skb, n, tp, fh,
+ tfilter_notify(net, skb, n, tp, q, parent, fh,
RTM_DELTFILTER, false);
tcf_proto_destroy(tp);
err = 0;
@@ -710,8 +714,8 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n,
}
break;
case RTM_DELTFILTER:
- err = tfilter_del_notify(net, skb, n, tp, fh, false,
- &last);
+ err = tfilter_del_notify(net, skb, n, tp, q, parent,
+ fh, false, &last);
if (err)
goto errout;
if (last) {
@@ -720,7 +724,7 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n,
}
goto errout;
case RTM_GETTFILTER:
- err = tfilter_notify(net, skb, n, tp, fh,
+ err = tfilter_notify(net, skb, n, tp, q, parent, fh,
RTM_NEWTFILTER, true);
goto errout;
default:
@@ -734,7 +738,8 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n,
if (err == 0) {
if (tp_created)
tcf_chain_tp_insert(chain, &chain_info, tp);
- tfilter_notify(net, skb, n, tp, fh, RTM_NEWTFILTER, false);
+ tfilter_notify(net, skb, n, tp, q, parent, fh,
+ RTM_NEWTFILTER, false);
} else {
if (tp_created)
tcf_proto_destroy(tp);
@@ -753,6 +758,8 @@ struct tcf_dump_args {
struct tcf_walker w;
struct sk_buff *skb;
struct netlink_callback *cb;
+ struct Qdisc *q;
+ u32 parent;
};
static int tcf_node_dump(struct tcf_proto *tp, void *n, struct tcf_walker *arg)
@@ -760,13 +767,14 @@ static int tcf_node_dump(struct tcf_proto *tp, void *n, struct tcf_walker *arg)
struct tcf_dump_args *a = (void *)arg;
struct net *net = sock_net(a->skb->sk);
- return tcf_fill_node(net, a->skb, tp, n, NETLINK_CB(a->cb->skb).portid,
+ return tcf_fill_node(net, a->skb, tp, a->q, a->parent,
+ n, NETLINK_CB(a->cb->skb).portid,
a->cb->nlh->nlmsg_seq, NLM_F_MULTI,
RTM_NEWTFILTER);
}
-static bool tcf_chain_dump(struct tcf_chain *chain, struct sk_buff *skb,
- struct netlink_callback *cb,
+static bool tcf_chain_dump(struct tcf_chain *chain, struct Qdisc *q, u32 parent,
+ struct sk_buff *skb, struct netlink_callback *cb,
long index_start, long *p_index)
{
struct net *net = sock_net(skb->sk);
@@ -788,7 +796,7 @@ static bool tcf_chain_dump(struct tcf_chain *chain, struct sk_buff *skb,
memset(&cb->args[1], 0,
sizeof(cb->args) - sizeof(cb->args[0]));
if (cb->args[1] == 0) {
- if (tcf_fill_node(net, skb, tp, 0,
+ if (tcf_fill_node(net, skb, tp, q, parent, 0,
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq, NLM_F_MULTI,
RTM_NEWTFILTER) <= 0)
@@ -801,6 +809,8 @@ static bool tcf_chain_dump(struct tcf_chain *chain, struct sk_buff *skb,
arg.w.fn = tcf_node_dump;
arg.skb = skb;
arg.cb = cb;
+ arg.q = q;
+ arg.parent = parent;
arg.w.stop = 0;
arg.w.skip = cb->args[1] - 1;
arg.w.count = 0;
@@ -826,6 +836,7 @@ static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
const struct Qdisc_class_ops *cops;
long index_start;
long index;
+ u32 parent;
int err;
if (nlmsg_len(cb->nlh) < sizeof(*tcm))
@@ -839,10 +850,13 @@ static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
if (!dev)
return skb->len;
- if (!tcm->tcm_parent)
+ parent = tcm->tcm_parent;
+ if (!parent) {
q = dev->qdisc;
- else
+ parent = q->handle;
+ } else {
q = qdisc_lookup(dev, TC_H_MAJ(tcm->tcm_parent));
+ }
if (!q)
goto out;
cops = q->ops->cl_ops;
@@ -866,7 +880,8 @@ static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
if (tca[TCA_CHAIN] &&
nla_get_u32(tca[TCA_CHAIN]) != chain->index)
continue;
- if (!tcf_chain_dump(chain, skb, cb, index_start, &index))
+ if (!tcf_chain_dump(chain, q, parent, skb, cb,
+ index_start, &index))
break;
}
--
2.9.5
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox