Mostly bugfixes or features developed in the past 6 months,
so this is a 10.1 candidate.
Basically no user API changes (some bugfixes in sys/net/netmap_user.h).
In detail:
1. netmap support for virtio-net, including in netmap mode.
Under bhyve and with a netmap backend [2] we reach over 1Mpps
with standard APIs (e.g. libpcap), and 5-8 Mpps in netmap mode.
2. (kernel) add support for multiple memory allocators, so we can
better partition physical and virtual interfaces giving access
to separate users. The most visible effect is one additional
argument to the various kernel functions to compute buffer
addresses. All netmap-supported drivers are affected, but changes
are mechanical and trivial
3. (kernel) simplify the prototype for *txsync() and *rxsync()
driver methods. All netmap drivers affected, changes mostly mechanical.
4. add support for netmap-monitor ports. Think of it as a mirroring
port on a physical switch: a netmap monitor port replicates traffic
present on the main port. Restrictions apply. Drive carefully.
5. if_lem.c: support for various paravirtualization features,
experimental and disabled by default.
Most of these are described in our ANCS'13 paper [1].
Paravirtualized support in netmap mode is new, and beats the
numbers in the paper by a large factor (under qemu-kvm,
we measured gues-host throughput up to 10-12 Mpps).
A lot of refactoring and additional documentation in the files
in sys/dev/netmap, but apart from #2 and #3 above, almost nothing
of this stuff is visible to other kernel parts.
Example programs in tools/tools/netmap have been updated with bugfixes
and to support more of the existing features.
This is meant to go into 10.1 so we plan an MFC before the Aug.22 deadline.
A lot of this code has been contributed by my colleagues at UNIPI,
including Giuseppe Lettieri, Vincenzo Maffione, Stefano Garzarella.
MFC after: 3 days.
* The way rings are updated changed with the last API bump.
Also sync ->head when moving slots in netmap_sw_to_nic().
* Remove a crashing selrecord() call.
* Unclog the logic surrounding netmap_rxsync_from_host().
* Add timestamping to RX host ring.
* Remove a couple of obsolete comments.
Submitted by: Franco Fichtner
MFC after: 3 days
Sponsored by: Packetwerk
interface allows the ifnet structure to be defined as an opaque
type in NIC drivers. This then allows the ifnet structure to be
changed without a need to change or recompile NIC drivers.
Put differently, NIC drivers can be written and compiled once and
be used with different network stack implementations, provided of
course that those network stack implementations have an API and
ABI compatible interface.
This commit introduces the 'if_t' type to replace 'struct ifnet *'
as the type of a network interface. The 'if_t' type is defined as
'void *' to enable the compiler to perform type conversion to
'struct ifnet *' and vice versa where needed and without warnings.
The functions that implement the API are the only functions that
need to have an explicit cast.
The MII code has been converted to use the driver API to avoid
unnecessary code churn. Code churn comes from having to work with
both converted and unconverted drivers in correlation with having
callback functions that take an interface. By converting the MII
code first, the callback functions can be defined so that the
compiler will perform the typecasts automatically.
As soon as all drivers have been converted, the if_t type can be
redefined as needed and the API functions can be fix to not need
an explicit cast.
The immediate benefactors of this change are:
1. Juniper Networks - The network stack implementation in Junos
is entirely different from FreeBSD's one and this change
allows Juniper to build "stock" NIC drivers that can be used
in combination with both the FreeBSD and Junos stacks.
2. FreeBSD - This change opens the door towards changing ifnet
and implementing new features and optimizations in the network
stack without it requiring a change in the many NIC drivers
FreeBSD has.
Submitted by: Anuranjan Shukla <anshukla@juniper.net>
Reviewed by: glebius@
Obtained from: Juniper Networks, Inc.
- intercept FIONBIO and FIOASYNC ioctls on netmap file descriptors.
libpcap calls them to set non blocking I/O on the file descriptor,
for netmap this is a no-op because there is no read/write,
but not intercepting would cause fcntl() to return -1
- rate limit and put under netmap.verbose some messages that occur
when threads use concurrently the same file descriptor.
- 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
add separate rx/tx ring indexes
add ring specifier in nm_open device name
netmap.c, netmap_vale.c
more consistent errno numbers
netmap_generic.c
correctly handle failure in registering interfaces.
tools/tools/netmap/
massive cleanup of the example programs
(a lot of common code is now in netmap_user.h.)
nm_util.[ch] are going away soon.
pcap.c will also go when i commit the native netmap support for libpcap.
Most relevant features:
- netmap emulation on any NIC, even those without native netmap support.
On the ixgbe we have measured about 4Mpps/core/queue in this mode,
which is still a lot more than with sockets/bpf.
- seamless interconnection of VALE switch, NICs and host stack.
If you disable accelerations on your NIC (say em0)
ifconfig em0 -txcsum -txcsum
you can use the VALE switch to connect the NIC and the host stack:
vale-ctl -h valeXX:em0
allowing sharing the NIC with other netmap clients.
- THE USER API HAS SLIGHTLY CHANGED (head/cur/tail pointers
instead of pointers/count as before). This was unavoidable to support,
in the future, multiple threads operating on the same rings.
Netmap clients require very small source code changes to compile again.
On the plus side, the new API should be easier to understand
and the internals are a lot simpler.
The manual page has been updated extensively to reflect the current
features and give some examples.
This is the result of work of several people including Giuseppe Lettieri,
Vincenzo Maffione, Michio Honda and myself, and has been financially
supported by EU projects CHANGE and OPENLAB, from NetApp University
Research Fund, NEC, and of course the Universita` di Pisa.
This includes the following:
- use separate memory regions for VALE ports
- locking fixes
- some simplifications in the NIC-specific routines
- performance improvements for the VALE switch
- some new features in the pkt-gen test program
- documentation updates
There are small API changes that require programs to be recompiled
(NETMAP_API has been bumped so you will detect old binaries at runtime).
In particular:
- struct netmap_slot now is 16 bytes to support an extra pointer,
which may save one data copy when using VALE ports or VMs;
- the struct netmap_if has two extra fields;
MFC after: 3 days
to this event, adding if_var.h to files that do need it. Also, include
all includes that now are included due to implicit pollution via if_var.h
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
- This version has support for the new Intel Avoton systems,
including 2.5Gb support, further it now has IPv6/TSO6 support as
well. Shared code has been updated where necessary as well. Thanks
to my new assistant Eric Joyner for doing the transmit path changes
to bring in the IPv6/TSO6 support. Thanks to Gleb for catching the
one bug and change needed in NETMAP.
Approved by: re
from each batch flowing on the VALE switch
- feature: add glue for 'indirect' buffers on the sender side:
if a slot has NS_INDIRECT set, the netmap buffer contains pointer(s)
to the actual userspace buffers, which are accessed with copyin().
The feature is not finalised yet, as it will likely need to deal
with some iovec variant for proper scatter/gather support.
This will save one copy for clients (e.g. qemu) that cannot
use the netmap buffer directly.
A curiosity: on amd64 copyin() appears to be 10-15% faster than pkt_copy()
or bcopy() at least for sizes of 256 and greater.
- the VALE switch now support up to 254 destinations per switch,
unicast or broadcast (multicast goes to all ports).
- we can attach hw interfaces and the host stack to a VALE switch,
which means we will be able to use it more or less as a native bridge
(minor tweaks still necessary).
A 'vale-ctl' program is supplied in tools/tools/netmap
to attach/detach ports the switch, and list current configuration.
- the lookup function in the VALE switch can be reassigned to
something else, similar to the pf hooks. This will enable
attaching the firewall, or other processing functions (e.g. in-kernel
openvswitch) directly on the netmap port.
The internal API used by device drivers does not change.
Userspace applications should be recompiled because we
bump NETMAP_API as we now use some fields in the struct nmreq
that were previously ignored -- otherwise, data structures
are the same.
Manpages will be committed separately.
- netmap_rx_irq()/netmap_tx_irq() can now be called by FreeBSD drivers
hiding the logic for handling NIC interrupts in netmap mode.
This also simplifies the case of NICs attached to VALE switches.
Individual drivers will be updated with separate commits.
- use the same refcount() API for FreeBSD and linux
- plus some comments, typos and formatting fixes
Portions contributed by Michio Honda
future further optimizations where the vm_object lock will be held
in read mode most of the time the page cache resident pool of pages
are accessed for reading purposes.
The change is mostly mechanical but few notes are reported:
* The KPI changes as follow:
- VM_OBJECT_LOCK() -> VM_OBJECT_WLOCK()
- VM_OBJECT_TRYLOCK() -> VM_OBJECT_TRYWLOCK()
- VM_OBJECT_UNLOCK() -> VM_OBJECT_WUNLOCK()
- VM_OBJECT_LOCK_ASSERT(MA_OWNED) -> VM_OBJECT_ASSERT_WLOCKED()
(in order to avoid visibility of implementation details)
- The read-mode operations are added:
VM_OBJECT_RLOCK(), VM_OBJECT_TRYRLOCK(), VM_OBJECT_RUNLOCK(),
VM_OBJECT_ASSERT_RLOCKED(), VM_OBJECT_ASSERT_LOCKED()
* The vm/vm_pager.h namespace pollution avoidance (forcing requiring
sys/mutex.h in consumers directly to cater its inlining functions
using VM_OBJECT_LOCK()) imposes that all the vm/vm_pager.h
consumers now must include also sys/rwlock.h.
* zfs requires a quite convoluted fix to include FreeBSD rwlocks into
the compat layer because the name clash between FreeBSD and solaris
versions must be avoided.
At this purpose zfs redefines the vm_object locking functions
directly, isolating the FreeBSD components in specific compat stubs.
The KPI results heavilly broken by this commit. Thirdy part ports must
be updated accordingly (I can think off-hand of VirtualBox, for example).
Sponsored by: EMC / Isilon storage division
Reviewed by: jeff
Reviewed by: pjd (ZFS specific review)
Discussed with: alc
Tested by: pho
By setting dev.netmap.fwd=1 (or enabling the feature with a per-ring flag),
packets are forwarded between the NIC and the host stack unless the
netmap client clears the NS_FORWARD flag on the individual descriptors.
This feature greatly simplifies applications where some traffic
(think of ARP, control traffic, ssh sessions...) must be processed
by the host stack, whereas the bulk is handled by the netmap process
which simply (un)marks packets that should not be forwarded.
The default is chosen so that now a netmap receiver operates
in a mode very similar to bpf.
Of course there is no free lunch: traffic to/from the host stack
still operates at OS speed (or less, as there is one extra copy in
one direction).
HOWEVER, since traffic goes to the user process before being
reinjected, and reinjection occurs in a user context, you get some
form of livelock protection for free.
two upcoming features:
semi-transparent mode:
when a device is opened in this mode, the
user program will be able to mark slots that must be forwarded
to the "other" side (i.e. from NIC to host stack, or viceversa),
and the forwarding will occur automatically at the next netmap syscall.
This saves the need to open another file descriptor and do
the forwarding manually.
direct-forwarding mode:
when operating with a VALE port, the user can specify in the slot
the actual destination port, overriding the forwarding decision
made by a lookup of the destination MAC. This can be useful to
implement packet dispatchers.
No API changes will be introduced.
No new functionality in this patch yet.
previous names, 'ptag' and 'pmap' -- p stands for packet.
This change reduces the difference between the code in stable/9
and head, and also helps using the same ixgbe_netmap.h on both branches.
Approved by: Jack Vogel
that revises the netmap memory allocator so that the
various parameters (number and size of buffers, rings, descriptors)
can be modified at runtime through sysctl variables.
The changes become effective when no netmap clients are active.
The API is mostly unchanged, although the NIOCUNREGIF ioctl now
does not bring the interface back to normal mode: and you
need to close the file descriptor for that.
This change was necessary to track who is using the mapped region,
and since it is a simplification of the API there was no
incentive in trying to preserve NIOCUNREGIF.
We will remove the ioctl from the kernel next time we need
a real API change (and version bump).
Among other things, buffer allocation when opening devices is
now much faster: it used to take O(N^2) time, now it is linear.
Submitted by: Giuseppe Lettieri
- Move destruction of per-ring locks to netmap_dtor_locked to mirror the
initialization that happens in NIOCREGIF. Otherwise unloading a netmap-
capable interface that was never put into netmap mode would try to
mtx_destroy an uninitialized mutex, and panic.
- Destroy core_lock in netmap_detach, mirroring init in netmap_attach.
- Also comment out the knlist_destroy for now as there is currently no
knlist_init.
Sponsored by: ADARA Networks
Reviewed by: luigi@
http://info.iet.unipi.it/~luigi/vale/
VALE lets you dynamically instantiate multiple software bridges
that talk the netmap API (and are *extremely* fast), so you can test
netmap applications without the need for high end hardware.
This is particularly useful as I am completing a netmap-aware
version of ipfw, and VALE provides an excellent testing platform.
Also, I also have netmap backends for qemu mostly ready for commit
to the port, and this too will let you interconnect virtual machines
at high speed without fiddling with bridges, tap or other slow solutions.
The API for applications is unchanged, so you can use the code
in tools/tools/netmap (which i will update soon) on the VALE ports.
This commit also syncs the code with the one in my internal repository,
so you will see some conditional code for other platforms.
The code should run mostly unmodified on stable/9 so people interested
in trying it can just copy sys/dev/netmap/ and sys/net/netmap*.h
from HEAD
VALE is joint work with my colleague Giuseppe Lettieri, and
is partly supported by the EU Projects CHANGE and OPENLAB
Contrarily to what i wrote in my previous commit, the 82599
does include the CRC in the length. The operating mode is
reset in ixgbe_init_locked() and so we need to hook into
the places where the two registers (HLREG0 and RDRXCTL) are
modified.
does not include the CRC irrespective of the setting
of CRCSTRIP. The 82599 data sheets (sec. 7.1.6) say differently.
Very strange. Need to check what happens on legacy descriptors,
but for the time being this restores functionality.
and make it easier to replace it with a different implementation.
On passing, also fix indentation.
NOTE: I know that #include "foo.c" is ugly, but the alternative
(add another entry to sys/conf/files, add a separate header with
structs and prototypes, and expose functions that are meant to
be private) looks even worse to me.
We need a more modular way to specify dependencies and build options.
- add a sysctl, dev.netmap.ix_crcstrip, to control whether ixgbe should
strip the CRC on received frames. Defaults to 0, which keeps the CRC.
and improves performance when receiving min-sized (64-byte) frames.
This matters because min-sized frames is one of the standard
benchmarks for switches and routers, some chipsets seem to issue
read-modify-write cycles for PCIe transactions that are not a
full cache line, and a min-sized frame triggers the bug, resulting
in reduced throughput -- 9.7 instead of 14.88 Mpps -- and heavy
bus load.
- for the time being, always look for incoming packets on a select/poll
even if there has not been an interrupt in the meantime. This is
only a temporary workaround for a probable race condition in keeping
track of rx interrupts.
Add a couple of diagnostic vars to help studying the problem.
USERSPACE:
1. add support for devices with different number of rx and tx queues;
2. add better support for zero-copy operation, adding an extra field
to the netmap ring to indicate how many buffers we have already processed
but not yet released (with help from Eddie Kohler);
3. The two changes above unfortunately require an API change, so while
at it add a version field and some spares to the ioctl() argument
to help detect mismatches.
4. update the manual page for the two changes above;
5. update sample applications in tools/tools/netmap
KERNEL:
1. simplify the internal structures moving the global wait queues
to the 'struct netmap_adapter';
2. simplify the functions that map kring<->nic ring indexes
3. normalize device-specific code, helps mainteinance;
4. start exploring the impact of micro-optimizations (prefetch etc.)
in the ixgbe driver.
Use 'legacy' descriptors on the tx ring and prefetch slots gives
about 20% speedup at 900 MHz. Another 7-10% would come from removing
the explict calls to bus_dmamap* in the core (they are effectively
NOPs in this case, but it takes expensive load of the per-buffer
dma maps to figure out that they are all NULL.
Rx performance not investigated.
I am postponing the MFC so i can import a few more improvements
before merging.
- remove the KEVENT code, which was incomplete and not compiled anyways;
- change some while() loops into for()
- adjust indentation
- remove extra whitespace
MFC after: 1 week
Introduce some functions to map NIC ring indexes into netmap ring
indexes and vice versa. This way we can implement the bound
checks only in one place (and hopefully in a correct way).
On passing, make the code and comments more uniform across the
various drivers.
txsync() and rxsync() callbacks, removing some variables made
useless by this change;
- add generic lock and irq handling routines. These can be useful
in case there are no driver locks that we can reuse;
- add a few macros to reduce differences with the Linux version.
TUNABLE variable (hw.netmap.buf_size) so we can experiment
with values different from 2048 which may give better cache performance.
- rearrange the memory allocation code so it will be easier
to replace it with a different implementation. The current code
relies on a single large contiguous chunk of memory obtained through
contigmalloc.
The new implementation (not committed yet) uses multiple
smaller chunks which are easier to fit in a fragmented address
space.
- remove experimental code for disabling CRC
- use the correct constant for conversion between interrupt rate
and EITR values (the previous values were off by a factor of 2)
- make dev.ix.N.queueM.interrupt_rate a RW sysctl variable.
Changing individual values affects the queue immediately,
and propagates to all interfaces at the next reinit.
- add dev.ix.N.queueM.irqs rdonly sysctl, to export the actual
interrupt counts
Netmap-related changes for ixgbe:
- use the "new" format for TX descriptors in netmap mode.
- pass interrupt mitigation delays to the user process doing poll()
on a netmap file descriptor.
On the RX side this means we will not check the ring more than once
per interrupt. This gives the process a chance to sleep and process
packets in larger batches, thus reducing CPU usage.
On the TX side we take this even further: completed transmissions are
reclaimed every half ring even if the NIC interrupts more often.
This saves even more CPU without any additional tx delays.
Generic Netmap-related changes:
- align the netmap_kring to cache lines so that there is no false sharing
(possibly useful for multiqueue NICs and MSIX interrupts, which are
handled by different cores). It's a minor improvement but it does not
cost anything.
Reviewed by: Jack Vogel
Approved by: Jack Vogel
1. as reported by Alexander Fiveg, the allocator was reporting
half of the allocated memory. Fix this by exiting from the
loop earlier (not too critical because this code is going
away soon).
2. following a discussion on freebsd-current
http://lists.freebsd.org/pipermail/freebsd-current/2012-January/031144.html
turns out that (re)loading the dmamap was expensive and not optimized.
This operation is in the critical path when doing zero-copy forwarding
between interfaces.
At least on netmap and i386/amd64, the bus_dmamap_load can be
completely bypassed if the map is NULL, so we do it.
The latter change gives an almost 3x improvement in forwarding
performance, from the previous 9.5Mpps at 2.9GHz to the current
line rate (14.2Mpps) at 1.733GHz. (this is for 64+4 byte packets,
in other configurations the PCIe bus is a bottleneck).
the memory allocator used by netmap. No functional change,
two small bug fixes:
- in if_re.c add a missing bus_dmamap_sync()
- in netmap.c comment out a spurious free() in an error handling block
