This disentangles "struct nl_dump" from "struct nl_sock", clearing the way
to make the use of either one thread-safe in an obviously correct manner.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Commit 7d7447 (netlink: Postpone choosing sequence numbers until send
time.) broke ovs-brcompatd because it prevented userspace replies to
kernel requests from using the correct sequence numbers. This commit fixes
it.
Atzm Watanabe found the root cause and provided an alternative patch to
avoid the problem.
Reported-by: André Ruß <andre.russ@hybris.com>
Reported-by: Atzm Watanabe <atzm@stratosphere.co.jp>
Tested-by: Atzm Watanabe <atzm@stratosphere.co.jp>
Signed-off-by: Ben Pfaff <blp@nicira.com>
Replaced all instances of Nicira Networks(, Inc) to Nicira, Inc.
Feature #10593
Signed-off-by: Raju Subramanian <rsubramanian@nicira.com>
Signed-off-by: Ben Pfaff <blp@nicira.com>
Typically an nl_sock client can stack-allocate the buffer for receiving
a Netlink message, which provides a performance boost.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Using poll() internally in dpif_linux_recv(), instead of relying
on the results of the main loop poll() call, brings netperf CRR
performance back within 1% of par versus the code base before the
poll_fd_woke() optimizations were introduced. It also increases
the ovs-benchmark results by about 5% versus that baseline, too.
My theory is that this is because the main loop takes long enough
that a significant number of packets can arrive during the main
loop itself, so this reduces the time before OVS gets to those
packets.
This reverts commit 1e276d1a10539a8cd97d2ad63c073a9a43f0f1ef.
The poll_fd_woke() and nl_sock_woke() function added in that commit are
no longer used, so there is no reason to keep them in the tree.
Each time we run through the poll loop, we check all file descriptors
that we were waiting on to see if there is data available. However,
this requires a system call and poll already provides information on
which FDs caused the wakeup so it is inefficient as the number of
active FDs grows. This provides a way to check whether a given FD
has data.
In the future, the kernel will use unicast messages instead of
multicast to send upcalls. As a result, we need to be able to
tell it where to direct the traffic. This adds a function to expose
the Netlink pid of a socket so it can be included in messages to the
kernel.
The nl_lookup_genl_mcgroup() function can fail on older kernels
which do not support the required netlink interface. Before this
patch, dpif-linux would refuse to create a datapath when this
happened. With this patch, it attempts to use a workaround. If
the workaround fails it simply disables the affected features
without completely disabling the dpif.
It's not safe to use a single Netlink fd to do multiple operations in an
synchronous way. Some of the limitations are fundamental; for example, the
kernel only supports a single "dump" operation at a time. Others are
limitations imposed by the OVS coding style; for example, our Netlink
library is not callback based, so nothing can be done about incoming
messages that can't be handled immediately. Regardless, in OVS multicast
groups, transactions, and dumps cannot coexist on a single nl_sock.
This is only mildly irritating at the moment, but it will become much worse
later on, when dpif-linux shifts to using Netlink dumps for listing various
kinds of datapath entities. When that happens, a dump will be in progress
in situations where the dpif-linux client might want to do other
operations. For example, it is reasonable for the client to list flows
and, in the middle, look up information on vports mentioned in those flows.
It might be possible to simply ban and avoid such nested operations--I have
not even audited the source tree to find out whether we do anything like
that already--but that seems like an unnecessary cramp on our coding style.
Furthermore, it's difficult to explain and justify without understanding
the implementation.
This patch takes another approach, by improving the Netlink socket library
to avoid artificial constraints. When an operation, or a dump, or joining
a multicast group would cause a problem, this patch makes the library
transparently create a separate Netlink socket. This solves the problem
without putting any onerous restrictions on use.
This commit also slightly simplifies netdev_vport_reset_names(). It had
been written to destroy the dump object before the Netlink socket that it
used, but this is no longer necessary and doing it in the opposite order
saved a few lines of code.
Reviewed by Ethan Jackson <ethan@nicira.com>.
When this library was originally implemented, support for Linux 2.4 was
important. The Netlink implementation in Linux only added support for
joining and leaving multicast groups after a socket is bound as of Linux
2.6.14, so the library did not support it either. But the current version
of Open vSwitch targets Linux 2.6.18 and over, so it's fine to add this
support now, and this commit does so.
This will be used more extensively in upcoming commits.
Reviewed by Justin Pettit.
The parts of the netlink module that are related to sockets are
Linux-specific, since only Linux has AF_NETLINK sockets. The rest can be
built anywhere. This commit breaks them into two modules, and builds the
generic one on all platforms.
Acked-by: Jesse Gross <jesse@nicira.com>
