Marek Lindner [Fri, 30 Sep 2016 13:21:05 +0000 (15:21 +0200)]
batman-adv: retrieve B.A.T.M.A.N. V WiFi neighbor stats from real interface
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
[sven.eckelmann@open-mesh.com: re-add batadv_get_real_netdev to take rtnl
semaphore for batadv_get_real_netdevice] Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Marek Lindner [Fri, 30 Sep 2016 13:21:04 +0000 (15:21 +0200)]
batman-adv: additional checks for virtual interfaces on top of WiFi
In a few situations batman-adv tries to determine whether a given interface
is a WiFi interface to enable specific WiFi optimizations. If the interface
batman-adv has been configured with is a virtual interface (e.g. VLAN) it
would not be properly detected as WiFi interface and thus not benefit from
the special WiFi treatment.
This patch changes that by peeking under the hood whenever a virtual
interface is in play.
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
[sven.eckelmann@open-mesh.com: integrate in wifi_flags caching, retrieve
namespace of link interface] Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Fri, 30 Sep 2016 13:21:03 +0000 (15:21 +0200)]
batman-adv: Cache the type of wifi device for each hardif
batman-adv is requiring the type of wifi device in different contexts. Some
of them can take the rtnl semaphore and some of them already have the
semaphore taken. But even others don't allow that the semaphore will be
taken.
The data has to be retrieved when the hardif is added to batman-adv because
some of the wifi information for an hardif will only be available with rtnl
lock. It can then be cached in the batadv_hard_iface and the functions
is_wifi_netdev and is_cfg80211_netdev can just compare the correct bits
without imposing extra locking requirements.
Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Marek Lindner [Fri, 30 Sep 2016 13:21:02 +0000 (15:21 +0200)]
batman-adv: refactor wifi interface detection
The ELP protocol requires cfg80211 to auto-detect the WiFi througput
to a given neighbor. Use batadv_is_cfg80211_netdev() to determine
whether or not an interface is eligible.
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch> Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Fri, 30 Sep 2016 13:21:00 +0000 (15:21 +0200)]
batman-adv: Return non-const ptr in batadv_getlink_net
The returned net_namespace of batadv_getlink_net may be used with functions
that potentially modify the struct. Thus it must return the pointer as
non-const like rtnl_link_ops::get_link_net does.
Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sat, 6 Aug 2016 15:04:23 +0000 (17:04 +0200)]
batman-adv: Reject unicast packet with zero/mcast dst address
An unicast batman-adv packet cannot be transmitted to a multicast or zero
mac address. So reject incoming packets which still have these classes of
addresses as destination mac address in the outer ethernet header.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sat, 6 Aug 2016 15:04:22 +0000 (17:04 +0200)]
batman-adv: Disallow zero and mcast src address for mgmt frames
The routing check for management frames is validating the source mac
address in the outer ethernet header. It rejects every source mac address
which is a broadcast address. But it also has to reject the zero-mac
address and multicast mac addresses.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sat, 6 Aug 2016 15:04:21 +0000 (17:04 +0200)]
batman-adv: Disallow mcast src address for data frames
The routing checks are validating the source mac address of the outer
ethernet header. They reject every source mac address which is a broadcast
address. But they also have to reject any multicast mac addresses.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
[sw@simonwunderlich.de: fix commit message typo] Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
No caller of batadv_send_skb_to_orig is expecting the results to be -1
(-EPERM) anymore when the skbuff was not consumed. They will instead expect
that the skbuff is always consumed. Having such return code filter is
therefore not needed anymore.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sun, 17 Jul 2016 19:04:04 +0000 (21:04 +0200)]
batman-adv: Consume skb in receive handlers
Receiving functions in Linux consume the supplied skbuff. Doing the same in
the batadv_rx_handler functions makes the behavior more similar to the rest
of the Linux network code.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sun, 17 Jul 2016 19:04:03 +0000 (21:04 +0200)]
batman-adv: Consume skb in batadv_send_skb_to_orig
Sending functions in Linux consume the supplied skbuff. Doing the same in
batadv_send_skb_to_orig avoids the hack of returning -1 (-EPERM) to signal
the caller that he is responsible for cleaning up the skb.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sun, 17 Jul 2016 19:04:02 +0000 (21:04 +0200)]
batman-adv: Consume skb in batadv_frag_send_packet
Sending functions in Linux consume the supplied skbuff. Doing the same in
batadv_frag_send_packet avoids the hack of returning -1 (-EPERM) to signal
the caller that he is responsible for cleaning up the skb.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sun, 17 Jul 2016 19:04:00 +0000 (21:04 +0200)]
batman-adv: use consume_skb for non-dropped packets
kfree_skb assumes that an skb is dropped after an failure and notes that.
consume_skb should be used in non-failure situations. Such information is
important for dropmonitor netlink which tells how many packets were dropped
and where this drop happened.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Linus Lüssing [Sun, 7 Aug 2016 10:34:19 +0000 (12:34 +0200)]
batman-adv: Simple (re)broadcast avoidance
With this patch, (re)broadcasting on a specific interfaces is avoided:
* No neighbor: There is no need to broadcast on an interface if there
is no node behind it.
* Single neighbor is source: If there is just one neighbor on an
interface and if this neighbor is the one we actually got this
broadcast packet from, then we do not need to echo it back.
* Single neighbor is originator: If there is just one neighbor on
an interface and if this neighbor is the originator of this
broadcast packet, then we do not need to echo it back.
Goodies for BATMAN V:
("Upgrade your BATMAN IV network to V now to get these for free!")
Thanks to the split of OGMv1 into two packet types, OGMv2 and ELP
that is, we can now apply the same optimizations stated above to OGMv2
packets, too.
Furthermore, with BATMAN V, rebroadcasts can be reduced in certain
multi interface cases, too, where BATMAN IV cannot. This is thanks to
the removal of the "secondary interface originator" concept in BATMAN V.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue> Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Linus Lüssing [Sat, 6 Aug 2016 20:23:16 +0000 (22:23 +0200)]
batman-adv: Use own timer for multicast TT and TVLV updates
Instead of latching onto the OGM period, this patch introduces a worker
dedicated to multicast TT and TVLV updates.
The reasoning is, that upon roaming especially the translation table
should be updated timely to minimize connectivity issues.
With BATMAN V, the idea is to greatly increase the OGM interval to
reduce overhead. Unfortunately, right now this could lead to
a bad user experience if multicast traffic is involved.
Therefore this patch introduces a fixed 500ms update interval for
multicast TT entries and the multicast TVLV.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue> Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Linus Lüssing [Sun, 7 Aug 2016 10:34:18 +0000 (12:34 +0200)]
batman-adv: Remove unused skb_reset_mac_header()
During broadcast queueing, the skb_reset_mac_header() sets the skb
to a place invalid for a MAC header, pointing right into the
batman-adv broadcast packet. Luckily, no one seems to actually use
eth_hdr(skb) afterwards until batadv_send_skb_packet() resets the
header to a valid position again.
Therefore removing this unnecessary, weird skb_reset_mac_header()
call.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue> Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Linus Lüssing [Sat, 6 Aug 2016 20:23:15 +0000 (22:23 +0200)]
batman-adv: Remove unnecessary lockdep in batadv_mcast_mla_list_free
batadv_mcast_mla_list_free() just frees some leftovers of a local feast
in batadv_mcast_mla_update(). No lockdep needed as it has nothing to do
with bat_priv->mcast.mla_list.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue> Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Sven Eckelmann [Sat, 29 Oct 2016 08:13:47 +0000 (10:13 +0200)]
batman-adv: Introduce missing headers for genetlink restructure
Fixes: 56989f6d8568 ("genetlink: mark families as __ro_after_init") Fixes: 2ae0f17df1cd ("genetlink: use idr to track families") Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
David S. Miller [Sat, 29 Oct 2016 20:19:41 +0000 (16:19 -0400)]
Merge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queue
Jeff Kirsher says:
====================
40GbE Intel Wired LAN Driver Updates 2016-10-28
This series contains updates to i40e and i40evf only.
Carolyn provides a couple of fixes, first resolving a problem in the
client interface that was causing random stack traces in the RDMA driver
which was due to a timing related NULL pointer dereference. Fixed a
problem where it could take a very long time to print the link down
notification, by changing how often we update link info from firmware.
Alex provides a number of changes, first is a re-write of the bust wait
loop in the Flow Director transmit function to reduce code size. Cleans
up unused code in favor of the same functionality which can be inlined.
Dropped the functionality for SCTP since we cannot currently support it.
Cleans up redundant code in the receive clean-up path. Finally cleaned
up the convoluted configuration for how the driver handled the debug
flags contained in msg_level.
Filip fixes an incorrect bit mask which was being used for testing the
"get link status". Cleaned up a workaround that is no longer needed
for production NICs and was causing frames to pass while disregarding
the VLAN tagging.
Mitch brings another fix for the client interface supporting the VF RDMA
driver to allow clients to recover from reset by re-opening existing
clients.
Alan fixes a bug in which a "perfect storm" can occur and cause interrupts
to fail to be correctly affinitized.
Lihong fixes a confusing dmesg reported when users were using ethtool -L
option.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Moves the function documentation to a single comment and renames all
helpers names in the list to conform to the bpf_ prefix notation so
they can be greped in the kernel source.
Signed-off-by: Thomas Graf <tgraf@suug.ch> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern [Wed, 26 Oct 2016 20:21:33 +0000 (13:21 -0700)]
net: dev: Fix non-RCU based lower dev walker
netdev_walk_all_lower_dev is not properly walking the lower device
list. Commit 1a3f060c1a47 made netdev_walk_all_lower_dev similar
to netdev_walk_all_upper_dev_rcu and netdev_walk_all_lower_dev_rcu
but failed to update its netdev_next_lower_dev iterator. This patch
fixes that.
Fixes: 1a3f060c1a47 ("net: Introduce new api for walking upper and
lower devices") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Wed, 26 Oct 2016 16:27:57 +0000 (09:27 -0700)]
tcp/dccp: drop SYN packets if accept queue is full
Per listen(fd, backlog) rules, there is really no point accepting a SYN,
sending a SYNACK, and dropping the following ACK packet if accept queue
is full, because application is not draining accept queue fast enough.
This behavior is fooling TCP clients that believe they established a
flow, while there is nothing at server side. They might then send about
10 MSS (if using IW10) that will be dropped anyway while server is under
stress.
Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Neal Cardwell <ncardwell@google.com> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Thomas Graf [Wed, 26 Oct 2016 08:53:16 +0000 (10:53 +0200)]
netlink: Add nla_memdup() to wrap kmemdup() use on nlattr
Wrap several common instances of:
kmemdup(nla_data(attr), nla_len(attr), GFP_KERNEL);
Signed-off-by: Thomas Graf <tgraf@suug.ch> Acked-by: Johannes Berg <johannes@sipsolutions.net> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Arnd Bergmann [Tue, 25 Oct 2016 15:53:22 +0000 (17:53 +0200)]
net: ip, diag: include net/inet_sock.h
The newly added raw_diag.c fails to build in some configurations
unless we include this header:
In file included from net/ipv4/raw_diag.c:6:0:
include/net/raw.h:71:21: error: field 'inet' has incomplete type
net/ipv4/raw_diag.c: In function 'raw_diag_dump':
net/ipv4/raw_diag.c:166:29: error: implicit declaration of function 'inet_sk'
Fixes: 432490f9d455 ("net: ip, diag -- Add diag interface for raw sockets") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Gao Feng [Tue, 25 Oct 2016 14:26:09 +0000 (22:26 +0800)]
driver: tun: Move tun check into the block of TUNSETIFF condition check
When cmd is TUNSETIFF and tun is not null, the original codes go ahead,
then reach the default case of switch(cmd) and set the ret is -EINVAL.
It is not clear for readers.
Now move the tun check into the block of TUNSETIFF condition check, and
return -EEXIST instead of -EINVAL when the tfile already owns one tun.
Signed-off-by: Gao Feng <fgao@ikuai8.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Aaron Conole [Tue, 25 Oct 2016 20:12:12 +0000 (16:12 -0400)]
virtio-net: Update the mtu code to match virtio spec
The virtio committee recently ratified a change, VIRTIO-152, which
defines the mtu field to be 'max' MTU, not simply desired MTU.
This commit brings the virtio-net device in compliance with VIRTIO-152.
Additionally, drop the max_mtu branch - it cannot be taken since the u16
returned by virtio_cread16 will never exceed the initial value of
max_mtu.
Signed-off-by: Aaron Conole <aconole@redhat.com> Acked-by: "Michael S. Tsirkin" <mst@redhat.com> Acked-by: Jarod Wilson <jarod@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Alexander Duyck [Fri, 30 Sep 2016 12:21:46 +0000 (08:21 -0400)]
i40e: Clean up handling of msglevel flags and debug parameter
So the i40e driver had a really convoluted configuration for how to handle
the debug flags contained in msg_level. Part of the issue is that the
driver has its own 32 bit mask that it was using to track a separate set of
debug features. From what I can tell it was trying to use the upper 4 bits
to determine if the value was meant to represent a bit-mask or the numeric
value provided by debug level.
What this patch does is clean this up by compressing those 4 bits into bit
31, as a result we just have to perform a check against the value being
negative to determine if we are looking at a debug level (positive), or a
debug mask (negative). The debug level will populate the msg_level, and
the debug mask will populate the debug_mask in the hardware struct.
I added similar logic for ethtool. If the value being provided has bit 31
set we assume the value being provided is a debug mask, otherwise we assume
it is a msg_enable mask. For displaying we only provide the msg_enable,
and if debug_mask is in use we will print it to the dmesg log.
Lastly I removed the debugfs interface. It is redundant with what we
already have in ethtool and really doesn't belong anyway.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
David Ertman [Fri, 30 Sep 2016 08:36:21 +0000 (01:36 -0700)]
i40e: Fix bit logic error in failure case
Patch a036244c0686 "i40e: Fix kernel panic on enable/disable LLDP"
introduced an error in bit logic.
Originally this bit manipulation was meant to clear two bits to indicate
that DCB was not enabled or capable. An "&" was incorrectly used instead
of an "|" bit operator to combine the two bitmasks into one. This also
created a static checker error since the resultant code was a no-op.
This patch fixes the error by using the correct bit-wise operator.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Filip Sadowski [Tue, 27 Sep 2016 18:28:53 +0000 (11:28 -0700)]
i40e: Removal of workaround for simple MAC address filter deletion
This is code refactoring. This patch removes the workaround which deleted
a default MAC filter added by the firmware when the interface was brought
up. This filter caused frames to pass disregarding the VLAN tagging.
It used to be automatically applied after reset in pre-SRA FW versions.
This workaround is not needed in production NICs and hence can be removed.
Change-ID: I129fe1aae1f17b5a224c9b29a996d916aa1be1ec Signed-off-by: Filip Sadowski <filip.sadowski@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes a problem where it could take a very
long time (>100 msec) to print the link down notification.
This problem is fixed by changing how often we update link
info from fw, when link is down. Without this patch, it can
take over 100msec to notify user link is down.
Change-ID: Ib876eb30834c7080792becd13ee093b9cbb35d78 Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Alexander Duyck [Tue, 27 Sep 2016 18:28:50 +0000 (11:28 -0700)]
i40e: Drop redundant Rx descriptor processing code
This patch cleans up several pieces of redundant code in the Rx clean-up
paths.
The first bit is that hdr_addr and the status_err_len portions of the Rx
descriptor represent the same value. As such there is no point in setting
them to 0 before setting them to 0. I'm dropping the second spot where we
are updating the value to 0 so that we only have 1 write for this value
instead of 2.
The second piece is the checking for the DD bit in the packet. We only
need to check for a non-zero value for the status_err_len because if the
device is done with the descriptor it will have written something back and
the DD is just one piece of it. In addition I have moved the reading of
the Rx descriptor bits related to rx_ptype down so that they are actually
below the dma_rmb() call so that we are guaranteed that we don't have any
funky 64b on 32b calls causing any ordering issues.
Change-ID: I256e44a025d3c64a7224aaaec37c852bfcb1871b Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Lihong Yang [Tue, 27 Sep 2016 18:28:48 +0000 (11:28 -0700)]
i40e: fix confusing dmesg info for ethtool -L option
Ethtool -L option with the combined parameter is for changing the number of
multi-purpose channels of the specified network device. The pre-set maximum
for the combined channels is cpu dependent. Currently, for an i40e device,
when the user sets a value between 64 and the maximum that the cpu can
support for the combined parameter, the i40e driver displays the confusing
info in dmesg to only show 64 as the RSS count regardless of what the
accepted user input is as long as it is larger than 64.
This patch fixes the message in the i40e driver when the user uses
ethtool -L to change the number of the combined channels to consistently
display the user requested value if it is valid and accepted by ethtool.
Change-ID: Ia80a68bc844b779a49e0f76e7d3dcc915032d9af Signed-off-by: Lihong Yang <lihong.yang@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Joe Perches [Tue, 27 Sep 2016 03:17:01 +0000 (20:17 -0700)]
i40e: Make struct i40e_stats const
Move some data to text
$ size drivers/net/ethernet/intel/i40e/i40e_ethtool.o*
text data bss dec hex filename
25012 0 32 25044 61d4 drivers/net/ethernet/intel/i40e/i40e_ethtool.o.new
22868 2120 32 25020 61bc drivers/net/ethernet/intel/i40e/i40e_ethtool.o.old
Signed-off-by: Joe Perches <joe@perches.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Alan Brady [Wed, 14 Sep 2016 23:24:38 +0000 (16:24 -0700)]
i40e/i40evf: fix interrupt affinity bug
There exists a bug in which a 'perfect storm' can occur and cause
interrupts to fail to be correctly affinitized. This causes unexpected
behavior and has a substantial impact on performance when it happens.
The bug occurs if there is heavy traffic, any number of CPUs that have
an i40e interrupt are pegged at 100%, and the interrupt afffinity for
those CPUs is changed. Instead of moving to the new CPU, the interrupt
continues to be polled while there is heavy traffic.
The bug is most readily realized as the driver is first brought up and
all interrupts start on CPU0. If there is heavy traffic and the
interrupt starts polling before the interrupt is affinitized, the
interrupt will be stuck on CPU0 until traffic stops. The bug, however,
can also be wrought out more simply by affinitizing all the interrupts
to a single CPU and then attempting to move any of those interrupts off
while there is heavy traffic.
This patch fixes the bug by registering for update notifications from
the kernel when the interrupt affinity changes. When that fires, we
cache the intended affinity mask. Then, while polling, if the cpu is
pegged at 100% and we failed to clean the rings, we check to make sure
we have the correct affinity and stop polling if we're firing on the
wrong CPU. When the kernel successfully moves the interrupt, it will
start polling on the correct CPU. The performance impact is minimal
since the only time this section gets executed is when performance is
already compromised by the CPU.
Change-ID: I4410a880159b9dba1f8297aa72bef36dca34e830 Signed-off-by: Alan Brady <alan.brady@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Group together the minimum set of offload capabilities that are always
supported by VF in base mode. This define would be used by PF to make
sure VF in base mode gets minimum of base capabilities .
Change-ID: Id5e8f22ba169c8f0a38d22fc36b2cb531c02582c Signed-off-by: Preethi Banala <preethi.banala@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Mitch Williams [Wed, 14 Sep 2016 23:24:36 +0000 (16:24 -0700)]
i40e: reopen client after reset
Allow the client interface to reopen existing clients if they were
closed. This allows clients to recover from reset, which is essential
for supporting VF RDMA. In one instance, the driver was not clearing the
open bit when the client was closed. Add the code to clear this bit so
that the state is accurate and the driver will not attempt to reopen
already-open clients. Remove the ref_cnt variable; it was just getting
in the way and was not being used consistently.
Change-ID: Ic71af4553b096963ac0c56a997f887c9a4ed162d Signed-off-by: Mitch Williams <mitch.a.williams@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Alexander Duyck [Wed, 14 Sep 2016 23:24:35 +0000 (16:24 -0700)]
i40e: Drop code for unsupported flow types
We cannot currently support SCTP in the hardware, and IPV4_FLOW is not used
anywhere by the software so we can go through and drop the functionality
related to these two flow types.
In addition we cannot support masking based on the protocol value so if the
user is expecting a value other than TCP or UDP we should simply return an
error rather then trying to allocate a filter for a rule that will only
partially match what the user requested.
Change-ID: I10d52bb97d8104d76255fe244551814ff9531a63 Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Alexander Duyck [Wed, 14 Sep 2016 23:24:34 +0000 (16:24 -0700)]
i40e: Remove unused function i40e_vsi_lookup
The function is not used so there is no need to carry it forward. I have
plans to add a slightly different function that can be inlined to handle
the same kind of functionality.
Change-ID: Ie2dfcb189dc75e5fbc156bac23003e3b4210ae0f Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Filip Sadowski [Wed, 14 Sep 2016 23:24:33 +0000 (16:24 -0700)]
i40e: Bit test mask correction
Incorrect bit mask was used for testing "get link status" response.
Instead of I40E_AQ_LSE_ENABLE (which is actually 0x03) it most probably
should be I40E_AQ_LSE_IS_ENABLED (which is defined as 0x01).
Change-ID: Ia199142906720507f847de3a33a25c61a9781b2f Signed-off-by: Filip Sadowski <filip.sadowski@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Alexander Duyck [Wed, 14 Sep 2016 23:24:32 +0000 (16:24 -0700)]
i40e: Rewrite Flow Director busy wait loop
We can reorder the busy wait loop at the start of the Flow Director
transmit function to reduce the overall code size while still retaining the
same functionality. As such I am taking advantage of the opportunity to do
so.
Change-ID: I34c403ca001953c6ac9816e65d5305e73d869026 Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes a problem in the client interface that
was causing random stack traces in RDMA driver load and
unload tests. This patch fixes the problem by checking
for an existing client before trying to open it. Without
this patch, there is a timing related null pointer deref.
Change-ID: Ib73d30671a27f6f9770dd53b3e5292b88d6b62da Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Jiri Pirko [Thu, 27 Oct 2016 13:12:59 +0000 (15:12 +0200)]
mlxsw: Move PCI id table definitions into driver modules
So far, mlxsw_pci.ko is the module that registers PCI table for all
drivers (spectrum and switchx2). That is problematic for example with
dracut. Since mlxsw_spectrum.ko and mlxsw_switchx2.ko are loaded
dynamically from within mlxsw_core.ko, dracut does not have track of
them and avoids them from being included in initramfs.
So make this in an ordinary way and define the PCI tables in individual
driver modules, so it can be properly loaded and included in dracut
initramfs image. As a side effect, this patch could remove no longer
necessary driver "kind" strings which were used to link PCI ids with
individual mlxsw drivers.
Suggested-by: Ivan Vecera <ivecera@redhat.com> Tested-by: Ivan Vecera <ivecera@redhat.com> Signed-off-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Acked-by: Ivan Vecera <ivecera@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
It's very critical to optimize the exit path for network namespaces,
because they are destroyed under net_mutex and many namespaces can be
destroyed for one iteration.
v2: use dev_set_uevent_suppress()
Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Andrei Vagin <avagin@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Stefan Richter [Mon, 24 Oct 2016 12:42:26 +0000 (14:42 +0200)]
ethernet: fix min/max MTU typos
Fixes: d894be57ca92('ethernet: use net core MTU range checking in more drivers') CC: Jarod Wilson <jarod@redhat.com> CC: Thomas Falcon <tlfalcon@linux.vnet.ibm.com> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Acked-by: Jarod Wilson <jarod@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Thu, 27 Oct 2016 20:16:14 +0000 (16:16 -0400)]
Merge branch 'genetlink-improvements'
Johannes Berg says:
====================
genetlink improvements
This series contains some generic netlink improvements, making
the API safer to use, and making the function pointers in the
family struct safer by allowing it to be __ro_after_init.
The first patch, introducing genl_family_attrbuf(), just ensures
that the users of family->attrbuf aren't actually racy, but making
them use the indirection function for obtaining a reference and
checking that the context can actually do so.
The second patch removes the more or less broken ability to have
a static family ID, the three IDs that need to be static because
it's simply needed (genl controller), or due to old API misused.
Everything else couldn't be static anyway, or could fail when the
family is registered, if somebody else already got a static ID.
The third patch statically initializes the families, mostly to save
some code. I wrote this initially because I thought I could make
them all const, but that ends up being very inefficient (it would
require always doing some kind of family -> id lookup), so now it's
just here because I had it already and it reduces the code size.
The fourth patch then, finally, lays the groundwork for what I had
really wanted - now with __ro_after_init instead of const; I remove
code there to do the ID->family hash table mapping in genetlink and
use IDR instead to both allocate and map the IDs, which again ends
up saving some code size.
Finally, the fifth patch updates all families, as it turns out, no
families exist that really dynamically register/unregister. This
last patch should perhaps be split up, I could submit it for each
subsystem separately, but it'd depend on the second and third to
go in first, so would take a while. I can do that though, if that
seems better to you.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg [Mon, 24 Oct 2016 12:40:05 +0000 (14:40 +0200)]
genetlink: mark families as __ro_after_init
Now genl_register_family() is the only thing (other than the
users themselves, perhaps, but I didn't find any doing that)
writing to the family struct.
In all families that I found, genl_register_family() is only
called from __init functions (some indirectly, in which case
I've add __init annotations to clarifly things), so all can
actually be marked __ro_after_init.
This protects the data structure from accidental corruption.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg [Mon, 24 Oct 2016 12:40:04 +0000 (14:40 +0200)]
genetlink: use idr to track families
Since generic netlink family IDs are small integers, allocated
densely, IDR is an ideal match for lookups. Replace the existing
hand-written hash-table with IDR for allocation and lookup.
This lets the families only be written to once, during register,
since the list_head can be removed and removal of a family won't
cause any writes.
It also slightly reduces the code size (by about 1.3k on x86-64).
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg [Mon, 24 Oct 2016 12:40:02 +0000 (14:40 +0200)]
genetlink: no longer support using static family IDs
Static family IDs have never really been used, the only
use case was the workaround I introduced for those users
that assumed their family ID was also their multicast
group ID.
Additionally, because static family IDs would never be
reserved by the generic netlink code, using a relatively
low ID would only work for built-in families that can be
registered immediately after generic netlink is started,
which is basically only the control family (apart from
the workaround code, which I also had to add code for so
it would reserve those IDs)
Thus, anything other than GENL_ID_GENERATE is flawed and
luckily not used except in the cases I mentioned. Move
those workarounds into a few lines of code, and then get
rid of GENL_ID_GENERATE entirely, making it more robust.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg [Mon, 24 Oct 2016 12:40:01 +0000 (14:40 +0200)]
genetlink: introduce and use genl_family_attrbuf()
This helper function allows family implementations to access
their family's attrbuf. This gets rid of the attrbuf usage
in families, and also adds locking validation, since it's not
valid to use the attrbuf with parallel_ops or outside of the
dumpit callback.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
skbedit: allow the user to specify bitmask for mark
The user may want to use only some bits of the skb mark in
his skbedit rules because the remaining part might be used by
something else.
Introduce the "mask" parameter to the skbedit actor in order
to implement such functionality.
When the mask is specified, only those bits selected by the
latter are altered really changed by the actor, while the
rest is left untouched.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com> Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Stefan Richter [Sun, 23 Oct 2016 14:30:56 +0000 (16:30 +0200)]
firewire: net: set initial MTU = 1500 unconditionally, fix IPv6 on some CardBus cards
firewire-net, like the older eth1394 driver, reduced the initial MTU to
less than 1500 octets if the local link layer controller's asynchronous
packet reception limit was lower.
This is bogus, since this reception limit does not have anything to do
with the transmission limit. Neither did this reduction affect the TX
path positively, nor could it prevent link fragmentation at the RX path.
Many FireWire CardBus cards have a max_rec of 9, causing an initial MTU
of 1024 - 16 = 1008. RFC 2734 and RFC 3146 allow a minimum max_rec = 8,
which would result in an initial MTU of 512 - 16 = 496. On such cards,
IPv6 could only be employed if the MTU was manually increased to 1280 or
more, i.e. IPv6 would not work without intervention from userland.
We now always initialize the MTU to 1500, which is the default according
to RFC 2734 and RFC 3146.
On a VIA VT6316 based CardBus card which was affected by this, changing
the MTU from 1008 to 1500 also increases TX bandwidth by 6 %.
RX remains unaffected.
CC: netdev@vger.kernel.org CC: linux1394-devel@lists.sourceforge.net CC: Jarod Wilson <jarod@redhat.com> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Stefan Richter [Sun, 23 Oct 2016 14:29:03 +0000 (16:29 +0200)]
firewire: net: fix maximum possible MTU
Commit b3e3893e1253 ("net: use core MTU range checking in misc drivers")
mistakenly introduced an upper limit for firewire-net's MTU based on the
local link layer controller's reception capability. Revert this. Neither
RFC 2734 nor our implementation impose any particular upper limit.
Actually, to be on the safe side and to make the code explicit, set
ETH_MAX_MTU = 65535 as upper limit now.
(I replaced sizeof(struct rfc2734_header) by the equivalent
RFC2374_FRAG_HDR_SIZE in order to avoid distracting long/int conversions.)
Fixes: b3e3893e1253('net: use core MTU range checking in misc drivers') CC: netdev@vger.kernel.org CC: linux1394-devel@lists.sourceforge.net CC: Jarod Wilson <jarod@redhat.com> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Acked-by: Jarod Wilson <jarod@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Wei Yongjun [Sat, 22 Oct 2016 14:36:36 +0000 (14:36 +0000)]
net: ena: use setup_timer() and mod_timer()
Use setup_timer() instead of init_timer(), being the preferred/standard
way to set a timer up.
Also, quoting the mod_timer() function comment:
-> mod_timer() is a more efficient way to update the expire field of an
active timer (if the timer is inactive it will be activated).
Use setup_timer and mod_timer to setup and arm a timer, to make the code
cleaner and easier to read.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Wei Yongjun [Sat, 22 Oct 2016 14:35:30 +0000 (14:35 +0000)]
amd-xgbe: Fix error return code in xgbe_probe()
Fix to return error code -ENODEV from the DMA is not supported error
handling case instead of 0, as done elsewhere in this function.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Wei Yongjun [Sat, 22 Oct 2016 14:34:55 +0000 (14:34 +0000)]
net: ns83820: use dev_kfree_skb_irq instead of kfree_skb
It is not allowed to call kfree_skb() from hardware interrupt
context or with interrupts being disabled, spin_lock_irqsave()
make sure always in irq disable context. So the kfree_skb()
should be replaced with dev_kfree_skb_irq().
This is detected by Coccinelle semantic patch.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Sven Eckelmann [Sat, 22 Oct 2016 07:46:24 +0000 (09:46 +0200)]
batman-adv: Revert "use core MTU range checking in misc drivers"
The maximum MTU is defined via the slave devices of an batman-adv
interface. Thus it is not possible to calculate the max_mtu during the
creation of the batman-adv device when no slave devices are attached. Doing
so would for example break non-fragmentation setups which then
(incorrectly) allow an MTU of 1500 even when underlying device cannot
transport 1500 bytes + batman-adv headers.
Checking the dynamically calculated max_mtu via the minimum of the slave
devices MTU during .ndo_change_mtu is also used by the bridge interface.
Cc: Jarod Wilson <jarod@redhat.com> Fixes: b3e3893e1253 ("net: use core MTU range checking in misc drivers") Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Xo Wang [Fri, 21 Oct 2016 17:20:13 +0000 (10:20 -0700)]
net: phy: broadcom: Add support for BCM54612E
This PHY has internal delays enabled after reset. This clears the
internal delay enables unless the interface specifically requests them.
Signed-off-by: Xo Wang <xow@google.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Joel Stanley <joel@jms.id.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Xo Wang [Fri, 21 Oct 2016 17:20:12 +0000 (10:20 -0700)]
net: phy: broadcom: Update Auxiliary Control Register macros
Add the RXD-to-RXC skew (delay) time bit in the Miscellaneous Control
shadow register and a mask for the shadow selector field.
Remove a re-definition of MII_BCM54XX_AUXCTL_SHDWSEL_AUXCTL.
Signed-off-by: Xo Wang <xow@google.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Joel Stanley <joel@jms.id.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Fri, 21 Oct 2016 16:02:43 +0000 (09:02 -0700)]
sch_htb: do not report fake rate estimators
When I prepared commit d250a5f90e53 ("pkt_sched: gen_estimator: Dont
report fake rate estimators"), htb still had an implicit rate estimator
for all its classes.
Then later, I made this rate estimator optional in commit 64153ce0a7b6
("net_sched: htb: do not setup default rate estimators"), but I forgot
to update htb use of gnet_stats_copy_rate_est()
After this patch, "tc -s qdisc ..." no longer report fake rate
estimators for HTB classes.
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Cyrill Gorcunov [Fri, 21 Oct 2016 10:03:44 +0000 (13:03 +0300)]
net: ip, diag -- Add diag interface for raw sockets
In criu we are actively using diag interface to collect sockets
present in the system when dumping applications. And while for
unix, tcp, udp[lite], packet, netlink it works as expected,
the raw sockets do not have. Thus add it.
v2:
- add missing sock_put calls in raw_diag_dump_one (by eric.dumazet@)
- implement @destroy for diag requests (by dsa@)
v3:
- add export of raw_abort for IPv6 (by dsa@)
- pass net-admin flag into inet_sk_diag_fill due to
changes in net-next branch (by dsa@)
v4:
- use @pad in struct inet_diag_req_v2 for raw socket
protocol specification: raw module carries sockets
which may have custom protocol passed from socket()
syscall and sole @sdiag_protocol is not enough to
match underlied ones
- start reporting protocol specifed in socket() call
when sockets are raw ones for the same reason: user
space tools like ss may parse this attribute and use
it for socket matching
v5 (by eric.dumazet@):
- use sock_hold in raw_sock_get instead of atomic_inc,
we're holding (raw_v4_hashinfo|raw_v6_hashinfo)->lock
when looking up so counter won't be zero here.
v6:
- use sdiag_raw_protocol() helper which will access @pad
structure used for raw sockets protocol specification:
we can't simply rename this member without breaking uapi
v7:
- sine sdiag_raw_protocol() helper is not suitable for
uapi lets rather make an alias structure with proper
names. __check_inet_diag_req_raw helper will catch
if any of structure unintentionally changed.
CC: David S. Miller <davem@davemloft.net> CC: Eric Dumazet <eric.dumazet@gmail.com> CC: David Ahern <dsa@cumulusnetworks.com> CC: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> CC: James Morris <jmorris@namei.org> CC: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org> CC: Patrick McHardy <kaber@trash.net> CC: Andrey Vagin <avagin@openvz.org> CC: Stephen Hemminger <stephen@networkplumber.org> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Andrey Vagin [Fri, 21 Oct 2016 02:45:43 +0000 (19:45 -0700)]
net: allow to kill a task which waits net_mutex in copy_new_ns
net_mutex can be locked for a long time. It may be because many
namespaces are being destroyed or many processes decide to create
a network namespace.
Both these operations are heavy, so it is better to have an ability to
kill a process which is waiting net_mutex.
Cc: "David S. Miller" <davem@davemloft.net> Cc: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Andrei Vagin <avagin@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Shmulik Ladkani [Thu, 20 Oct 2016 21:18:08 +0000 (00:18 +0300)]
net/sched: em_meta: Fix 'meta vlan' to correctly recognize zero VID frames
META_COLLECTOR int_vlan_tag() assumes that if the accel tag (vlan_tci)
is zero, then no vlan accel tag is present.
This is incorrect for zero VID vlan accel packets, making the following
match fail:
tc filter add ... basic match 'meta(vlan mask 0xfff eq 0)' ...
Apparently 'int_vlan_tag' was implemented prior VLAN_TAG_PRESENT was
introduced in 05423b2 "vlan: allow null VLAN ID to be used"
(and at time introduced, the 'vlan_tx_tag_get' call in em_meta was not
adapted).
Fix, testing skb_vlan_tag_present instead of testing skb_vlan_tag_get's
value.
Fixes: 05423b2413 ("vlan: allow null VLAN ID to be used") Fixes: 1a31f2042e ("netsched: Allow meta match on vlan tag on receive") Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com> Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Jiri Pirko [Fri, 21 Oct 2016 14:07:23 +0000 (16:07 +0200)]
mlxsw: Convert resources into array
Since the number of resources is going to get much bigger, ease up the
addition by simly defining IDs. Convert the existing structure members
to a set array, one for validity, one for values. Introduce a set of
getters and setters for easy access.
Signed-off-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Fri, 21 Oct 2016 10:46:33 +0000 (12:46 +0200)]
bpf: add helper for retrieving current numa node id
Use case is mainly for soreuseport to select sockets for the local
numa node, but since generic, lets also add this for other networking
and tracing program types.
Suggested-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
This patch series refactor the udp memory accounting, replacing the
generic implementation with a custom one, in order to remove the needs for
locking the socket on the enqueue and dequeue operations. The socket backlog
usage is dropped, as well.
The first patch factor out pieces of some queue and memory management
socket helpers, so that they can later be used by the udp memory accounting
functions.
The second patch adds the memory account helpers, without using them.
The third patch replacse the old rx memory accounting path for udp over ipv4 and
udp over ipv6. In kernel UDP users are updated, as well.
The memory accounting schema is described in detail in the individual patch
commit message.
The performance gain depends on the specific scenario; with few flows (and
little contention in the original code) the differences are in the noise range,
while with several flows contending the same socket, the measured speed-up
is relevant (e.g. even over 100% in case of extreme contention)
Many thanks to Eric Dumazet for the reiterated reviews and suggestions.
v5 -> v6:
- do not orphan the skb on enqueue, skb_steal_sock() already did
the work for us
v4 -> v5:
- use the receive queue spin lock to protect the memory accounting
- several minor clean-up
v3 -> v4:
- simplified the locking schema, always use a plain spinlock
v2 -> v3:
- do not set the now unsed backlog_rcv callback
v1 -> v2:
- changed slighly the memory accounting schema, we now perform lazy reclaim
- fixed forward_alloc updating issue
- fixed memory counter integer overflows
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Paolo Abeni [Fri, 21 Oct 2016 11:55:47 +0000 (13:55 +0200)]
udp: use it's own memory accounting schema
Completely avoid default sock memory accounting and replace it
with udp-specific accounting.
Since the new memory accounting model encapsulates completely
the required locking, remove the socket lock on both enqueue and
dequeue, and avoid using the backlog on enqueue.
Be sure to clean-up rx queue memory on socket destruction, using
udp its own sk_destruct.
Tested using pktgen with random src port, 64 bytes packet,
wire-speed on a 10G link as sender and udp_sink as the receiver,
using an l4 tuple rxhash to stress the contention, and one or more
udp_sink instances with reuseport.
v2 -> v3:
- do not set the now unsed backlog_rcv callback
v1 -> v2:
- add memory pressure support
- fixed dropwatch accounting for ipv6
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Paolo Abeni [Fri, 21 Oct 2016 11:55:46 +0000 (13:55 +0200)]
udp: implement memory accounting helpers
Avoid using the generic helpers.
Use the receive queue spin lock to protect the memory
accounting operation, both on enqueue and on dequeue.
On dequeue perform partial memory reclaiming, trying to
leave a quantum of forward allocated memory.
On enqueue use a custom helper, to allow some optimizations:
- use a plain spin_lock() variant instead of the slightly
costly spin_lock_irqsave(),
- avoid dst_force check, since the calling code has already
dropped the skb dst
- avoid orphaning the skb, since skb_steal_sock() already did
the work for us
The above needs custom memory reclaiming on shutdown, provided
by the udp_destruct_sock().
v5 -> v6:
- don't orphan the skb on enqueue
v4 -> v5:
- replace the mem_lock with the receive queue spin lock
- ensure that the bh is always allowed to enqueue at least
a skb, even if sk_rcvbuf is exceeded
v3 -> v4:
- reworked memory accunting, simplifying the schema
- provide an helper for both memory scheduling and enqueuing
v1 -> v2:
- use a udp specific destrctor to perform memory reclaiming
- remove a couple of helpers, unneeded after the above cleanup
- do not reclaim memory on dequeue if not under memory
pressure
- reworked the fwd accounting schema to avoid potential
integer overflow
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / karo-tx-linux.git / log