Simon Horman [Fri, 12 Oct 2012 00:26:28 +0000 (09:26 +0900)]
ofproto: Do not add flow on flow mod if new_cookie is UINT64_MAX
Internally a new_cookie value UINT64_MAX is used for
an OpenFlow 1.2 flow mod whose command is not Add.
Open Flow 1.2 does not permit adding flows for such commands.
Also, UINT64_MAX is a prohibited cookie value, and the
existing code created a flow with that value as for the cookie.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Simon Horman [Fri, 12 Oct 2012 00:26:29 +0000 (09:26 +0900)]
ofp-print: Do not display flow mod cookie if it is UINT64_MAX
Internally UINT64_MAX is used to denote that the cookie
should not be changed, so it seems appropriate
to omit this value when pretty printing flow mods.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Simon Horman [Fri, 12 Oct 2012 00:26:27 +0000 (09:26 +0900)]
ofp-util: Correct setting of Flow Mod cookie on encode
When the command of a Flow Mod messages is ADD the cookie should be set as
fm->new_cookie, this is the new cookie value to be set. Otherwise it
should be set as fm->cookie, internally this is the cookie value to match.
Also remove 'XXX' marker from the matching code in the decoder.
I am now comfortable with the implementation.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Simon Horman [Fri, 12 Oct 2012 04:00:38 +0000 (13:00 +0900)]
ofp-msgs: Update for OpenFlow 1.1 & 1.2 Port Desc Stats
* Split OFPRAW_OFPST_PORT_DESC_REPLY into OpenFlow 1.0
and OpenFlow 1.1+ versions.
* Allow OFPRAW_OFPST_PORT_DESC_REQUEST to be used for
OpenFlow 1.0+ instead of only OpenFlow 1.0.
This appears to be necessary and sufficient to allow encoding of Port Desc
Stats reply messages and decoding of Port Desc Stats request messages. Or
in other words both the client and server side of ovs-ofctl dump-ports-desc
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Ralf Spenneberg [Tue, 2 Oct 2012 18:31:55 +0000 (14:31 -0400)]
build: Add support for building RPMs for Fedora Linux
Add RPM specfiles for building OVS for Fedora Linux. This
allows users of the upstream project the ability to
generate RPMs for their Open vSwitch needs.
Signed-off-by: Ralf Spenneberg <ralf@spenneberg.net>
[kmestery@cisco.com made substantial changes including
writing documentation] Signed-off-by: Kyle Mestery <kmestery@cisco.com> Signed-off-by: Ben Pfaff <blp@nicira.com>
Pravin B Shelar [Tue, 9 Oct 2012 23:08:39 +0000 (16:08 -0700)]
datapath: Fix GRE tunnel without any key.
Commit 2de795adb96 (datapath: 64-bit GRE support) introduced a bug
for tunnels with no key. Following patch fixes it by setting tunnel
type to GRE type.
Ethan Jackson [Fri, 5 Oct 2012 20:24:21 +0000 (13:24 -0700)]
config: Add explicit support for building on ESX.
The ESX userspace looks quite a bit like linux, but has some key
differences which need to be specially handled in the build. To
distinguish between ESX and systems which use the linux datapath
module, this patch adds two new macros "ESX" and "LINUX_DATAPATH".
It uses these macros to disable building code on ESX which only
applies to a true Linux environment. In addition, it adds a new
route-table-stub implementation which is required for the build to
complete successfully on ESX.
Ethan Jackson [Fri, 5 Oct 2012 01:38:59 +0000 (18:38 -0700)]
timeval: Recover from failed timer_create() calls.
The timer_create() system call is not supported in ESX and returns
an error when called. Aborting when this system call fails seems a
bit extreme. So instead, this patch simply falls back to disabling
the cached time optimization.
Ethan Jackson [Fri, 5 Oct 2012 00:57:16 +0000 (17:57 -0700)]
timeval: Don't require signals for time_alarm().
Before this patch, time_alarm() used the SIGALRM handler to notify
the poll loop that it should exit the program. Instead, this patch
simply implements time_alarm() directly in the pool loop. This
significantly simplifies the code, while removing a call to
timer_create() which is not currently supported on ESX.
Ben Pfaff [Thu, 4 Oct 2012 22:11:39 +0000 (15:11 -0700)]
ofproto-dpif: Avoid zeroing tunnel info in handle_miss_upcalls().
Commit 296e07ace0f (flow: Extend struct flow to contain tunnel outer
header.) changed the tunnel ID parameter of flow_extract() from an integer
passed by value to a structure passed by pointer. Before flow_extract()
reads the tunnel ID, it zeros the entire flow parameter. This means that,
if a caller passes the address of the tunnel member of the flow as the
tunnel ID, then flow_extract() zeros the tunnel data before it reads and
copies the tunnel data (that it just zeroed). The result is that the
tunnel data is ignored.
This commit fixes the problem by making the caller that did this use a
separate flow structure instead of trying to be clever.
Bug #13461. CC: Pankaj Thakkar <thakkar@nicira.com> Reported-by: Michael Hu <mhu@nicira.com> Signed-off-by: Ben Pfaff <blp@nicira.com>
2012-09-14T05:38:26Z|00001|jsonrpc|WARN|tcp:127.0.0.1:6634: receive error: Con
ovsdb-client: transaction failed (Connection reset by peer)
NOTE: This occurs intermittently depending on how ovsdb-server runs.
Running ovsdb-client on a remote machine increases the possibility.
This is because ovsdb-server closes newly accepted tcp connection.
The following changesets caused it. struct jsonrpc_session::dscp isn't set
based on listening socket's dscp value.
- ovsdb-server creates passive connection and listens on it.
- ovsdb-server accepts connection by ovsdb_jsonrpc_server_run().
The accepted socket inherits from the listening sockets.
ovsdb_jsonrpc_server_run() creates json session, but leaves dscp of
struct jsonrpc_session zero.
- On calling reconfigure_from_db(), it resets dscp value to
all jsonrpc sessions. Eventually jsonrpc_session_set_dscp() is called.
Then jsonrpc_session_force_reconnect() closes the connection.
With this patch,
- struct jsonrpc_session::dscp is correctly set based on
listening sockets dscp value.
- dscp of listening socket is changed dynamically by setsockopt.
This leaves a window where accepted socket may have old dscp.
But it is ignored for now because it would complicates codes
too much.
Ben Pfaff [Thu, 4 Oct 2012 22:55:43 +0000 (15:55 -0700)]
Avoid requiring autom4te when it won't necessarily be needed.
Commit 7fd3e43db07 (tests: Allow environment specified autom4te.) added a
requirement for Autom4te without anyone noticing it in advance. This
caused a build failure in the XenServer DDK, which doesn't have Autom4te.
Since Autom4te is only needed if the testsuite is modified, this commit
makes it optional.
(Autom4te is part of Autoconf.)
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Ethan Jackson <ethan@nicira.com>
Ben Pfaff [Thu, 4 Oct 2012 22:53:14 +0000 (15:53 -0700)]
Fix vswitch-idl.ovsidl build rule with separate source and build dirs.
Before commit e3a880272 (lib: Specify idl location more precisely.), the
files named in VSWITCH_IDL_FILES were relative to the source directory.
That commit made them inconsistent: one remained relative to the source
directory, the other became relative to the build directory. This meant
that if the source and build directories differed, the ovsdb-idlc
invocation had no change of succeeding.
This commit fixes the problem by making the file names consistently
relative to the build directory and then adjusting the ovsdb-idlc
invocation to expect that.
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Ethan Jackson <ethan@nicira.com>
Jesse Gross [Thu, 4 Oct 2012 20:34:36 +0000 (13:34 -0700)]
datapath: Print device when warning about over MTU packets.
If an attempt is made to transmit a packet that is over the device's
MTU then we log it using the datapath's name. However, it is much
more helpful to use the device name instead.
Ethan Jackson [Wed, 3 Oct 2012 22:58:54 +0000 (15:58 -0700)]
tests: Allow environment specified autom4te.
In the ESX build system, autom4te is in a funny location specified
by an environment variable. Before this patch, this environment
variable was overridden by the tests automake file.
The new struct flow_tnl contains an extra four bytes of padding on
64-bit machines but we currently assert that the total struct flow
is a fixed size. The size difference isn't actually a problem
because both are multiples of 4 and the build assertion is only
intended to remind people to update FLOW_WC_SEQ when new fields are
added. This changes the assertion to fix just the non-tunnel field
size.
Suggested-by: Ben Pfaff <blp@nicira.com> Signed-off-by: Jesse Gross <jesse@nicira.com>
flow: Extend struct flow to contain tunnel outer header.
Soon the kernel will begin supplying the information about the outer
IP header for tunneled packets and userspace will need to be able to
track it as part of the flow. For the time being this is only used
internally by OVS and not exposed outwards to OpenFlow. As a result,
this threads the information throughout userspace but simply stores
the existing tun_id in it.
Ben Pfaff [Mon, 1 Oct 2012 20:37:47 +0000 (13:37 -0700)]
ovs-ctl: Add support for glibc malloc debugging.
Unlike valgrind, glibc's built-in features for malloc debugging are cheap
enough that one can run with them enabled all the time, at least in test
scenarios.
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Kyle Mestery <kmestery@cisco.com>
Ethan Jackson [Thu, 20 Sep 2012 18:13:15 +0000 (11:13 -0700)]
idl: Optionally warn when writing to read-write columns.
ovs-vswitchd should only write to write-only columns. Furthermore,
writing to a column which is not write-only can cause serious
performance degradations. This patch causes ovs-vswitchd to log
and reject writes to read-write columns.
python/ovs/db/idl: getattr(Row) raises TypeError, not AttributeError.
In some cases getattr(Row instance, attrname) doesn't raise AttributeError,
but TypeError
> File "python/ovs/db/idl.py", line 554, in __getattr__
> datum = self._data[column_name]
> TypeError: 'NoneType' object has no attribute '__getitem__'
So getattr(Row instance, attrname, default value) doesn't work.
This occurs when row._changes doesn't include attrname and row._data is None.
So teach Row.__getattr__ _data=None case.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp> Signed-off-by: Ben Pfaff <blp@nicira.com>
ofp-actions: Add support for OpenFlow 1.2 "set-field" action.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp> Signed-off-by: Simon Horman <horms@verge.net.au>
[blp@nicira.com extracted this code from a larger patch by above, so:] Signed-off-by: Ben Pfaff <blp@nicira.com>
[regarding final version of patch:] Reviewed-by: Simon Horman <horms@verge.net.au>
Ben Pfaff [Mon, 24 Sep 2012 20:18:38 +0000 (13:18 -0700)]
ofp-actions: Allow OF1.1+ actions to be variable-length.
Previously there was no need for this, because all implemented standard
OpenFlow actions had a fixed length, but the OF1.2 "set-field" action (soon
to be implemented) is variable length.
Signed-off-by: Ben Pfaff <blp@nicira.com> Reviewed-by: Simon Horman <horms@verge.net.au>
Ben Pfaff [Tue, 25 Sep 2012 17:23:38 +0000 (10:23 -0700)]
ofp-actions: Prepare to treat OF1.2 actions as OF1.1 actions.
The numbering of OpenFlow 1.0 actions overlaps with the numbering
of OpenFlow 1.1+ actions, so the two sets of actions have to be
distinguished for input and output. But OpenFlow 1.1 and 1.2
actions are numbered to avoid this problem, so there is no need
to distinguish them in the same way. Therefore, this commit
prepares to treat them together.
Signed-off-by: Ben Pfaff <blp@nicira.com> Reviewed-by: Simon Horman <horms@verge.net.au>
Ben Pfaff [Mon, 24 Sep 2012 20:11:37 +0000 (13:11 -0700)]
openflow-1.2: Remove OFPAT12_* definitions that duplicate OFPAT11_* ones.
OpenFlow 1.1 and 1.2 action numbering is compatible, in that no
OpenFlow 1.2 action uses an OpenFlow 1.1 action number in a different
way from OpenFlow 1.1. So it's confusing and unnecessary to have
separate definitions for these numbers.
Signed-off-by: Ben Pfaff <blp@nicira.com> Reviewed-by: Simon Horman <horms@verge.net.au>
When unparsing the kernel tunnel configuration, TTL was incorrectly
converted to "tos". Although it leads to confusing configuration
output, actual operation is not affected.
Ben Pfaff [Fri, 7 Sep 2012 17:07:03 +0000 (10:07 -0700)]
ovsdb-server: Add support for multiple databases.
The OVSDB protocol has supported multiple databases for a long time, but
the ovsdb-server implementation only supported one database at a time.
This commit adds support for multiple databases.
Feature #12353. Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Ethan Jackson <ethan@nicira.com>
Ben Pfaff [Fri, 21 Sep 2012 18:12:39 +0000 (11:12 -0700)]
ovsdb-server: Fix null pointer deref when bool "is_connected" is empty.
The ovsdb-server supports obtaining its remote connection targets from a
database table and updating that table with connection status information.
One of the supported connection status columns is a boolean column named
"is_connected". The code in ovsdb-server blindly assigned a bool into
this column without checking that it actually had space allocated for one.
This was and is fine with the ovs-vswitchd schema, which always has exactly
one value in this column. However, if a database schema makes this column
optional, and there are actually no values in it, then this assignment
dereferences a null pointer.
This commit fixes the problem by allocating space for a bool if none has
yet been allocated.
Noticed while adding an extra test for the connection status feature.
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Ethan Jackson <ethan@nicira.com>
Ethan Jackson [Thu, 20 Sep 2012 02:21:06 +0000 (19:21 -0700)]
bridge: Omit alerts on the cfm_remote_opstate column.
This column should be write only, otherwise every call to update it
has to make a trip to the database. Since this column is updated
every time through the run loop as part of refresh_instant_stats(),
this patch fixes a significant performance degradation.
Ben Pfaff [Thu, 20 Sep 2012 15:40:29 +0000 (08:40 -0700)]
ovs-ofctl: Accept port keywords, OF1.1 port numbers, reject port number 0.
OpenFlow 1.0 has special reserved ports in the range 0xfff8 to 0xffff.
OpenFlow 1.1 and later has the same ports in the range 0xfffffff8 to
0xffffffff and allows the OF1.0 range to be used for ordinary ("physical")
switch ports. This means that, naively, the meaning of a port number in
the range 0xfff8 to 0xffff given on the ovs-ofctl command line depends on
the protocol in use. This commit implements something a little smarter:
- Accept keyword names (e.g. LOCAL) for special reserved ports
everywhere that such a port can plausibly be used (previously they
were only accepted in some places).
- Translate 0xfff8...0xffff to 0xfffffff8...0xffffffff for now, since
OF1.1+ isn't in widespread use and those particular ports aren't
likely to be in use in OF1.1+ anyway.
- Log warnings about those ports when they are specified by number, to
allow users to fix their invocations.
Also:
- Accept the OF1.1+ port numbers for these ports, without warning, for
compatibility with the upcoming OF1.1+ support.
- Stop accepting port number 0, which has never been a valid port
number in OpenFlow 1.0 and later. (This required fixing some tests
that inadvertently used this port number).
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Simon Horman <horms@verge.net.au>
Ben Pfaff [Fri, 14 Sep 2012 20:09:33 +0000 (13:09 -0700)]
jsonrpc: Fix Python implementation of inactivity logic.
When a JSON-RPC session receives bytes, or when it successfully sends
queued bytes, then it should count that as activity. However, the code
here was reversed, in that it used the wrong check in each place. That is,
when it tried to receive data, it would check whether data had just been
sent, and when it tried to send data, it would check whether data had just
been received. Neither one makes sense and doesn't work.
Bug #13214. Reported-by: Luca Giraudo <lgiraudo@nicira.com> CC: James Schmidt <jschmidt@nicira.com> Signed-off-by: Ben Pfaff <blp@nicira.com>
datapath: Add version info for out-of-tree modules.
The upstream version of the module always has the version of the running kernel
but for out-of-tree modules it can be difficult to tell the current version.
This adds the information to the module where it can be read using modinfo for
the on-disk version or from /sys/module/openvswitch/version for the currently
loaded module.
Ben Pfaff [Fri, 14 Sep 2012 20:04:15 +0000 (13:04 -0700)]
tests: Fix sensitivity to record ordering in test-netflow output.
The order of records in a NetFlow message is essentially random, but the
test case was picky about it. I started getting failures when I modified
apparently unrelated code, so here's a fix.
Simon Horman [Wed, 12 Sep 2012 04:47:27 +0000 (21:47 -0700)]
ofp-util: Allow decoding of Open Flow 1.1 & 1.2 Table Statistics Request Messages
Signed-off-by: Simon Horman <horms@verge.net.au>
[blp@nicira.com then made substantial changes that were then:] Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
In the above case, uuid in "row" aren't replaced by "named-uuid" because
the function doesn't look into elements of lists.
When list/tuple is found, look into elements recursively.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp> Signed-off-by: Ben Pfaff <blp@nicira.com>
Ben Pfaff [Wed, 5 Sep 2012 20:34:35 +0000 (13:34 -0700)]
jsonrpc: Treat receiving part of a message as activity.
Until now, the jsonrpc code has only counted receiving a full JSON-RPC
messages as activity. This could theoretically time out, then, while a
very long message is in transit or if a slow link is involved. This commit
changes this code to count receiving any part of a message as activity.
This isn't a problem for OpenFlow connections because OpenFlow messages are
at most 64 kB in size.
This problem hasn't actually been observed in practice.
Bug #12789. Signed-off-by: Ben Pfaff <blp@nicira.com>
Ben Pfaff [Fri, 7 Sep 2012 17:50:15 +0000 (10:50 -0700)]
jsonrpc: Treat draining data from send queue as activity.
Until now, the jsonrpc module has used messages received from the
remote peer as the sole means to determine that the JSON-RPC
connection is up. This could in theory interact badly with a
remote peer that stops reading and processing messages from the
receive queue when there is a backlog in the send queue for a
given connection (ovsdb-server is an example of a program that
behaves this way). This commit fixes the problem by expanding
the definition of "activity" to include successfully sending
JSON-RPC data that was previously queued.
The above change is exactly analogous to the similar change
made to the rconn library in commit 133f2dc95454 (rconn: Treat
draining a message from the send queue as activity.).
Bug #12789. Signed-off-by: Ben Pfaff <blp@nicira.com>
Ben Pfaff [Wed, 8 Aug 2012 20:32:57 +0000 (13:32 -0700)]
reconnect: Rename reconnect_received() to reconnect_activity().
Receiving data is not the only reasonable way to verify that a connection
is up. For example, on a TCP connection, receiving an acknowledgment that
the remote side has accepted data that we sent is also a reasonable means.
Therefore, this commit generalizes the naming.
Also, similarly for the Python implementation: Reconnect.received() becomes
Reconnect.activity().
Joe Stringer [Wed, 5 Sep 2012 03:25:58 +0000 (15:25 +1200)]
third-party: Fix tcpdump patch
Other parts of OVS have moved on since the tcpdump patch was created. This
commit brings the patch up to date and will compile cleanly against
tcpdump-4.3.0.
Signed-off-by: Joe Stringer <joe@wand.net.nz> Signed-off-by: Justin Pettit <jpettit@nicira.com>
Ben Pfaff [Wed, 5 Sep 2012 17:35:20 +0000 (10:35 -0700)]
ovsdb: Enforce immutability of immutable columns.
OVSDB has always had the ability to mark a column as "immutable", so that
its value cannot be changed in a given row after that row is initially
inserted. However, we discovered recently that ovsdb-server has never
enforced this constraint. This commit implements enforcement.
Reported-by: Paul Ingram <paul@nicira.com> Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Kyle Mestery <kmestery@cisco.com>
Simon Horman [Wed, 5 Sep 2012 02:50:38 +0000 (11:50 +0900)]
ofp-errors: Use OFPERR_OFPBRC_BAD_TABLE_ID
* In the case of OpenFlow 1.1+ OFPERR_OFPBRC_BAD_TABLE_ID is defined
in the specification and seems to be the most appropriate error
to use when an unknown table id is encountered.
* In the case of OpenFlow 1.0 no appropriate error message
seems to exist. Perhaps because an invalid port is not possible?
I'm unsure.
In any case, make use of a non-standard error code (1,512).
This was formerly known as OFPERR_NXBRC_BAD_TABLE_ID but
has been rolled into OFPERR_OFPBRC_BAD_TABLE_ID to allow the
latter to be used without concern for the prevailing Open Flow version.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Simon Horman [Wed, 5 Sep 2012 02:50:37 +0000 (11:50 +0900)]
ofp-errors: Use OFPERR_OFPBRC_BAD_PORT
* In the case of OpenFlow 1.2+ OFPERR_OFPBRC_BAD_PORT is defined
in the specification and seems to be the most appropriate error
to use when an invalid port is encountered in a Packet Out request.
* In the case of OpenFlow 1.0 and 1.1 no appropriate error message
seems to exist. Perhaps because an invalid port is not possible?
I'm unsure.
In any case, make use of a non-standard error code (1,514).
This was formerly known as OFPERR_NXBRC_BAD_IN_PORT but
has been rolled into OFPERR_NXBRC_BAD_IN_PORT to allow the
latter to be used without concern for the prevailing Open Flow version.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Ben Pfaff [Wed, 5 Sep 2012 17:18:56 +0000 (10:18 -0700)]
extract-ofp-errors: Check that error codes are in the expected ranges.
All real OpenFlow error codes are small numbers, and for Nicira extensions
we've intentionally chosen large numbers. This commit adds a check that
standard and extension codes are properly designated in the ofp-errors.h
header.
Simon Horman [Wed, 5 Sep 2012 02:50:36 +0000 (11:50 +0900)]
ofp-errors: Ignore text enclosed in square brackets
Enhance to extract-ofp-errors to omit text enclosed in square brackets from
error description. This allows some commentary other than
the error description to be supplied in ofp-errors.h
As suggested by Ben Pfaff <blp@nicira.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
[blp@nicira.com added the large comment on enum ofperr.] Signed-off-by: Ben Pfaff <blp@nicira.com>
python/ovs/poller: use select.select instead of select.poll.
eventlet/gevent doesn't work well with select.poll because select.poll blocks
python interpreter as a whole instead of switching from the current thread
which is about to block to other runnable thread.
So ovsdb python binding can't be used with eventlet/gevent.
Emulate select.poll with select.select because using python means that
performance isn't so important.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp> Signed-off-by: Ben Pfaff <blp@nicira.com>
OFPERR_NXFMFC_GROUPS_NOT_SUPPORTED is currently only used in paths which
are part of a non-NX extension portions of the Open Flow 1.1+
implementation.
After recent discussion it has been decided to attempt to only use
standardised, albeit lest-specify, errors unless errors arise from use of
an NX extension.
With the above in mind it seems appropriate to:
* Use OFPERR_OFPFMFC_UNKNOWN in place of OFPERR_NXFMFC_GROUPS_NOT_SUPPORTED.
* Remove OFPERR_NXFMFC_GROUPS_NOT_SUPPORTED as it is no longer used.
An unfortunate side-effect of this change is that the error for
the case in question is now less-specific.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Simon Horman [Tue, 4 Sep 2012 06:51:59 +0000 (15:51 +0900)]
ofp-errors: Remove OFPERR_NXBIC_DUP_TYPE
OFPERR_NXBIC_DUP_TYPE is currently only used in
decode_openflow11_instructions() which is part of a non-NX extension
portion of the Open Flow 1.1+ implementation.
After recent discussion it has been decided to attempt to only use
standardised, albeit less-specific, errors unless errors arise from use of
an NX extension.
With the above in mind it seems appropriate to:
* Use OFPERR_OFPIT_BAD_INSTRUCTION in place of OFPERR_NXBIC_DUP_TYPE.
* Remove OFPERR_NXBIC_DUP_TYPE as it is no longer used.
An unfortunate side-effect of this change is that the error for
the case in question is now less-specific.
Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Ben Pfaff <blp@nicira.com>
Ben Pfaff [Tue, 4 Sep 2012 19:43:53 +0000 (12:43 -0700)]
Introduce sparse flows and masks, to reduce memory usage and improve speed.
A cls_rule is 324 bytes on i386 now. The cost of a flow table lookup is
currently proportional to this size, which is going to continue to grow.
However, the required cost of a flow table lookup, with the classifier that
we currently use, is only proportional to the number of bits that a rule
actually matches. This commit implements that optimization by replacing
the match inside "struct cls_rule" by a sparse representation.
This reduces struct cls_rule to 100 bytes on i386.
There is still some headroom for further optimization following this
commit:
- I suspect that adding an 'n' member to struct miniflow would make
miniflow operations faster, since popcount() has some cost.
- It's probably possible to replace the "struct minimatch" in cls_rule
by just a "struct miniflow", since the cls_rule's cls_table has a
copy of the minimask.
- Some of the miniflow operations aren't well-optimized.
Ben Pfaff [Tue, 21 Aug 2012 21:26:23 +0000 (14:26 -0700)]
hash: Introduce an implementation of murmurhash.
Murmurhash is generally superior to the Jenkins lookup3 hash according to
the available figures. Perhaps we should generally replace our current
hashes by murmurhash.
For now, I'm introducing a parallel implementation to allow it to be used
in cases where an incremental one-word-at-a-time hash is desirable. The
first user will be added in an upcoming commit.
Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Ethan Jackson <ethan@nicira.com>
projects / mirror_ovs.git / log