Add utility functions to convert a 32 bits key into a 64 bits tunnel and
vice versa.
These functions will be used instead of cloning code in GRE and VXLAN,
and in tc act_iptunnel which will be introduced in a following patch in
this patchset.
Signed-off-by: NAmir Vadai <amir@vadai.me>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Reviewed-by: NShmulik Ladkani <shmulik.ladkani@gmail.com>
Acked-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

* Multiplications for the size determination of memory allocations
  indicated that array data structures should be processed.
  Thus use the corresponding function "kmalloc_array".

  This issue was detected by using the Coccinelle software.

* Replace the specification of data types by pointer dereferences
  to make the corresponding size determination a bit safer according to
  the Linux coding style convention.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tobias Regnery says:

====================
alx: add msi-x support

This patchset adds msi-x support to the alx driver. It is a preparatory
series for multi queue support, which I am currently working on. As there
is no advantage over msi interrupts without multi queue support, msi-x
interrupts are disabled by default. In order to test for regressions, a
new module parameter is added to enable msi-x interrupts.

Based on information of the downstream driver at github.com/qca/alx
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

msi-x support is default disabled in the alx driver. In order to test msi-x
interrupts for regressions add a module parameter to the driver.
Signed-off-by: NTobias Regnery <tobias.regnery@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add msi-x support to the alx driver. This is in preparation for multi queue
support.

msi-x interrupts are disabled by default because without multi queue support
there is no advantage over msi interrupts. The performance numbers observed
with iperf stay the same.

Based on information of the downstream driver at github.com/qca/alx
Signed-off-by: NTobias Regnery <tobias.regnery@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Factor out the handling of misc interrupts into a new function.
This function can be reused later for msi-x interrupts.
Signed-off-by: NTobias Regnery <tobias.regnery@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Introduce a new flag field for the advanced interrupt capatibilities and add
new functions to enable and disable msi interrupts. These functions will be
extended later to cover msi-x interrupts.

We enable msi interrupts earlier in alx_init_intr because with msi-x and multi
queue support the number of queues must be set before we allocate resources for
the rx and tx paths.
Signed-off-by: NTobias Regnery <tobias.regnery@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We get a few warnings when building kernel with W=1:
drivers/net/ethernet/qlogic/qed/qed_l2.c:112:5: warning: no previous prototype for 'qed_sp_vport_start' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:110:6: warning: no previous prototype for 'qed_iov_is_valid_vfid' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:188:5: warning: no previous prototype for 'qed_iov_post_vf_bulletin' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:578:6: warning: no previous prototype for 'qed_iov_set_vfs_to_disable' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:1135:28: warning: no previous prototype for 'qed_iov_get_public_vf_info' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:1148:6: warning: no previous prototype for 'qed_iov_clean_vf' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:2444:5: warning: no previous prototype for 'qed_iov_chk_ucast' [-Wmissing-prototypes]
drivers/net/ethernet/qlogic/qed/qed_sriov.c:2762:5: warning: no previous prototype for 'qed_iov_vf_flr_cleanup' [-Wmissing-prototypes]
....

In fact, these functions are only used in the file in which they are
declared and don't need a declaration, but can be made static.
so this patch marks these functions with 'static'.
Signed-off-by: NBaoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
Some BPF helper cleanups

This series contains a couple of misc cleanups and improvements
for BPF helpers. For details please see individual patches. We
let this also sit for a few days with Fengguang's kbuild test
robot, and there were no issues seen (besides one false positive,
see last one for details).
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This work adds BPF_CALL_<n>() macros and converts all the eBPF helper functions
to use them, in a similar fashion like we do with SYSCALL_DEFINE<n>() macros
that are used today. Motivation for this is to hide all the register handling
and all necessary casts from the user, so that it is done automatically in the
background when adding a BPF_CALL_<n>() call.

This makes current helpers easier to review, eases to write future helpers,
avoids getting the casting mess wrong, and allows for extending all helpers at
once (f.e. build time checks, etc). It also helps detecting more easily in
code reviews that unused registers are not instrumented in the code by accident,
breaking compatibility with existing programs.

BPF_CALL_<n>() internals are quite similar to SYSCALL_DEFINE<n>() ones with some
fundamental differences, for example, for generating the actual helper function
that carries all u64 regs, we need to fill unused regs, so that we always end up
with 5 u64 regs as an argument.

I reviewed several 0-5 generated BPF_CALL_<n>() variants of the .i results and
they look all as expected. No sparse issue spotted. We let this also sit for a
few days with Fengguang's kbuild test robot, and there were no issues seen. On
s390, it barked on the "uses dynamic stack allocation" notice, which is an old
one from bpf_perf_event_output{,_tp}() reappearing here due to the conversion
to the call wrapper, just telling that the perf raw record/frag sits on stack
(gcc with s390's -mwarn-dynamicstack), but that's all. Did various runtime tests
and they were fine as well. All eBPF helpers are now converted to use these
macros, getting rid of a good chunk of all the raw castings.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When fetching ifindex, we don't need to test dev for being NULL since
we're always guaranteed to have a valid dev for clsact programs. Thus,
avoid this test in fast path.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add BPF_SIZEOF() and BPF_FIELD_SIZEOF() macros to improve the code a bit
which otherwise often result in overly long bytes_to_bpf_size(sizeof())
and bytes_to_bpf_size(FIELD_SIZEOF()) lines. So place them into a macro
helper instead. Moreover, we currently have a BUILD_BUG_ON(BPF_FIELD_SIZEOF())
check in convert_bpf_extensions(), but we should rather make that generic
as well and add a BUILD_BUG_ON() test in all BPF_SIZEOF()/BPF_FIELD_SIZEOF()
users to detect any rewriter size issues at compile time. Note, there are
currently none, but we want to assert that it stays this way.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some minor misc cleanups, f.e. use sizeof(__u32) instead of hardcoding
and in __bpf_skb_max_len(), I missed that we always have skb->dev valid
anyway, so we can drop the unneeded test for dev; also few more other
misc bits addressed here.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If skb has a valid l4 hash, there is no point clearing hash and force
a further flow dissection when a tunnel encapsulation is added.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

* Multiplications for the size determination of memory allocations
  indicated that array data structures should be processed.
  Thus use the corresponding function "kmalloc_array".

  This issue was detected by using the Coccinelle software.

* Replace the specification of data types by pointer dereferences
  to make the corresponding size determination a bit safer according to
  the Linux coding style convention.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

* A multiplication for the size determination of a memory allocation
  indicated that an array data structure should be processed.
  Thus use the corresponding function "kmalloc_array".

  This issue was detected by using the Coccinelle software.

* Replace the specification of a data structure by a pointer dereference
  to make the corresponding size determination a bit safer according to
  the Linux coding style convention.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Merge tag 'rxrpc-rewrite-20160908' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs

David Howells says:

====================
rxrpc: Rewrite data and ack handling

This patch set constitutes the main portion of the AF_RXRPC rewrite.  It
consists of five fix/helper patches:

 (1) Fix ASSERTCMP's and ASSERTIFCMP's handling of signed values.

 (2) Update some protocol definitions slightly.

 (3) Use of an hlist for RCU purposes.

 (4) Removal of per-call sk_buff accounting (not really needed when skbs
     aren't being queued on the main queue).

 (5) Addition of a tracepoint to log incoming packets in the data_ready
     callback and to log the end of the data_ready callback.

And then there are two patches that form the main part:

 (6) Preallocation of resources for incoming calls so that in patch (7) the
     data_ready handler can be made to fully instantiate an incoming call
     and make it live.  This extends through into AFS so that AFS can
     preallocate its own incoming call resources.

     The preallocation size is capped at the listen() backlog setting - and
     that is capped at a sysctl limit which can be set between 4 and 32.

     The preallocation is (re)charged either by accepting/rejecting pending
     calls or, in the case of AFS, manually.  If insufficient preallocation
     resources exist, a BUSY packet will be transmitted.

     The advantage of using this preallocation is that once a call is set
     up in the data_ready handler, DATA packets can be queued on it
     immediately rather than the DATA packets being queued for a background
     work item to do all the allocation and then try and sort out the DATA
     packets whilst other DATA packets may still be coming in and going
     either to the background thread or the new call.

 (7) Rewrite the handling of DATA, ACK and ABORT packets.

     In the receive phase, DATA packets are now held in per-call circular
     buffers with deduplication, out of sequence detection and suchlike
     being done in data_ready.  Since there is only one producer and only
     once consumer, no locks need be used on the receive queue.

     Received ACK and ABORT packets are now parsed and discarded in
     data_ready to recycle resources as fast as possible.

     sk_buffs are no longer pulled, trimmed or cloned, but rather the
     offset and size of the content is tracked.  This particularly affects
     jumbo DATA packets which need insertion into the receive buffer in
     multiple places.  Annotations are kept to track which bit is which.

     Packets are no longer queued on the socket receive queue; rather,
     calls are queued.  Dummy packets to convey events therefore no longer
     need to be invented and metadata packets can be discarded as soon as
     parsed rather then being pushed onto the socket receive queue to
     indicate terminal events.

     The preallocation facility added in (6) is now used to set up incoming
     calls with very little locking required and no calls to the allocator
     in data_ready.

     Decryption and verification is now handled in recvmsg() rather than in
     a background thread.  This allows for the future possibility of
     decrypting directly into the user buffer.

     With this patch, the code is a lot simpler and most of the mass of
     call event and state wangling code in call_event.c is gone.

With this, the majority of the AF_RXRPC rewrite is complete.  However,
there are still things to be done, including:

 (*) Limit the number of active service calls to prevent an attacker from
     filling up a server's memory.

 (*) Limit the number of calls on the rebuff-with-BUSY queue.

 (*) Transmit delayed/deferred ACKs from recvmsg() if possible, rather than
     punting to the background thread.  Ideally, the background thread
     shouldn't run at all, but data_ready can't call kernel_sendmsg() and
     we can't rely on recvmsg() attending to the call in a timely fashion.

 (*) Prevent the call at the front of the socket queue from hogging
     recvmsg()'s attention if there's a sufficiently continuous supply of
     data.

 (*) Distribute ICMP errors by connection rather than by call.  Possibly
     parse the ICMP packet to try and pin down the exact connection and
     call.

 (*) Encrypt/decrypt directly between user buffers and socket buffers where
     possible.

 (*) IPv6.

 (*) Service ID upgrade.  This is a facility whereby a special flag bit is
     set in the DATA packet header when making a call that tells the server
     that it is allowed to change the service ID to an upgraded one and
     reply with an equivalent call from the upgraded service.

     This is used, for example, to override certain AFS calls so that IPv6
     addresses can be returned.

 (*) Allow userspace to preallocate call user IDs for incoming calls.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

tdinfo->skb_dma is a 7 element array of dma_addr_t hence cannot be
null, so the pull pointer check on tdinfo->skb_dma  is redundant.
Remove it.
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Acked-by: NFrancois Romieu <romieu@fr.zoreil.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We get 1 warning when building kernel with W=1:
drivers/net/ethernet/qlogic/qede/qede_main.c:2113:5: warning: no previous prototype for 'qede_set_features' [-Wmissing-prototypes]

In fact, this function is only used in the file in which it is
declared and don't need a declaration, but can be made static.
so this patch marks this function with 'static'.
Signed-off-by: NBaoyou Xie <baoyou.xie@linaro.org>
Acked-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The existing VSC85xx PHY driver did not follow the coding style and caused "checkpatch" to complain. This commit fixes this.
Signed-off-by: NRaju Lakkaraju <Raju.Lakkaraju@microsemi.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

dtefacs.calling_ae and called_ae are both 20 element __u8 arrays and
cannot be null and hence are redundant checks. Remove these.
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The netdevice type structure for macsec was being defined but never used.
To set the network device type the macro SET_NETDEV_DEVTYPE must be called.
Compile tested only, I don't use macsec.
Signed-off-by: NStephen Hemminger <stephen@networkplumber.org>
Acked-by: NSabrina Dubroca <sd@queasysnail.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This structure is defined but never used. Flagged with W=1
Signed-off-by: NStephen Hemminger <stephen@networkplumber.org>
Acked-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Guillaume Nault says:

====================
ip: fix creation flags reported in RTM_NEWROUTE events

Netlink messages sent to user-space upon RTM_NEWROUTE events have their
nlmsg_flags field inconsistently set. While the NLM_F_REPLACE and
NLM_F_APPEND bits are correctly handled, NLM_F_CREATE and NLM_F_EXCL
are always 0.

This series sets the NLM_F_CREATE and NLM_F_EXCL bits when applicable,
for IPv4 and IPv6.

Since IPv6 ignores the NLM_F_APPEND flags in requests, this flag isn't
reported in RTM_NEWROUTE IPv6 events. This keeps IPv6 internal
consistency (same flag semantic for user requests and kernel events) at
the cost of bringing different flag interpretation for IPv4 and IPv6.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since commit 37a1d361 ("ipv6: include NLM_F_REPLACE in route
replace notifications"), RTM_NEWROUTE notifications have their
NLM_F_REPLACE flag set if the new route replaced a preexisting one.
However, other flags aren't set.

This patch reports the missing NLM_F_CREATE and NLM_F_EXCL flag bits.

NLM_F_APPEND is not reported, because in ipv6 a NLM_F_CREATE request
is interpreted as an append request (contrary to ipv4, "prepend" is not
supported, so if NLM_F_EXCL is not set then NLM_F_APPEND is implicit).

As a result, the possible flag combination can now be reported
(iproute2's terminology into parentheses):

  * NLM_F_CREATE | NLM_F_EXCL: route didn't exist, exclusive creation
    ("add").
  * NLM_F_CREATE: route did already exist, new route added after
    preexisting ones ("append").
  * NLM_F_REPLACE: route did already exist, new route replaced the
    first preexisting one ("change").
Signed-off-by: NGuillaume Nault <g.nault@alphalink.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

fib_table_insert() inconsistently fills the nlmsg_flags field in its
notification messages.

Since commit b8f55831 ("[RTNETLINK]: Fix sending netlink message
when replace route."), the netlink message has its nlmsg_flags set to
NLM_F_REPLACE if the route replaced a preexisting one.

Then commit a2bb6d7d ("ipv4: include NLM_F_APPEND flag in append
route notifications") started setting nlmsg_flags to NLM_F_APPEND if
the route matched a preexisting one but was appended.

In other cases (exclusive creation or prepend), nlmsg_flags is 0.

This patch sets ->nlmsg_flags in all situations, preserving the
semantic of the NLM_F_* bits:

  * NLM_F_CREATE: a new fib entry has been created for this route.
  * NLM_F_EXCL: no other fib entry existed for this route.
  * NLM_F_REPLACE: this route has overwritten a preexisting fib entry.
  * NLM_F_APPEND: the new fib entry was added after other entries for
    the same route.

As a result, the possible flag combination can now be reported
(iproute2's terminology into parentheses):

  * NLM_F_CREATE | NLM_F_EXCL: route didn't exist, exclusive creation
    ("add").
  * NLM_F_CREATE | NLM_F_APPEND: route did already exist, new route
    added after preexisting ones ("append").
  * NLM_F_CREATE: route did already exist, new route added before
    preexisting ones ("prepend").
  * NLM_F_REPLACE: route did already exist, new route replaced the
    first preexisting one ("change").
Signed-off-by: NGuillaume Nault <g.nault@alphalink.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In commit f02db315 ("ipv4: IP_TOS and IP_TTL can be specified as
ancillary data") Francesco added IP_TOS values specified as integer.

However, kernel sends to userspace (at recvmsg() time) an IP_TOS value
in a single byte, when IP_RECVTOS is set on the socket.

It can be very useful to reflect all ancillary options as given by the
kernel in a subsequent sendmsg(), instead of aborting the sendmsg() with
EINVAL after Francesco patch.

So this patch extends IP_TOS ancillary to accept an u8, so that an UDP
server can simply reuse same ancillary block without having to mangle
it.

Jesper can then augment
https://github.com/netoptimizer/network-testing/blob/master/src/udp_example02.c
to add TOS reflection ;)

Fixes: f02db315 ("ipv4: IP_TOS and IP_TTL can be specified as ancillary data")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Francesco Fusco <ffusco@redhat.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

LLVM can generate code that tests for direct packet access via
skb->data/data_end in a way that currently gets rejected by the
verifier, example:

  [...]
   7: (61) r3 = *(u32 *)(r6 +80)
   8: (61) r9 = *(u32 *)(r6 +76)
   9: (bf) r2 = r9
  10: (07) r2 += 54
  11: (3d) if r3 >= r2 goto pc+12
   R1=inv R2=pkt(id=0,off=54,r=0) R3=pkt_end R4=inv R6=ctx
   R9=pkt(id=0,off=0,r=0) R10=fp
  12: (18) r4 = 0xffffff7a
  14: (05) goto pc+430
  [...]

  from 11 to 24: R1=inv R2=pkt(id=0,off=54,r=0) R3=pkt_end R4=inv
                 R6=ctx R9=pkt(id=0,off=0,r=0) R10=fp
  24: (7b) *(u64 *)(r10 -40) = r1
  25: (b7) r1 = 0
  26: (63) *(u32 *)(r6 +56) = r1
  27: (b7) r2 = 40
  28: (71) r8 = *(u8 *)(r9 +20)
  invalid access to packet, off=20 size=1, R9(id=0,off=0,r=0)

The reason why this gets rejected despite a proper test is that we
currently call find_good_pkt_pointers() only in case where we detect
tests like rX > pkt_end, where rX is of type pkt(id=Y,off=Z,r=0) and
derived, for example, from a register of type pkt(id=Y,off=0,r=0)
pointing to skb->data. find_good_pkt_pointers() then fills the range
in the current branch to pkt(id=Y,off=0,r=Z) on success.

For above case, we need to extend that to recognize pkt_end >= rX
pattern and mark the other branch that is taken on success with the
appropriate pkt(id=Y,off=0,r=Z) type via find_good_pkt_pointers().
Since eBPF operates on BPF_JGT (>) and BPF_JGE (>=), these are the
only two practical options to test for from what LLVM could have
generated, since there's no such thing as BPF_JLT (<) or BPF_JLE (<=)
that we would need to take into account as well.

After the fix:

  [...]
   7: (61) r3 = *(u32 *)(r6 +80)
   8: (61) r9 = *(u32 *)(r6 +76)
   9: (bf) r2 = r9
  10: (07) r2 += 54
  11: (3d) if r3 >= r2 goto pc+12
   R1=inv R2=pkt(id=0,off=54,r=0) R3=pkt_end R4=inv R6=ctx
   R9=pkt(id=0,off=0,r=0) R10=fp
  12: (18) r4 = 0xffffff7a
  14: (05) goto pc+430
  [...]

  from 11 to 24: R1=inv R2=pkt(id=0,off=54,r=54) R3=pkt_end R4=inv
                 R6=ctx R9=pkt(id=0,off=0,r=54) R10=fp
  24: (7b) *(u64 *)(r10 -40) = r1
  25: (b7) r1 = 0
  26: (63) *(u32 *)(r6 +56) = r1
  27: (b7) r2 = 40
  28: (71) r8 = *(u8 *)(r9 +20)
  29: (bf) r1 = r8
  30: (25) if r8 > 0x3c goto pc+47
   R1=inv56 R2=imm40 R3=pkt_end R4=inv R6=ctx R8=inv56
   R9=pkt(id=0,off=0,r=54) R10=fp
  31: (b7) r1 = 1
  [...]

Verifier test cases are also added in this work, one that demonstrates
the mentioned example here and one that tries a bad packet access for
the current/fall-through branch (the one with types pkt(id=X,off=Y,r=0),
pkt(id=X,off=0,r=0)), then a case with good and bad accesses, and two
with both test variants (>, >=).

Fixes: 969bf05e ("bpf: direct packet access")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Over the years, TCP BDP has increased by several orders of magnitude,
and some people are considering to reach the 2 Gbytes limit.

Even with current window scale limit of 14, ~1 Gbytes maps to ~740,000
MSS.

In presence of packet losses (or reorders), TCP stores incoming packets
into an out of order queue, and number of skbs sitting there waiting for
the missing packets to be received can be in the 10^5 range.

Most packets are appended to the tail of this queue, and when
packets can finally be transferred to receive queue, we scan the queue
from its head.

However, in presence of heavy losses, we might have to find an arbitrary
point in this queue, involving a linear scan for every incoming packet,
throwing away cpu caches.

This patch converts it to a RB tree, to get bounded latencies.

Yaogong wrote a preliminary patch about 2 years ago.
Eric did the rebase, added ofo_last_skb cache, polishing and tests.

Tested with network dropping between 1 and 10 % packets, with good
success (about 30 % increase of throughput in stress tests)

Next step would be to also use an RB tree for the write queue at sender
side ;)
Signed-off-by: NYaogong Wang <wygivan@google.com>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-By: NIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Eric Garver says:

====================
openvswitch: add 802.1ad support

This series adds 802.1ad support to openvswitch. It is a continuation of the
work originally started by Thomas F Herbert - hence the large rev number.

The extra VLAN is implemented by using an additional level of the
OVS_KEY_ATTR_ENCAP netlink attribute.
In OVS flow speak, this looks like

   eth_type(0x88a8),vlan(vid=100),encap(eth_type(0x8100), vlan(vid=200),
                                        encap(eth_type(0x0800), ...))

The userspace counterpart has also seen recent activity on the ovs-dev mailing
lists. There are some new 802.1ad OVS tests being added - also on the ovs-dev
list. This patch series has been tested using the most recent version of
userspace (v3) and tests (v2).

v22 changes:
  - merge patch 4 into patch 3
  - fix checkpatch.pl errors
    - Still some 80 char warnings for long string literals
  - refresh pointer after pskb_may_pull()
  - refactor vlan nlattr parsing to remove some double checks
  - introduce ovs_nla_put_vlan()
  - move triple VLAN check to after ethertype serialization
  - WARN_ON_ONCE() on triple VLAN and unexpected encap values

v21 changes:
  - Fix (and simplify) netlink attribute parsing
  - re-add handling of truncated VLAN tags
  - fix if/else dangling assignment in {push,pop}_vlan()
  - simplify parse_vlan()
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for 802.1ad including the ability to push and pop double
tagged vlans. Add support for 802.1ad to netlink parsing and flow
conversion. Uses double nested encap attributes to represent double
tagged vlan. Inner TPID encoded along with ctci in nested attributes.

This is based on Thomas F Herbert's original v20 patch. I made some
small clean ups and bug fixes.
Signed-off-by: NThomas F Herbert <thomasfherbert@gmail.com>
Signed-off-by: NEric Garver <e@erig.me>
Acked-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is to simplify using double tagged vlans. This function allows all
valid vlan ethertypes to be checked in a single function call.
Also replace some instances that check for both ETH_P_8021Q and
ETH_P_8021AD.

Patch based on one originally by Thomas F Herbert.
Signed-off-by: NThomas F Herbert <thomasfherbert@gmail.com>
Signed-off-by: NEric Garver <e@erig.me>
Acked-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

openvswitch: Add support for 8021.AD

Change the description of the VLAN tpid field.
Signed-off-by: NThomas F Herbert <thomasfherbert@gmail.com>
Acked-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This adds the capability for a process that has CAP_NET_ADMIN on
a socket to see the socket mark in socket dumps.

Commit a52e95ab ("net: diag: allow socket bytecode filters to
match socket marks") recently gave privileged processes the
ability to filter socket dumps based on mark. This patch is
complementary: it ensures that the mark is also passed to
userspace in the socket's netlink attributes.  It is useful for
tools like ss which display information about sockets.

Tested: https://android-review.googlesource.com/270210Signed-off-by: NLorenzo Colitti <lorenzo@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The MIPS based xilfpga platform uses this driver.
Enable it for MIPS
Signed-off-by: NZubair Lutfullah Kakakhel <Zubair.Kakakhel@imgtec.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Steffen Klassert says:

====================
ipsec-next 2016-09-08

1) Constify the xfrm_replay structures. From Julia Lawall

2) Protect xfrm state hash tables with rcu, lookups
   can be done now without acquiring xfrm_state_lock.
   From Florian Westphal.

3) Protect xfrm policy hash tables with rcu, lookups
   can be done now without acquiring xfrm_policy_lock.
   From Florian Westphal.

4) We don't need to have a garbage collector list per
   namespace anymore, so use a global one instead.
   From Florian Westphal.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Rewrite the data and ack handling code such that:

 (1) Parsing of received ACK and ABORT packets and the distribution and the
     filing of DATA packets happens entirely within the data_ready context
     called from the UDP socket.  This allows us to process and discard ACK
     and ABORT packets much more quickly (they're no longer stashed on a
     queue for a background thread to process).

 (2) We avoid calling skb_clone(), pskb_pull() and pskb_trim().  We instead
     keep track of the offset and length of the content of each packet in
     the sk_buff metadata.  This means we don't do any allocation in the
     receive path.

 (3) Jumbo DATA packet parsing is now done in data_ready context.  Rather
     than cloning the packet once for each subpacket and pulling/trimming
     it, we file the packet multiple times with an annotation for each
     indicating which subpacket is there.  From that we can directly
     calculate the offset and length.

 (4) A call's receive queue can be accessed without taking locks (memory
     barriers do have to be used, though).

 (5) Incoming calls are set up from preallocated resources and immediately
     made live.  They can than have packets queued upon them and ACKs
     generated.  If insufficient resources exist, DATA packet #1 is given a
     BUSY reply and other DATA packets are discarded).

 (6) sk_buffs no longer take a ref on their parent call.

To make this work, the following changes are made:

 (1) Each call's receive buffer is now a circular buffer of sk_buff
     pointers (rxtx_buffer) rather than a number of sk_buff_heads spread
     between the call and the socket.  This permits each sk_buff to be in
     the buffer multiple times.  The receive buffer is reused for the
     transmit buffer.

 (2) A circular buffer of annotations (rxtx_annotations) is kept parallel
     to the data buffer.  Transmission phase annotations indicate whether a
     buffered packet has been ACK'd or not and whether it needs
     retransmission.

     Receive phase annotations indicate whether a slot holds a whole packet
     or a jumbo subpacket and, if the latter, which subpacket.  They also
     note whether the packet has been decrypted in place.

 (3) DATA packet window tracking is much simplified.  Each phase has just
     two numbers representing the window (rx_hard_ack/rx_top and
     tx_hard_ack/tx_top).

     The hard_ack number is the sequence number before base of the window,
     representing the last packet the other side says it has consumed.
     hard_ack starts from 0 and the first packet is sequence number 1.

     The top number is the sequence number of the highest-numbered packet
     residing in the buffer.  Packets between hard_ack+1 and top are
     soft-ACK'd to indicate they've been received, but not yet consumed.

     Four macros, before(), before_eq(), after() and after_eq() are added
     to compare sequence numbers within the window.  This allows for the
     top of the window to wrap when the hard-ack sequence number gets close
     to the limit.

     Two flags, RXRPC_CALL_RX_LAST and RXRPC_CALL_TX_LAST, are added also
     to indicate when rx_top and tx_top point at the packets with the
     LAST_PACKET bit set, indicating the end of the phase.

 (4) Calls are queued on the socket 'receive queue' rather than packets.
     This means that we don't need have to invent dummy packets to queue to
     indicate abnormal/terminal states and we don't have to keep metadata
     packets (such as ABORTs) around

 (5) The offset and length of a (sub)packet's content are now passed to
     the verify_packet security op.  This is currently expected to decrypt
     the packet in place and validate it.

     However, there's now nowhere to store the revised offset and length of
     the actual data within the decrypted blob (there may be a header and
     padding to skip) because an sk_buff may represent multiple packets, so
     a locate_data security op is added to retrieve these details from the
     sk_buff content when needed.

 (6) recvmsg() now has to handle jumbo subpackets, where each subpacket is
     individually secured and needs to be individually decrypted.  The code
     to do this is broken out into rxrpc_recvmsg_data() and shared with the
     kernel API.  It now iterates over the call's receive buffer rather
     than walking the socket receive queue.

Additional changes:

 (1) The timers are condensed to a single timer that is set for the soonest
     of three timeouts (delayed ACK generation, DATA retransmission and
     call lifespan).

 (2) Transmission of ACK and ABORT packets is effected immediately from
     process-context socket ops/kernel API calls that cause them instead of
     them being punted off to a background work item.  The data_ready
     handler still has to defer to the background, though.

 (3) A shutdown op is added to the AF_RXRPC socket so that the AFS
     filesystem can shut down the socket and flush its own work items
     before closing the socket to deal with any in-progress service calls.

Future additional changes that will need to be considered:

 (1) Make sure that a call doesn't hog the front of the queue by receiving
     data from the network as fast as userspace is consuming it to the
     exclusion of other calls.

 (2) Transmit delayed ACKs from within recvmsg() when we've consumed
     sufficiently more packets to avoid the background work item needing to
     run.
Signed-off-by: NDavid Howells <dhowells@redhat.com>

Make it possible for the data_ready handler called from the UDP transport
socket to completely instantiate an rxrpc_call structure and make it
immediately live by preallocating all the memory it might need.  The idea
is to cut out the background thread usage as much as possible.

[Note that the preallocated structs are not actually used in this patch -
 that will be done in a future patch.]

If insufficient resources are available in the preallocation buffers, it
will be possible to discard the DATA packet in the data_ready handler or
schedule a BUSY packet without the need to schedule an attempt at
allocation in a background thread.

To this end:

 (1) Preallocate rxrpc_peer, rxrpc_connection and rxrpc_call structs to a
     maximum number each of the listen backlog size.  The backlog size is
     limited to a maxmimum of 32.  Only this many of each can be in the
     preallocation buffer.

 (2) For userspace sockets, the preallocation is charged initially by
     listen() and will be recharged by accepting or rejecting pending
     new incoming calls.

 (3) For kernel services {,re,dis}charging of the preallocation buffers is
     handled manually.  Two notifier callbacks have to be provided before
     kernel_listen() is invoked:

     (a) An indication that a new call has been instantiated.  This can be
     	 used to trigger background recharging.

     (b) An indication that a call is being discarded.  This is used when
     	 the socket is being released.

     A function, rxrpc_kernel_charge_accept() is called by the kernel
     service to preallocate a single call.  It should be passed the user ID
     to be used for that call and a callback to associate the rxrpc call
     with the kernel service's side of the ID.

 (4) Discard the preallocation when the socket is closed.

 (5) Temporarily bump the refcount on the call allocated in
     rxrpc_incoming_call() so that rxrpc_release_call() can ditch the
     preallocation ref on service calls unconditionally.  This will no
     longer be necessary once the preallocation is used.

Note that this does not yet control the number of active service calls on a
client - that will come in a later patch.

A future development would be to provide a setsockopt() call that allows a
userspace server to manually charge the preallocation buffer.  This would
allow user call IDs to be provided in advance and the awkward manual accept
stage to be bypassed.
Signed-off-by: NDavid Howells <dhowells@redhat.com>

Add two tracepoints:

 (1) Record the RxRPC protocol header of packets retrieved from the UDP
     socket by the data_ready handler.

 (2) Record the outcome of the data_ready handler.
Signed-off-by: NDavid Howells <dhowells@redhat.com>

Remove the sk_buff count from the rxrpc_call struct as it's less useful
once we stop queueing sk_buffs.
Signed-off-by: NDavid Howells <dhowells@redhat.com>