There's no need to allocate a portid structure and then, for
each of those, walk the interfaces - we can just add a flag
to each interface and walk those directly. Due to padding in
the struct, we can even do it without any memory cost, and
it even simplifies the code.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The monitor mode delivery logic makes it hard to add any
kind of filtering in an efficient way, because the monitor
SKB is created first and then passed to all interfaces.
Rewrite the logic to create the monitor SKB the first time
it's actually needed, and then keep delivering it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When part of a bigger bandwidth (160 MHz) channel falls in DFS
channel range it is possible that the center frequency may not
necessarily be a radar channel. Remove the sanity check on channel
flag for IEEE80211_CHAN_RADAR in regulatory_propagate_dfs_state(),
this should fix the dfs state propagation for non-DFS center freq
which has DFS channels in it's bandwidth, should also fix unnecessary
WARN_ON() spam in regulatory_propagate_dfs_state().
Fixes: 8976672736 ("cfg80211: Share Channel DFS state across wiphys of same DFS domain")
Reported-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qti.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Fix to return error code -ENODEV from the no PHY found error
handling case instead of 0, as done elsewhere in this function.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This provides a generic SKB based non-optimized XDP path which is used
if either the driver lacks a specific XDP implementation, or the user
requests it via a new IFLA_XDP_FLAGS value named XDP_FLAGS_SKB_MODE.
It is arguable that perhaps I should have required something like
this as part of the initial XDP feature merge.
I believe this is critical for two reasons:
1) Accessibility. More people can play with XDP with less
dependencies. Yes I know we have XDP support in virtio_net, but
that just creates another depedency for learning how to use this
facility.
I wrote this to make life easier for the XDP newbies.
2) As a model for what the expected semantics are. If there is a pure
generic core implementation, it serves as a semantic example for
driver folks adding XDP support.
One thing I have not tried to address here is the issue of
XDP_PACKET_HEADROOM, thanks to Daniel for spotting that. It seems
incredibly expensive to do a skb_cow(skb, XDP_PACKET_HEADROOM) or
whatever even if the XDP program doesn't try to push headers at all.
I think we really need the verifier to somehow propagate whether
certain XDP helpers are used or not.
v5:
- Handle both negative and positive offset after running prog
- Fix mac length in XDP_TX case (Alexei)
- Use rcu_dereference_protected() in free_netdev (kbuild test robot)
v4:
- Fix MAC header adjustmnet before calling prog (David Ahern)
- Disable LRO when generic XDP is installed (Michael Chan)
- Bypass qdisc et al. on XDP_TX and record the event (Alexei)
- Do not perform generic XDP on reinjected packets (DaveM)
v3:
- Make sure XDP program sees packet at MAC header, push back MAC
header if we do XDP_TX. (Alexei)
- Elide GRO when generic XDP is in use. (Alexei)
- Add XDP_FLAG_SKB_MODE flag which the user can use to request generic
XDP even if the driver has an XDP implementation. (Alexei)
- Report whether SKB mode is in use in rtnl_xdp_fill() via XDP_FLAGS
attribute. (Daniel)
v2:
- Add some "fall through" comments in switch statements based
upon feedback from Andrew Lunn
- Use RCU for generic xdp_prog, thanks to Johannes Berg.
Tested-by: Andy Gospodarek <andy@greyhouse.net>
Tested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Tested-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sizeof() when applied to a pointer typed expression gives the
size of the pointer, not that of the pointed data.
Fixes: b5a9ee7cf3 ("qed: Revise QM configuration")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When iterating through a map, we need to find a key that does not exist
in the map so map_get_next_key will give us the first key of the map.
This often requires a lot of guessing in production systems.
This patch makes map_get_next_key return the first key when the key
pointer in the parameter is NULL.
Signed-off-by: Teng Qin <qinteng@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
The error return falue form sock_fanout_open is -1, not zero. One test
case was checking for 0 instead of -1.
Tested: Built and tested in clean client.
Signed-off-by: Mike Maloney <maloney@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This code is unused and probably was unintentionally left while
moving completion queue mapping in submit function.
Signed-off-by: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Manish Chopra says:
====================
qed/qede: VF tunnelling support
With this series VFs can run vxlan/geneve/gre tunnels over it.
Please consider applying this series to "net-next"
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds hardware channel APIs support between
VF and PF for tunnelling configuration for the VFs.
According to that configuration VFs can run VXLAN/GENEVE/GRE
tunnels over it with tunnel features offloaded.
Using these APIs VF can also request for UDP ports configuration
to the PF, although PF and it's child VFs share the same port.
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for UDP ports in bulletin board
to notify UDP ports change to the VFs
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch configures UDP ports locally instead of
configuring them in deferred context which would be
helpful in synchronizing UDP ports configuration for VFs
which will be enabled in further patches.
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch disables tunnel offloads via ndo_features_check()
if given UDP port is not offloaded to hardware. This in turn
allows to run multiple tunnel interfaces using different UDP ports.
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch enables tunnel feature offloads based on hw configuration
at initialization time instead of enabling them always.
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch changes the tunnel APIs to use per tunnel
info instead of using bitmasks for all tunnels and also
uses single struct to hold the data to prepare multiple
variant of tunnel configuration ramrods to be sent to the hardware.
Signed-off-by: Manish Chopra <manish.chopra@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Guillaume Nault says:
====================
l2tp: add informations about l2tpeth interfaces in /sys
Patch #1 lets userspace retrieve the naming scheme of an l2tpeth
interface, using /sys/class/net/<iface>/name_assign_type.
Patch #2 adds the DEVTYPE field in /sys/class/net/<iface>/uevent so
that userspace can reliably know if a device is an l2tpeth interface.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Export type of l2tpeth interfaces to userspace
(/sys/class/net/<iface>/uevent).
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Acked-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Export naming scheme used when creating l2tpeth interfaces
(/sys/class/net/<iface>/name_assign_type). This let userspace know if
the device's name has been generated automatically or defined manually.
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Acked-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
per discussion at netconf/netdev:
When we have an action that is capable of branching (example a policer),
we can achieve a continuation of the action graph by programming a
"continue" where we find an exact replica of the same filter rule with a lower
priority and the remainder of the action graph. When you have 100s of thousands
of filters which require such a feature it gets very inefficient to do two
lookups.
This patch completes a leftover feature of action codes. Its time has come.
Example below where a user labels packets with a different skbmark on ingress
of a port depending on whether they have/not exceeded the configured rate.
This mark is then used to make further decisions on some egress port.
#rate control, very low so we can easily see the effect
sudo $TC actions add action police rate 1kbit burst 90k \
conform-exceed pipe/jump 2 index 10
# skbedit index 11 will be used if the user conforms
sudo $TC actions add action skbedit mark 11 ok index 11
# skbedit index 12 will be used if the user does not conform
sudo $TC actions add action skbedit mark 12 ok index 12
#lets bind the user ..
sudo $TC filter add dev $ETH parent ffff: protocol ip prio 8 u32 \
match ip dst 127.0.0.8/32 flowid 1:10 \
action police index 10 \
action skbedit index 11 \
action skbedit index 12
#run a ping -f and see what happens..
#
jhs@foobar:~$ sudo $TC -s filter ls dev $ETH parent ffff: protocol ip
filter pref 8 u32
filter pref 8 u32 fh 800: ht divisor 1
filter pref 8 u32 fh 800::800 order 2048 key ht 800 bkt 0 flowid 1:10 (rule hit 2800 success 1005)
match 7f000008/ffffffff at 16 (success 1005 )
action order 1: police 0xa rate 1Kbit burst 23440b mtu 2Kb action pipe/jump 2 overhead 0b
ref 2 bind 1 installed 207 sec used 122 sec
Action statistics:
Sent 84420 bytes 1005 pkt (dropped 0, overlimits 721 requeues 0)
backlog 0b 0p requeues 0
action order 2: skbedit mark 11 pass
index 11 ref 2 bind 1 installed 204 sec used 122 sec
Action statistics:
Sent 60564 bytes 721 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
action order 3: skbedit mark 12 pass
index 12 ref 2 bind 1 installed 201 sec used 122 sec
Action statistics:
Sent 23856 bytes 284 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
Not bad, about 28% non-conforming packets..
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
-----BEGIN PGP SIGNATURE-----
iQFHBAABCgAxFiEE4bay/IylYqM/npjQHv7KIOw4HPYFAlj/Cw8THG1rbEBwZW5n
dXRyb25peC5kZQAKCRAe/sog7Dgc9keOB/9dFFKUSqXEbevRCVj8Hc/tpmUnAYDP
xcIpz/8GRHPrkOx/tpqtzkAQjeiNzcrT5LDPLDoMSpubZDJjNTKGfcb5sOvGqK9P
IyY4dv0DO9/z1zxdpkK7CkR+g9Z3w9mEdQl2OS0yxbOXRPgX5Sl44Tp5xWgvJhOc
s60m/Y60PQ22CSee7EBYWCwvJPfLIdsr5AIM6wtbEveZU13afAFbyIqoP/97RZKF
sJ8NfGwQmRcD+AHw1nB/YfhNh4NEBE52IiBKf3zgC8Y8lDId/Wve/j/MnJGzeK48
eXPU3g1QaJTuEddn4xC0RRMKycR9klwfGJkY1cKSLtCWQxd1cfHKLdWY
=P474
-----END PGP SIGNATURE-----
Merge tag 'linux-can-next-for-4.12-20170425' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next
Marc Kleine-Budde says:
====================
pull-request: can-next 2017-04-25
this is a pull request of 21 patches for net-next/master.
There are 4 patches by Stephane Grosjean for the PEAK PCAN-PCIe FD
CAN-FD boards. The next 7 patches are by Mario Huettel, which add
support for M_CAN IP version >= v3.1.x to the m_can driver. A patch by
Remigiusz Kołłątaj adds support for the Microchip CAN BUS Analyzer. 8
patches by Oliver Hartkopp complete the initial CAN network namespace
support. Wei Yongjun's patch for the ti_hecc driver fixes the return
value check in the probe function.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
commit 4fbae7d83c ("ipvlan: Introduce l3s mode") added
registration of netfilter hooks via nf_register_hooks().
This API provides the illusion of 'global' netfilter hooks by placing the
hooks in all current and future network namespaces.
In case of ipvlan the hook appears to be only needed in the namespace
that contains the ipvlan master device (i.e., usually init_net), so
placing them in all namespaces is not needed.
This switches ipvlan driver to pernet operations, and then only registers
hooks in namespaces where a ipvlan master device is set to l3s mode.
Extra care has to be taken when the master device is moved to another
namespace, as we might have to 'move' the netfilter hooks too.
This is done by storing the namespace the ipvlan port was created in.
On REGISTER event, do (un)register operations in the old/new namespaces.
This will also allow removal of the nf_register_hooks() in a future patch.
Cc: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case of error, the function devm_ioremap_resource() returns ERR_PTR()
and never returns NULL. The NULL test in the return value check should
be replaced with IS_ERR().
Fixes: dabf54dd1c ("can: ti_hecc: Convert TI HECC driver to DT only driver")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Autoload the vcan module when a vcan instance is to be created by
'ip link add type vcan'
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Similar to the virtual ethernet driver veth, vxcan implements a
local CAN traffic tunnel between two virtual CAN network devices.
See Kconfig entry for details.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The CAN gateway was not implemented as per-net in the initial network
namespace support by Mario Kicherer (8e8cda6d73).
This patch enables the CAN gateway to be used in different namespaces.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The CAN_BCM protocol and its procfs entries were not implemented as per-net
in the initial network namespace support by Mario Kicherer (8e8cda6d73).
This patch adds the missing per-net functionality for the CAN BCM.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The statistics and its proc output was not implemented as per-net in the
initial network namespace support by Mario Kicherer (8e8cda6d73).
This patch adds the missing per-net statistics for the CAN subsystem.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
can_rx_alldev_list is a per-net data structure now. Remove it's definition
here and can_rx_dev_list too.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The namespace support for the CAN subsystem does not need any additional
memory. So when ".size = 0" there's no extra memory allocated by the system.
And therefore ".id" is obsolete too.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The can_rx_alldev_list is a per-net data structure now and allocated in
can_pernet_init(). Make sure the memory is free'd in can_pernet_exit() too.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
SocketCAN driver for Microchip CAN BUS Analyzer
(http://www.microchip.com/development-tools/)
Changes in v4:
- possible memory leak fixed in mcba_usb_write_bulk_callback
- LED support added
- failure handling in mcba_usb_probe improved
- C99 initializers for structs on stack
Changes in v3:
- improved/simplified CAN ID conversion
- functions for transmission of skb and cmd separated
- fixed/improved netif_stop_queue handling
- style/cosmetic corrections
Changes in v2:
- Termination handling reimplemented to fit new netlink API
(IFLA_CAN_TERMINATION)
- Bitrate handling reimplemented to fit new netlink API
(IFLA_CAN_BITRATE)
- CAN ID conversion refactored (changed from macro to inline functions)
- CAN DLC handling using get_can_dlc()
- Endianness handling for can_speed introduced
- Debugging removed
- Redundant error prints removed
- Style/cosmetic corrections (i.e. macro names, redefs, inits etc.)
Signed-off-by: Remigiusz Kołłątaj <remigiusz.kollataj@mobica.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
* Added defines for TX Event FIFO Element
* Adapted ndo_start_xmit function.
For versions >= v3.1.x it uses the TX FIFO to optimize the data
throughput. It stores the echo skb at the same index as in the
M_CAN's TX FIFO. The frame's message marker is set to this index.
This message marker is received in the TX Event FIFO after
the message was successfully transmitted. It is used to echo the
correct echo skb back to the network stack.
* Added m_can_echo_tx_event function. It reads all received
message markers in the TX Event FIFO and loops back the
corresponding echo skbs.
* ISR checks for new TX Event Entry interrupt for version >= 3.1.x.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Reviewed-by: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
* TX/TX Event FIFO sizes are configured for version >= v3.1.x
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
This patch adapts the initialization of the M_CAN. So it can be used
with all versions >= 3.0.x.
Changes:
* Added version element to m_can_priv structure to hold M_CAN version.
* Renamed bittiming structs for version 3.0.x
* Added new bittiming structs for version >= 3.1.x
* Function alloc_m_can_dev takes 2 new arguments. The TX FIFO size and the
base address of the module.
* Chip configuration for CAN_CTRLMODE_LOOPBACK is changed: Enabled
CCCR_MON bit. In combination with TEST_LBCK it activates the internal
loopback mode. Leaving CCCR_MON '0' results in external loopback mode.
* Clocks are temporarily enabled by platform_propbe function in order to
allow read access to the Core Release register and the Control Register.
Registers are used to detect M_CAN version and optional Non-ISO Feature.
Initialization of M_CAN for version >= 3.1.x:
* TX FIFO of M_CAN is used to transmit frames. The driver does not need to
stop the tx queue after each frame sent.
* Initialization of TX Event FIFO is added.
* NON-ISO is fixed for all M_CAN versions < 3.2.x. Version 3.2.x _can_ have
the NISO (Non-ISO) bit which can switch the mode of the M_CAN to Non-ISO
mode. This bit does not have to be writeable. Therefore it is checked.
If it is writable Non-ISO support is added to the controllers supported
CAN modes.
New Functions:
* Function to check the Core Release version. The read value determines the
behaviour of the driver.
* Function to check if the NISO bit for version >= 3.2.x is implemented.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Reviewed-by: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
* Updated register defines to newest M_CAN version (v3.2.1).
* Changed defines in the whole code.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Reviewed-by: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The virtual address of the device was printed. I removed it because it
leaks internal information.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
FIFO water marks disabled because the driver doesn't handle water mark
events.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Reviewed-by: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
* Disabled interrupt line 1. The driver didn't use it.
Signed-off-by: Mario Huettel <mario.huettel@gmx.net>
Tested-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
This patch adds the support of the PCAN-PCI Express FD boards made
by PEAK-System, for computers using the PCI Express slot.
The PCAN-PCI Express FD has one or two CAN FD channels, depending
on the model. A galvanic isolation of the CAN ports protects
the electronics of the card and the respective computer against
disturbances of up to 500 Volts. The PCAN-PCI Express FD can be operated
with ambient temperatures in a range of -40 to +85 °C.
Such boards run an extented version of the CAN-FD IP running into USB
CAN-FD interfaces from PEAK-System, so this patch adds several new commands
and their corresponding data types to the PEAK CAN-FD common definitions
header file too.
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The CAN-FD IP from PEAK-System runs into several kinds of PC CAN-FD
interfaces. Up to now, only the USB CAN-FD adapters were supported by
the Kernel. In order to prepare the adding of some new non-USB CAN-FD
interfaces, this patch moves - and rename - the IP definitions file
from its private (usb) sub-directory into a - newly created - CAN specific
one.
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Fixes the usage of the const qualifier in the memory pointer arguments
of the declared inline functions. By changing the line containing "const",
this patch also changes the name of the arg into a more usual one.
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
This patch fixes the wrong usage of a specific USB data type into a common
header file. This common header file is intended to define the common data
types and values that define access to the PEAK-System CAN-FD IP, whatever
the PC interface is.
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Willem de Bruijn says:
====================
virtio-net tx napi
Add napi for virtio-net transmit completion processing.
Changes:
v2 -> v3:
- convert __netif_tx_trylock to __netif_tx_lock on tx napi poll
ensure that the handler always cleans, to avoid deadlock
- unconditionally clean in start_xmit
avoid adding an unnecessary "if (use_napi)" branch
- remove virtqueue_disable_cb in patch 5/5
a noop in the common event_idx based loop
- document affinity_hint_set constraint
v1 -> v2:
- disable by default
- disable unless affinity_hint_set
because cache misses add up to a third higher cycle cost,
e.g., in TCP_RR tests. This is not limited to the patch
that enables tx completion cleaning in rx napi.
- use trylock to avoid contention between tx and rx napi
- keep interrupts masked during xmit_more (new patch 5/5)
this improves cycles especially for multi UDP_STREAM, which
does not benefit from cleaning tx completions on rx napi.
- move free_old_xmit_skbs (new patch 3/5)
to avoid forward declaration
not changed:
- deduplicate virnet_poll_tx and virtnet_poll_txclean
they look similar, but have differ too much to make it
worthwhile.
- delay netif_wake_subqueue for more than 2 + MAX_SKB_FRAGS
evaluated, but made no difference
- patch 1/5
RFC -> v1:
- dropped vhost interrupt moderation patch:
not needed and likely expensive at light load
- remove tx napi weight
- always clean all tx completions
- use boolean to toggle tx-napi, instead
- only clean tx in rx if tx-napi is enabled
- then clean tx before rx
- fix: add missing braces in virtnet_freeze_down
- testing: add 4KB TCP_RR + UDP test results
Based on previous patchsets by Jason Wang:
[RFC V7 PATCH 0/7] enable tx interrupts for virtio-net
http://lkml.iu.edu/hypermail/linux/kernel/1505.3/00245.html
Before commit b0c39dbdc2 ("virtio_net: don't free buffers in xmit
ring") the virtio-net driver would free transmitted packets on
transmission of new packets in ndo_start_xmit and, to catch the edge
case when no new packet is sent, also in a timer at 10HZ.
A timer can cause long stalls. VIRTIO_F_NOTIFY_ON_EMPTY avoids stalls
due to low free descriptor count. It does not address a stalls due to
low socket SO_SNDBUF. Increasing timer frequency decreases that stall
time, but increases interrupt rate and, thus, cycle count.
Currently, with no timer, packets are freed only at ndo_start_xmit.
Latency of consume_skb is now unbounded. To avoid a deadlock if a sock
reaches SO_SNDBUF, packets are orphaned on tx. This breaks TCP small
queues.
Reenable TCP small queues by removing the orphan. Instead of using a
timer, convert the driver to regular tx napi. This does not have the
unresolved stall issue and does not have any frequency to tune.
By keeping interrupts enabled by default, napi increases tx
interrupt rate. VIRTIO_F_EVENT_IDX avoids sending an interrupt if
one is already unacknowledged, so makes this more feasible today.
Combine that with an optimization that brings interrupt rate
back in line with the existing version for most workloads:
Tx completion cleaning on rx interrupts elides most explicit tx
interrupts by relying on the fact that many rx interrupts fire.
Tested by running {1, 10, 100} {TCP, UDP} STREAM, RR, 4K_RR benchmarks
from a guest to a server on the host, on an x86_64 Haswell. The guest
runs 4 vCPUs pinned to 4 cores. vhost and the test server are
pinned to a core each.
All results are the median of 5 runs, with variance well < 10%.
Used neper (github.com/google/neper) as test process.
Napi increases single stream throughput, but increases cycle cost.
The optimizations bring this down. The previous patchset saw a
regression with UDP_STREAM, which does not benefit from cleaning tx
interrupts in rx napi. This regression is now gone for 10x, 100x.
Remaining difference is higher 1x TCP_STREAM, lower 1x UDP_STREAM.
The latest results are with process, rx napi and tx napi affine to
the same core. All numbers are lower than the previous patchset.
upstream napi
TCP_STREAM:
1x:
Mbps 27816 39805
Gcycles 274 285
10x:
Mbps 42947 42531
Gcycles 300 296
100x:
Mbps 31830 28042
Gcycles 279 269
TCP_RR Latency (us):
1x:
p50 21 21
p99 27 27
Gcycles 180 167
10x:
p50 40 39
p99 52 52
Gcycles 214 211
100x:
p50 281 241
p99 411 337
Gcycles 218 226
TCP_RR 4K:
1x:
p50 28 29
p99 34 36
Gcycles 177 167
10x:
p50 70 71
p99 85 134
Gcycles 213 214
100x:
p50 442 611
p99 802 785
Gcycles 237 216
UDP_STREAM:
1x:
Mbps 29468 26800
Gcycles 284 293
10x:
Mbps 29891 29978
Gcycles 285 312
100x:
Mbps 30269 30304
Gcycles 318 316
UDP_RR:
1x:
p50 19 19
p99 23 23
Gcycles 180 173
10x:
p50 35 40
p99 54 64
Gcycles 245 237
100x:
p50 234 286
p99 484 473
Gcycles 224 214
Note that GSO is enabled, so 4K RR still translates to one packet
per request.
Lower throughput at 100x vs 10x can be (at least in part)
explained by looking at bytes per packet sent (nstat). It likely
also explains the lower throughput of 1x for some variants.
upstream:
N=1 bytes/pkt=16581
N=10 bytes/pkt=61513
N=100 bytes/pkt=51558
at_rx:
N=1 bytes/pkt=65204
N=10 bytes/pkt=65148
N=100 bytes/pkt=56840
====================
Acked-by: Michael S. Tsirkin <mst@redhat.com>
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: Willem de Bruijn <willemb@google.com>
Signed-off-by: David 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: Willem de Bruijn <willemb@google.com>
Signed-off-by: David 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: Willem de Bruijn <willemb@google.com>
Signed-off-by: David 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: Willem de Bruijn <willemb@google.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David 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: Willem de Bruijn <willemb@google.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Doing a full 64-bit decomposition is really stupid especially for
simple values like 0 and -1.
But if we are going to optimize this, go all the way and try for all 2
and 3 instruction sequences not requiring a temporary register as
well.
First we do the easy cases where it's a zero or sign extended 32-bit
number (sethi+or, sethi+xor, respectively).
Then we try to find a range of set bits we can load simply then shift
up into place, in various ways.
Then we try negating the constant and see if we can do a simple
sequence using that with a xor at the end. (f.e. the range of set
bits can't be loaded simply, but for the negated value it can)
The final optimized strategy involves 4 instructions sequences not
needing a temporary register.
Otherwise we sadly fully decompose using a temp..
Example, from ALU64_XOR_K: 0x0000ffffffff0000 ^ 0x0 = 0x0000ffffffff0000:
0000000000000000 <foo>:
0: 9d e3 bf 50 save %sp, -176, %sp
4: 01 00 00 00 nop
8: 90 10 00 18 mov %i0, %o0
c: 13 3f ff ff sethi %hi(0xfffffc00), %o1
10: 92 12 63 ff or %o1, 0x3ff, %o1 ! ffffffff <foo+0xffffffff>
14: 93 2a 70 10 sllx %o1, 0x10, %o1
18: 15 3f ff ff sethi %hi(0xfffffc00), %o2
1c: 94 12 a3 ff or %o2, 0x3ff, %o2 ! ffffffff <foo+0xffffffff>
20: 95 2a b0 10 sllx %o2, 0x10, %o2
24: 92 1a 60 00 xor %o1, 0, %o1
28: 12 e2 40 8a cxbe %o1, %o2, 38 <foo+0x38>
2c: 9a 10 20 02 mov 2, %o5
30: 10 60 00 03 b,pn %xcc, 3c <foo+0x3c>
34: 01 00 00 00 nop
38: 9a 10 20 01 mov 1, %o5 ! 1 <foo+0x1>
3c: 81 c7 e0 08 ret
40: 91 eb 40 00 restore %o5, %g0, %o0
Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg
Vasanthakumar Thiagarajan
Wei Yongjun
David S. Miller
Teng Qin
Mike Maloney
Ivan Khoronzhuk
Chopra, Manish
Guillaume Nault
Jamal Hadi Salim
Florian Westphal
Oliver Hartkopp
Remigiusz Kołłątaj
Mario Huettel
Stephane Grosjean
Willem de Bruijn