* [PATCH 0/14 net-next] more TIPC fixes and updates
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
With net-next reopened, here is another batch of TIPC patches primarily
from the out-of-kernel repo. There is no real binding between any of the
commits aside from taking what appear to be stand-alone bits and migrating
them across and fixing up minor coding style things etc. I believe there
will be a chance to get a lot more of this done in this dev window, so
hopefully this will be just one batch of several that I end up sending.
Basic runtime sanity test done on v2.6.36-rc1, and all patches git am
cleanly on the net-next of today as well.
Thanks,
Paul.
^ permalink raw reply
* [PATCH net-next 01/14] tipc: Fix log buffer memory leak if initialization fails
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Anders Kaseorg <[andersk@ksplice.com]>
Moves log buffer cleanup into tipc_core_stop() so that memory allocated
for the log buffer is freed if tipc_core_start() is unsuccessful.
Signed-off-by: Anders Kaseorg <andersk@ksplice.com>
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/core.c | 6 ++++--
1 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/net/tipc/core.c b/net/tipc/core.c
index 6964681..466b861 100644
--- a/net/tipc/core.c
+++ b/net/tipc/core.c
@@ -169,6 +169,7 @@ void tipc_core_stop(void)
tipc_nametbl_stop();
tipc_ref_table_stop();
tipc_socket_stop();
+ tipc_log_resize(0);
}
/**
@@ -203,7 +204,9 @@ static int __init tipc_init(void)
{
int res;
- tipc_log_resize(CONFIG_TIPC_LOG);
+ if (tipc_log_resize(CONFIG_TIPC_LOG) != 0)
+ warn("Unable to create log buffer\n");
+
info("Activated (version " TIPC_MOD_VER
" compiled " __DATE__ " " __TIME__ ")\n");
@@ -230,7 +233,6 @@ static void __exit tipc_exit(void)
tipc_core_stop_net();
tipc_core_stop();
info("Deactivated\n");
- tipc_log_resize(0);
}
module_init(tipc_init);
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 03/14] tipc: Provide correct error code for unsupported connect() operation
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Modify TIPC to return EOPNOTSUPP if an application attempts to perform
a non-blocking connect() operation, which is not supported by TIPC.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/socket.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 69d0fd1..b89c7b1 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -1380,7 +1380,7 @@ static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
/* For now, TIPC does not support the non-blocking form of connect() */
if (flags & O_NONBLOCK) {
- res = -EWOULDBLOCK;
+ res = -EOPNOTSUPP;
goto exit;
}
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 04/14] tipc: correct problems with misleading flags returned using poll()
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Prevent TIPC from incorrectly setting returned flags to poll()
in the following cases:
- an unconnected socket no longer indicates that it is always readable
- an unconnected, connecting, or listening socket no longer indicates
that it is always writable
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/socket.c | 61 ++++++++++++++++++++++++++++++++++------------------
1 files changed, 40 insertions(+), 21 deletions(-)
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index b89c7b1..7b81fdd 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -429,36 +429,55 @@ static int get_name(struct socket *sock, struct sockaddr *uaddr,
* to handle any preventable race conditions, so TIPC will do the same ...
*
* TIPC sets the returned events as follows:
- * a) POLLRDNORM and POLLIN are set if the socket's receive queue is non-empty
- * or if a connection-oriented socket is does not have an active connection
- * (i.e. a read operation will not block).
- * b) POLLOUT is set except when a socket's connection has been terminated
- * (i.e. a write operation will not block).
- * c) POLLHUP is set when a socket's connection has been terminated.
- *
- * IMPORTANT: The fact that a read or write operation will not block does NOT
- * imply that the operation will succeed!
+ *
+ * socket state flags set
+ * ------------ ---------
+ * unconnected no read flags
+ * no write flags
+ *
+ * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
+ * no write flags
+ *
+ * connected POLLIN/POLLRDNORM if data in rx queue
+ * POLLOUT if port is not congested
+ *
+ * disconnecting POLLIN/POLLRDNORM/POLLHUP
+ * no write flags
+ *
+ * listening POLLIN if SYN in rx queue
+ * no write flags
+ *
+ * ready POLLIN/POLLRDNORM if data in rx queue
+ * [connectionless] POLLOUT (since port cannot be congested)
+ *
+ * IMPORTANT: The fact that a read or write operation is indicated does NOT
+ * imply that the operation will succeed, merely that it should be performed
+ * and will not block.
*/
static unsigned int poll(struct file *file, struct socket *sock,
poll_table *wait)
{
struct sock *sk = sock->sk;
- u32 mask;
+ u32 mask = 0;
poll_wait(file, sk_sleep(sk), wait);
- if (!skb_queue_empty(&sk->sk_receive_queue) ||
- (sock->state == SS_UNCONNECTED) ||
- (sock->state == SS_DISCONNECTING))
- mask = (POLLRDNORM | POLLIN);
- else
- mask = 0;
-
- if (sock->state == SS_DISCONNECTING)
- mask |= POLLHUP;
- else
- mask |= POLLOUT;
+ switch ((int)sock->state) {
+ case SS_READY:
+ case SS_CONNECTED:
+ if (!tipc_sk_port(sk)->congested)
+ mask |= POLLOUT;
+ /* fall thru' */
+ case SS_CONNECTING:
+ case SS_LISTENING:
+ if (!skb_queue_empty(&sk->sk_receive_queue))
+ mask |= (POLLIN | POLLRDNORM);
+ break;
+ case SS_DISCONNECTING:
+ mask = (POLLIN | POLLRDNORM | POLLHUP);
+ break;
+ }
return mask;
}
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 02/14] tipc: add SO_RCVLOWAT support to stream socket receive path
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Florian Westphal <fw@strlen.de>
Add support for the SO_RCVLOWAT socket option to TIPC's stream socket
type.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/socket.c | 7 ++++---
1 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 66e889b..69d0fd1 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -1026,9 +1026,8 @@ static int recv_stream(struct kiocb *iocb, struct socket *sock,
struct sk_buff *buf;
struct tipc_msg *msg;
unsigned int sz;
- int sz_to_copy;
+ int sz_to_copy, target, needed;
int sz_copied = 0;
- int needed;
char __user *crs = m->msg_iov->iov_base;
unsigned char *buf_crs;
u32 err;
@@ -1050,6 +1049,8 @@ static int recv_stream(struct kiocb *iocb, struct socket *sock,
goto exit;
}
+ target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len);
+
restart:
/* Look for a message in receive queue; wait if necessary */
@@ -1138,7 +1139,7 @@ restart:
if ((sz_copied < buf_len) && /* didn't get all requested data */
(!skb_queue_empty(&sk->sk_receive_queue) ||
- (flags & MSG_WAITALL)) && /* and more is ready or required */
+ (sz_copied < target)) && /* and more is ready or required */
(!(flags & MSG_PEEK)) && /* and aren't just peeking at data */
(!err)) /* and haven't reached a FIN */
goto restart;
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 05/14] tipc: Check for disabled bearer when processing incoming messages
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Add a check to tipc_recv_msg() to ensure it discards messages
arriving on a newly disabled bearer. This is needed to deal with a
race condition that can arise if the bearer is in the midst of being
disabled when it receives a message. Performing the check after
tipc_net_lock has been taken ensures that TIPC's bearers are in a
stable state while the message is being processed.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/link.c | 14 ++++++++++++++
1 files changed, 14 insertions(+), 0 deletions(-)
diff --git a/net/tipc/link.c b/net/tipc/link.c
index a3616b9..9d18c9b 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -1802,6 +1802,15 @@ static int link_recv_buf_validate(struct sk_buff *buf)
return pskb_may_pull(buf, hdr_size);
}
+/**
+ * tipc_recv_msg - process TIPC messages arriving from off-node
+ * @head: pointer to message buffer chain
+ * @tb_ptr: pointer to bearer message arrived on
+ *
+ * Invoked with no locks held. Bearer pointer must point to a valid bearer
+ * structure (i.e. cannot be NULL), but bearer can be inactive.
+ */
+
void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *tb_ptr)
{
read_lock_bh(&tipc_net_lock);
@@ -1819,6 +1828,11 @@ void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *tb_ptr)
head = head->next;
+ /* Ensure bearer is still enabled */
+
+ if (unlikely(!b_ptr->active))
+ goto cont;
+
/* Ensure message is well-formed */
if (unlikely(!link_recv_buf_validate(buf)))
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 06/14] tipc: Prevent crash when broadcast link cannot send to all nodes
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Allow TIPC's broadcast link to continue operation when it is unable
to send a message to all nodes in the cluster. Previously, the
broadcast link attempted to put the broadcast pseudo-bearer into a
blocked state; however, this caused a crash because the associated
bearer structure is only partially initialized. Further
investigation has revealed some conceptual problems with blocking
the pseudo-bearer; consequently, this functionality has been
disabled for the time being and the undelivered message is
eventually resent by the broadcast link's existing message
retransmission mechanism (if possible).
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bcast.c | 10 ++++++----
1 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index a008c66..42b1737 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -609,11 +609,13 @@ static int tipc_bcbearer_send(struct sk_buff *buf,
bcbearer->remains = bcbearer->remains_new;
}
- /* Unable to reach all targets */
+ /*
+ * Unable to reach all targets (indicate success, since currently
+ * there isn't code in place to properly block & unblock the
+ * pseudo-bearer used by the broadcast link)
+ */
- bcbearer->bearer.publ.blocked = 1;
- bcl->stats.bearer_congs++;
- return 1;
+ return TIPC_OK;
}
/**
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 07/14] tipc: Fix premature broadcast advertisement by sending node
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Prevent a TIPC node from sending out a LINK_STATE message
advertising a broadcast message that it is in the process
of sending, but has not yet actually sent. Previously, it was
possible for a link timeout to occur in between the time the
broadcast link updated its "last message sent" counter and the
time the broadcast message was passed to the broadcast bearer
for transmission. This ensures that the code which issues
the LINK_STATE message isn't informed of the new message until
the broadcast bearer has had a chance to send it.
Note: The "last message sent" value is stored in the "fsm_msg_count"
field of the link structure used by the broadcast link. Since the
broadcast link doesn't utilize the normal link FSM, this field can
be re-used rather than adding a new field to the broadcast link.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bcast.c | 30 +++++++++++++++++++-----------
1 files changed, 19 insertions(+), 11 deletions(-)
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index 42b1737..eefdd1a 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -143,6 +143,19 @@ static void bcbuf_decr_acks(struct sk_buff *buf)
}
+static void bclink_set_last_sent(void)
+{
+ if (bcl->next_out)
+ bcl->fsm_msg_cnt = mod(buf_seqno(bcl->next_out) - 1);
+ else
+ bcl->fsm_msg_cnt = mod(bcl->next_out_no - 1);
+}
+
+u32 tipc_bclink_get_last_sent(void)
+{
+ return bcl->fsm_msg_cnt;
+}
+
/**
* bclink_set_gap - set gap according to contents of current deferred pkt queue
*
@@ -237,8 +250,10 @@ void tipc_bclink_acknowledge(struct tipc_node *n_ptr, u32 acked)
/* Try resolving broadcast link congestion, if necessary */
- if (unlikely(bcl->next_out))
+ if (unlikely(bcl->next_out)) {
tipc_link_push_queue(bcl);
+ bclink_set_last_sent();
+ }
if (unlikely(released && !list_empty(&bcl->waiting_ports)))
tipc_link_wakeup_ports(bcl, 0);
spin_unlock_bh(&bc_lock);
@@ -394,8 +409,10 @@ int tipc_bclink_send_msg(struct sk_buff *buf)
res = tipc_link_send_buf(bcl, buf);
if (unlikely(res == -ELINKCONG))
buf_discard(buf);
- else
+ else {
bcl->stats.sent_info++;
+ bclink_set_last_sent();
+ }
if (bcl->out_queue_size > bcl->stats.max_queue_sz)
bcl->stats.max_queue_sz = bcl->out_queue_size;
@@ -529,15 +546,6 @@ receive:
tipc_node_unlock(node);
}
-u32 tipc_bclink_get_last_sent(void)
-{
- u32 last_sent = mod(bcl->next_out_no - 1);
-
- if (bcl->next_out)
- last_sent = mod(buf_seqno(bcl->next_out) - 1);
- return last_sent;
-}
-
u32 tipc_bclink_acks_missing(struct tipc_node *n_ptr)
{
return (n_ptr->bclink.supported &&
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 09/14] tipc: Remove per-connection sequence number logic
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Remove validation of the per-connection sequence numbers on routable
connections, since routable connections are not supported by TIPC.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/port.c | 15 +++------------
1 files changed, 3 insertions(+), 12 deletions(-)
diff --git a/net/tipc/port.c b/net/tipc/port.c
index 0737680..ebcbc21 100644
--- a/net/tipc/port.c
+++ b/net/tipc/port.c
@@ -588,19 +588,10 @@ void tipc_port_recv_proto_msg(struct sk_buff *buf)
if (!p_ptr) {
err = TIPC_ERR_NO_PORT;
} else if (p_ptr->publ.connected) {
- if (port_peernode(p_ptr) != msg_orignode(msg))
+ if ((port_peernode(p_ptr) != msg_orignode(msg)) ||
+ (port_peerport(p_ptr) != msg_origport(msg))) {
err = TIPC_ERR_NO_PORT;
- if (port_peerport(p_ptr) != msg_origport(msg))
- err = TIPC_ERR_NO_PORT;
- if (!err && msg_routed(msg)) {
- u32 seqno = msg_transp_seqno(msg);
- u32 myno = ++p_ptr->last_in_seqno;
- if (seqno != myno) {
- err = TIPC_ERR_NO_PORT;
- abort_buf = port_build_self_abort_msg(p_ptr, err);
- }
- }
- if (msg_type(msg) == CONN_ACK) {
+ } else if (msg_type(msg) == CONN_ACK) {
int wakeup = tipc_port_congested(p_ptr) &&
p_ptr->publ.congested &&
p_ptr->wakeup;
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 08/14] tipc: Fix bug in broadcast link transmit statistics computation
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Modify TIPC's broadcast link so that it counts each piece of a
fragmented message individually, rather than as treating the group
as a single message. This ensures that proper correlation of sent
and received traffic can be done when the broadcast link statistics
are displayed, and is consistent with the way fragments are counted
by TIPC's unicast links.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bcast.c | 5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index eefdd1a..b11248c 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -409,10 +409,8 @@ int tipc_bclink_send_msg(struct sk_buff *buf)
res = tipc_link_send_buf(bcl, buf);
if (unlikely(res == -ELINKCONG))
buf_discard(buf);
- else {
- bcl->stats.sent_info++;
+ else
bclink_set_last_sent();
- }
if (bcl->out_queue_size > bcl->stats.max_queue_sz)
bcl->stats.max_queue_sz = bcl->out_queue_size;
@@ -578,6 +576,7 @@ static int tipc_bcbearer_send(struct sk_buff *buf,
msg = buf_msg(buf);
msg_set_non_seq(msg, 1);
msg_set_mc_netid(msg, tipc_net_id);
+ bcl->stats.sent_info++;
}
/* Send buffer over bearers until all targets reached */
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 10/14] tipc: Optimize tipc_node_has_active_links()
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Eliminate unnecessary checking for null node pointer and redundant
check of second active link array entry.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/node.c | 3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/net/tipc/node.c b/net/tipc/node.c
index b634942..9408517 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -237,8 +237,7 @@ void tipc_node_link_down(struct tipc_node *n_ptr, struct link *l_ptr)
int tipc_node_has_active_links(struct tipc_node *n_ptr)
{
- return (n_ptr &&
- ((n_ptr->active_links[0]) || (n_ptr->active_links[1])));
+ return n_ptr->active_links[0] != NULL;
}
int tipc_node_has_redundant_links(struct tipc_node *n_ptr)
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 12/14] tipc: Minor enhancements to name table display format
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Eliminate printing of dashes after name table column headers
(to adhere more closely to the standard format used in tipc-config),
and simplify name table display logic using array lookups rather
than if-then-else logic.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/name_table.c | 44 ++++++++++++++++----------------------------
1 files changed, 16 insertions(+), 28 deletions(-)
diff --git a/net/tipc/name_table.c b/net/tipc/name_table.c
index 8ba7962..d504e49 100644
--- a/net/tipc/name_table.c
+++ b/net/tipc/name_table.c
@@ -877,7 +877,7 @@ static void subseq_list(struct sub_seq *sseq, struct print_buf *buf, u32 depth,
u32 index)
{
char portIdStr[27];
- char *scopeStr;
+ const char *scope_str[] = {"", " zone", " cluster", " node"};
struct publication *publ = sseq->zone_list;
tipc_printf(buf, "%-10u %-10u ", sseq->lower, sseq->upper);
@@ -893,15 +893,8 @@ static void subseq_list(struct sub_seq *sseq, struct print_buf *buf, u32 depth,
tipc_node(publ->node), publ->ref);
tipc_printf(buf, "%-26s ", portIdStr);
if (depth > 3) {
- if (publ->node != tipc_own_addr)
- scopeStr = "";
- else if (publ->scope == TIPC_NODE_SCOPE)
- scopeStr = "node";
- else if (publ->scope == TIPC_CLUSTER_SCOPE)
- scopeStr = "cluster";
- else
- scopeStr = "zone";
- tipc_printf(buf, "%-10u %s", publ->key, scopeStr);
+ tipc_printf(buf, "%-10u %s", publ->key,
+ scope_str[publ->scope]);
}
publ = publ->zone_list_next;
@@ -951,24 +944,19 @@ static void nameseq_list(struct name_seq *seq, struct print_buf *buf, u32 depth,
static void nametbl_header(struct print_buf *buf, u32 depth)
{
- tipc_printf(buf, "Type ");
-
- if (depth > 1)
- tipc_printf(buf, "Lower Upper ");
- if (depth > 2)
- tipc_printf(buf, "Port Identity ");
- if (depth > 3)
- tipc_printf(buf, "Publication");
-
- tipc_printf(buf, "\n-----------");
-
- if (depth > 1)
- tipc_printf(buf, "--------------------- ");
- if (depth > 2)
- tipc_printf(buf, "-------------------------- ");
- if (depth > 3)
- tipc_printf(buf, "------------------");
-
+ const char *header[] = {
+ "Type ",
+ "Lower Upper ",
+ "Port Identity ",
+ "Publication Scope"
+ };
+
+ int i;
+
+ if (depth > 4)
+ depth = 4;
+ for (i = 0; i < depth; i++)
+ tipc_printf(buf, header[i]);
tipc_printf(buf, "\n");
}
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 11/14] tipc: Eliminate useless linked list initialization
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Eliminate the initialization of the port_list field of a newly created
TIPC port, since the port is immediately added to TIPC's global list
of ports (which alters the port_list field).
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/port.c | 1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/net/tipc/port.c b/net/tipc/port.c
index ebcbc21..4f4f763 100644
--- a/net/tipc/port.c
+++ b/net/tipc/port.c
@@ -253,7 +253,6 @@ struct tipc_port *tipc_createport_raw(void *usr_handle,
k_init_timer(&p_ptr->timer, (Handler)port_timeout, ref);
spin_lock_bh(&tipc_port_list_lock);
INIT_LIST_HEAD(&p_ptr->publications);
- INIT_LIST_HEAD(&p_ptr->port_list);
list_add_tail(&p_ptr->port_list, &ports);
spin_unlock_bh(&tipc_port_list_lock);
return &(p_ptr->publ);
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 13/14] tipc: Allow connect() to wait indefinitely
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Cause a socket whose TIPC_CONN_TIMEOUT option is zero to wait
indefinitely for a response to a connection request using connect().
Previously, specifying a timeout of 0 ms resulted in an immediate
timeout, which was inconsistent with the behavior specified by Posix
for a socket's receive and send timeout.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/socket.c | 11 +++++++----
1 files changed, 7 insertions(+), 4 deletions(-)
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 7b81fdd..f7ac94d 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -64,6 +64,7 @@ struct tipc_sock {
struct sock sk;
struct tipc_port *p;
struct tipc_portid peer_name;
+ long conn_timeout;
};
#define tipc_sk(sk) ((struct tipc_sock *)(sk))
@@ -240,9 +241,9 @@ static int tipc_create(struct net *net, struct socket *sock, int protocol,
sock->state = state;
sock_init_data(sock, sk);
- sk->sk_rcvtimeo = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT);
sk->sk_backlog_rcv = backlog_rcv;
tipc_sk(sk)->p = tp_ptr;
+ tipc_sk(sk)->conn_timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT);
spin_unlock_bh(tp_ptr->lock);
@@ -1385,6 +1386,7 @@ static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
struct msghdr m = {NULL,};
struct sk_buff *buf;
struct tipc_msg *msg;
+ long timeout;
int res;
lock_sock(sk);
@@ -1445,11 +1447,12 @@ static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
+ timeout = tipc_sk(sk)->conn_timeout;
release_sock(sk);
res = wait_event_interruptible_timeout(*sk_sleep(sk),
(!skb_queue_empty(&sk->sk_receive_queue) ||
(sock->state != SS_CONNECTING)),
- sk->sk_rcvtimeo);
+ timeout ? timeout : MAX_SCHEDULE_TIMEOUT);
lock_sock(sk);
if (res > 0) {
@@ -1712,7 +1715,7 @@ static int setsockopt(struct socket *sock,
res = tipc_set_portunreturnable(tport->ref, value);
break;
case TIPC_CONN_TIMEOUT:
- sk->sk_rcvtimeo = msecs_to_jiffies(value);
+ tipc_sk(sk)->conn_timeout = msecs_to_jiffies(value);
/* no need to set "res", since already 0 at this point */
break;
default:
@@ -1767,7 +1770,7 @@ static int getsockopt(struct socket *sock,
res = tipc_portunreturnable(tport->ref, &value);
break;
case TIPC_CONN_TIMEOUT:
- value = jiffies_to_msecs(sk->sk_rcvtimeo);
+ value = jiffies_to_msecs(tipc_sk(sk)->conn_timeout);
/* no need to set "res", since already 0 at this point */
break;
case TIPC_NODE_RECVQ_DEPTH:
--
1.7.2.1
^ permalink raw reply related
* [PATCH net-next 14/14] tipc: Prevent missing name table entries when link flip-flops rapidly
From: Paul Gortmaker @ 2010-08-17 21:00 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <allan.stephens@windriver.com>
Ensure that TIPC does not re-establish communication with a
neighboring node until it has finished updating all data structures
containing information about that node to reflect the earlier loss of
contact. Previously, it was possible for TIPC to perform its purge of
name table entries relating to the node once contact had already been
re-established, resulting in the unwanted removal of valid name table
entries.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/discover.c | 8 ++++++++
net/tipc/link.c | 11 ++++++++++-
net/tipc/node.c | 19 +++++++++++++++++++
net/tipc/node.h | 2 ++
4 files changed, 39 insertions(+), 1 deletions(-)
diff --git a/net/tipc/discover.c b/net/tipc/discover.c
index fc1fcf5..f28d1ae 100644
--- a/net/tipc/discover.c
+++ b/net/tipc/discover.c
@@ -203,6 +203,14 @@ void tipc_disc_recv_msg(struct sk_buff *buf, struct bearer *b_ptr)
return;
}
spin_lock_bh(&n_ptr->lock);
+
+ /* Don't talk to neighbor during cleanup after last session */
+
+ if (n_ptr->cleanup_required) {
+ spin_unlock_bh(&n_ptr->lock);
+ return;
+ }
+
link = n_ptr->links[b_ptr->identity];
if (!link) {
dbg("creating link\n");
diff --git a/net/tipc/link.c b/net/tipc/link.c
index 9d18c9b..a6a3102 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -1869,13 +1869,22 @@ void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *tb_ptr)
goto cont;
}
- /* Locate unicast link endpoint that should handle message */
+ /* Locate neighboring node that sent message */
n_ptr = tipc_node_find(msg_prevnode(msg));
if (unlikely(!n_ptr))
goto cont;
tipc_node_lock(n_ptr);
+ /* Don't talk to neighbor during cleanup after last session */
+
+ if (n_ptr->cleanup_required) {
+ tipc_node_unlock(n_ptr);
+ goto cont;
+ }
+
+ /* Locate unicast link endpoint that should handle message */
+
l_ptr = n_ptr->links[b_ptr->identity];
if (unlikely(!l_ptr)) {
tipc_node_unlock(n_ptr);
diff --git a/net/tipc/node.c b/net/tipc/node.c
index 9408517..b702c7b 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -383,6 +383,20 @@ static void node_established_contact(struct tipc_node *n_ptr)
tipc_highest_allowed_slave);
}
+static void node_cleanup_finished(unsigned long node_addr)
+{
+ struct tipc_node *n_ptr;
+
+ read_lock_bh(&tipc_net_lock);
+ n_ptr = tipc_node_find(node_addr);
+ if (n_ptr) {
+ tipc_node_lock(n_ptr);
+ n_ptr->cleanup_required = 0;
+ tipc_node_unlock(n_ptr);
+ }
+ read_unlock_bh(&tipc_net_lock);
+}
+
static void node_lost_contact(struct tipc_node *n_ptr)
{
struct cluster *c_ptr;
@@ -457,6 +471,11 @@ static void node_lost_contact(struct tipc_node *n_ptr)
tipc_k_signal((Handler)ns->handle_node_down,
(unsigned long)ns->usr_handle);
}
+
+ /* Prevent re-contact with node until all cleanup is done */
+
+ n_ptr->cleanup_required = 1;
+ tipc_k_signal((Handler)node_cleanup_finished, n_ptr->addr);
}
/**
diff --git a/net/tipc/node.h b/net/tipc/node.h
index 6f990da..45f3db3 100644
--- a/net/tipc/node.h
+++ b/net/tipc/node.h
@@ -52,6 +52,7 @@
* @active_links: pointers to active links to node
* @links: pointers to all links to node
* @working_links: number of working links to node (both active and standby)
+ * @cleanup_required: non-zero if cleaning up after a prior loss of contact
* @link_cnt: number of links to node
* @permit_changeover: non-zero if node has redundant links to this system
* @routers: bitmap (used for multicluster communication)
@@ -78,6 +79,7 @@ struct tipc_node {
struct link *links[MAX_BEARERS];
int link_cnt;
int working_links;
+ int cleanup_required;
int permit_changeover;
u32 routers[512/32];
int last_router;
--
1.7.2.1
^ permalink raw reply related
* Re: [GIT] Networking
From: David Miller @ 2010-08-17 22:14 UTC (permalink / raw)
To: eric.dumazet; +Cc: torvalds, akpm, netdev, linux-kernel, kaber, peterz
In-Reply-To: <1281990130.2487.70.camel@edumazet-laptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Mon, 16 Aug 2010 22:22:10 +0200
> [PATCH] netfilter: {ip,ip6,arp}_tables: avoid lockdep false positive
>
> After commit 24b36f019 (netfilter: {ip,ip6,arp}_tables: dont block
> bottom half more than necessary), lockdep can raise a warning
> because we attempt to lock a spinlock with BH enabled, while
> the same lock is usually locked by another cpu in a softirq context.
>
> Disable again BH to avoid these lockdep warnings.
>
> Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
> Diagnosed-by: David S. Miller <davem@davemloft.net>
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Since Patrick seems to be busy, I'll merge this directly into
net-2.6, thanks everyone!
^ permalink raw reply
* Re: [PATCH] net sched: fix some kernel memory leaks
From: David Miller @ 2010-08-17 22:14 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev
In-Reply-To: <1282025062.2487.696.camel@edumazet-laptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Tue, 17 Aug 2010 08:04:22 +0200
> We leak at least 32bits of kernel memory to user land in tc dump,
> because we dont init all fields (capab ?) of the dumped structure.
>
> Use C99 initializers so that holes and non explicit fields are zeroed.
>
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Applied, thanks for fixing this Eric.
^ permalink raw reply
* tcp_shift_skb_data uses wrong mss in non-gso case?
From: Robert Evans @ 2010-08-17 22:35 UTC (permalink / raw)
To: 'netdev@vger.kernel.org'
Hello all,
In reading through the latest SACK code introduced by 832d11c, I have noticed
that for the in_sack case tcp_shift_skb will take mss = tcp_skb_seglen(skb).
This seems to be wrong since the queue might contain small packets (f.e.
TCP_NODELAY). If the collapse succeeds, the resulting skb will have an
arbitrarily small gso_size equal to the original skb length.
8ed88d4:net/ipv4/tcp_input.c
1506 in_sack = !after(start_seq, TCP_SKB_CB(skb)->seq) &&
1507 !before(end_seq, TCP_SKB_CB(skb)->end_seq);
1508
1509 if (in_sack) {
1510 len = skb->len;
1511 pcount = tcp_skb_pcount(skb);
1512 mss = tcp_skb_seglen(skb); /* possibly wrong? */
1513
1514 /* TODO: Fix DSACKs to not fragment already SACKed and w
1514 e can
1515 * drop this restriction as unnecessary
1516 */
1517 if (mss != tcp_skb_seglen(prev))
1518 goto fallback;
1519 } else {
This ends up being troublesome if the segment is later retransmitted and the
device driver has trouble with very small gso_size (e1000e seems to be an
example).
Is the small gso_size the correct and/or desired behavior? Or am I missing
something else that prevents this from being a problem?
-Bob Evans
^ permalink raw reply
* [PATCH v2] net/sched: add ACT_CSUM action to update packets checksums
From: Grégoire Baron @ 2010-08-17 23:59 UTC (permalink / raw)
To: netdev
net/sched: add ACT_CSUM action to update packets checksums
ACT_CSUM can be called just after ACT_PEDIT in order to re-compute some
altered checksums in IPv4 and IPv6 packets. The following checksums are
supported by this patch:
- IPv4: IPv4 header, ICMP, IGMP, TCP, UDP & UDPLite
- IPv6: ICMPv6, TCP, UDP & UDPLite
It's possible to request in the same action to update different kind of
checksums, if the packets flow mix TCP, UDP and UDPLite, ...
An example of usage is done in the associated iproute2 patch.
Version 2 changes:
- coding style correction
- remove useless arguments of some functions
- use stack in tcf_csum_dump()
- add tcf_csum_skb_nextlayer() to factor code
Signed-off-by: Gregoire Baron <baronchon@n7mm.org>
---
include/linux/tc_act/Kbuild | 1 +
include/linux/tc_act/tc_csum.h | 32 ++
include/net/tc_act/tc_csum.h | 15 +
net/sched/Kconfig | 10 +
net/sched/Makefile | 1 +
net/sched/act_csum.c | 619 ++++++++++++++++++++++++++++++++++++++++
6 files changed, 678 insertions(+), 0 deletions(-)
create mode 100644 include/linux/tc_act/tc_csum.h
create mode 100644 include/net/tc_act/tc_csum.h
create mode 100644 net/sched/act_csum.c
diff --git a/include/linux/tc_act/Kbuild b/include/linux/tc_act/Kbuild
index 7699093..67b501c 100644
--- a/include/linux/tc_act/Kbuild
+++ b/include/linux/tc_act/Kbuild
@@ -4,3 +4,4 @@ header-y += tc_mirred.h
header-y += tc_pedit.h
header-y += tc_nat.h
header-y += tc_skbedit.h
+header-y += tc_csum.h
diff --git a/include/linux/tc_act/tc_csum.h b/include/linux/tc_act/tc_csum.h
new file mode 100644
index 0000000..a047c49
--- /dev/null
+++ b/include/linux/tc_act/tc_csum.h
@@ -0,0 +1,32 @@
+#ifndef __LINUX_TC_CSUM_H
+#define __LINUX_TC_CSUM_H
+
+#include <linux/types.h>
+#include <linux/pkt_cls.h>
+
+#define TCA_ACT_CSUM 16
+
+enum {
+ TCA_CSUM_UNSPEC,
+ TCA_CSUM_PARMS,
+ TCA_CSUM_TM,
+ __TCA_CSUM_MAX
+};
+#define TCA_CSUM_MAX (__TCA_CSUM_MAX - 1)
+
+enum {
+ TCA_CSUM_UPDATE_FLAG_IPV4HDR = 1,
+ TCA_CSUM_UPDATE_FLAG_ICMP = 2,
+ TCA_CSUM_UPDATE_FLAG_IGMP = 4,
+ TCA_CSUM_UPDATE_FLAG_TCP = 8,
+ TCA_CSUM_UPDATE_FLAG_UDP = 16,
+ TCA_CSUM_UPDATE_FLAG_UDPLITE = 32
+};
+
+struct tc_csum {
+ tc_gen;
+
+ __u32 update_flags;
+};
+
+#endif /* __LINUX_TC_CSUM_H */
diff --git a/include/net/tc_act/tc_csum.h b/include/net/tc_act/tc_csum.h
new file mode 100644
index 0000000..9e8710b
--- /dev/null
+++ b/include/net/tc_act/tc_csum.h
@@ -0,0 +1,15 @@
+#ifndef __NET_TC_CSUM_H
+#define __NET_TC_CSUM_H
+
+#include <linux/types.h>
+#include <net/act_api.h>
+
+struct tcf_csum {
+ struct tcf_common common;
+
+ u32 update_flags;
+};
+#define to_tcf_csum(pc) \
+ container_of(pc,struct tcf_csum,common)
+
+#endif /* __NET_TC_CSUM_H */
diff --git a/net/sched/Kconfig b/net/sched/Kconfig
index 2f691fb..522d5a9 100644
--- a/net/sched/Kconfig
+++ b/net/sched/Kconfig
@@ -518,6 +518,16 @@ config NET_ACT_SKBEDIT
To compile this code as a module, choose M here: the
module will be called act_skbedit.
+config NET_ACT_CSUM
+ tristate "Checksum Updating"
+ depends on NET_CLS_ACT
+ ---help---
+ Say Y here to update some common checksum after some direct
+ packet alterations.
+
+ To compile this code as a module, choose M here: the
+ module will be called act_csum.
+
config NET_CLS_IND
bool "Incoming device classification"
depends on NET_CLS_U32 || NET_CLS_FW
diff --git a/net/sched/Makefile b/net/sched/Makefile
index f14e71b..960f5db 100644
--- a/net/sched/Makefile
+++ b/net/sched/Makefile
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET_ACT_NAT) += act_nat.o
obj-$(CONFIG_NET_ACT_PEDIT) += act_pedit.o
obj-$(CONFIG_NET_ACT_SIMP) += act_simple.o
obj-$(CONFIG_NET_ACT_SKBEDIT) += act_skbedit.o
+obj-$(CONFIG_NET_ACT_CSUM) += act_csum.o
obj-$(CONFIG_NET_SCH_FIFO) += sch_fifo.o
obj-$(CONFIG_NET_SCH_CBQ) += sch_cbq.o
obj-$(CONFIG_NET_SCH_HTB) += sch_htb.o
diff --git a/net/sched/act_csum.c b/net/sched/act_csum.c
new file mode 100644
index 0000000..2d35bed
--- /dev/null
+++ b/net/sched/act_csum.c
@@ -0,0 +1,619 @@
+/*
+ * Checksum updating actions
+ *
+ * Copyright (c) 2010 Gregoire Baron <baronchon@n7mm.org>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation; either version 2 of the License, or (at your option)
+ * any later version.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/spinlock.h>
+
+#include <linux/netlink.h>
+#include <net/netlink.h>
+#include <linux/rtnetlink.h>
+
+#include <linux/skbuff.h>
+
+#include <net/ip.h>
+#include <net/ipv6.h>
+#include <net/icmp.h>
+#include <linux/icmpv6.h>
+#include <linux/igmp.h>
+#include <net/tcp.h>
+#include <net/udp.h>
+
+#include <net/act_api.h>
+
+#include <linux/tc_act/tc_csum.h>
+#include <net/tc_act/tc_csum.h>
+
+#define CSUM_TAB_MASK 15
+static struct tcf_common *tcf_csum_ht[CSUM_TAB_MASK + 1];
+static u32 csum_idx_gen;
+static DEFINE_RWLOCK(csum_lock);
+
+static struct tcf_hashinfo csum_hash_info = {
+ .htab = tcf_csum_ht,
+ .hmask = CSUM_TAB_MASK,
+ .lock = &csum_lock,
+};
+
+static const struct nla_policy csum_policy[TCA_CSUM_MAX + 1] = {
+ [TCA_CSUM_PARMS] = { .len = sizeof(struct tc_csum), },
+};
+
+static int tcf_csum_init(struct nlattr *nla, struct nlattr *est,
+ struct tc_action *a, int ovr, int bind)
+{
+ struct nlattr *tb[TCA_CSUM_MAX + 1];
+ struct tc_csum *parm;
+ struct tcf_common *pc;
+ struct tcf_csum *p;
+ int ret = 0, err;
+
+ if (nla == NULL)
+ return -EINVAL;
+
+ err = nla_parse_nested(tb, TCA_CSUM_MAX, nla,csum_policy);
+ if (err < 0)
+ return err;
+
+ if (tb[TCA_CSUM_PARMS] == NULL)
+ return -EINVAL;
+ parm = nla_data(tb[TCA_CSUM_PARMS]);
+
+ pc = tcf_hash_check(parm->index, a, bind, &csum_hash_info);
+ if (!pc) {
+ pc = tcf_hash_create(parm->index, est, a, sizeof(*p), bind, &csum_idx_gen, &csum_hash_info);
+ if (IS_ERR(pc))
+ return PTR_ERR(pc);
+ p = to_tcf_csum(pc);
+ ret = ACT_P_CREATED;
+ } else {
+ p = to_tcf_csum(pc);
+ if (!ovr) {
+ tcf_hash_release(pc, bind, &csum_hash_info);
+ return -EEXIST;
+ }
+ }
+
+ spin_lock_bh(&p->tcf_lock);
+ p->tcf_action = parm->action;
+ p->update_flags = parm->update_flags;
+ spin_unlock_bh(&p->tcf_lock);
+
+ if (ret == ACT_P_CREATED)
+ tcf_hash_insert(pc, &csum_hash_info);
+
+ return ret;
+}
+
+static int tcf_csum_cleanup(struct tc_action *a, int bind)
+{
+ struct tcf_csum *p = a->priv;
+ return tcf_hash_release(&p->common, bind, &csum_hash_info);
+}
+
+/**
+ * tcf_csum_skb_nextlayer - Get next layer pointer
+ * @skb: sk_buff to use
+ * @ihl: previous summed headers length
+ * @ipl: complete packet length
+ * @jhl: next header length
+ *
+ * Check the expected next layer availability in the specified sk_buff.
+ * Return the next layer pointer if pass, NULL otherwise.
+ */
+static void *tcf_csum_skb_nextlayer(struct sk_buff *skb,
+ unsigned int ihl, unsigned int ipl,
+ unsigned int jhl)
+{
+ int ntkoff = skb_network_offset(skb);
+ int hl = ihl + jhl;
+
+ if (!pskb_may_pull(skb, ipl + ntkoff) || (ipl < hl) ||
+ (skb_cloned(skb) &&
+ !skb_clone_writable(skb, hl + ntkoff) &&
+ pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
+ return NULL;
+ else
+ return (void *)(skb_network_header(skb) + ihl);
+}
+
+static int tcf_csum_ipv4_icmp(struct sk_buff *skb,
+ unsigned int ihl, unsigned int ipl)
+{
+ struct icmphdr *icmph;
+
+ icmph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*icmph));
+ if (icmph == NULL)
+ goto fail;
+
+ icmph->checksum = 0;
+ skb->csum = csum_partial(icmph, ipl - ihl, 0);
+ icmph->checksum = csum_fold(skb->csum);
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv4_igmp(struct sk_buff *skb,
+ unsigned int ihl, unsigned int ipl)
+{
+ struct igmphdr *igmph;
+
+ igmph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*igmph));
+ if (igmph == NULL)
+ goto fail;
+
+ igmph->csum = 0;
+ skb->csum = csum_partial(igmph, ipl - ihl, 0);
+ igmph->csum = csum_fold(skb->csum);
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv6_icmp(struct sk_buff *skb, struct ipv6hdr *ip6h,
+ unsigned int ihl, unsigned int ipl)
+{
+ struct icmp6hdr *icmp6h;
+
+ icmp6h = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*icmp6h));
+ if (icmp6h == NULL)
+ goto fail;
+
+ icmp6h->icmp6_cksum = 0;
+ skb->csum = csum_partial(icmp6h, ipl - ihl, 0);
+ icmp6h->icmp6_cksum = csum_ipv6_magic(&ip6h->saddr, &ip6h->daddr,
+ ipl - ihl, IPPROTO_ICMPV6,
+ skb->csum);
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv4_tcp(struct sk_buff *skb, struct iphdr *iph,
+ unsigned int ihl, unsigned int ipl)
+{
+ struct tcphdr *tcph;
+
+ tcph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*tcph));
+ if (tcph == NULL)
+ goto fail;
+
+ tcph->check = 0;
+ skb->csum = csum_partial(tcph, ipl - ihl, 0);
+ tcph->check = tcp_v4_check(ipl - ihl,
+ iph->saddr, iph->daddr, skb->csum);
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv6_tcp(struct sk_buff *skb, struct ipv6hdr *ip6h,
+ unsigned int ihl, unsigned int ipl)
+{
+ struct tcphdr *tcph;
+
+ tcph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*tcph));
+ if (tcph == NULL)
+ goto fail;
+
+ tcph->check = 0;
+ skb->csum = csum_partial(tcph, ipl - ihl, 0);
+ tcph->check = csum_ipv6_magic(&ip6h->saddr, &ip6h->daddr,
+ ipl - ihl, IPPROTO_TCP,
+ skb->csum);
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv4_udp(struct sk_buff *skb, struct iphdr *iph,
+ unsigned int ihl, unsigned int ipl, int udplite)
+{
+ struct udphdr *udph;
+ u16 ul;
+
+ /* Support both UDP and UDPLITE checksum algorithms,
+ * Don't use udph->len to get the real length without any protocol check,
+ * UDPLITE uses udph->len for another thing,
+ * Use iph->tot_len, or just ipl.
+ */
+
+ udph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*udph));
+ if (udph == NULL)
+ goto fail;
+
+ ul = ntohs(udph->len);
+
+ if (udplite || udph->check) {
+
+ udph->check = 0;
+
+ if (udplite) {
+ if (ul == 0)
+ skb->csum = csum_partial(udph, ipl - ihl, 0);
+
+ else if ((ul >= sizeof(*udph)) && (ul <= ipl - ihl))
+ skb->csum = csum_partial(udph, ul, 0);
+
+ else
+ goto ignore_obscure_skb;
+ } else {
+ if (ul != ipl - ihl)
+ goto ignore_obscure_skb;
+
+ skb->csum = csum_partial(udph, ul, 0);
+ }
+
+ udph->check = csum_tcpudp_magic(iph->saddr, iph->daddr,
+ ul, iph->protocol,
+ skb->csum);
+
+ if (!udph->check)
+ udph->check = CSUM_MANGLED_0;
+ }
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ignore_obscure_skb:
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv6_udp(struct sk_buff *skb, struct ipv6hdr *ip6h,
+ unsigned int ihl, unsigned int ipl, int udplite)
+{
+ struct udphdr *udph;
+ u16 ul;
+
+ /* Support both UDP and UDPLITE checksum algorithms,
+ * Don't use udph->len to get the real length without any protocol check,
+ * UDPLITE uses udph->len for another thing,
+ * Use ip6h->payload_len + sizeof(*ip6h) ... , or just ipl.
+ */
+
+ udph = tcf_csum_skb_nextlayer(skb, ihl, ipl, sizeof(*udph));
+ if (udph == NULL)
+ goto fail;
+
+ ul = ntohs(udph->len);
+
+ udph->check = 0;
+
+ if (udplite) {
+ if (ul == 0)
+ skb->csum = csum_partial(udph, ipl - ihl, 0);
+
+ else if ((ul >= sizeof(*udph)) && (ul <= ipl - ihl))
+ skb->csum = csum_partial(udph, ul, 0);
+
+ else
+ goto ignore_obscure_skb;
+ } else {
+ if (ul != ipl - ihl)
+ goto ignore_obscure_skb;
+
+ skb->csum = csum_partial(udph, ul, 0);
+ }
+
+ udph->check = csum_ipv6_magic(&ip6h->saddr, &ip6h->daddr, ul,
+ udplite ? IPPROTO_UDPLITE : IPPROTO_UDP,
+ skb->csum);
+
+ if (!udph->check)
+ udph->check = CSUM_MANGLED_0;
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ignore_obscure_skb:
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv4(struct sk_buff *skb, u32 update_flags)
+{
+ struct iphdr *iph;
+ int ntkoff;
+
+ ntkoff = skb_network_offset(skb);
+
+ if (!pskb_may_pull(skb, sizeof(*iph) + ntkoff))
+ goto fail;
+
+ iph = ip_hdr(skb);
+
+ switch (iph->frag_off & htons(IP_OFFSET) ? 0 : iph->protocol) {
+ case IPPROTO_ICMP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_ICMP)
+ if (!tcf_csum_ipv4_icmp(skb,
+ iph->ihl * 4, ntohs(iph->tot_len)))
+ goto fail;
+ break;
+ case IPPROTO_IGMP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_IGMP)
+ if (!tcf_csum_ipv4_igmp(skb,
+ iph->ihl * 4, ntohs(iph->tot_len)))
+ goto fail;
+ break;
+ case IPPROTO_TCP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_TCP)
+ if (!tcf_csum_ipv4_tcp(skb, iph,
+ iph->ihl * 4, ntohs(iph->tot_len)))
+ goto fail;
+ break;
+ case IPPROTO_UDP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_UDP)
+ if (!tcf_csum_ipv4_udp(skb, iph,
+ iph->ihl * 4, ntohs(iph->tot_len), 0))
+ goto fail;
+ break;
+ case IPPROTO_UDPLITE:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_UDPLITE)
+ if (!tcf_csum_ipv4_udp(skb, iph,
+ iph->ihl * 4, ntohs(iph->tot_len), 1))
+ goto fail;
+ break;
+ }
+
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_IPV4HDR) {
+ if (skb_cloned(skb) &&
+ !skb_clone_writable(skb, sizeof(*iph) + ntkoff) &&
+ pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
+ goto fail;
+
+ ip_send_check(iph);
+ }
+
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv6_hopopts(struct ipv6_opt_hdr *ip6xh,
+ unsigned int ixhl, unsigned int *pl)
+{
+ int off, len, optlen;
+ unsigned char *xh = (void *)ip6xh;
+
+ off = sizeof(*ip6xh);
+ len = ixhl - off;
+
+ while (len > 0) {
+ switch (xh[off])
+ {
+ case IPV6_TLV_PAD0:
+ optlen = 1;
+ break;
+ case IPV6_TLV_JUMBO:
+ optlen = xh[off + 1] + 2;
+ if (xh[off + 1] != 4 || (off & 3) != 2)
+ /* wrong jumbo option length/alignment */
+ goto fail;
+ *pl = ntohl(*(__be32 *)(xh + off + 2));
+ goto done;
+ default:
+ optlen = xh[off + 1] + 2;
+ if (optlen > len)
+ goto ignore;
+ break;
+ }
+ off += optlen;
+ len -= optlen;
+ }
+
+done:
+ignore:
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum_ipv6(struct sk_buff *skb, u32 update_flags)
+{
+ struct ipv6hdr *ip6h;
+ struct ipv6_opt_hdr *ip6xh;
+ unsigned int hl, ixhl;
+ unsigned int pl;
+ int ntkoff;
+ u8 nexthdr;
+
+ ntkoff = skb_network_offset(skb);
+
+ hl = sizeof(*ip6h);
+
+ if (!pskb_may_pull(skb, hl + ntkoff))
+ goto fail;
+
+ ip6h = ipv6_hdr(skb);
+
+ pl = ntohs(ip6h->payload_len);
+ nexthdr = ip6h->nexthdr;
+
+ do {
+ switch (nexthdr) {
+ case NEXTHDR_FRAGMENT:
+ goto ignore_skb;
+ case NEXTHDR_ROUTING:
+ case NEXTHDR_HOP:
+ case NEXTHDR_DEST:
+ if (!pskb_may_pull(skb, hl + sizeof(*ip6xh) + ntkoff))
+ goto fail;
+ ip6xh = (void *)(skb_network_header(skb) + hl);
+ ixhl = ipv6_optlen(ip6xh);
+ if (!pskb_may_pull(skb, hl + ixhl + ntkoff))
+ goto fail;
+ if ((nexthdr == NEXTHDR_HOP) &&
+ !(tcf_csum_ipv6_hopopts(ip6xh, ixhl, &pl)))
+ goto fail;
+ nexthdr = ip6xh->nexthdr;
+ hl += ixhl;
+ break;
+ case IPPROTO_ICMPV6:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_ICMP)
+ if (!tcf_csum_ipv6_icmp(skb, ip6h,
+ hl, pl + sizeof(*ip6h)))
+ goto fail;
+ goto done;
+ case IPPROTO_TCP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_TCP)
+ if (!tcf_csum_ipv6_tcp(skb, ip6h,
+ hl, pl + sizeof(*ip6h)))
+ goto fail;
+ goto done;
+ case IPPROTO_UDP:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_UDP)
+ if (!tcf_csum_ipv6_udp(skb, ip6h,
+ hl, pl + sizeof(*ip6h), 0))
+ goto fail;
+ goto done;
+ case IPPROTO_UDPLITE:
+ if (update_flags & TCA_CSUM_UPDATE_FLAG_UDPLITE)
+ if (!tcf_csum_ipv6_udp(skb, ip6h,
+ hl, pl + sizeof(*ip6h), 1))
+ goto fail;
+ goto done;
+ default:
+ goto ignore_skb;
+ }
+ } while (pskb_may_pull(skb, hl + 1 + ntkoff));
+
+done:
+ignore_skb:
+ return 1;
+
+fail:
+ return 0;
+}
+
+static int tcf_csum(struct sk_buff *skb,
+ struct tc_action *a, struct tcf_result *res)
+{
+ struct tcf_csum *p = a->priv;
+ int action;
+ u32 update_flags;
+
+ spin_lock(&p->tcf_lock);
+ p->tcf_tm.lastuse = jiffies;
+ p->tcf_bstats.bytes += qdisc_pkt_len(skb);
+ p->tcf_bstats.packets++;
+ action = p->tcf_action;
+ update_flags = p->update_flags;
+ spin_unlock(&p->tcf_lock);
+
+ if (unlikely(action == TC_ACT_SHOT))
+ goto drop;
+
+ switch (skb->protocol) {
+ case cpu_to_be16(ETH_P_IP):
+ if (!tcf_csum_ipv4(skb, update_flags))
+ goto drop;
+ break;
+ case cpu_to_be16(ETH_P_IPV6):
+ if (!tcf_csum_ipv6(skb, update_flags))
+ goto drop;
+ break;
+ }
+
+ return action;
+
+drop:
+ spin_lock(&p->tcf_lock);
+ p->tcf_qstats.drops++;
+ spin_unlock(&p->tcf_lock);
+ return TC_ACT_SHOT;
+}
+
+static int tcf_csum_dump(struct sk_buff *skb,
+ struct tc_action *a, int bind, int ref)
+{
+ unsigned char *b = skb_tail_pointer(skb);
+ struct tcf_csum *p = a->priv;
+ struct tc_csum opt = {
+ .update_flags = p->update_flags,
+
+ .index = p->tcf_index,
+ .action = p->tcf_action,
+ .refcnt = p->tcf_refcnt - ref,
+ .bindcnt = p->tcf_bindcnt - bind,
+ };
+ struct tcf_t t;
+
+ NLA_PUT(skb, TCA_CSUM_PARMS, sizeof(opt), &opt);
+ t.install = jiffies_to_clock_t(jiffies - p->tcf_tm.install);
+ t.lastuse = jiffies_to_clock_t(jiffies - p->tcf_tm.lastuse);
+ t.expires = jiffies_to_clock_t(p->tcf_tm.expires);
+ NLA_PUT(skb, TCA_CSUM_TM, sizeof(t), &t);
+
+ return skb->len;
+
+nla_put_failure:
+ nlmsg_trim(skb, b);
+ return -1;
+}
+
+static struct tc_action_ops act_csum_ops = {
+ .kind = "csum",
+ .hinfo = &csum_hash_info,
+ .type = TCA_ACT_CSUM,
+ .capab = TCA_CAP_NONE,
+ .owner = THIS_MODULE,
+ .act = tcf_csum,
+ .dump = tcf_csum_dump,
+ .cleanup = tcf_csum_cleanup,
+ .lookup = tcf_hash_search,
+ .init = tcf_csum_init,
+ .walk = tcf_generic_walker
+};
+
+MODULE_DESCRIPTION("Checksum updating actions");
+MODULE_LICENSE("GPL");
+
+static int __init csum_init_module(void)
+{
+ return tcf_register_action(&act_csum_ops);
+}
+
+static void __exit csum_cleanup_module(void)
+{
+ tcf_unregister_action(&act_csum_ops);
+}
+
+module_init(csum_init_module);
+module_exit(csum_cleanup_module);
^ permalink raw reply related
* Re: [PATCH] net/sched: add ACT_CSUM action to update packets checksums
From: Grégoire Baron @ 2010-08-18 0:04 UTC (permalink / raw)
To: netdev; +Cc: Eric Dumazet, jamal, David Miller
In-Reply-To: <1282048602.5765.752.camel@bigi>
Thank you for your help!
I've just sent a new version (v2) of this patch.
Regards,
Grégoire
^ permalink raw reply
* Re: [PATCH 1/5] ptp: Added a brand new class driver for ptp clocks.
From: john stultz @ 2010-08-18 0:22 UTC (permalink / raw)
To: Richard Cochran
Cc: netdev, linux-kernel, linuxppc-dev, devicetree-discuss,
linux-arm-kernel, Krzysztof Halasa, Rodolfo Giometti,
Arnd Bergmann
In-Reply-To: <20100817085324.GB3330@riccoc20.at.omicron.at>
On Tue, 2010-08-17 at 10:53 +0200, Richard Cochran wrote:
> On Mon, Aug 16, 2010 at 12:24:48PM -0700, john stultz wrote:
> > 3) I'm not sure I see the benefit of being able to have multiple
> > frequency corrected time domains. In other words, what benefit would
> > you get from adjusting a PTP clock's frequency instead of just
> > adjusting the system's time freq? Having the PTP time as a reference
> > to correct the system time seems reasonable, but I'm not sure I see
> > why userland would want to adjust the PTP clock's freq.
>
> For PTP enabled hardware, the timestamp on the network packet comes
> from from the PTP clock, not from the system time.
>
> Of course, you can always just leave the PTP clock alone, figure the
> needed correction, and apply it whenever needed. But this has some
> disadvantages. First of all, the (one and only) open source PTPd does
> not do it that way. Also, only one program (the PTPd or equivalent)
> will know the correct time. Other programs will not be able to ask the
> operating system for time services. Instead, they would need to use
> IPC to the PTPd.
Why would system time not be adjusted to the PTP time?
This is my main concern, that we're presenting a fractured API to
userland. Suddenly there isn't just system time, but ptp time as well,
and possibly multiple different ptp times.
Forgive me, as I suspect I'm missing something key here.
> The PTP protocol (and some PTP hardware) offers a "one step" method,
> where the timestamps are inserted by the hardware on the fly. Here you
> really do need the PTP clock to be correctly adjusted.
>
> All of the PTP hardware that I am familiar with provides a clock
> adjustment method, so it simpler and cleaner just to use this facility
> to tune the PTP clock.
Hmm. So trying to recap here to see if I'm understanding properly:
The PTP clock is a bit of hardware (usually on the NIC) that can put
timestamps on packets (both incoming or outgoing?).
The need to offset/freq correct the PTP clock is important so that the
timestamps (incoming and outgoing) are correct, which makes future
offset calculations simpler.
Hmm. So I'm actually starting to come around to the chardev interface.
In a way, it seems it has some similarities to how the RTC device is
used in NTPd. Granted, NTPd doesn't correct the RTC (the kernel does
when we're synced, but its not a perfect parallel), but it can be used
as an input the steer the clock.
So while to me, it think it would be more ideal (or maybe just less
different) to have a read-only interface (like the RTC), leaving PTPd to
manage offset calculations and use that to steer the system time. I can
acknowledge the need to have some way to correct the freq so the packet
timestamps are corrected.
I still feel a little concerned over the timer/alarm related interfaces.
Could you explain why the alarm interface is necessary?
So really I think my initial negative gut reaction to this was mostly
out of the fact that you introduced a char dev that provides almost 100%
coverage of the posix-time interface. That is duplication we definitely
don't want.
Also I think the documentation I've read about PTP (likely just due to
the engineering focus) has an odd inverted sense of priority, focusing
on keeping obscure hardware clocks on NIC cards in sync, rather then the
the more tangible feature of keeping the system time in sync.
This could be comically interpreted as trying to create a shadow-time on
the system that is the "real time" and "yea, maybe we'll let the system
know what time it is, but user-apps who want to know the score can send
a magic ioctl to /dev/something and get the real deal". ;) I'm sure
that's not the case, but I'd like to keep any confusion in userland
about which time is the best time to a minimum (ie: use the system
time).
thanks
-john
^ permalink raw reply
* Re: [PATCH net-next 11/14] tipc: Eliminate useless linked list initialization
From: David Miller @ 2010-08-18 0:29 UTC (permalink / raw)
To: paul.gortmaker; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-12-git-send-email-paul.gortmaker@windriver.com>
From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Tue, 17 Aug 2010 17:00:13 -0400
> From: Allan Stephens <allan.stephens@windriver.com>
>
> Eliminate the initialization of the port_list field of a newly created
> TIPC port, since the port is immediately added to TIPC's global list
> of ports (which alters the port_list field).
>
> Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
This is not correct.
INIT_LIST_HEAD() is mandatory for all new objects, since it may
initialize debugging knobs and whatnot inside of the list_head object
when certain debugging options are enabled.
Therefore it may never be elided.
^ permalink raw reply
* Re: [PATCH 0/14 net-next] more TIPC fixes and updates
From: David Miller @ 2010-08-18 0:33 UTC (permalink / raw)
To: paul.gortmaker; +Cc: netdev, allan.stephens
In-Reply-To: <1282078816-11844-1-git-send-email-paul.gortmaker@windriver.com>
From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Tue, 17 Aug 2010 17:00:02 -0400
> With net-next reopened, here is another batch of TIPC patches primarily
> from the out-of-kernel repo. There is no real binding between any of the
> commits aside from taking what appear to be stand-alone bits and migrating
> them across and fixing up minor coding style things etc. I believe there
> will be a chance to get a lot more of this done in this dev window, so
> hopefully this will be just one batch of several that I end up sending.
>
> Basic runtime sanity test done on v2.6.36-rc1, and all patches git am
> cleanly on the net-next of today as well.
All applied except patch #11, as explained in my reply to that patch
posting.
^ permalink raw reply
* Re: [PATCH] net: Fix a memmove bug in dev_gro_receive()
From: David Miller @ 2010-08-18 0:37 UTC (permalink / raw)
To: jarkao2; +Cc: xiaohui.xin, netdev, herbert
In-Reply-To: <20100811120210.GA24019@ff.dom.local>
From: Jarek Poplawski <jarkao2@gmail.com>
Date: Wed, 11 Aug 2010 12:02:10 +0000
>>Xin Xiaohui wrote:
>> I looked into the code dev_gro_receive(), found the code here:
>> if the frags[0] is pulled to 0, then the page will be released,
>> and memmove() frags left.
>> Is that right? I'm not sure if memmove do right or not, but
>> frags[0].size is never set after memove at least. what I think
>> a simple way is not to do anything if we found frags[0].size == 0.
>> The patch is as followed.
> ...
>
> This version of the patch fixes the bug directly in memmove.
>
> Reported-by: "Xin, Xiaohui" <xiaohui.xin@intel.com>
> Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Applied thanks a lot Jarek.
^ permalink raw reply
* Re: [PATCH 1/5] ptp: Added a brand new class driver for ptp clocks.
From: john stultz @ 2010-08-18 0:40 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Richard Cochran, linuxppc-dev, Rodolfo Giometti, netdev,
devicetree-discuss, linux-kernel, linux-arm-kernel,
Krzysztof Halasa
In-Reply-To: <201008171336.29375.arnd@arndb.de>
On Tue, 2010-08-17 at 13:36 +0200, Arnd Bergmann wrote:
> On Tuesday 17 August 2010, Richard Cochran wrote:
> > On Tue, Aug 17, 2010 at 09:25:55AM +0000, Arnd Bergmann wrote:
> > > Another difference is that we generally use ioctl for devices that can
> > > be enumerated, while syscalls are for system services that are not tied to
> > > a specific device. This argument works both ways for PTP IMHO: On the one
> > > hand you want to have a reliable clock that you can use without knowing
> > > where it comes from, on the other you might have multiple PTP sources that
> > > you need to differentiate.
> >
> > Yes, I agree. In normal use, there will be only one PTP clock in a
> > system. However, for research purposes, it would be nice to have more
> > than one.
> >
> > I've been looking at offering the PTP clock as a posix clock, and it
> > is not as hard as I first thought. The PTP clock or clocks just have
> > to be registered as one of the posix_clocks[MAX_CLOCKS] in
> > posix-timers.c.
>
> Ok sounds good.
Oh no. I'm starting to waffle here. :)
So while I'm not a fan of the duplication of the posix clocks interface
that the proposed chardev interface introduced, I'm not sure if
absorbing the PTP clocks as a clockid is much better.
Mainly due to the fact that userland apps now have to chose between two
clockids that supposedly represent the same thing (the current wall
time).
> > My suggestion would be to reserve three clock ids in time.h,
> > CLOCK_PTP0, CLOCK_PTP1, and CLOCK_PTP2. The first one would be the
> > same as CLOCK_REALTIME, for SW timestamping, and the other two would
> > allow two different PTP clocks at the same time, for the research use
> > case.
Why would you have CLOCK_PTP0 == CLOCK_REALTIME? Whats the point of
that?
> I don't think there is a point in making exactly two independent sources
> available. The clockid_t space is not really limited, so we could define
> an arbitrary range of ids for ptp sources that could be used simultaneously,
> as long as we have space more more ids with a fixed number.
>
> Would it be reasonable to assume that on a machine with a large number
> of NICs, you'd want a separate ptp source for each of them for timestamping?
> Or would you preferably define just one source in such a setup?
>
> I think both could be done with the use of class device attributes in
> sysfs for configuration. Maybe you can have one CLOCK_PTP value for one
> global PTP source and use sysfs to configure which device that is.
>
> If you also need simultaneous access to the specific clocks, you could
> have run-time configured clockid numbers in a sysfs attribute for each
> ptp class device.
>
> > Using the clock id will bring another advantage, since it will then be
> > possible for user space to specify the desired timestamp source for
> > SO_TIMESTAMPING.
So how exactly would one map CLOCK_PTPx to an eth device?
So this is a little bit further out there, but assuming PTPd can sync
the PTP clock correctly, why could the kernel itself not sync the PTP
clock to system time?
So instead of the sync path looking like:
1) packet from master arrives on NIC, is timestamped by PTP clock
2) PTPd calculates offset between PTP clock and master time
3) PTPd corrects PTP clock freq/offset
4) PTPd corrects system time freq/offset
It would look like:
1) packet from master arrives on NIC, is timestamped by PTP clock
2) PTPd calculates offset between PTP clock and master time
3) PTPd corrects system time freq/offset
4) kernel corrects PTP clock freq/offset
I'm guessing the indirect PTP clock sync would have greater error, but
it avoids the fractured sense of time.
thanks
-john
^ permalink raw reply
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