r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are
still not MPSAFE (or already are but aren’t properly marked).
Use it in preparation for a general review of all nodes.
This is non-functional change that adds annotations to SYSCTL_NODE and
SYSCTL_PROC nodes using one of the soon-to-be-required flags.
Mark all obvious cases as MPSAFE. All entries that haven't been marked
as MPSAFE before are by default marked as NEEDGIANT
Approved by: kib (mentor, blanket)
Commented by: kib, gallatin, melifaro
Differential Revision: https://reviews.freebsd.org/D23718
switch over to opt-in instead of opt-out for epoch.
Instead of IFF_NEEDSEPOCH, provide IFF_KNOWSEPOCH. If driver marks
itself with IFF_KNOWSEPOCH, then ether_input() would not enter epoch
when processing its packets.
Now this will create recursive entrance in epoch in >90% network
drivers, but will guarantee safeness of the transition.
Mark several tested drivers as IFF_KNOWSEPOCH.
Reviewed by: hselasky, jeff, bz, gallatin
Differential Revision: https://reviews.freebsd.org/D23674
The patch could be simplier, using only the second chunk to
vtnet_rxq_eof(), that passes full mbufs to pfil(9). Packet
filter would m_free() them in case of returning PFIL_DROPPED.
However, we pretend to be a hardware driver, so we first try
to pass a memory buffer via PFIL_MEMPTR feature. This is mostly
done for debugging purposes, so that one can experiment in bhyve
with packet filters utilizing same features as a true driver.
Don't wait until the vtnet_debugnet_init() call happens, because at that
point we might already have allocated something from
vtnet_tx_header_zone.
Some systems showed this panic:
vtnet0: link state changed to UP
panic: keg vtnet_tx_hdr initialization after use.
cpuid = 5
time = 1578427700
KDB: stack backtrace:
db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 0xfffffe004db427f0
vpanic() at vpanic+0x17e/frame 0xfffffe004db42850
panic() at panic+0x43/frame 0xfffffe004db428b0
uma_zone_reserve() at uma_zone_reserve+0xf6/frame 0xfffffe004db428f0
vtnet_debugnet_init() at vtnet_debugnet_init+0x77/frame 0xfffffe004db42930
debugnet_any_ifnet_update() at debugnet_any_ifnet_update+0x42/frame 0xfffffe004db42980
do_link_state_change() at do_link_state_change+0x1b3/frame 0xfffffe004db429d0
taskqueue_run_locked() at taskqueue_run_locked+0x178/frame 0xfffffe004db42a30
taskqueue_run() at taskqueue_run+0x4d/frame 0xfffffe004db42a50
ithread_loop() at ithread_loop+0x1d6/frame 0xfffffe004db42ab0
fork_exit() at fork_exit+0x80/frame 0xfffffe004db42af0
fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe004db42af0
--- trap 0, rip = 0, rsp = 0, rbp = 0 ---
KDB: enter: panic
[ thread pid 12 tid 100011 ]
Stopped at kdb_enter+0x37: movq $0,0x1084eb6(%rip)
db>
Reviewed by: cem, markj
Differential Revision: https://reviews.freebsd.org/D23073
This patch is part of an effort to make bhyve networking (in particular TCP)
faster. The key strategy to enhance TCP throughput is to let the whole packet
datapath work with TSO/LRO packets (up to 64KB each), so that the per-packet
overhead is amortized over a large number of bytes.
This capability is supported in the guest by means of the vtnet(4) driver,
which is able to handle TSO/LRO packets leveraging the virtio-net header
(see struct virtio_net_hdr and struct virtio_net_hdr_mrg_rxbuf).
A bhyve VM exchanges packets with the host through a network backend,
which can be vale(4) or if_tap(4).
While vale(4) supports TSO/LRO packets, if_tap(4) does not.
This patch extends if_tap(4) with the ability to understand the virtio-net
header, so that a tapX interface can process TSO/LRO packets.
A couple of ioctl commands have been added to configure and probe the
virtio-net header. Once the virtio-net header is set, the tapX interface
acquires all the IFCAP capabilities necessary for TSO/LRO.
Reviewed by: kevans
Differential Revision: https://reviews.freebsd.org/D21263
Debugnet is a simplistic and specialized panic- or debug-time reliable
datagram transport. It can drive a single connection at a time and is
currently unidirectional (debug/panic machine transmit to remote server
only).
It is mostly a verbatim code lift from netdump(4). Netdump(4) remains
the only consumer (until the rest of this patch series lands).
The INET-specific logic has been extracted somewhat more thoroughly than
previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as
much as possible as is protocol-independent, remains in debugnet.c. The
separation is not perfect and future improvement is welcome. Supporting
INET6 is a long-term goal.
Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to
'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the
generic module would be more confusing than the refactoring.
The only functional change here is the mbuf allocation / tracking. Instead
of initiating solely on netdump-configured interface(s) at dumpon(8)
configuration time, we watch for any debugnet-enabled NIC for link
activation and query it for mbuf parameters at that time. If they exceed
the existing high-water mark allocation, we re-allocate and track the new
high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone.
In a future patch in this series, this will allow initiating netdump from
panic ddb(4) without pre-panic configuration.
No other functional change intended.
Reviewed by: markj (earlier version)
Some discussion with: emaste, jhb
Objection from: marius
Differential Revision: https://reviews.freebsd.org/D21421
Register MODULE_PNP_INFO for virtio devices using the newbus PNP information
provided by the previous commit. Matching can be quite simple; existing
probe routines only matched on bus (implicit) and device_type. The same
matching criteria are retained exactly, but is now also available to
devmatch(8).
Reviewed by: bryanv, markj; imp (earlier version)
Differential Revision: https://reviews.freebsd.org/D20407
Checksum offloading for SCTP is not currently specified for virtio.
If the hypervisor announces checksum offloading support, it means TCP
and UDP checksum offload. If an SCTP packet is sent and the host announced
checksum offload support, the hypervisor inserts the IP checksum (16-bit)
at the correct offset, but this is not the right checksum, which is a CRC32c.
This results in all outgoing packets having the wrong checksum and therefore
breaking SCTP based communications.
This patch removes SCTP checksum offloading support from the virtio
network interface.
Thanks to Felix Weinrank for making me aware of the issue.
Reviewed by: bryanv@
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D20147
Because of a typo, the code was mistakenly resetting the
vtnrx_vq pointer rather than vtntx_tq.
Reviewed by: bryanv
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D19015
netmap(4) support for vtnet(4) was incomplete and had multiple bugs.
This commit fixes those bugs to bring netmap on vtnet in a functional state.
Changelist:
- handle errors returned by virtqueue_enqueue() properly (they were
previously ignored)
- make sure netmap XOR rest of the kernel access each virtqueue.
- compute the number of netmap slots for TX and RX separately, according to
whether indirect descriptors are used or not for a given virtqueue.
- make sure sglist are freed according to their type (mbufs or netmap
buffers)
- add support for mulitiqueue and netmap host (aka sw) rings.
- intercept VQ interrupts directly instead of intercepting them in txq_eof
and rxq_eof. This simplifies the code and makes it easier to make sure
taskqueues are not running for a VQ while it is in netmap mode.
- implement vntet_netmap_config() to cope with changes in the number of queues.
Reviewed by: bryanv
Approved by: gnn (mentor)
MFC after: 3 days
Sponsored by: Sunny Valley Networks
Differential Revision: https://reviews.freebsd.org/D17916
Uses of mallocarray(9).
The use of mallocarray(9) has rocketed the required swap to build FreeBSD.
This is likely caused by the allocation size attributes which put extra pressure
on the compiler.
Given that most of these checks are superfluous we have to choose better
where to use mallocarray(9). We still have more uses of mallocarray(9) but
hopefully this is enough to bring swap usage to a reasonable level.
Reported by: wosch
PR: 225197
Focus on code where we are doing multiplications within malloc(9). None of
these is likely to overflow, however the change is still useful as some
static checkers can benefit from the allocation attributes we use for
mallocarray.
This initial sweep only covers malloc(9) calls with M_NOWAIT. No good
reason but I started doing the changes before r327796 and at that time it
was convenient to make sure the sorrounding code could handle NULL values.
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.
The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
Mainly focus on files that use BSD 3-Clause license.
The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
Special thanks to Wind River for providing access to "The Duke of
Highlander" tool: an older (2014) run over FreeBSD tree was useful as a
starting point.
Currently in Virtio driver without TSO/GSO features enabled, the max scatter
gather segments for the TX path can be 4, which limits the support for 9K JUMBO
frames. 9K JUMBO frames results in more than 4 scatter gather segments and
virtio driver fails to send the frame down to host OS. With TSO/GSO feature
enabled max scatter gather segments can be 64, then 9K JUMBO frames are fine,
this is making virtio driver to support JUMBO frames only with TSO/GSO.
Increasing the VTNET_MIN_TX_SEGS which is the case for non TSO/GSO to 32 to
support upto 64K JUMBO frames to Host.
Submitted by: Lohith Bellad <lohithbsd@gmail.com>
Reviewed by: adrian
Differential Revision: https://reviews.freebsd.org/D8803
Correctly limit npairs passed to vtnet_ctrl_mq_cmd. This ensures that
VQ_ALLOC_INFO_INIT is called with the correct value, preventing the system
from hanging when max_virtqueue_pairs > VTNET_MAX_QUEUE_PAIRS.
Add new sysctl requested_vq_pairs which allow the user to configure
the requested number of virtqueue pairs. The actual value will still take
into account the system limits.
Also missing sysctls for the current tunables so their values can be seen.
PR: 207446
Reported by: Andy Carrel
MFC after: 3 days
Relnotes: Yes
Sponsored by: Multiplay
Since r276367 added the virtio_mmio support vtnet_modevent() gets called twice.
This resulted in a memory leak during load and a panic on unload.
Count the loads so we only initialise once (just like cxgbe(4)), and only clean
up in the final unload.
PR: 209428
Submitted by: novel@FreeBSD.org
MFC after: 1 week
vtnet interfaces are always in promiscuous mode (at least if the
VIRTIO_NET_F_CTRL_RX feature is not negotiated with the host). if_promisc() on
a vtnet interface returned ENOTSUP although it has IFF_PROMISC set. This
confused the bridge code. Instead we now accept all enable/disable promiscuous
commands (and always keep IFF_PROMISC set).
There are also two issues with the if_bridge error handling.
If if_promisc() fails it uses bridge_delete_member() to clean up. This tries to
disable promiscuous mode on the interface. That runs into an assert, because
promiscuous mode was never set in the first place. (That's the panic reported in
PR 200210.)
We can only unset promiscuous mode if the interface actually is promiscuous.
This goes against the reference counting done by if_promisc(), but only the
first/last if_promic() calls can actually fail, so this is safe.
A second issue is a double free of bif. It's already freed by
bridge_delete_member().
PR: 200210
Differential Revision: https://reviews.freebsd.org/D2804
Reviewed by: philip (mentor)
Currently if you ifconfig down a vtnet interface while it is being used
via netmap, the kernel panics due to trying to treat the cookie values
in the virtio rings as mbufs to be freed. When netmap is enabled, these
cookie values are pointers to something else.
Note that other netmap-aware drivers don't seem to need this as they
store the mbuf pointers in the software rings that mirror the hardware
descriptor rings, and since netmap doesn't touch those, the software
state always has NULL mbuf pointers causing the loops to free mbufs to
not do anything. However, vtnet reuses the same state area for both
netmap and non-netmap mode, so it needs to explicitly avoid looking at
the rings and treating the cookie values as mbufs if netmap is
enabled.
Differential Revision: https://reviews.freebsd.org/D2348
Reviewed by: adrian, bryanv, luigi
MFC after: 1 week
Sponsored by: Norse Corp, Inc.
from the FreeBSD network code. The flag is still kept around in the
"sys/mbuf.h" header file, but does no longer have any users. Instead
the "m_pkthdr.rsstype" field in the mbuf structure is now used to
decide the meaning of the "m_pkthdr.flowid" field. To modify the
"m_pkthdr.rsstype" field please use the existing "M_HASHTYPE_XXX"
macros as defined in the "sys/mbuf.h" header file.
This patch introduces new behaviour in the transmit direction.
Previously network drivers checked if "M_FLOWID" was set in "m_flags"
before using the "m_pkthdr.flowid" field. This check has now now been
replaced by checking if "M_HASHTYPE_GET(m)" is different from
"M_HASHTYPE_NONE". In the future more hashtypes will be added, for
example hashtypes for hardware dedicated flows.
"M_HASHTYPE_OPAQUE" indicates that the "m_pkthdr.flowid" value is
valid and has no particular type. This change removes the need for an
"if" statement in TCP transmit code checking for the presence of a
valid flowid value. The "if" statement mentioned above is now a direct
variable assignment which is then later checked by the respective
network drivers like before.
Additional notes:
- The SCTP code changes will be committed as a separate patch.
- Removal of the "M_FLOWID" flag will also be done separately.
- The FreeBSD version has been bumped.
MFC after: 1 month
Sponsored by: Mellanox Technologies
The prior change to not enable LRO by default has confused several
people. The configurations where LRO is problematic is not the
typical use case for VirtIO, and due to other issues, this often
requires checksum offloading to be disabled anyways.
PR: 185864
MFC after: 2 weeks
- Do not accumulate statistics on every tick.
- Accumulate statistics in vtnet_setup_stat_sysctl()
and in vtnet_get_counter().
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
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 Tx interrupt is now kept disabled in the common case, only
enabled when the number of free descriptors in the queue falls
below a threshold. Transmitted frames are cleared from the VQ
before subsequent transmit, or in the watchdog timer.
This was a very big performance improvement for an experimental
Netmap bhyve backend.
MFC after: 1 month
The header structure consists of two 1-byte elements, but it must always
be describable by a single SG entry. Note for consistency, specify the
alignment everywhere, even if the structure has the appropriate natural
alignment since it contains a uint16_t.
Obtained from: DragonFlyBSD
MFC after: 1 week
A similar fix should be applied to vmxnet, ixgbe, igb, i40e.
(some of them previously reported by Michael Tuexen)
Drivers using if_transmit are correct, and so are most of the
other drivers that reassing if_transmit.
Among other things, this bug causes panics when using netmap emulation
on top of generic drivers.
Approved by: bryanv
MFC after: 3 days
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.
The sglist segment array has grown to a bit over 512 bytes (on
64-bit system) which is more than ideally should be put on the
stack. Instead allocate an appropriately sized sglist and hang
it off each Rx/Tx queue structure.
Bump the maximum number of Tx segments to 64 to make it unlikely
we'll have defragment an mbuf chain. Our previous count was
rounded up to this value since it is the next power of two, so
effective memory usage should not change.
Also only allocate the maximum number of Tx segments if TSO was
negotiated.
