* memory is now allocated as early as possible, without holding locks.
* sysctl users are now guaranteed to get a response (M_WAITOK buffer prealloc).
* socket users are more likely to use on-stack buffer for replies.
* standard kernel malloc/free functions are now used instead of radix wrappers.
rt_msg2() has been renamed to rtsock_msg_buffer().
MFC after: 1 month
RTM_CHANGE is now handled inside route.c:rtrequest1_fib() as it should be.
Note change change handler is a separate function rtrequest1_fib_change().
MFC after: 1 month
These two bugs are closely related. The root cause is that ifa_ifwithnet
does not consider FIBs when searching for an interface address.
sys/net/if_var.h
sys/net/if.c
Add a fib argument to ifa_ifwithnet and ifa_ifwithdstadddr. Those
functions will only return an address whose interface fib equals the
argument.
sys/net/route.c
Update calls to ifa_ifwithnet and ifa_ifwithdstaddr with fib
arguments.
sys/netinet/in.c
Update in_addprefix to consider the interface fib when adding
prefixes. This will prevent it from not adding a subnet route when
one already exists on a different fib.
sys/net/rtsock.c
sys/netinet/in_pcb.c
sys/netinet/ip_output.c
sys/netinet/ip_options.c
sys/netinet6/nd6.c
Add RT_DEFAULT_FIB arguments to ifa_ifwithdstaddr and ifa_ifwithnet.
In some cases it there wasn't a clear specific fib number to use.
In others, I was unable to test those functions so I chose
RT_DEFAULT_FIB to minimize divergence from current behavior. I will
fix some of the latter changes along with PR kern/187553.
tests/sys/netinet/fibs_test.sh
tests/sys/netinet/udp_dontroute.c
tests/sys/netinet/Makefile
Revert r263738. The udp_dontroute test was right all along.
However, bugs kern/187550 and kern/187553 cancelled each other out
when it came to this test. Because of kern/187553, ifa_ifwithnet
searched the default fib instead of the requested one, but because
of kern/187550, there was an applicable subnet route on the default
fib. The new test added in r263738 doesn't work right, however. I
can verify with dtrace that ifa_ifwithnet returned the wrong address
before I applied this commit, but route(8) miraculously found the
correct interface to use anyway. I don't know how.
Clear expected failure messages for kern/187550 and kern/187552.
PR: kern/187550
PR: kern/187552
Reviewed by: melifaro
MFC after: 3 weeks
Sponsored by: Spectra Logic
sys/net/route.c
In rtinit1, use the interface fib instead of the process fib. The
latter wasn't very useful because ifconfig(8) is usually invoked
with the default process fib. Changing ifconfig(8) to use setfib(2)
would be redundant, because it already sets the interface fib.
tests/sys/netinet/fibs_test.sh
Clear the expected ATF failure
sys/net/if.c
Pass the interface fib in calls to rtrequest1_fib and rtalloc1_fib
sys/netinet/in.c
sys/net/if_var.h
Add a fibnum argument to ifa_switch_loopback_route, a subroutine of
in_scrubprefix. Pass it the interface fib.
PR: kern/187549
Reviewed by: melifaro
MFC after: 3 weeks
Sponsored by: Spectra Logic Corporation
De-virtualize UMA zone pf_mtag_z and move to global initialization part.
The m_tag struct does not know about vnet context and the pf_mtag_free()
callback is called unaware of current vnet. This causes a panic.
MFC after: 1 week
network interfaces limited to 32 transmit segments, there
are two known issues.
The more serious one is that for an I/O of slightly less than 64K,
the net device driver prepends an ethernet header, resulting in a
TSO segment slightly larger than 64K. Since m_defrag() copies this
into 33 mbuf clusters, the transmit fails with EFBIG.
A tester indicated observing a similar failure using iSCSI.
The second less critical problem is that the network
device driver must copy the mbuf chain via m_defrag()
(m_collapse() is not sufficient), resulting in measurable overhead.
This patch reduces the default size of if_hw_tsomax
slightly, so that the first issue is avoided.
Fixing the second issue will require a way for the
network device driver to inform tcp_output() that it
is limited to 32 transmit segments.
Reported and tested by: csforgeron@gmail.com, markus.gebert@hostpoint.ch
MFC after: 2 weeks
was stacked on top of a network interface that set if_hw_tsomax,
tcp_output() would see the default value instead of the value
set by the network interface. This patch modifies vlan so that
it sets if_hw_tsomax to the value of the parent interface.
Reviewed by: glebius
MFC after: 2 weeks
was stacked on top of network interfaces that set if_hw_tsomax,
tcp_output() would see the default value instead of the value
set by the network interface(s). This patch modifies lagg so that
it sets if_hw_tsomax to the minimum of the value(s) for the
underlying network interfaces.
Reviewed by: glebius
MFC after: 2 weeks
This KASSERT() existed as a sanity check that upper layers in the network
stack (e.g. inet, inet6) had released their reference to the underlying
driver's multicast memberships (ifmultiaddr{}). However it assumes the
lifecycle of the driver membership corresponds to the lifecycle of the
network layer membership.
In the submitter's case, ieee80211_ioctl_updatemulti() attempts to
reprogram the (parent, physical) ifnet{} memberships in response
to a change in membership on the (child, virtual) VAP ifnet, using
a batched update mechanism. These updates happen independently from
the network layer, causing a "false negative" assertion failure.
There are possibly other use cases where this KASSERT() may be triggered
by other networking stack activity (e.g. where a nesting relationship
exists between multiple ifnet{} instances). This suggests that further
review of FreeBSD's approach to nested ifnet relationships is needed.
MFC after: 6 weeks
Submitted by: adrian@
The m_tag struct does not know about vnet context and the pf_mtag_free()
callback is called unaware of current vnet. This causes a panic.
Reviewed by: Nikos Vassiliadis, trociny@
ifi_oqdrops. This is a temporary workaround until ifqueue(9) vanishes.
While here, remove the pointless ifi_vhid assignment. It has
sense only when we are exporting ifaddrs, not ifnets.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
linking NIC Receive Side Scaling (RSS) to the network stack's
connection-group implementation. This prototype (and derived patches)
are in use at Juniper and several other FreeBSD-using companies, so
despite some reservations about its maturity, merge the patch to the
base tree so that it can be iteratively refined in collaboration rather
than maintained as a set of gradually diverging patch sets.
(1) Merge a software implementation of the Toeplitz hash specified in
RSS implemented by David Malone. This is used to allow suitable
pcbgroup placement of connections before the first packet is
received from the NIC. Software hashing is generally avoided,
however, due to high cost of the hash on general-purpose CPUs.
(2) In in_rss.c, maintain authoritative versions of RSS state intended
to be pushed to each NIC, including keying material, hash
algorithm/ configuration, and buckets. Provide software-facing
interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both
the RSS standardised Toeplitz and a 'naive' variation with a hash
efficient in software but with poor distribution properties.
Implement rss_m2cpuid()to be used by netisr and other load
balancing code to look up the CPU on which an mbuf should be
processed.
(3) In the Ethernet link layer, allow netisr distribution using RSS as
a source of policy as an alternative to source ordering; continue
to default to direct dispatch (i.e., don't try and requeue packets
for processing on the 'right' CPU if they arrive in a directly
dispatchable context).
(4) Allow RSS to control tuning of connection groups in order to align
groups with RSS buckets. If a packet arrives on a protocol using
connection groups, and contains a suitable hardware-generated
hash, use that hash value to select the connection group for pcb
lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz
hash is available, we fall back on regular PCB lookup risking
contention rather than pay the cost of Toeplitz in software --
this is a less scalable but, at my last measurement, faster
approach. As core counts go up, we may want to revise this
strategy despite CPU overhead.
Where device drivers suitably configure NICs, and connection groups /
RSS are enabled, this should avoid both lock and line contention during
connection lookup for TCP. This commit does not modify any device
drivers to tune device RSS configuration to the global RSS
configuration; patches are in circulation to do this for at least
Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device
drivers is not particularly robust, nor aware of more advanced features
such as runtime reconfiguration/rebalancing. This will hopefully prove
a useful starting point for refinement.
No MFC is scheduled as we will first want to nail down a more mature
and maintainable KPI/KBI for device drivers.
Sponsored by: Juniper Networks (original work)
Sponsored by: EMC/Isilon (patch update and merge)
AppleTalk was a network transport protocol for Apple Macintosh devices
in 80s and then 90s. Starting with Mac OS X in 2000 the AppleTalk was
a legacy protocol and primary networking protocol is TCP/IP. The last
Mac OS X release to support AppleTalk happened in 2009. The same year
routing equipment vendors (namely Cisco) end their support.
Thus, AppleTalk won't be supported in FreeBSD 11.0-RELEASE.
IPX was a network transport protocol in Novell's NetWare network operating
system from late 80s and then 90s. The NetWare itself switched to TCP/IP
as default transport in 1998. Later, in this century the Novell Open
Enterprise Server became successor of Novell NetWare. The last release
that claimed to still support IPX was OES 2 in 2007. Routing equipment
vendors (e.g. Cisco) discontinued support for IPX in 2011.
Thus, IPX won't be supported in FreeBSD 11.0-RELEASE.
interface, in the r241616 a crutch was provided. It didn't work well, and
finally we decided that it is time to break ABI and simply make if_baudrate
a 64-bit value. Meanwhile, the entire struct if_data was reviewed.
o Remove the if_baudrate_pf crutch.
o Make all fields of struct if_data fixed machine independent size. The
notion of data (packet counters, etc) are by no means MD. And it is a
bug that on amd64 we've got a 64-bit counters, while on i386 32-bit,
which at modern speeds overflow within a second.
This also removes quite a lot of COMPAT_FREEBSD32 code.
o Give 16 bit for the ifi_datalen field. This field was provided to
make future changes to if_data less ABI breaking. Unfortunately the
8 bit size of it had effectively limited sizeof if_data to 256 bytes.
o Give 32 bits to ifi_mtu and ifi_metric.
o Give 64 bits to the rest of fields, since they are counters.
__FreeBSD_version bumped.
Discussed with: emax
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
after r262763 it started to return locked mutexes to UMA. To fix that,
conditionally unlock the mutex in the destructor.
Tested by: "Sergey V. Dyatko" <sergey.dyatko@gmail.com>
- Use counter(9) for rt_pksent (former rt_rmx.rmx_pksent). This
removes another cache trashing ++ from packet forwarding path.
- Create zini/fini methods for the rtentry UMA zone. Via initialize
mutex and counter in them.
- Fix reporting of rmx_pksent to routing socket.
- Fix netstat(1) to report "Use" both in kvm(3) and sysctl(3) mode.
The change is mostly targeted for stable/10 merge. For head,
rt_pksent is expected to just disappear.
Discussed with: melifaro
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
RADIX_NODE_HEAD_LOCK as grabbing the write lock,
but RADIX_NODE_HEAD_LOCK_ASSERT as checking the read lock.
Submitted by: Vijay Singh <vijju.singh at gmail.com>
MFC after: 1 month
were primarily used to size the sysctl name list macros that were removed
in r254295. A few other constants either did not have an associated
sysctl node, or the associated node used OID_AUTO instead.
PR: ports/184525 (exp-run)
tested and is unfinished. However, I've tested my version,
it works okay. As before it is unfinished: timeout aren't
driven by TCP session state. To enable the HASH_ALL mode,
one needs in kernel config:
options FLOWTABLE_HASH_ALL
o Reduce the alignment on flentry to 64 bytes. Without
the FLOWTABLE_HASH_ALL option, twice less memory would
be consumed by flows.
o API to ip_output()/ip6_output() got even more thin: 1 liner.
o Remove unused unions. Simply use fle->f_key[].
o Merge all IPv4 code into flowtable_lookup_ipv4(), and do same
flowtable_lookup_ipv6(). Stop copying data to on stack
sockaddr structures, simply use key[] on stack.
o Move code from flowtable_lookup_common() that actually works
on insertion into flowtable_insert().
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
- netmap pipes, providing bidirectional blocking I/O while moving
100+ Mpps between processes using shared memory channels
(no mistake: over one hundred million. But mind you, i said
*moving* not *processing*);
- kqueue support (BHyVe needs it);
- improved user library. Just the interface name lets you select a NIC,
host port, VALE switch port, netmap pipe, and individual queues.
The upcoming netmap-enabled libpcap will use this feature.
- optional extra buffers associated to netmap ports, for applications
that need to buffer data yet don't want to make copies.
- segmentation offloading for the VALE switch, useful between VMs.
and a number of bug fixes and performance improvements.
My colleagues Giuseppe Lettieri and Vincenzo Maffione did a substantial
amount of work on these features so we owe them a big thanks.
There are some external repositories that can be of interest:
https://code.google.com/p/netmap
our public repository for netmap/VALE code, including
linux versions and other stuff that does not belong here,
such as python bindings.
https://code.google.com/p/netmap-libpcap
a clone of the libpcap repository with netmap support.
With this any libpcap client has access to most netmap
feature with no recompilation. E.g. tcpdump can filter
packets at 10-15 Mpps.
https://code.google.com/p/netmap-ipfw
a userspace version of ipfw+dummynet which uses netmap
to send/receive packets. Speed is up in the 7-10 Mpps
range per core for simple rulesets.
Both netmap-libpcap and netmap-ipfw will be merged upstream at some
point, but while this happens it is useful to have access to them.
And yes, this code will be merged soon. It is infinitely better
than the version currently in 10 and 9.
MFC after: 3 days
insert flow entry. During the route lookup the critical section is
exited. It may happen, that after route lookup we will be executed
on an other CPU that already has such flowentry. Before this change
we simply freed the flowentry and returned to ip_output() with
failure.
Actually there is nothing wrong with using previously allocated
flow entry, updating it properly. Thus, make flowentry_insert()
return the new either old fle, and make use of it.
Count reuses as "collisions" and real inserts as "inserts".
Reviewed by: adrian
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
race prone. Some just gather statistics, but some are later used in
different calculations.
A real problem was the race provoked underflow of the states_cur counter
on a rule. Once it goes below zero, it wraps to UINT32_MAX. Later this
value is used in pf_state_expires() and any state created by this rule
is immediately expired.
Thus, make fields states_cur, states_tot and src_nodes of struct
pf_rule be counter(9)s.
Thanks to Dennis for providing me shell access to problematic box and
his help with reproducing, debugging and investigating the problem.
Thanks to: Dennis Yusupoff <dyr smartspb.net>
Also reported by: dumbbell, pgj, Rambler
Sponsored by: Nginx, Inc.
Some of the collisions that are occuring are due to flowtable lookups
that succeed but have an invalid lle - typically because the L2 adjacency
lookup hasn't completed. This would lead to a follow-up insert which
would then fail (ie, collision) and the code would fall through to doing
a slow-path L2/L3 lookup in the netinet/netinet6 code.
This patch simply aborts storing a new flowtable entry if the lle isn't
yet valid.
Whilst I'm here, add a new pcpu counter for the item so the number of
failures can be tracked separately from generic "collisions."
Reviewed by: glebius
MFC after: 10 days
Sponsored by: Netflix, Inc.
