With TLP, the peer may reply to a probe with an
ACK+D-SACK, with ack value set to tlp_high_seq. In the current code,
such ACK+DSACK will be missed and only at next, higher ack will the TLP
episode be considered done. Since the DSACK is not present anymore,
this will cost a cwnd reduction.
This patch ensures that this scenario does not cause a cwnd reduction, since
receiving an ACK+DSACK indicates that both the initial segment and the probe
have been received by the peer.
The following packetdrill test, from Neal Cardwell, validates this patch:
// Establish a connection.
0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
+0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+0 bind(3, ..., ...) = 0
+0 listen(3, 1) = 0
+0 < S 0:0(0) win 32792 <mss 1000,sackOK,nop,nop,nop,wscale 7>
+0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 6>
+.020 < . 1:1(0) ack 1 win 257
+0 accept(3, ..., ...) = 4
// Send 1 packet.
+0 write(4, ..., 1000) = 1000
+0 > P. 1:1001(1000) ack 1
// Loss probe retransmission.
// packets_out == 1 => schedule PTO in max(2*RTT, 1.5*RTT + 200ms)
// In this case, this means: 1.5*RTT + 200ms = 230ms
+.230 > P. 1:1001(1000) ack 1
+0 %{ assert tcpi_snd_cwnd == 10 }%
// Receiver ACKs at tlp_high_seq with a DSACK,
// indicating they received the original packet and probe.
+.020 < . 1:1(0) ack 1001 win 257 <sack 1:1001,nop,nop>
+0 %{ assert tcpi_snd_cwnd == 10 }%
// Send another packet.
+0 write(4, ..., 1000) = 1000
+0 > P. 1001:2001(1000) ack 1
// Receiver ACKs above tlp_high_seq, which should end the TLP episode
// if we haven't already. We should not reduce cwnd.
+.020 < . 1:1(0) ack 2001 win 257
+0 %{ assert tcpi_snd_cwnd == 10, tcpi_snd_cwnd }%
Credits:
-Gregory helped in finding that tcp_process_tlp_ack was where the cwnd
got reduced in our MPTCP tests.
-Neal wrote the packetdrill test above
-Yuchung reworked the patch to make it more readable.
Cc: Gregory Detal <gregory.detal@uclouvain.be>
Cc: Nandita Dukkipati <nanditad@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Sébastien Barré <sebastien.barre@uclouvain.be>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This work adds the possibility to define a per route/destination
congestion control algorithm. Generally, this opens up the possibility
for a machine with different links to enforce specific congestion
control algorithms with optimal strategies for each of them based
on their network characteristics, even transparently for a single
application listening on all links.
For our specific use case, this additionally facilitates deployment
of DCTCP, for example, applications can easily serve internal
traffic/dsts in DCTCP and external one with CUBIC. Other scenarios
would also allow for utilizing e.g. long living, low priority
background flows for certain destinations/routes while still being
able for normal traffic to utilize the default congestion control
algorithm. We also thought about a per netns setting (where different
defaults are possible), but given its actually a link specific
property, we argue that a per route/destination setting is the most
natural and flexible.
The administrator can utilize this through ip-route(8) by appending
"congctl [lock] <name>", where <name> denotes the name of a
congestion control algorithm and the optional lock parameter allows
to enforce the given algorithm so that applications in user space
would not be allowed to overwrite that algorithm for that destination.
The dst metric lookups are being done when a dst entry is already
available in order to avoid a costly lookup and still before the
algorithms are being initialized, thus overhead is very low when the
feature is not being used. While the client side would need to drop
the current reference on the module, on server side this can actually
even be avoided as we just got a flat-copied socket clone.
Joint work with Florian Westphal.
Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the minimum necessary for the RTAX_CC_ALGO congestion
control metric to be set up and dumped back to user space.
While the internal representation of RTAX_CC_ALGO is handled as a u32
key, we avoided to expose this implementation detail to user space, thus
instead, we chose the netlink attribute that is being exchanged between
user space to be the actual congestion control algorithm name, similarly
as in the setsockopt(2) API in order to allow for maximum flexibility,
even for 3rd party modules.
It is a bit unfortunate that RTAX_QUICKACK used up a whole RTAX slot as
it should have been stored in RTAX_FEATURES instead, we first thought
about reusing it for the congestion control key, but it brings more
complications and/or confusion than worth it.
Joint work with Florian Westphal.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds necessary infrastructure to the congestion control
framework for later per route congestion control support.
For a per route congestion control possibility, our aim is to store
a unique u32 key identifier into dst metrics, which can then be
mapped into a tcp_congestion_ops struct. We argue that having a
RTAX key entry is the most simple, generic and easy way to manage,
and also keeps the memory footprint of dst entries lower on 64 bit
than with storing a pointer directly, for example. Having a unique
key id also allows for decoupling actual TCP congestion control
module management from the FIB layer, i.e. we don't have to care
about expensive module refcounting inside the FIB at this point.
We first thought of using an IDR store for the realization, which
takes over dynamic assignment of unused key space and also performs
the key to pointer mapping in RCU. While doing so, we stumbled upon
the issue that due to the nature of dynamic key distribution, it
just so happens, arguably in very rare occasions, that excessive
module loads and unloads can lead to a possible reuse of previously
used key space. Thus, previously stale keys in the dst metric are
now being reassigned to a different congestion control algorithm,
which might lead to unexpected behaviour. One way to resolve this
would have been to walk FIBs on the actually rare occasion of a
module unload and reset the metric keys for each FIB in each netns,
but that's just very costly.
Therefore, we argue a better solution is to reuse the unique
congestion control algorithm name member and map that into u32 key
space through jhash. For that, we split the flags attribute (as it
currently uses 2 bits only anyway) into two u32 attributes, flags
and key, so that we can keep the cacheline boundary of 2 cachelines
on x86_64 and cache the precalculated key at registration time for
the fast path. On average we might expect 2 - 4 modules being loaded
worst case perhaps 15, so a key collision possibility is extremely
low, and guaranteed collision-free on LE/BE for all in-tree modules.
Overall this results in much simpler code, and all without the
overhead of an IDR. Due to the deterministic nature, modules can
now be unloaded, the congestion control algorithm for a specific
but unloaded key will fall back to the default one, and on module
reload time it will switch back to the expected algorithm
transparently.
Joint work with Florian Westphal.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We can just move this to an extra function and make the code
a bit more readable, no functional change.
Joint work with Florian Westphal.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add ip_cmsg_recv_offset function which takes an offset argument
that indicates the starting offset in skb where data is being received
from. This will be useful in the case of UDP and provided checksum
to user space.
ip_cmsg_recv is an inline call to ip_cmsg_recv_offset with offset of
zero.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add ip_cmsg_recv_offset function which takes an offset argument
that indicates the starting offset in skb where data is being received
from. This will be useful in the case of UDP and provided checksum
to user space.
ip_cmsg_recv is an inline call to ip_cmsg_recv_offset with offset of
zero.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move the IP_CMSG_* constants from ip_sockglue.c to inet_sock.h so that
they can be referenced in other source files.
Restructure ip_cmsg_recv to not go through flags using shift, check
for flags by 'and'. This eliminates both the shift and a conditional
per flag check.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move convert_csum from udp_sock to inet_sock. This allows the
possibility that we can use convert checksum for different types
of sockets and also allows convert checksum to be enabled from
inet layer (what we'll want to do when enabling IP_CHECKSUM cmsg).
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When searching for an existing socket to reuse, the address family
is not taken into account - only port number. This means that an
IPv4 socket could be used for IPv6 traffic and vice versa, which
is sure to cause problems when passing packets.
It is not possible to trigger this problem currently because the
only user of Geneve creates just IPv4 sockets. However, that is
likely to change in the near future.
Signed-off-by: Jesse Gross <jesse@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The hash table for open Geneve ports is used only on creation and
deletion time. It is not performance critical and is not likely to
grow to a large number of items. Therefore, this can be changed
to use a simple linked list.
Signed-off-by: Jesse Gross <jesse@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The existing Geneve locking scheme was pulled over directly from
VXLAN. However, VXLAN has a number of built in mechanisms which make
the locking more complex and are unlikely to be necessary with Geneve.
This simplifies the locking to use a basic scheme of a mutex
when doing updates plus RCU on receive.
In addition to making the code easier to read, this also avoids the
possibility of a race when creating or destroying sockets since
UDP sockets and the list of Geneve sockets are protected by different
locks. After this change, the entire operation is atomic.
Signed-off-by: Jesse Gross <jesse@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The work queue is used only to free the UDP socket upon destruction.
This is not necessary with Geneve and generally makes the code more
difficult to reason about. It also introduces nondeterministic
behavior such as when a socket is rapidly deleted and recreated, which
could fail as the the deletion happens asynchronously.
Signed-off-by: Jesse Gross <jesse@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Thomas Jarosch reported IPsec TCP stalls when a PMTU event occurs.
In fact the problem was completely unrelated to IPsec. The bug is
also reproducible if you just disable TSO/GSO.
The problem is that when the MSS goes down, existing queued packet
on the TX queue that have not been transmitted yet all look like
TSO packets and get treated as such.
This then triggers a bug where tcp_mss_split_point tells us to
generate a zero-sized packet on the TX queue. Once that happens
we're screwed because the zero-sized packet can never be removed
by ACKs.
Fixes: 1485348d24 ("tcp: Apply device TSO segment limit earlier")
Reported-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Cheers,
Signed-off-by: David S. Miller <davem@davemloft.net>
This results in an approximately 30% increase in throughput
when handling encapsulated bulk traffic.
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change adds a tracking value for the maximum suffix length of all
prefixes stored in any given tnode. With this value we can determine if we
need to backtrace or not based on if the suffix is greater than the pos
value.
By doing this we can reduce the CPU overhead for lookups in the local table
as many of the prefixes there are 32b long and have a suffix length of 0
meaning we can immediately backtrace to the root node without needing to
test any of the nodes between it and where we ended up.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For some reason the compiler doesn't seem to understand that when we are in
a loop that runs from tnode_child_length - 1 to 0 we don't expect the value
of tn->bits to change. As such every call to tnode_get_child was rerunning
tnode_chile_length which ended up consuming quite a bit of space in the
resultant assembly code.
I have gone though and verified that in all cases where tnode_get_child
is used we are either winding though a fixed loop from tnode_child_length -
1 to 0, or are in a fastpath case where we are verifying the value by
either checking for any remaining bits after shifting index by bits and
testing for leaf, or by using tnode_child_length.
size net/ipv4/fib_trie.o
Before:
text data bss dec hex filename
15506 376 8 15890 3e12 net/ipv4/fib_trie.o
After:
text data bss dec hex filename
14827 376 8 15211 3b6b net/ipv4/fib_trie.o
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change pulls the node_set_parent functionality out of put_child_reorg
and instead leaves that to the function to take care of as well. By doing
this we can fully construct the new cluster of tnodes and all of the
pointers out of it before we start routing pointers into it.
I am suspecting this will likely fix some concurency issues though I don't
have a good test to show as such.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change pushes the tnode freeing down into the inflate and halve
functions. It makes more sense here as we have a better grasp of what is
going on and when a given cluster of nodes is ready to be freed.
I believe this may address a bug in the freeing logic as well. For some
reason if the freelist got to a certain size we would call
synchronize_rcu(). I'm assuming that what they meant to do is call
synchronize_rcu() after they had handed off that much memory via
call_rcu(). As such that is what I have updated the behavior to be.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change makes it so that the assignment of the tnode to the parent is
handled directly within whatever function is currently handling the node be
it inflate, halve, or resize. By doing this we can avoid some of the need
to set NULL pointers in the tree while we are resizing the subnodes.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change pulls the logic for if we should inflate/halve the nodes out
into separate functions. It also addresses what I believe is a bug where 1
full node is all that is needed to keep a node from ever being halved.
Simple script to reproduce the issue:
modprobe dummy; ifconfig dummy0 up
for i in `seq 0 255`; do ifconfig dummy0:$i 10.0.${i}.1/24 up; done
ifconfig dummy0:256 10.0.255.33/16 up
for i in `seq 0 254`; do ifconfig dummy0:$i down; done
Results from /proc/net/fib_triestat
Before:
Local:
Aver depth: 3.00
Max depth: 4
Leaves: 17
Prefixes: 18
Internal nodes: 11
1: 8 2: 2 10: 1
Pointers: 1048
Null ptrs: 1021
Total size: 11 kB
After:
Local:
Aver depth: 3.41
Max depth: 5
Leaves: 17
Prefixes: 18
Internal nodes: 12
1: 8 2: 3 3: 1
Pointers: 36
Null ptrs: 8
Total size: 3 kB
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change consists of a cut/paste of resize to behind inflate and halve
so that I could remove the two function prototypes.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change is to start cleaning up some of the rcu_read_lock/unlock
handling. I realized while reviewing the code there are several spots that
I don't believe are being handled correctly or are masking warnings by
locally calling rcu_read_lock/unlock instead of calling them at the correct
level.
A common example is a call to fib_get_table followed by fib_table_lookup.
The rcu_read_lock/unlock ought to wrap both but there are several spots where
they were not wrapped.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change makes it so that anything that can be shifted by, or compared
to a value shifted by bits is updated to be an unsigned long. This is
mostly a precaution against an insanely huge address space that somehow
starts coming close to the 2^32 root node size which would require
something like 1.5 billion addresses.
I chose unsigned long instead of unsigned long long since I do not believe
it is possible to allocate a 32 bit tnode on a 32 bit system as the memory
consumed would be 16GB + 28B which exceeds the addressible space for any
one process.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change moves the pos value to the other side of the "bits" field. By
doing this it actually simplifies a significant amount of code in the trie.
For example when halving a tree we know that the bit lost exists at
oldnode->pos, and if we inflate the tree the new bit being add is at
tn->pos. Previously to find those bits you would have to subtract pos and
bits from the keylength or start with a value of (1 << 31) and then shift
that.
There are a number of spots throughout the code that benefit from this. In
the case of the hot-path searches the main advantage is that we can drop 2
or more operations from the search path as we no longer need to compute the
value for the index to be shifted by and can instead just use the raw pos
value.
In addition the tkey_extract_bits is now defunct and can be replaced by
get_index since the two operations were doing the same thing, but now
get_index does it much more quickly as it is only an xor and shift versus a
pair of shifts and a subtraction.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch updates the fib_table_insert function to take advantage of the
changes made to improve the performance of fib_table_lookup. As a result
the code should be smaller and run faster then the original.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch makes use of the same features I made use of for
fib_table_lookup to streamline fib_find_node. The resultant code should be
smaller and run faster than the original.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is meant to reduce the complexity of fib_table_lookup by reducing
the number of variables to the bare minimum while still keeping the same if
not improved functionality versus the original.
Most of this change was started off by the desire to rid the function of
chopped_off and current_prefix_length as they actually added very little to
the function since they only applied when computing the cindex. I was able
to replace them mostly with just a check for the prefix match. As long as
the prefix between the key and the node being tested was the same we know
we can search the tnode fully versus just testing cindex 0.
The second portion of the change ended up being a massive reordering.
Originally the calls to check_leaf were up near the start of the loop, and
the backtracing and descending into lower levels of tnodes was later. This
didn't make much sense as the structure of the tree means the leaves are
always the last thing to be tested. As such I reordered things so that we
instead have a loop that will delve into the tree and only exit when we
have either found a leaf or we have exhausted the tree. The advantage of
rearranging things like this is that we can fully inline check_leaf since
there is now only one reference to it in the function.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change makes it so that leaf and tnode are the same struct. As a
result there is no need for rt_trie_node anymore since everyting can be
merged into tnode.
On 32b systems this results in the leaf being 4 bytes larger, however I
don't know if that is really an issue as this and an eariler patch that
added bits & pos have increased the size from 20 to 28. If I am not
mistaken slub/slab allocate on power of 2 sizes so 20 was likely being
rounded up to 32 anyway.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Both the leaf and the tnode had an rcu_head in them, but they had them in
slightly different places. Since we now have them in the same spot and
know that any node with bits == 0 is a leaf and the rest are either vmalloc
or kmalloc tnodes depending on the value of bits it makes it easy to combine
the functions and reduce overhead.
In addition I have taken advantage of the rcu_head pointer to go ahead and
put together a simple linked list instead of using the tnode pointer as
this way we can merge either type of structure for freeing.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change makes some fundamental changes to the way leaves and tnodes are
constructed. The big differences are:
1. Leaves now populate pos and bits indicating their full key size.
2. Trie nodes now mask out their lower bits to be consistent with the leaf
3. Both structures have been reordered so that rt_trie_node now consisists
of a much larger region including the pos, bits, and rcu portions of
the tnode structure.
On 32b systems this will result in the leaf being 4B larger as the pos and
bits values were added to a hole created by the key as it was only 4B in
length.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The trie usage stats were currently being shared by all threads that were
calling fib_table_lookup. As a result when multiple threads were
performing lookups simultaneously the trie would begin to cache bounce
between those threads.
In order to prevent this I have updated the usage stats to use a set of
percpu variables. By doing this we should be able to avoid the cache
bouncing and still make use of these stats.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The GRE tap device supports Ethernet over GRE, but doesn't
care about the source address of the tunnel, therefore it
can be changed without bring device down.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Today vport-send has complex error handling because it involves
freeing skb and updating stats depending on return value from
vport send implementation.
This can be simplified by delegating responsibility of freeing
skb to the vport implementation for all cases. So that
vport-send needs just update stats.
Fixes: 91b7514cdf ("openvswitch: Unify vport error stats
handling")
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, searching for a socket to add a reference to is not
synchronized with deletion of sockets. This can result in use
after free if there is another operation that is removing a
socket at the same time. Solving this requires both holding the
appropriate lock and checking the refcount to ensure that it
has not already hit zero.
Inspired by a related (but not exactly the same) issue in the
VXLAN driver.
Fixes: 0b5e8b8e ("net: Add Geneve tunneling protocol driver")
CC: Andy Zhou <azhou@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Sockets aren't currently removed from the the global list when
they are destroyed. In addition, offload handlers need to be cleaned
up as well.
Fixes: 0b5e8b8e ("net: Add Geneve tunneling protocol driver")
CC: Andy Zhou <azhou@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
The encap->type comes straight from Netlink. Validate it against
max supported encap types just like ip_encap_hlen() already does.
Fixes: a8c5f9 ("ip_tunnel: Ops registration for secondary encap (fou, gue)")
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
The symbols are exported and could be used by external modules.
Fixes: a8c5f9 ("ip_tunnel: Ops registration for secondary encap (fou, gue)")
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
The NBMA GRE tunnels temporarily push GRE header that contain the
per-packet NBMA destination on the skb via header ops early in xmit
path. It is the later pulled before the real GRE header is constructed.
The inner mac was thus set differently in nbma case: the GRE header
has been pushed by neighbor layer, and mac header points to beginning
of the temporary gre header (set by dev_queue_xmit).
Now that the offloads expect mac header to point to the gre payload,
fix the xmit patch to:
- pull first the temporary gre header away
- and reset mac header to point to gre payload
This fixes tso to work again with nbma tunnels.
Fixes: 14051f0452 ("gre: Use inner mac length when computing tunnel length")
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Cc: Tom Herbert <therbert@google.com>
Cc: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch addresses an issue with the level compression of the fib_trie.
Specifically in the case of adding a new leaf that triggers a new node to
be added that takes the place of the old node. The result is a trie where
the 1 child tnode is on one side and one leaf is on the other which gives
you a very deep trie. Below is the script I used to generate a trie on
dummy0 with a 10.X.X.X family of addresses.
ip link add type dummy
ipval=184549374
bit=2
for i in `seq 1 23`
do
ifconfig dummy0:$bit $ipval/8
ipval=`expr $ipval - $bit`
bit=`expr $bit \* 2`
done
cat /proc/net/fib_triestat
Running the script before the patch:
Local:
Aver depth: 10.82
Max depth: 23
Leaves: 29
Prefixes: 30
Internal nodes: 27
1: 26 2: 1
Pointers: 56
Null ptrs: 1
Total size: 5 kB
After applying the patch and repeating:
Local:
Aver depth: 4.72
Max depth: 9
Leaves: 29
Prefixes: 30
Internal nodes: 12
1: 3 2: 2 3: 7
Pointers: 70
Null ptrs: 30
Total size: 4 kB
What this fix does is start the rebalance at the newly created tnode
instead of at the parent tnode. This way if there is a gap between the
parent and the new node it doesn't prevent the new tnode from being
coalesced with any pre-existing nodes that may have been pushed into one
of the new nodes child branches.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking updates from David Miller:
1) New offloading infrastructure and example 'rocker' driver for
offloading of switching and routing to hardware.
This work was done by a large group of dedicated individuals, not
limited to: Scott Feldman, Jiri Pirko, Thomas Graf, John Fastabend,
Jamal Hadi Salim, Andy Gospodarek, Florian Fainelli, Roopa Prabhu
2) Start making the networking operate on IOV iterators instead of
modifying iov objects in-situ during transfers. Thanks to Al Viro
and Herbert Xu.
3) A set of new netlink interfaces for the TIPC stack, from Richard
Alpe.
4) Remove unnecessary looping during ipv6 routing lookups, from Martin
KaFai Lau.
5) Add PAUSE frame generation support to gianfar driver, from Matei
Pavaluca.
6) Allow for larger reordering levels in TCP, which are easily
achievable in the real world right now, from Eric Dumazet.
7) Add a variable of napi_schedule that doesn't need to disable cpu
interrupts, from Eric Dumazet.
8) Use a doubly linked list to optimize neigh_parms_release(), from
Nicolas Dichtel.
9) Various enhancements to the kernel BPF verifier, and allow eBPF
programs to actually be attached to sockets. From Alexei
Starovoitov.
10) Support TSO/LSO in sunvnet driver, from David L Stevens.
11) Allow controlling ECN usage via routing metrics, from Florian
Westphal.
12) Remote checksum offload, from Tom Herbert.
13) Add split-header receive, BQL, and xmit_more support to amd-xgbe
driver, from Thomas Lendacky.
14) Add MPLS support to openvswitch, from Simon Horman.
15) Support wildcard tunnel endpoints in ipv6 tunnels, from Steffen
Klassert.
16) Do gro flushes on a per-device basis using a timer, from Eric
Dumazet. This tries to resolve the conflicting goals between the
desired handling of bulk vs. RPC-like traffic.
17) Allow userspace to ask for the CPU upon what a packet was
received/steered, via SO_INCOMING_CPU. From Eric Dumazet.
18) Limit GSO packets to half the current congestion window, from Eric
Dumazet.
19) Add a generic helper so that all drivers set their RSS keys in a
consistent way, from Eric Dumazet.
20) Add xmit_more support to enic driver, from Govindarajulu
Varadarajan.
21) Add VLAN packet scheduler action, from Jiri Pirko.
22) Support configurable RSS hash functions via ethtool, from Eyal
Perry.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1820 commits)
Fix race condition between vxlan_sock_add and vxlan_sock_release
net/macb: fix compilation warning for print_hex_dump() called with skb->mac_header
net/mlx4: Add support for A0 steering
net/mlx4: Refactor QUERY_PORT
net/mlx4_core: Add explicit error message when rule doesn't meet configuration
net/mlx4: Add A0 hybrid steering
net/mlx4: Add mlx4_bitmap zone allocator
net/mlx4: Add a check if there are too many reserved QPs
net/mlx4: Change QP allocation scheme
net/mlx4_core: Use tasklet for user-space CQ completion events
net/mlx4_core: Mask out host side virtualization features for guests
net/mlx4_en: Set csum level for encapsulated packets
be2net: Export tunnel offloads only when a VxLAN tunnel is created
gianfar: Fix dma check map error when DMA_API_DEBUG is enabled
cxgb4/csiostor: Don't use MASTER_MUST for fw_hello call
net: fec: only enable mdio interrupt before phy device link up
net: fec: clear all interrupt events to support i.MX6SX
net: fec: reset fep link status in suspend function
net: sock: fix access via invalid file descriptor
net: introduce helper macro for_each_cmsghdr
...
Introduce helper macro for_each_cmsghdr as a wrapper of the enumerating
cmsghdr from msghdr, just cleanup.
Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Merge first patchbomb from Andrew Morton:
- a few minor cifs fixes
- dma-debug upadtes
- ocfs2
- slab
- about half of MM
- procfs
- kernel/exit.c
- panic.c tweaks
- printk upates
- lib/ updates
- checkpatch updates
- fs/binfmt updates
- the drivers/rtc tree
- nilfs
- kmod fixes
- more kernel/exit.c
- various other misc tweaks and fixes
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (190 commits)
exit: pidns: fix/update the comments in zap_pid_ns_processes()
exit: pidns: alloc_pid() leaks pid_namespace if child_reaper is exiting
exit: exit_notify: re-use "dead" list to autoreap current
exit: reparent: call forget_original_parent() under tasklist_lock
exit: reparent: avoid find_new_reaper() if no children
exit: reparent: introduce find_alive_thread()
exit: reparent: introduce find_child_reaper()
exit: reparent: document the ->has_child_subreaper checks
exit: reparent: s/while_each_thread/for_each_thread/ in find_new_reaper()
exit: reparent: fix the cross-namespace PR_SET_CHILD_SUBREAPER reparenting
exit: reparent: fix the dead-parent PR_SET_CHILD_SUBREAPER reparenting
exit: proc: don't try to flush /proc/tgid/task/tgid
exit: release_task: fix the comment about group leader accounting
exit: wait: drop tasklist_lock before psig->c* accounting
exit: wait: don't use zombie->real_parent
exit: wait: cleanup the ptrace_reparented() checks
usermodehelper: kill the kmod_thread_locker logic
usermodehelper: don't use CLONE_VFORK for ____call_usermodehelper()
fs/hfs/catalog.c: fix comparison bug in hfs_cat_keycmp
nilfs2: fix the nilfs_iget() vs. nilfs_new_inode() races
...
Memory is internally accounted in bytes, using spinlock-protected 64-bit
counters, even though the smallest accounting delta is a page. The
counter interface is also convoluted and does too many things.
Introduce a new lockless word-sized page counter API, then change all
memory accounting over to it. The translation from and to bytes then only
happens when interfacing with userspace.
The removed locking overhead is noticable when scaling beyond the per-cpu
charge caches - on a 4-socket machine with 144-threads, the following test
shows the performance differences of 288 memcgs concurrently running a
page fault benchmark:
vanilla:
18631648.500498 task-clock (msec) # 140.643 CPUs utilized ( +- 0.33% )
1,380,638 context-switches # 0.074 K/sec ( +- 0.75% )
24,390 cpu-migrations # 0.001 K/sec ( +- 8.44% )
1,843,305,768 page-faults # 0.099 M/sec ( +- 0.00% )
50,134,994,088,218 cycles # 2.691 GHz ( +- 0.33% )
<not supported> stalled-cycles-frontend
<not supported> stalled-cycles-backend
8,049,712,224,651 instructions # 0.16 insns per cycle ( +- 0.04% )
1,586,970,584,979 branches # 85.176 M/sec ( +- 0.05% )
1,724,989,949 branch-misses # 0.11% of all branches ( +- 0.48% )
132.474343877 seconds time elapsed ( +- 0.21% )
lockless:
12195979.037525 task-clock (msec) # 133.480 CPUs utilized ( +- 0.18% )
832,850 context-switches # 0.068 K/sec ( +- 0.54% )
15,624 cpu-migrations # 0.001 K/sec ( +- 10.17% )
1,843,304,774 page-faults # 0.151 M/sec ( +- 0.00% )
32,811,216,801,141 cycles # 2.690 GHz ( +- 0.18% )
<not supported> stalled-cycles-frontend
<not supported> stalled-cycles-backend
9,999,265,091,727 instructions # 0.30 insns per cycle ( +- 0.10% )
2,076,759,325,203 branches # 170.282 M/sec ( +- 0.12% )
1,656,917,214 branch-misses # 0.08% of all branches ( +- 0.55% )
91.369330729 seconds time elapsed ( +- 0.45% )
On top of improved scalability, this also gets rid of the icky long long
types in the very heart of memcg, which is great for 32 bit and also makes
the code a lot more readable.
Notable differences between the old and new API:
- res_counter_charge() and res_counter_charge_nofail() become
page_counter_try_charge() and page_counter_charge() resp. to match
the more common kernel naming scheme of try_do()/do()
- res_counter_uncharge_until() is only ever used to cancel a local
counter and never to uncharge bigger segments of a hierarchy, so
it's replaced by the simpler page_counter_cancel()
- res_counter_set_limit() is replaced by page_counter_limit(), which
expects its callers to serialize against themselves
- res_counter_memparse_write_strategy() is replaced by
page_counter_limit(), which rounds down to the nearest page size -
rather than up. This is more reasonable for explicitely requested
hard upper limits.
- to keep charging light-weight, page_counter_try_charge() charges
speculatively, only to roll back if the result exceeds the limit.
Because of this, a failing bigger charge can temporarily lock out
smaller charges that would otherwise succeed. The error is bounded
to the difference between the smallest and the biggest possible
charge size, so for memcg, this means that a failing THP charge can
send base page charges into reclaim upto 2MB (4MB) before the limit
would have been reached. This should be acceptable.
[akpm@linux-foundation.org: add includes for WARN_ON_ONCE and memparse]
[akpm@linux-foundation.org: add includes for WARN_ON_ONCE, memparse, strncmp, and PAGE_SIZE]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull VFS changes from Al Viro:
"First pile out of several (there _definitely_ will be more). Stuff in
this one:
- unification of d_splice_alias()/d_materialize_unique()
- iov_iter rewrite
- killing a bunch of ->f_path.dentry users (and f_dentry macro).
Getting that completed will make life much simpler for
unionmount/overlayfs, since then we'll be able to limit the places
sensitive to file _dentry_ to reasonably few. Which allows to have
file_inode(file) pointing to inode in a covered layer, with dentry
pointing to (negative) dentry in union one.
Still not complete, but much closer now.
- crapectomy in lustre (dead code removal, mostly)
- "let's make seq_printf return nothing" preparations
- assorted cleanups and fixes
There _definitely_ will be more piles"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (63 commits)
copy_from_iter_nocache()
new helper: iov_iter_kvec()
csum_and_copy_..._iter()
iov_iter.c: handle ITER_KVEC directly
iov_iter.c: convert copy_to_iter() to iterate_and_advance
iov_iter.c: convert copy_from_iter() to iterate_and_advance
iov_iter.c: get rid of bvec_copy_page_{to,from}_iter()
iov_iter.c: convert iov_iter_zero() to iterate_and_advance
iov_iter.c: convert iov_iter_get_pages_alloc() to iterate_all_kinds
iov_iter.c: convert iov_iter_get_pages() to iterate_all_kinds
iov_iter.c: convert iov_iter_npages() to iterate_all_kinds
iov_iter.c: iterate_and_advance
iov_iter.c: macros for iterating over iov_iter
kill f_dentry macro
dcache: fix kmemcheck warning in switch_names
new helper: audit_file()
nfsd_vfs_write(): use file_inode()
ncpfs: use file_inode()
kill f_dentry uses
lockd: get rid of ->f_path.dentry->d_sb
...
Conflicts:
drivers/net/ethernet/amd/xgbe/xgbe-desc.c
drivers/net/ethernet/renesas/sh_eth.c
Overlapping changes in both conflict cases.
Signed-off-by: David S. Miller <davem@davemloft.net>
When I cooked commit c3658e8d0f ("tcp: fix possible NULL dereference in
tcp_vX_send_reset()") I missed other spots we could deref a NULL
skb_dst(skb)
Again, if a socket is provided, we do not need skb_dst() to get a
pointer to network namespace : sock_net(sk) is good enough.
Reported-by: Dann Frazier <dann.frazier@canonical.com>
Bisected-by: Dann Frazier <dann.frazier@canonical.com>
Tested-by: Dann Frazier <dann.frazier@canonical.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: ca777eff51 ("tcp: remove dst refcount false sharing for prequeue mode")
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 95bd09eb27 ("tcp: TSO packets automatic sizing") tried to
control TSO size, but did this at the wrong place (sendmsg() time)
At sendmsg() time, we might have a pessimistic view of flow rate,
and we end up building very small skbs (with 2 MSS per skb).
This is bad because :
- It sends small TSO packets even in Slow Start where rate quickly
increases.
- It tends to make socket write queue very big, increasing tcp_ack()
processing time, but also increasing memory needs, not necessarily
accounted for, as fast clones overhead is currently ignored.
- Lower GRO efficiency and more ACK packets.
Servers with a lot of small lived connections suffer from this.
Lets instead fill skbs as much as possible (64KB of payload), but split
them at xmit time, when we have a precise idea of the flow rate.
skb split is actually quite efficient.
Patch looks bigger than necessary, because TCP Small Queue decision now
has to take place after the eventual split.
As Neal suggested, introduce a new tcp_tso_autosize() helper, so that
tcp_tso_should_defer() can be synchronized on same goal.
Rename tp->xmit_size_goal_segs to tp->gso_segs, as this variable
contains number of mss that we can put in GSO packet, and is not
related to the autosizing goal anymore.
Tested:
40 ms rtt link
nstat >/dev/null
netperf -H remote -l -2000000 -- -s 1000000
nstat | egrep "IpInReceives|IpOutRequests|TcpOutSegs|IpExtOutOctets"
Before patch :
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/s
87380 2000000 2000000 0.36 44.22
IpInReceives 600 0.0
IpOutRequests 599 0.0
TcpOutSegs 1397 0.0
IpExtOutOctets 2033249 0.0
After patch :
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
87380 2000000 2000000 0.36 44.27
IpInReceives 221 0.0
IpOutRequests 232 0.0
TcpOutSegs 1397 0.0
IpExtOutOctets 2013953 0.0
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
