Since virtio does not provide it's own ndo_features_check handler,
TSO, and now checksum offload, are disabled for stacked vlans.
Re-enable the support and let the host take care of it.  This
restores/improves Guest-to-Guest performance over Q-in-Q vlans.
Acked-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We are printing a decimal value for truesize so we shouldn't use an "0x"
prefix.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We already do a reset once in remove_vq_common -
there appears to be no point in doing another one
when we add/remove XDP.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

When ring size is small (<32 entries) making buffers smaller means a
full ring might not be able to hold enough buffers to fit a single large
packet.

Make sure a ring full of buffers is large enough to allow at least one
packet of max size.

Fixes: 2613af0e ("virtio_net: migrate mergeable rx buffers to page frag allocators")
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We don't need to align length to any particular
value anymore. Aligning to L1 cache size probably
sill makes sense to reduce false sharing.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Use the new _ctx virtio API to maintain true length for each buffer.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

With mergeable buffers we never use s/g for rx,
so allow specifying context in that case.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Small follow-up to d74a32ac ("xdp: use netlink extended ACK reporting")
in order to let drivers all use the same NL_SET_ERR_MSG_MOD() helper macro
for reporting. This also ensures that we consistently add the driver's
prefix for dumping the report in user space to indicate that the error
message is driver specific and not coming from core code. Furthermore,
NL_SET_ERR_MSG_MOD() now reuses NL_SET_ERR_MSG() and thus makes all macros
check the pointer as suggested.

References: https://www.spinics.net/lists/netdev/msg433267.htmlSigned-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We are going to add more parameters to find_vqs, let's wrap the call so
we don't need to tweak all drivers every time.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Try to carry error messages to the user via the netlink extended
ack message attribute.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Avoid hashing the tx napi struct into napi_hash[], which is used for
busy polling receive queues.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As of tx napi, device down (`ip link set dev $dev down`) hangs unless
tx napi is enabled. Else napi_enable is not called, so napi_disable
will spin on test_and_set_bit NAPI_STATE_SCHED.

Only call napi_disable if tx napi is enabled.

Fixes: 5a719c2552ca ("virtio-net: transmit napi")
Reported-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tx napi mode increases the rate of transmit interrupts. Suppress some
by masking interrupts while more packets are expected. The interrupts
will be reenabled before the last packet is sent.

This optimization reduces the througput drop with tx napi for
unidirectional flows such as UDP_STREAM that do not benefit from
cleaning tx completions in the the receive napi handler.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Amortize the cost of virtual interrupts by doing both rx and tx work
on reception of a receive interrupt if tx napi is enabled. With
VIRTIO_F_EVENT_IDX, this suppresses most explicit tx completion
interrupts for bidirectional workloads.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

An upcoming patch will call free_old_xmit_skbs indirectly from
virtnet_poll. Move the function above this to avoid having to
introduce a forward declaration.

This is a pure move: no code changes.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Convert virtio-net to a standard napi tx completion path. This enables
better TCP pacing using TCP small queues and increases single stream
throughput.

The virtio-net driver currently cleans tx descriptors on transmission
of new packets in ndo_start_xmit. Latency depends on new traffic, so
is unbounded. To avoid deadlock when a socket reaches its snd limit,
packets are orphaned on tranmission. This breaks socket backpressure,
including TSQ.

Napi increases the number of interrupts generated compared to the
current model, which keeps interrupts disabled as long as the ring
has enough free descriptors. Keep tx napi optional and disabled for
now. Follow-on patches will reduce the interrupt cost.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Prepare virtio-net for tx napi by converting existing napi code to
use helper functions. This also deduplicates some logic.
Signed-off-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

virtio attempts to clear the MTU feature bit if the value is out of the
supported range, but this has no real effect since FEATURES_OK has
already been set.

Fix this up by checking the MTU in the new validate callback.

Fixes: 14de9d11 ("virtio-net: Add initial MTU advice feature")
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

If one enables e.g. jumbo frames without mergeable
buffers, packets won't fit in 1500 byte buffers
we use. Switch to big packet mode instead.
TODO: make sizing more exact, possibly extend small
packet mode to use larger pages.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

The ethtool api {get|set}_settings is deprecated.
We move this driver to new api {get|set}_link_ksettings.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Declaring the factor is counter-intuitive, and people are prone
to using small(-ish) values even when that makes no sense.

Change the DECLARE_EWMA() macro to take the fractional precision,
in bits, rather than a factor, and update all users.

While at it, add some more documentation.
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Add a struct irq_affinity pointer to the find_vqs methods, which if set
is used to tell the PCI layer to create the MSI-X vectors for our I/O
virtqueues with the proper affinity from the start.  Compared to after
the fact affinity hints this gives us an instantly working setup and
allows to allocate the irq descritors node-local and avoid interconnect
traffic.  Last but not least this will allow blk-mq queues are created
based on the interrupt affinity for storage drivers.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

This patch switch to use build_skb() for small buffer which can have
better performance for both TCP and XDP (since we can work at page
before skb creation). It also remove lots of XDP codes since both
mergeable and small buffer use page frag during refill now.

                       Before   | After
XDP_DROP(xdp1) 64B  :  11.1Mpps | 14.4Mpps

Tested with xdp1/xdp2/xdp_ip_tx_tunnel and netperf.
Signed-off-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We set queues before reset which will cause a crash[1]. This is
because is_xdp_raw_buffer_queue() depends on the old xdp queue pairs
number to do the correct detection. So fix this by

- passing xdp queue pairs and current queue pairs to virtnet_reset()
- change vi->xdp_qp after reset but before refill, to make sure both
  free_unused_bufs() and refill can make correct detection of XDP.
- remove the duplicated queue pairs setting before virtnet_reset()
  since we will do it inside virtnet_reset()

[1]

[   74.328168] general protection fault: 0000 [#1] SMP
[   74.328625] Modules linked in: nfsd xfs libcrc32c virtio_net virtio_pci
[   74.329117] CPU: 0 PID: 2849 Comm: xdp2 Not tainted 4.10.0-rc7+ #499
[   74.329577] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.10.1-0-g8891697-prebuilt.qemu-project.org 04/01/2014
[   74.330424] task: ffff88007a894000 task.stack: ffffc90004388000
[   74.330844] RIP: 0010:skb_release_head_state+0x28/0x80
[   74.331298] RSP: 0018:ffffc9000438b8d0 EFLAGS: 00010206
[   74.331676] RAX: 0000000000000000 RBX: ffff88007ad96300 RCX: 0000000000000000
[   74.332217] RDX: ffff88007fc137a8 RSI: ffff88007fc0db28 RDI: 0001bf00000001be
[   74.332758] RBP: ffffc9000438b8d8 R08: 000000000005008f R09: 00000000000005f9
[   74.333274] R10: ffff88007d001700 R11: ffffffff820a8a4d R12: ffff88007ad96300
[   74.333787] R13: 0000000000000002 R14: ffff880036604000 R15: 000077ff80000000
[   74.334308] FS:  00007fc70d8a7b40(0000) GS:ffff88007fc00000(0000) knlGS:0000000000000000
[   74.334891] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   74.335314] CR2: 00007fff4144a710 CR3: 000000007ab56000 CR4: 00000000003406f0
[   74.335830] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   74.336373] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   74.336895] Call Trace:
[   74.337086]  skb_release_all+0xd/0x30
[   74.337356]  consume_skb+0x2c/0x90
[   74.337607]  free_unused_bufs+0x1ff/0x270 [virtio_net]
[   74.337988]  ? vp_synchronize_vectors+0x3b/0x60 [virtio_pci]
[   74.338398]  virtnet_xdp+0x21e/0x440 [virtio_net]
[   74.338741]  dev_change_xdp_fd+0x101/0x140
[   74.339048]  do_setlink+0xcf4/0xd20
[   74.339304]  ? symcmp+0xf/0x20
[   74.339529]  ? mls_level_isvalid+0x52/0x60
[   74.339828]  ? mls_range_isvalid+0x43/0x50
[   74.340135]  ? nla_parse+0xa0/0x100
[   74.340400]  rtnl_setlink+0xd4/0x120
[   74.340664]  ? cpumask_next_and+0x30/0x50
[   74.340966]  rtnetlink_rcv_msg+0x7f/0x1f0
[   74.341259]  ? sock_has_perm+0x59/0x60
[   74.341586]  ? napi_consume_skb+0xe2/0x100
[   74.342010]  ? rtnl_newlink+0x890/0x890
[   74.342435]  netlink_rcv_skb+0x92/0xb0
[   74.342846]  rtnetlink_rcv+0x23/0x30
[   74.343277]  netlink_unicast+0x162/0x210
[   74.343677]  netlink_sendmsg+0x2db/0x390
[   74.343968]  sock_sendmsg+0x33/0x40
[   74.344233]  SYSC_sendto+0xee/0x160
[   74.344482]  ? SYSC_bind+0xb0/0xe0
[   74.344806]  ? sock_alloc_file+0x92/0x110
[   74.345106]  ? fd_install+0x20/0x30
[   74.345360]  ? sock_map_fd+0x3f/0x60
[   74.345586]  SyS_sendto+0x9/0x10
[   74.345790]  entry_SYSCALL_64_fastpath+0x1a/0xa9
[   74.346086] RIP: 0033:0x7fc70d1b8f6d
[   74.346312] RSP: 002b:00007fff4144a708 EFLAGS: 00000246 ORIG_RAX: 000000000000002c
[   74.346785] RAX: ffffffffffffffda RBX: 00000000ffffffff RCX: 00007fc70d1b8f6d
[   74.347244] RDX: 000000000000002c RSI: 00007fff4144a720 RDI: 0000000000000003
[   74.347683] RBP: 0000000000000003 R08: 0000000000000000 R09: 0000000000000000
[   74.348544] R10: 0000000000000000 R11: 0000000000000246 R12: 00007fff4144bd90
[   74.349082] R13: 0000000000000002 R14: 0000000000000002 R15: 00007fff4144cda0
[   74.349607] Code: 00 00 00 55 48 89 e5 53 48 89 fb 48 8b 7f 58 48 85 ff 74 0e 40 f6 c7 01 74 3d 48 c7 43 58 00 00 00 00 48 8b 7b 68 48 85 ff 74 05 <f0> ff 0f 74 20 48 8b 43 60 48 85 c0 74 14 65 8b 15 f3 ab 8d 7e
[   74.351008] RIP: skb_release_head_state+0x28/0x80 RSP: ffffc9000438b8d0
[   74.351625] ---[ end trace fe6e19fd11cfc80b ]---

Fixes: 2de2f7f4 ("virtio_net: XDP support for adjust_head")
Cc: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We already have counters for sent/recv packets and sent/recv bytes.
Doing a batched update to reduce the number of
u64_stats_update_begin/end().

Take care not to bother with stats update when called
speculatively.

Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for XDP adjust head by allocating a 256B header region
that XDP programs can grow into. This is only enabled when a XDP
program is loaded.

In order to ensure that we do not have to unwind queue headroom push
queue setup below bpf_prog_add. It reads better to do a prog ref
unwind vs another queue setup call.

At the moment this code must do a full reset to ensure old buffers
without headroom on program add or with headroom on program removal
are not used incorrectly in the datapath. Ideally we would only
have to disable/enable the RX queues being updated but there is no
API to do this at the moment in virtio so use the big hammer. In
practice it is likely not that big of a problem as this will only
happen when XDP is enabled/disabled changing programs does not
require the reset. There is some risk that the driver may either
have an allocation failure or for some reason fail to correctly
negotiate with the underlying backend in this case the driver will
be left uninitialized. I have not seen this ever happen on my test
systems and for what its worth this same failure case can occur
from probe and other contexts in virtio framework.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For XDP we will need to reset the queues to allow for buffer headroom
to be configured. In order to do this we need to essentially run the
freeze()/restore() code path. Unfortunately the locking requirements
between the freeze/restore and reset paths are different however so
we can not simply reuse the code.

This patch refactors the code path and adds a reset helper routine.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Factor out qp assignment.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

At this point the do_xdp_prog is mostly if/else branches handling
the different modes of virtio_net. So remove it and handle running
the program in the per mode handlers.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For XDP use case and to allow ethtool reset tests it is useful to be
able to use reset paths from contexts where rtnl lock is already
held.

This requries updating virtnet_set_queues and free_receive_bufs the
two places where rtnl_lock is taken in virtio_net. To do this we
use the following pattern,

	_foo(...) { do stuff }
	foo(...) { rtnl_lock(); _foo(...); rtnl_unlock()};

this allows us to use freeze()/restore() flow from both contexts.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since commit 364b6055 ("net: busy-poll: return busypolling status to
drivers"), napi_complete_done() returns a boolean that can be used
by drivers to conditionally rearm interrupts.

This patch changes virtio_net to use this boolean to avoid a bit of
overhead for busy-poll users.

Jason reports about 1.1% improvement for 1 byte TCP_RR (burst 100).
Signed-off-by: NEric Dumazet <edumazet@google.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Generic NAPI busy polling allows us to remove custom implementations
found in drivers.

It is possible further optimization could be done by testing
napi_complete_done() return value.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 17bedab2 ("bpf: xdp: Allow head adjustment in XDP prog")
added a new XDP helper to prepend and remove data from a frame.
Make virtio_net reject programs making use of this helper until
proper support is added.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the small buffer case during driver unload we currently use
put_page instead of dev_kfree_skb. Resolve this by adding a check
for virtnet mode when checking XDP queue type. Also name the
function so that the code reads correctly to match the additional
check.

Fixes: bb91accf ("virtio-net: XDP support for small buffers")
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This work adds a number of tracepoints to paths that are either
considered slow-path or exception-like states, where monitoring or
inspecting them would be desirable.

For bpf(2) syscall, tracepoints have been placed for main commands
when they succeed. In XDP case, tracepoint is for exceptions, that
is, f.e. on abnormal BPF program exit such as unknown or XDP_ABORTED
return code, or when error occurs during XDP_TX action and the packet
could not be forwarded.

Both have been split into separate event headers, and can be further
extended. Worst case, if they unexpectedly should get into our way in
future, they can also removed [1]. Of course, these tracepoints (like
any other) can be analyzed by eBPF itself, etc. Example output:

  # ./perf record -a -e bpf:* sleep 10
  # ./perf script
  sock_example  6197 [005]   283.980322:      bpf:bpf_map_create: map type=ARRAY ufd=4 key=4 val=8 max=256 flags=0
  sock_example  6197 [005]   283.980721:       bpf:bpf_prog_load: prog=a5ea8fa30ea6849c type=SOCKET_FILTER ufd=5
  sock_example  6197 [005]   283.988423:   bpf:bpf_prog_get_type: prog=a5ea8fa30ea6849c type=SOCKET_FILTER
  sock_example  6197 [005]   283.988443: bpf:bpf_map_lookup_elem: map type=ARRAY ufd=4 key=[06 00 00 00] val=[00 00 00 00 00 00 00 00]
  [...]
  sock_example  6197 [005]   288.990868: bpf:bpf_map_lookup_elem: map type=ARRAY ufd=4 key=[01 00 00 00] val=[14 00 00 00 00 00 00 00]
       swapper     0 [005]   289.338243:    bpf:bpf_prog_put_rcu: prog=a5ea8fa30ea6849c type=SOCKET_FILTER

  [1] https://lwn.net/Articles/705270/Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

I don't have any guests with PAGE_SIZE > 64k but the
code seems to be clearly broken in that case
as PAGE_SIZE / MERGEABLE_BUFFER_ALIGN will need
more than 8 bit and so the code in mergeable_ctx_to_buf_address
does not give us the actual true size.

Cc: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 501db511 ("virtio: don't set VIRTIO_NET_HDR_F_DATA_VALID on
xmit") in fact disables VIRTIO_HDR_F_DATA_VALID on receiving path too,
fixing this by adding a hint (has_data_valid) and set it only on the
receiving path.

Cc: Rolf Neugebauer <rolf.neugebauer@docker.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Acked-by: NRolf Neugebauer <rolf.neugebauer@docker.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The network device operation for reading statistics is only called
in one place, and it ignores the return value. Having a structure
return value is potentially confusing because some future driver could
incorrectly assume that the return value was used.

Fix all drivers with ndo_get_stats64 to have a void function.
Signed-off-by: NStephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

when some other buffer is immediately copied into allocated region.
Replace calls to kmalloc followed by a memcpy with a direct
call to kmemdup.
Signed-off-by: NShyam Saini <mayhs11saini@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the state names got added a script was used to add the extra argument
to the calls. The script basically converted the state constant to a
string, but the cleanup to convert these strings into meaningful ones did
not happen.

Replace all the useless strings with 'subsys/xxx/yyy:state' strings which
are used in all the other places already.
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Siewior <bigeasy@linutronix.de>
Link: http://lkml.kernel.org/r/20161221192112.085444152@linutronix.deSigned-off-by: NThomas Gleixner <tglx@linutronix.de>