Ben Pfaff [Tue, 23 Dec 2008 23:03:37 +0000 (15:03 -0800)]
Make flow_hash() use hash_lookup3(), for speed and hash quality.
Ben Pfaff [Tue, 23 Dec 2008 23:01:25 +0000 (15:01 -0800)]
Add faster and better-quality hash function hash_lookup3().
Ben Pfaff [Mon, 22 Dec 2008 06:19:17 +0000 (22:19 -0800)]
Remove misplaced comment.
Ben Pfaff [Sat, 20 Dec 2008 00:33:31 +0000 (16:33 -0800)]
vswitch: Implement basic bonding.
Rebalancing and link failure detection are missing, but the basics are
there (and work OK in simple testing).
Ben Pfaff [Fri, 19 Dec 2008 23:10:18 +0000 (15:10 -0800)]
vswitch: Pass --monitor to secchan processes, to allow monitoring them.
Justin Pettit [Sat, 20 Dec 2008 00:06:11 +0000 (16:06 -0800)]
Add #include <limits.h> to fix build problem with undefined "_POSIX_PIPE_BUF".
Ben Pfaff [Thu, 18 Dec 2008 22:26:29 +0000 (14:26 -0800)]
vswitchd: Fix stupid thinko.
Ben Pfaff [Thu, 18 Dec 2008 22:00:59 +0000 (14:00 -0800)]
vswitchd: Basic working VLAN support.
Ben Pfaff [Thu, 18 Dec 2008 22:01:35 +0000 (14:01 -0800)]
New functions put_openflow() and put_openflow_xid().
Ben Pfaff [Thu, 18 Dec 2008 22:00:23 +0000 (14:00 -0800)]
Add support for VLAN tags to the MAC learning library.
vswitchd needs to keep separate per-VLAN MAC learning tables, so this adds
a VLAN tag to each MAC learning table entry. The existing users of the
MAC learning table don't care about VLANs, so they always pass in a VLAN
of 0.
There is a very good chance that vswitchd will need additional features in
its MAC learning table that don't fit well into the existing library. In
that case this commit will probably be reverted and a separate MAC learning
implementation added in the vswitch directory.
Ben Pfaff [Thu, 18 Dec 2008 20:43:57 +0000 (12:43 -0800)]
cfg: Fix functions for retrieving keys.
They didn't work. At all.
Ben Pfaff [Wed, 17 Dec 2008 01:19:09 +0000 (17:19 -0800)]
Initial, skeletal implementation of vswitchd.
Ben Pfaff [Wed, 17 Dec 2008 00:39:20 +0000 (16:39 -0800)]
New function ds_get_line().
Ben Pfaff [Wed, 17 Dec 2008 00:39:01 +0000 (16:39 -0800)]
New function ds_put_buffer().
Ben Pfaff [Wed, 17 Dec 2008 00:38:01 +0000 (16:38 -0800)]
New functions for verify OpenFlow message types and lengths.
Ben Pfaff [Wed, 17 Dec 2008 00:36:25 +0000 (16:36 -0800)]
New function ofp_message_type_to_string().
Ben Pfaff [Wed, 17 Dec 2008 00:28:50 +0000 (16:28 -0800)]
Make secchan exit if the connection drops and can never be reconnected.
In particular this happens if the controller is specified using a fd vconn,
since once disconnected a fd vconn can never be reconnected.
This behavior is useful for vswitchd, which connects to secchan over a
fd vconn.
Ben Pfaff [Wed, 17 Dec 2008 00:57:40 +0000 (16:57 -0800)]
Add new "fd" vconn, which takes a file descriptor number as argument.
This requires introducing the concept of a "reconnectable" vconn. A vconn
is reconnectable if, when it is closed, it is possible to try to reconnect
to it using the name that was originally used. This is the case for all
existing vconn types, but it is not true for fd vconns, because closing a
fd vconn closes the specified file descriptor, and thus attempting to
reopen it will get an EBADF error.
An rconn is not reliable if it is connected to a vconn that is not
reconnectable.
Ben Pfaff [Wed, 17 Dec 2008 00:19:36 +0000 (16:19 -0800)]
Introduce x2nrealloc() helper function, and use it.
Also fixes a bug in read_cert_file() in vconn-ssl.c: "sizeof *certs"
should have been "sizeof **certs". In fact the sizes will be identical
in all practical cases since both *certs and **certs are pointer types,
so this is not an important fix.
Ben Pfaff [Wed, 17 Dec 2008 00:13:36 +0000 (16:13 -0800)]
Make datapath interface library more flexible.
A single netlink socket can address any kernel datapath, but the datapath
interface library was artificially restricting a dpif to address a single
datapath that was selected at dpif creation time. This commit removes
that restriction.
This is useful for vswitchd, which can create and destroy any number of
datapaths and add and remove ports on them. It's more convenient to use
a single dpif for that instead of many.
Ben Pfaff [Wed, 17 Dec 2008 00:10:31 +0000 (16:10 -0800)]
Move DP_MAX into public header file.
vswitchd wants to know the maximum number of datapaths, so DP_MAX needs to
be exported.
(This is not an ideal situation. Really the kernel should not have an
explicit limit on the number of datapaths, and userspace should be able
to query the current maximum.)
Ben Pfaff [Wed, 17 Dec 2008 00:35:56 +0000 (16:35 -0800)]
Improve formatting of process termination messages in secchan logging.
My expectation was that strsignal() returns the signal's name, e.g.
SIGTERM. It actually returns an English explanation, so that the existing
code would log a message like "terminated by signal Terminated". This
commit changes the message to the more sensible "terminated by signal 15
(Terminated)".
Also, the recently approved POSIX 2008 standardized strsignal() and in
particular says that it may return NULL for unknown signal numbers, so
this commit fixes the behavior on NULL return.
Ben Pfaff [Wed, 17 Dec 2008 00:32:27 +0000 (16:32 -0800)]
vlog: Optimize logging disabled, rate-limited messages.
Previously the arguments of rate-limited messages were always evaluated
and passed to vlog_rate_limit(). This commit short-circuits the case
where the vlog levels would prevent any messages from being logged even
without considering rate limiting.
Ben Pfaff [Wed, 17 Dec 2008 00:29:36 +0000 (16:29 -0800)]
Reduce rate limit for short-packet messages even further.
These messages appear a lot in a virtualized environment, and they really
are not meaningful there.
Ben Pfaff [Wed, 17 Dec 2008 00:30:19 +0000 (16:30 -0800)]
Fix typo in comment.
Ben Pfaff [Wed, 17 Dec 2008 00:07:41 +0000 (16:07 -0800)]
Add reminder comment to lib/learning-switch.c.
(This doesn't change the semantics of the code at all.)
Ben Pfaff [Fri, 12 Dec 2008 22:56:57 +0000 (14:56 -0800)]
Fix mangled header guards in ofp-print.h.
Ben Pfaff [Fri, 12 Dec 2008 01:14:18 +0000 (17:14 -0800)]
Improve secchan.8 manpage.
Ben Pfaff [Fri, 12 Dec 2008 01:00:43 +0000 (17:00 -0800)]
Fix typo in error message.
Ben Pfaff [Fri, 12 Dec 2008 00:41:25 +0000 (16:41 -0800)]
Style fix: f(x) is better than f((x))
Ben Pfaff [Thu, 11 Dec 2008 22:07:08 +0000 (14:07 -0800)]
Add userspace datapath to openflow.
Ben Pfaff [Thu, 11 Dec 2008 22:06:34 +0000 (14:06 -0800)]
Update required Autoconf version in INSTALL.
The AC_PREREQ was updated some time ago but the documentation lagged
behind.
Ben Pfaff [Thu, 11 Dec 2008 17:35:24 +0000 (09:35 -0800)]
Make ds_put_char_multiple() use the provided character instead of a space.
Fortunately this function was only used for formatting log messages, in
a context where spaces were the normal padding character.
Found by Chris Eagle.
Justin Pettit [Wed, 3 Dec 2008 07:55:12 +0000 (23:55 -0800)]
Cleanup some formatting in openflow.h.
Justin Pettit [Wed, 3 Dec 2008 06:17:36 +0000 (22:17 -0800)]
Change references in comments from "ofp_vendor" to "ofp_vendor_header".
Ben Pfaff [Tue, 2 Dec 2008 16:27:31 +0000 (08:27 -0800)]
Update prerequisite Autoconf version to 2.60 (from 2.59).
We use the MKDIR_P variable that only Autoconf 2.60 and
later supports, but we were only requiring Autoconf 2.59
or later. This fixes the problem.
Also, we were not explicitly invoking AC_PROG_MKDIR_P from
configure.ac, even though we were using $(MKDIR_P). Some
other macro must have been internally invoking
AC_PROG_MKDIR_P, but it's not good to depend on that, so
this commit adds an explicit call to it.
Justin Pettit [Tue, 2 Dec 2008 06:23:14 +0000 (22:23 -0800)]
Fix compatibility back to Linux 2.6.15 and 2.4.32.
Ben Pfaff [Tue, 2 Dec 2008 05:26:57 +0000 (21:26 -0800)]
Add missing "%s" to VLOG_WARN_RL call.
Thanks to Justin for pointing out the problem.
Justin Pettit [Tue, 2 Dec 2008 05:08:43 +0000 (21:08 -0800)]
Officially call this code version 0.8.9~1.
Justin Pettit [Mon, 1 Dec 2008 22:34:23 +0000 (14:34 -0800)]
Add support for listing and deleting entries based on an output port.
To support this, an "out_port" field has been added to the "ofp_flow_mod",
"ofp_flow_stats_request", and "ofp_aggregate_stats_request" messages. If an
"out_port" contains a value other than "OFPP_NONE", it introduces a constraint
when matching. This constraint is that the rule must contain an output action
directed at that port. Other constraints such as ofp_match structs and
priorities are still used; this is purely an *additional* constraint. Note
that to get previous behavior, though, "out_port" must be set to "OFPP_NONE",
since "0" is a valid port id. This only applies to the delete and
delete_strict flow mod commands; the field is ignored by add, modify, and
modify_strict.
Ben Pfaff [Wed, 26 Nov 2008 19:38:28 +0000 (11:38 -0800)]
Add DATAPATH_ID to switch packaging, to enable persistent DPIDs.
Ben Pfaff [Wed, 26 Nov 2008 19:27:06 +0000 (11:27 -0800)]
Allow the datapath ID to be configured by changing the ofX MAC address.
Previously, you could change the MAC address but that had no effect on
the datapath ID, so that there was no way to change the datapath ID at all.
Ben Pfaff [Thu, 27 Nov 2008 00:18:16 +0000 (16:18 -0800)]
Simplify code in stream_recv().
There is no need to test whether retval is nonzero at this point in the
code: we already know that it is negative from the two previous "if"
branches.
Ben Pfaff [Thu, 27 Nov 2008 00:07:56 +0000 (16:07 -0800)]
Make dpif_close() accept a null pointer.
It's customary for destructor functions to accept and ignore null pointers.
This commit does not fix any known bug.
Ben Pfaff [Thu, 27 Nov 2008 00:27:21 +0000 (16:27 -0800)]
rconn_destroy() should close monitoring connections, to avoid a leak.
Our current code base never closes an rconn that has a monitoring
connection, so this does not fix any larger problem.
Ben Pfaff [Thu, 27 Nov 2008 00:25:55 +0000 (16:25 -0800)]
Free txbuf, rxbuf in vconn-ssl and vconn-stream destructors.
This bug was causing secchan to leak a little bit of memory on almost every
connection close. This is most visible when openflow-monitor is running,
because openflow-monitor connects to secchan once per second.
Justin Pettit [Thu, 27 Nov 2008 04:57:54 +0000 (20:57 -0800)]
Add support for understanding ICMP type and code in flow entries.
Ben Pfaff [Fri, 21 Nov 2008 21:05:37 +0000 (13:05 -0800)]
Maintain separate async and sync connections to nl:0 in secchan.
When a network link is flooded with traffic, secchan's OpenFlow socket
queue becomes congested with traffic. This leaves no room for replies
to OpenFlow requests relayed to that socket by secchan.
This commit modifies secchan to use separate sockets for asynchronous
traffic and for OpenFlow requests/replies, which should avoid the problem.
Ben Pfaff [Fri, 21 Nov 2008 20:35:20 +0000 (12:35 -0800)]
Dynamically allocate switch status categories.
I got tired of increasing the statically allocated number of categories
whenever we exceeded it, so this will make things simpler.
Ben Pfaff [Fri, 21 Nov 2008 20:23:01 +0000 (12:23 -0800)]
Get rid of unused parameter to rate_limit_start().
Ben Pfaff [Fri, 21 Nov 2008 18:48:00 +0000 (10:48 -0800)]
Don't rate-limit packet_ins from flows that the controller set up.
Ben Pfaff [Thu, 20 Nov 2008 18:10:59 +0000 (10:10 -0800)]
Prevent accidentally passing an integer value to IP_ARGS.
Ben Pfaff [Thu, 20 Nov 2008 18:07:22 +0000 (10:07 -0800)]
Fix printing of IP addresses in ofp_print_action().
IP_ARGS takes a pointer, not a raw IP address.
Thanks to kk yap for reporting the problem.
Ben Pfaff [Thu, 20 Nov 2008 17:53:27 +0000 (09:53 -0800)]
Use "%zu" to print size_t, not "%"PRIu16.
Ben Pfaff [Thu, 13 Nov 2008 20:48:02 +0000 (12:48 -0800)]
Copy skbs when we save them.
Ben Pfaff [Thu, 13 Nov 2008 20:50:03 +0000 (12:50 -0800)]
Simplify code for constructing skb in recv_packet_out().
Ben Pfaff [Thu, 13 Nov 2008 19:29:20 +0000 (11:29 -0800)]
Fix double-free: NF_HOOK sometimes frees the sk_buff passed in.
NF_HOOK is supposed to *always* consume the sk_buff passed in, either
internally or through the okfn argument. We assumed that it never
consumed its sk_buff, which was OK in the case where it called okfn,
since our okfn (snat_pre_route_finish) never freed its sk_buff, but
not when one of the netfilter hooks dropped or stole the packet, because
then we'd assume that it still existed and free it a second time.
The other users of NF_HOOK in this file, in snat_skb() and
snat_skb_finish(), do not need to be fixed because they always pass a
copy of their sk_buff argument to NF_HOOK and expect it to be freed.
Ben Pfaff [Thu, 13 Nov 2008 18:26:15 +0000 (10:26 -0800)]
No need to test argument of kfree_skb() for non-null.
Ben Pfaff [Thu, 13 Nov 2008 18:25:06 +0000 (10:25 -0800)]
Never free an skb that has been passed to genlmsg_reply().
genlmsg_reply() always consumes its argument, not just in the success case.
Ben Pfaff [Thu, 13 Nov 2008 20:44:35 +0000 (12:44 -0800)]
Pull arp header before grabbing the pointer to it.
Ben Pfaff [Wed, 12 Nov 2008 22:42:51 +0000 (14:42 -0800)]
Pull data into headers properly, and checksum ICMP replies correctly.
Ben Pfaff [Wed, 12 Nov 2008 00:51:56 +0000 (16:51 -0800)]
Reset mac header in flow_extract(), because it might initially be null.
__alloc_skb() sets the mac header to null, and not all call chains
reset it.
Ben Pfaff [Wed, 12 Nov 2008 00:18:32 +0000 (16:18 -0800)]
Use skb_copy_bits() to copy data that might not be pulled into headers.
Ben Pfaff [Tue, 11 Nov 2008 23:41:04 +0000 (15:41 -0800)]
Verify in fwd_port_input() that 'skb' has no destructor.
Ben Pfaff [Tue, 11 Nov 2008 23:40:48 +0000 (15:40 -0800)]
Verify in execute_actions() that 'skb' is not shared.
Ben Pfaff [Tue, 11 Nov 2008 23:40:28 +0000 (15:40 -0800)]
Fix datapath make_writable() function.
- We weren't necessarily pulling enough into the headers in the non-shared
case, since the IP header by itself can be longer than 40 bytes.
- The skb is guaranteed not to have a destructor at this point.
Ben Pfaff [Tue, 11 Nov 2008 23:35:26 +0000 (15:35 -0800)]
Make datapath's flow_extract() properly pull data into the headers.
Otherwise we might be reading headers that aren't really there.
Ben Pfaff [Tue, 11 Nov 2008 23:48:08 +0000 (15:48 -0800)]
Add comment.
Ben Pfaff [Tue, 11 Nov 2008 02:26:25 +0000 (18:26 -0800)]
Don't oops in dp_output_control() for skbs with null ->dev.
This shouldn't ordinarily happen, since normal packets are received from
a real device, but some packets can be spontaneously generated within
the switch and thus have null ->dev. (Possibly that is itself a bug
that we should track down; not sure.)
Ben Pfaff [Tue, 11 Nov 2008 22:03:31 +0000 (14:03 -0800)]
Verify in fwd_port_input() that we are not passed packets that are shared.
Ben Pfaff [Tue, 11 Nov 2008 21:57:07 +0000 (13:57 -0800)]
When a packet arrives via bridging, clone it if it is shared.
Otherwise we will mangle the packet for anyone who came before us
(e.g. tcpdump via AF_PACKET).
Ben Pfaff [Tue, 11 Nov 2008 04:56:13 +0000 (20:56 -0800)]
Make sure that do_port_input() always puts a MAC header on packets.
Prompted by persistent oopses on packets received by e1000e, in which
skb_mac_header() for the packet always returned null.
Ben Pfaff [Tue, 11 Nov 2008 21:33:24 +0000 (13:33 -0800)]
In handle_arp_snat() and snat_pre_route() pull enough payload into the headers.
There is no guarantee that the device put any data at all into the header.
The e1000 device, for example, appears to not put any data into the header
when the packet is longer than its configured copybreak value, which is
256 bytes by default. So we need to do it ourselves.
Ben Pfaff [Tue, 11 Nov 2008 20:58:52 +0000 (12:58 -0800)]
In update_mapping(), update m->hw_addr unconditionally.
There is no benefit to comparing it first, since we already know that that
cache line will be dirtied by the assignment to m->used.
Ben Pfaff [Tue, 11 Nov 2008 20:52:24 +0000 (12:52 -0800)]
Factor common SNAT code into new functions.
Ben Pfaff [Tue, 11 Nov 2008 21:01:12 +0000 (13:01 -0800)]
Make snat_skb() skb argument const.
After all, it doesn't modify it.
Ben Pfaff [Tue, 11 Nov 2008 04:06:45 +0000 (20:06 -0800)]
Don't process packets in NAT that aren't destined for us.
This time for sure!
Ben Pfaff [Tue, 11 Nov 2008 22:08:41 +0000 (14:08 -0800)]
Orphan and clone packets transmitted on dp_dev (proper fix for bug #478).
We need to orphan packets transmitted on dp_dev so that any socket that
the packets came from is no longer charged for them. If we don't do
this, then the sk_buff will be destructed eventually, but it is
harder to predict when.
We need to clone packets because we are sticking them on
&dp_dev->xmit_queue, which is done by modifying the sk_buff, which we
can only do if it is not shared.
Ben Pfaff [Wed, 12 Nov 2008 22:58:44 +0000 (14:58 -0800)]
Always compile openflow modules with debug information.
Ben Pfaff [Tue, 11 Nov 2008 20:13:49 +0000 (12:13 -0800)]
Check kmalloc() return value.
Ben Pfaff [Tue, 11 Nov 2008 20:11:31 +0000 (12:11 -0800)]
Fix typo: = should be ==
Ben Pfaff [Tue, 11 Nov 2008 01:45:23 +0000 (17:45 -0800)]
Stopgap fix for bug #478, where kernel panics on SNAT to input port.
Ben Pfaff [Mon, 10 Nov 2008 21:56:11 +0000 (13:56 -0800)]
Fix " is running" and " is not running" messages from openflow-switch init.d
With the "secchan" name in there it makes a lot more sense!
Ben Pfaff [Mon, 10 Nov 2008 18:43:18 +0000 (10:43 -0800)]
Make openflow-switch depend on dmidecode, procps instead of acpi-support.
acpi-support pulls in a lot of stuff we really don't want or need, such
as various X11-related stuff. Being more specific reduces our
dependencies.
Ben Pfaff [Mon, 10 Nov 2008 18:10:02 +0000 (10:10 -0800)]
Add build number to kernel modules built via Debian packaging.
Ben Pfaff [Fri, 7 Nov 2008 01:26:42 +0000 (17:26 -0800)]
Add --with-build-number configure argument and support in debian/rules.
Justin Pettit [Thu, 6 Nov 2008 20:42:07 +0000 (12:42 -0800)]
Add --monitor flag to default init scripts for secchan.
Ben Pfaff [Thu, 6 Nov 2008 18:25:08 +0000 (10:25 -0800)]
New package openflow-dbg for debugging symbols for the rest of OpenFlow.
Ben Pfaff [Thu, 6 Nov 2008 18:07:41 +0000 (10:07 -0800)]
Fix typo in code to set core limit that broke the openflow-switch init script.
Ben Pfaff [Wed, 5 Nov 2008 23:47:44 +0000 (15:47 -0800)]
Add openflow-switch settings for the limit on core files.
Ben Pfaff [Wed, 5 Nov 2008 23:46:25 +0000 (15:46 -0800)]
Add new Debian package "corekeeper" to centralize and reap core files.
Justin Pettit [Wed, 5 Nov 2008 23:24:29 +0000 (15:24 -0800)]
Rename switchmon to switchui.
Ben Pfaff [Wed, 5 Nov 2008 22:07:15 +0000 (14:07 -0800)]
Ignore openflow-monitor and openflow-wdt build directories.
Justin Pettit [Wed, 5 Nov 2008 21:01:05 +0000 (13:01 -0800)]
Add dist_sbin_SCRIPTS make target.
Justin Pettit [Wed, 5 Nov 2008 21:00:45 +0000 (13:00 -0800)]
Have git ignore the monitor and wdt packaging links.
Ben Pfaff [Wed, 5 Nov 2008 18:22:41 +0000 (10:22 -0800)]
When a new OpenFlow kernel module package is installed, restart the switch.
This helps to ensure that the secchan and the kernel module are exactly
the same version.
Ben Pfaff [Wed, 5 Nov 2008 18:20:04 +0000 (10:20 -0800)]
Unload kernel module when stopping switch, in Debian init scripts.
This way, we have a better chance at ensuring that the secchan and the
kernel module are exactly the same version.
Ben Pfaff [Wed, 5 Nov 2008 18:08:47 +0000 (10:08 -0800)]
Add required "subst" script to EXTRA_DIST.
Ben Pfaff [Wed, 5 Nov 2008 18:04:31 +0000 (10:04 -0800)]
Move "update" remote command from base secchan into Debian packaging.
The update command only works with Debian anyway, so there's no point
in installing it elsewhere.
Ben Pfaff [Wed, 5 Nov 2008 18:03:49 +0000 (10:03 -0800)]
New switch remote command for setting configuration parameters.