This patch allows the user to set and retrieve speed and duplex of the
virtio_net device via ethtool. Having this functionality is very helpful
for simulating different environments and also enables the virtio_net
device to participate in operations where proper speed and duplex are
required (e.g. currently bonding lacp mode requires full duplex). Custom
speed and duplex are not allowed, the user-supplied settings are validated
before applying.

Example:
$ ethtool eth1
Settings for eth1:
...
	Speed: Unknown!
	Duplex: Unknown! (255)
$ ethtool -s eth1 speed 1000 duplex full
$ ethtool eth1
Settings for eth1:
...
	Speed: 1000Mb/s
	Duplex: Full

Based on a patch by Roopa Prabhu.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Once virtio starts using the DMA API, we won't be able to safely DMA
from the stack.  virtio-net does a couple of config DMA requests
from small stack buffers -- switch to using dynamically-allocated
memory.

This should have no effect on any performance-critical code paths.
Reported-by: NAndy Lutomirski <luto@kernel.org>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NAndy Lutomirski <luto@kernel.org>

NAPI drivers no longer need to observe a particular protocol
to benefit from busy polling (CONFIG_NET_RX_BUSY_POLL=y)

napi_hash_add() and napi_hash_del() are automatically called
from core networking stack, respectively from
netif_napi_add() and netif_napi_del()

This patch depends on free_netdev() and netif_napi_del() being
called from process context, which seems to be the norm.

Drivers might still prefer to call napi_hash_del() on their
own, since they might combine all the rcu grace periods into
a single one, knowing their NAPI structures lifetime, while
core networking stack has no idea of a possible combining.

Once this patch proves to not bring serious regressions,
we will cleanup drivers to either remove napi_hash_del()
or provide appropriate rcu grace periods combining.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We would like to automatically provide busy polling support
to all NAPI drivers, without them having to implement anything.

skb_mark_napi_id() can be called from napi_gro_receive() and
napi_get_frags().

Few drivers are still calling skb_mark_napi_id() because
they use netif_receive_skb(). They should eventually call
napi_gro_receive() instead. I will leave this to drivers
maintainers.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Usually an skb does not have up to MAX_SKB_FRAGS frags. So no need to
initialize the unuse part of sg. This patch initialize the sg based on
the real number it will used:

- during xmit, it could be inferred from nr_frags and can_push.
- for small receive buffer, it will also be 2.

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of using the out-of-line EWMA calculation, use DECLARE_EWMA()
to create static inlines. On x86/64 this results in no change in code
size for me, but reduces the struct receive_queue size by the two
unsigned long values that store the parameters.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

virtio declares support for NETIF_F_FRAGLIST, but assumes
that there are at most MAX_SKB_FRAGS + 2 fragments which isn't
always true with a fraglist.

A longer fraglist in the skb will make the call to skb_to_sgvec overflow
the sg array, leading to memory corruption.

Drop NETIF_F_FRAGLIST so we only get what we can handle.

Cc: Michael S. Tsirkin <mst@redhat.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>

Straightforward patch to add GRO processing to virtio_net.

napi_complete_done() usage allows more aggressive aggregation,
opted-in by setting /sys/class/net/xxx/gro_flush_timeout

Tested:

Setting /sys/class/net/xxx/gro_flush_timeout to 1000 nsec,
Rick Jones reported following results.

One VM of each on a pair of OpenStack compute nodes with E5-2650Lv3 CPUs
and Intel 82599ES-based NICs. So, two "before" and two "after" VMs.
The OpenStack compute nodes were running OpenStack Kilo, with VxLAN
encapsulation being used through OVS so no GRO coming-up the host
stack.  The compute nodes themselves were running a 3.14-based kernel.

Single-stream netperf, CPU utilizations and thus service demands are
based on intra-guest reported CPU.

Throughput Mbit/s, bigger is better
        Min     Median  Average Max
4.2.0-rc3+      1364    1686    1678    1938
4.2.0-rc3+flush1k       1824    2269    2275    2647

Send Service Demand, smaller is better
        Min     Median  Average Max
4.2.0-rc3+      0.236   0.558   0.524   0.802
4.2.0-rc3+flush1k       0.176   0.503   0.471   0.738

Receive Service Demand, smaller is better.
        Min     Median  Average Max
4.2.0-rc3+      1.906   2.188   2.191   2.531
4.2.0-rc3+flush1k       0.448   0.529   0.533   0.692
Signed-off-by: NEric Dumazet <edumazet@google.com>
Tested-by: NRick Jones <rick.jones2@hp.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ANY_LAYOUT is a compatibility feature. It's implied
for VERSION_1 devices, and non-transitional devices
might not offer it. Change code to behave accordingly.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NPaolo Bonzini <pbonzini@redhat.com>
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit d631b94e
    virtio: change comment in transmit

started clarifying the logic behind queue state management,
but introduced an inaccuracy: TX_BUSY does not cause
a BUG message.

Clean this up some more, explaining the tradeoffs in detail.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

received is 0, no need to minus it and use "+=" to reassign it
Signed-off-by: NLi RongQing <roy.qing.li@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The original comment was not really informative or funny
as well as sexist. Replace it with a better explanation of
why the driver does stop and what the impacts are.
Signed-off-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We don't delete napi from hash list during module exit. This will
cause the following panic when doing module load and unload:

BUG: unable to handle kernel paging request at 0000004e00000075
IP: [<ffffffff816bd01b>] napi_hash_add+0x6b/0xf0
PGD 3c5d5067 PUD 0
Oops: 0000 [#1] SMP
...
Call Trace:
[<ffffffffa0a5bfb7>] init_vqs+0x107/0x490 [virtio_net]
[<ffffffffa0a5c9f2>] virtnet_probe+0x562/0x791815639d880be [virtio_net]
[<ffffffff8139e667>] virtio_dev_probe+0x137/0x200
[<ffffffff814c7f2a>] driver_probe_device+0x7a/0x250
[<ffffffff814c81d3>] __driver_attach+0x93/0xa0
[<ffffffff814c8140>] ? __device_attach+0x40/0x40
[<ffffffff814c6053>] bus_for_each_dev+0x63/0xa0
[<ffffffff814c7a79>] driver_attach+0x19/0x20
[<ffffffff814c76f0>] bus_add_driver+0x170/0x220
[<ffffffffa0a60000>] ? 0xffffffffa0a60000
[<ffffffff814c894f>] driver_register+0x5f/0xf0
[<ffffffff8139e41b>] register_virtio_driver+0x1b/0x30
[<ffffffffa0a60010>] virtio_net_driver_init+0x10/0x12 [virtio_net]

This patch fixes this by doing this in virtnet_free_queues(). And also
don't delete napi in virtnet_freeze() since it will call
virtnet_free_queues() which has already did this.

Fixes 91815639 ("virtio-net: rx busy polling support")
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts commit 3d0ad094.

Now that GSO functionality can correctly track if the fragment
id has been selected and select a fragment id if necessary,
we can re-enable UFO on tap/macvap and virtio devices.
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch enables the use of software timestamping via the virtio_net
driver.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Some devices might not implement config space access
(e.g. remoteproc used not to - before 3.9).
virtio/net needs config space access so make it
fail gracefully if not there.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

There's no need to do header check for virtio-net since:

- Host sets dodgy for all gso packets from guest and check the header.
- Host should be prepared for all kinds of evil packets from guest, since
  malicious guest can send any kinds of packet.

So this patch sets NETIF_F_GSO_ROBUST for virtio-net to skip the check.

Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The commit d75b1ade (net: less
interrupt masking in NAPI) breaks virtio_net in an insidious way.

It is now required that if the entire budget is consumed when poll
returns, the napi poll_list must remain empty.  However, like some
other drivers virtio_net tries to do a last-ditch check and if
there is more work it will call napi_schedule and then immediately
process some of this new work.  Should the entire budget be consumed
while processing such new work then we will violate the new caller
contract.

This patch fixes this by not touching any work when we reschedule
in virtio_net.

The worst part of this bug is that the list corruption causes other
napi users to be moved off-list.  In my case I was chasing a stall
in IPsec (IPsec uses netif_rx) and I only belatedly realised that it
was virtio_net which caused the stall even though the virtio_net
poll was still functioning perfectly after IPsec stalled.
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Core activates this bit automatically now,
drop it from drivers that set it explicitly.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Now that we have completed 1.0 support, enable it in our driver.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

The spec states that mac in config space is only driver-writable in the
legacy case.  Fence writing it in virtnet_set_mac_address() in the
virtio 1.0 case.
Suggested-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>

With VERSION_1 virtio_net uses same header size
whether mergeable buffers are enabled or not.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NJason Wang <jasowang@redhat.com>

Our buffer length check is not strict enough for mergeable
buffers: buffer can still be shorter that header + address
by 2 bytes.

Fix that up.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NJason Wang <jasowang@redhat.com>

virtio 1.0 doesn't use virtio_net_hdr anymore, and in fact, it's not
really useful since virtio_net_hdr_mrg_rxbuf includes that as the first
field anyway.

Let's drop it, precalculate header len and store within vi instead.

This way we can also remove struct skb_vnet_hdr.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NJason Wang <jasowang@redhat.com>

Too many places poke at [rs]q->vq->vdev->priv just to get
the vi structure.  Let's just pass the pointer around: seems
cleaner, and might even be faster.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Based on patches by Rusty Russell, Cornelia Huck.
Note: more code changes are needed for 1.0 support
(due to different header size).
So we don't advertize support for 1.0 yet.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We currently trigger BUG when VIRTIO_NET_F_CTRL_VQ
is not set but one of features depending on it is.
That's not a friendly way to report errors to
hypervisors.
Let's check, and fail probe instead.

Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IPv6 does not allow fragmentation by routers, so there is no
fragmentation ID in the fixed header.  UFO for IPv6 requires the ID to
be passed separately, but there is no provision for this in the virtio
net protocol.

Until recently our software implementation of UFO/IPv6 generated a new
ID, but this was a bug.  Now we will use ID=0 for any UFO/IPv6 packet
passed through a tap, which is even worse.

Unfortunately there is no distinction between UFO/IPv4 and v6
features, so disable UFO on taps and virtio_net completely until we
have a proper solution.

We cannot depend on VM managers respecting the tap feature flags, so
keep accepting UFO packets but log a warning the first time we do
this.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Fixes: 916e4cf4 ("ipv6: reuse ip6_frag_id from ip6_ufo_append_data")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 0b725a2c
    net: Remove ndo_xmit_flush netdev operation, use signalling instead.

added code that looks at skb->xmit_more after the skb has
been put in TX VQ. Since some paths process the ring and free the skb
immediately, this can cause use after free.

Fix by storing xmit_more in a local variable.

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after restore returns, virtio net violated this
rule by using receive VQs within restore.

To fix, call virtio_device_ready before using VQs.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

In the extremely unlikely event that driver initialization fails after
RX buffers are added, virtio net frees RX buffers while VQs are
still active, potentially causing device to use a freed buffer.

To fix, reset device first - same as we do on device removal.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after probe returns, virtio net violated this
rule by using receive VQs within probe.

To fix, call virtio_device_ready before using VQs.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

	goto done;
done:
	return;
is ugly, it was put there to make diff review easier.
replace by open-coded return.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

config_mutex served two purposes: prevent multiple concurrent config
change handlers, and synchronize access to config_enable flag.

Since commit dbf2576e
    workqueue: make all workqueues non-reentrant
all workqueues are non-reentrant, and config_enable
is now gone.

Get rid of the unnecessary lock.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

Now that virtio core ensures config changes don't arrive during probing,
drop config_enable flag in virtio net.
On removal, flush is now sufficient to guarantee that no change work is
queued.

This help simplify the driver, and will allow setting DRIVER_OK earlier
without losing config change notifications.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

This is the only driver which doesn't hand virtqueue_add_inbuf and
virtqueue_add_outbuf a well-formed, well-terminated sg.  Fix it,
so we can make virtio_add_* simpler.

pktgen results:
	modprobe pktgen
	echo 'add_device eth0' > /proc/net/pktgen/kpktgend_0
	echo nowait 1 > /proc/net/pktgen/eth0
	echo count 1000000 > /proc/net/pktgen/eth0
	echo clone_skb 100000 > /proc/net/pktgen/eth0
	echo dst_mac 4e:14:25:a9:30:ac > /proc/net/pktgen/eth0
	echo dst 192.168.1.2 > /proc/net/pktgen/eth0
	for i in `seq 20`; do echo start > /proc/net/pktgen/pgctrl; tail -n1 /proc/net/pktgen/eth0; done

Before:
  746547-793084(786421+/-9.6e+03)pps 346-367(364.4+/-4.4)Mb/sec (346397808-367990976(3.649e+08+/-4.5e+06)bps) errors: 0

After:
  767390-792966(785159+/-6.5e+03)pps 356-367(363.75+/-2.9)Mb/sec (356068960-367936224(3.64314e+08+/-3e+06)bps) errors: 0
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Mirror the changes made to ixgbe in commit 2367a173
("ixgbe: flush when in xmit_more mode and under descriptor pressure")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As reported by Jesper Dangaard Brouer, for high packet rates the
overhead of having another indirect call in the TX path is
non-trivial.

There is the indirect call itself, and then there is all of the
reloading of the state to refetch the tail pointer value and
then write the device register.

Move to a more passive scheme, which requires very light modifications
to the device drivers.

The signal is a new skb->xmit_more value, if it is non-zero it means
that more SKBs are pending to be transmitted on the same queue as the
current SKB.  And therefore, the driver may elide the tail pointer
update.

Right now skb->xmit_more is always zero.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add basic support for rx busy polling. Instead of introducing new
states and spinlock to synchronize between NAPI and polling method,
this patch just reuse NAPI state to avoid extra overhead for fast path
and simplified the codes.

Test was done between a kvm guest and an external host. Two hosts were
connected through 40gb mlx4 cards. With both busy_poll and busy_read
are set to 50 in guest, 1 byte netperf tcp_rr shows 127% improvement:
transaction rate was increased from 8353.33 to 18966.87.

Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Vlad Yasevich <vyasevic@redhat.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>