The transition from markov state "3 => lost packets within a burst
period" to "1 => successfully transmitted packets within a gap period"
has no *additional* loss event. The loss already happen for transition
from 1 -> 3, this additional loss will make things go wild.
E.g. transition probabilities:
p13: 10%
p31: 100%
Expected:
Ploss = p13 / (p13 + p31)
Ploss = ~9.09%
... but it isn't. Even worse: we get a double loss - each time.
So simple don't return true to indicate loss, rather break and return
false.
Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Stefano Salsano <stefano.salsano@uniroma2.it>
Cc: Fabio Ludovici <fabio.ludovici@yahoo.it>
Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
- data structure reshaping to accommodate multiple routing protocol
implementations
- routing protocol API enhancement
- send to userspace the event "batman-adv Gateway loss" in case of soft-iface
destruction and a "batman-adv Gateway" was configured
- improve the TT component to support and advertise runtime flag changes
- minor code refactoring
- make the ICMP kernel-to-userspace communication more generic
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)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=sYKR
-----END PGP SIGNATURE-----
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
Antonio Quartulli says:
====================
this is another set of changes intended for net-next/linux-3.13.
(probably our last pull request for this cycle)
Patches 1 and 2 reshape two of our main data structures in a way that they can
easily be extended in the future to accommodate new routing protocols.
Patches from 3 to 9 improve our routing protocol API and its users so that all
the protocol-related code is not mixed up with the other components anymore.
Patch 10 limits the local Translation Table maximum size to a value such that it
can be fully transfered over the air if needed. This value depends on
fragmentation being enabled or not and on the mtu values.
Patch 11 makes batman-adv send a uevent in case of soft-interface destruction
while a "bat-Gateway" was configured (this informs userspace about the GW not
being available anymore).
Patches 13 and 14 enable the TT component to detect non-mesh client flag
changes at runtime (till now those flags where set upon client detection and
were not changed anymore).
Patch 16 is a generalisation of our user-to-kernel space communication (and
viceversa) used to exchange ICMP packets to send/received to/from the mesh
network. Now it can easily accommodate new ICMP packet types without breaking
the existing userspace API anymore.
Remaining patches are minor changes and cleanups.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
All fragmentation hash secrets now get initialized by their
corresponding hash function with net_get_random_once. Thus we can
eliminate the initial seeding.
Also provide a comment that hash secret seeding happens at the first
call to the corresponding hashing function.
Cc: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Defer the fragmentation hash secret initialization for IPv6 like the
previous patch did for IPv4.
Because the netfilter logic reuses the hash secret we have to split it
first. Thus introduce a new nf_hash_frag function which takes care to
seed the hash secret.
Cc: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Defer the generation of the first hash secret for the ipv4 fragmentation
cache as late as possible.
ip4_frags.rnd gets initial seeded by inet_frags_init and regulary
reseeded by inet_frag_secret_rebuild. Either we call ipqhashfn directly
from ip_fragment.c in which case we initialize the secret directly.
If we first get called by inet_frag_secret_rebuild we install a new secret
by a manual call to get_random_bytes. This secret will be overwritten
as soon as the first call to ipqhashfn happens. This is safe because we
won't race while publishing the new secrets with anyone else.
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/usb/qmi_wwan.c
include/net/dst.h
Trivial merge conflicts, both were overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently net_secret_init does not get inlined, so we always have a call
to net_secret_init even in the fast path.
Let's specify net_secret_init as __always_inline so we have the nop in
the fast-path without the call to net_secret_init and the unlikely path
at the epilogue of the function.
jump_labels handle the inlining correctly.
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of handling icmp packets only up to length of icmp_packet_rr,
the code should handle any icmp length size. Therefore the length
truncating is moved to when the packet is actually sent to userspace
(this does not support lengths longer than icmp_packet_rr yet). Longer
packets are forwarded without truncating.
This patch also cleans up some parts where the icmp header struct could
be used instead of other icmp_packet(_rr) structs to make the code more
readable.
Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <antonio@meshcoding.com>
Flags covered by TT_SYNC_MASK are kept in sync among the
nodes in the network and therefore they have to be
considered while computing the global/local table CRC.
In this way a generic originator is able to understand if
its table contains the correct flags or not.
Bits from 4 to 7 in the TT flags fields are now reserved for
"synchronized" flags only.
This allows future developers to add more flags of this type
without breaking compatibility.
It's important to note that not all the remote TT flags are
synchronised. This comes from the fact that some flags are
used to inject an information once only.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Some flags (i.e. the WIFI flag) may change after that the
related client has already been announced. However it is
useful to informa the rest of the network about this change.
Add a runtime-flag-switch detection mechanism and
re-announce the related TT entry to advertise the new flag
value.
This mechanism can be easily exploited by future flags that
may need the same treatment.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Upcoming changes need to perform other checks on the
incoming net_device struct.
To avoid performing dev_get_by_index() for each and every
check, it is better to move it outside of is_wifi_iface()
and search the netdev object once only.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
In case of soft_iface destruction send a GW DEL event to
userspace so that applications which are listening for GW
events are informed about the lost of connectivity and can
react accordingly.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
The local translation table size is limited by what can be
transferred from one node to another via a full table request.
The number of entries fitting into a full table request depend
on whether the fragmentation is enabled or not. Therefore this
patch introduces a max table size check and refuses to add
more local clients when that size is reached. Moreover, if the
max full table packet size changes (MTU change or fragmentation
is disabled) the local table is downsized instantaneously.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Some operations executed on an orig_node depends on the
current routing algorithm being used. To easily make this
mechanism routing algorithm agnostic add a orig_node
specific API that each algorithm can populate with its own
routines.
Such routines are then invoked by the code when needed,
without knowing which routing algorithm is currently in use
With this patch 3 API functions are added:
- orig_free (to free routing depending internal structs)
- orig_add_if (to change the inner state of an orig_node
when a new hard interface is added)
- orig_del_if (to change the inner state of an orig_node
when an hard interface is removed)
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Each routing protocol has its own metric semantic and
therefore is the protocol itself the only component able to
compare two metrics to check their "similarity".
This new API allows each routing protocol to implement its
own logic and make the external code protocol agnostic.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
This new API allows to compare the two neighbours based on
the metric avoiding the user to deal with any routing
algorithm specific detail
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Each routing protocol has its own metric and private
variables, therefore it is useful to introduce a new API
for originator information printing.
This API needs to be implemented by each protocol in order
to provide its specific originator table output.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
some of the struct batadv_orig_node members are B.A.T.M.A.N. IV
specific and therefore they are moved in a algorithm specific
substruct in order to make batadv_orig_node routing algorithm
agnostic
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
some of the fields in struct batadv_neigh_node are strictly
related to the B.A.T.M.A.N. IV algorithm. In order to
make the struct usable by any routing algorithm it has to be
split and made more generic
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
While this commit was a good attempt to fix issues occuring when no
multicast querier is present, this commit still has two more issues:
1) There are cases where mdb entries do not expire even if there is a
querier present. The bridge will unnecessarily continue flooding
multicast packets on the according ports.
2) Never removing an mdb entry could be exploited for a Denial of
Service by an attacker on the local link, slowly, but steadily eating up
all memory.
Actually, this commit became obsolete with
"bridge: disable snooping if there is no querier" (b00589af3b)
which included fixes for a few more cases.
Therefore reverting the following commits (the commit stated in the
commit message plus three of its follow up fixes):
====================
Revert "bridge: update mdb expiration timer upon reports."
This reverts commit f144febd93.
Revert "bridge: do not call setup_timer() multiple times"
This reverts commit 1faabf2aab.
Revert "bridge: fix some kernel warning in multicast timer"
This reverts commit c7e8e8a8f7.
Revert "bridge: only expire the mdb entry when query is received"
This reverts commit 9f00b2e7cf.
====================
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Linus Lüssing <linus.luessing@web.de>
Reviewed-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
What sk_reset_txq() does is just calls function sk_tx_queue_reset(),
and sk_reset_txq() is used only in sock.h, by dst_negative_advice().
Let dst_negative_advice() calls sk_tx_queue_reset() directly so we
can remove unneeded sk_reset_txq().
Signed-off-by: ZHAO Gang <gamerh2o@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For passive TCP connections, upon receiving the ACK that completes the
3WHS, make sure we set our pacing rate after we get our first RTT
sample.
On passive TCP connections, when we receive the ACK completing the
3WHS we do not take an RTT sample in tcp_ack(), but rather in
tcp_synack_rtt_meas(). So upon receiving the ACK that completes the
3WHS, tcp_ack() leaves sk_pacing_rate at its initial value.
Originally the initial sk_pacing_rate value was 0, so passive-side
connections defaulted to sysctl_tcp_min_tso_segs (2 segs) in skbuffs
made in the first RTT. With a default initial cwnd of 10 packets, this
happened to be correct for RTTs 5ms or bigger, so it was hard to
see problems in WAN or emulated WAN testing.
Since 7eec4174ff ("pkt_sched: fq: fix non TCP flows pacing"), the
initial sk_pacing_rate is 0xffffffff. So after that change, passive
TCP connections were keeping this value (and using large numbers of
segments per skbuff) until receiving an ACK for data.
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Routes need to be probed asynchronous otherwise the call stack gets
exhausted when the kernel attemps to deliver another skb inline, like
e.g. xt_TEE does, and we probe at the same time.
We update neigh->updated still at once, otherwise we would send to
many probes.
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now ipv6_gso_segment() is stackable, its relatively easy to
implement GSO/TSO support for SIT tunnels
Performance results, when segmentation is done after tunnel
device (as no NIC is yet enabled for TSO SIT support) :
Before patch :
lpq84:~# ./netperf -H 2002:af6:1153:: -Cc
MIGRATED TCP STREAM TEST from ::0 (::) port 0 AF_INET6 to 2002:af6:1153:: () port 0 AF_INET6
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 16384 16384 10.00 3168.31 4.81 4.64 2.988 2.877
After patch :
lpq84:~# ./netperf -H 2002:af6:1153:: -Cc
MIGRATED TCP STREAM TEST from ::0 (::) port 0 AF_INET6 to 2002:af6:1153:: () port 0 AF_INET6
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 16384 16384 10.00 5525.00 7.76 5.17 2.763 1.840
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to support GSO on SIT tunnels, we need to make
inet_gso_segment() stackable.
It should not assume network header starts right after mac
header.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow unprivileged users to use:
/proc/sys/net/ipv4/icmp_echo_ignore_all
/proc/sys/net/ipv4/icmp_echo_ignore_broadcasts
/proc/sys/net/ipv4/icmp_ignore_bogus_error_response
/proc/sys/net/ipv4/icmp_errors_use_inbound_ifaddr
/proc/sys/net/ipv4/icmp_ratelimit
/proc/sys/net/ipv4/icmp_ratemask
/proc/sys/net/ipv4/ping_group_range
/proc/sys/net/ipv4/tcp_ecn
/proc/sys/net/ipv4/ip_local_ports_range
These are occassionally handy and after a quick review I don't see
any problems with unprivileged users using them.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simplify maintenance of ipv4_net_table by using math to point the per
net sysctls into the appropriate struct net, instead of manually
reassinging all of the variables into hard coded table slots.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace the pointers in struct cg_proto with actual data fields and kill
struct tcp_memcontrol as it is not fully redundant.
This removes a confusing, unnecessary layer of abstraction.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The code that is implemented is per memory cgroup not per netns, and
having per netns bits is just confusing. Remove the per netns bits to
make it easier to see what is really going on.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The code is broken and does not constrain sysctl_tcp_mem as
tcp_update_limit does. With the result that it allows the cgroup tcp
memory limits to be bypassed.
The semantics are broken as the settings are not per netns and are in a
per netns table, and instead looks at current.
Since the code is broken in both design and implementation and does not
implement the functionality for which it was written remove it.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This function is never called. Remove it.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now when rt6_nexthop() can return nexthop address we can use it
for proper nexthop comparison of directly connected destinations.
For more information refer to commit bbb5823cf7
("netfilter: nf_conntrack: fix rt_gateway checks for H.323 helper").
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure rt6i_gateway contains nexthop information in
all routes returned from lookup or when routes are directly
attached to skb for generated ICMP packets.
The effect of this patch should be a faster version of
rt6_nexthop() and the consideration of local addresses as
nexthop.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
- email addresses update in documentation, source files and MAINTAINERS
- make the TT component distinguish non-mesh clients based on the VLAN they
belong to
- improve all the internal components to properly work on a per-VLAN basis
(enabled by the new TT-VLAN feature)
- enhance the sysfs interface in order to provide behaviour switches on a
per-VLAN basis (enabled by the new TT-VLAN feature)
- improve TT lock mechanism
- improve unicast transmission APIs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)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=sEEg
-----END PGP SIGNATURE-----
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
Antonio Quartulli says:
====================
this is another batch intended for net-next/linux-3.13.
This pull request is a bit bigger than usual, but 6 patches are very small
(three of them are about email updates)..
Patch 1 is fixing a previous merge conflict resolution that went wrong
(I realised that only now while checking other patches..).
Patches from 2 to 4 that are updating our emails in all the proper files
(Documentation/, headers and MAINTAINERS).
Patches 5, 6 and 7 are bringing a big improvement to the TranslationTable
component: it is now able to group non-mesh clients based on the VLAN they
belong to. In this way a lot a new enhancements are now possible thanks to the
fact that each batman-adv behaviour can be applied on a per VLAN basis.
And, of course, in patches from 8 to 12 you have some of the enhancements I was
talking about:
- make the batman-Gateway selection VLAN dependent
- make DAT (Distributed ARP Table) group ARP entries on a VLAN basis (this
allows DAT to work even when the admin decided to use the same IP subnet on
different VLANs)
- make the AP-Isolation behaviour switchable on each VLAN independently
- export VLAN specific attributes via sysfs. Switches like the AP-Isolation are
now exported once per VLAN (backward compatibility of the sysfs interface has
been preserved)
Patches 13 and 14 are small code cleanups.
Patch 15 is a minor improvement in the TT locking mechanism.
Patches 16 and 17 are other enhancements to the TT component. Those allow a
node to parse a "non-mesh client announcement message" and accept only those
TT entries belonging to certain VLANs.
Patch 18 exploits this parse&accept mechanism to make the Bridge Loop Avoidance
component reject only TT entries connected to the VLAN where it is operating.
Previous to this change, BLA was rejecting all the entries coming from any other
Backbone node, regardless of the VLAN (for more details about how the Bridge
Loop Avoidance works please check [1]).
[1] http://www.open-mesh.org/projects/batman-adv/wiki/Bridge-loop-avoidance-II
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Changed key initialization of tcp_fastopen cookies to net_get_random_once.
If the user sets a custom key net_get_random_once must be called at
least once to ensure we don't overwrite the user provided key when the
first cookie is generated later on.
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Initialize the ehash and ipv6_hash_secrets with net_get_random_once.
Each compilation unit gets its own secret now:
ipv4/inet_hashtables.o
ipv4/udp.o
ipv6/inet6_hashtables.o
ipv6/udp.o
rds/connection.o
The functions still get inlined into the hashing functions. In the fast
path we have at most two (needed in ipv6) if (unlikely(...)).
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch splits the secret key for syncookies for ipv4 and ipv6 and
initializes them with net_get_random_once. This change was the reason I
did this series. I think the initialization of the syncookie_secret is
way to early.
Cc: Florian Westphal <fw@strlen.de>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
net_get_random_once is a new macro which handles the initialization
of secret keys. It is possible to call it in the fast path. Only the
initialization depends on the spinlock and is rather slow. Otherwise
it should get used just before the key is used to delay the entropy
extration as late as possible to get better randomness. It returns true
if the key got initialized.
The usage of static_keys for net_get_random_once is a bit uncommon so
it needs some further explanation why this actually works:
=== In the simple non-HAVE_JUMP_LABEL case we actually have ===
no constrains to use static_key_(true|false) on keys initialized with
STATIC_KEY_INIT_(FALSE|TRUE). So this path just expands in favor of
the likely case that the initialization is already done. The key is
initialized like this:
___done_key = { .enabled = ATOMIC_INIT(0) }
The check
if (!static_key_true(&___done_key)) \
expands into (pseudo code)
if (!likely(___done_key > 0))
, so we take the fast path as soon as ___done_key is increased from the
helper function.
=== If HAVE_JUMP_LABELs are available this depends ===
on patching of jumps into the prepared NOPs, which is done in
jump_label_init at boot-up time (from start_kernel). It is forbidden
and dangerous to use net_get_random_once in functions which are called
before that!
At compilation time NOPs are generated at the call sites of
net_get_random_once. E.g. net/ipv6/inet6_hashtable.c:inet6_ehashfn (we
need to call net_get_random_once two times in inet6_ehashfn, so two NOPs):
71: 0f 1f 44 00 00 nopl 0x0(%rax,%rax,1)
76: 0f 1f 44 00 00 nopl 0x0(%rax,%rax,1)
Both will be patched to the actual jumps to the end of the function to
call __net_get_random_once at boot time as explained above.
arch_static_branch is optimized and inlined for false as return value and
actually also returns false in case the NOP is placed in the instruction
stream. So in the fast case we get a "return false". But because we
initialize ___done_key with (enabled != (entries & 1)) this call-site
will get patched up at boot thus returning true. The final check looks
like this:
if (!static_key_true(&___done_key)) \
___ret = __net_get_random_once(buf, \
expands to
if (!!static_key_false(&___done_key)) \
___ret = __net_get_random_once(buf, \
So we get true at boot time and as soon as static_key_slow_inc is called
on the key it will invert the logic and return false for the fast path.
static_key_slow_inc will change the branch because it got initialized
with .enabled == 0. After static_key_slow_inc is called on the key the
branch is replaced with a nop again.
=== Misc: ===
The helper defers the increment into a workqueue so we don't
have problems calling this code from atomic sections. A seperate boolean
(___done) guards the case where we enter net_get_random_once again before
the increment happend.
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch splits the inet6_ehashfn into separate ones in
ipv6/inet6_hashtables.o and ipv6/udp.o to ease the introduction of
seperate secrets keys later.
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This duplicates a bit of code but let's us easily introduce
separate secret keys later. The separate compilation units are
ipv4/inet_hashtabbles.o, ipv4/udp.o and rds/connection.o.
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now inet_gso_segment() is stackable, its relatively easy to
implement GSO/TSO support for IPIP
Performance results, when segmentation is done after tunnel
device (as no NIC is yet enabled for TSO IPIP support) :
Before patch :
lpq83:~# ./netperf -H 7.7.9.84 -Cc
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 7.7.9.84 () port 0 AF_INET
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 16384 16384 10.00 3357.88 5.09 3.70 2.983 2.167
After patch :
lpq83:~# ./netperf -H 7.7.9.84 -Cc
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 7.7.9.84 () port 0 AF_INET
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 16384 16384 10.00 7710.19 4.52 6.62 1.152 1.687
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to support GSO on IPIP, we need to make
inet_gso_segment() stackable.
It should not assume network header starts right after mac
header.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch makes gre_handle_offloads() more generic
and rename it to iptunnel_handle_offloads()
This will be used to add GSO/TSO support to IPIP tunnels.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While implementing GSO/TSO support for IPIP, I found skb_segment()
was assuming network header was immediately following mac header.
Its not really true in the case inet_gso_segment() is stacked :
By the time tcp_gso_segment() is called, network header points
to the inner IP header.
Let's instead assume nothing and pick the current offsets found in
original skb, we have skb_headers_offset_update() helper for that.
Also move the csum_start update inside skb_headers_offset_update()
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now, if user application does:
sendto len<mtu flag MSG_MORE
sendto len>mtu flag 0
The skb is not treated as fragmented one because it is not initialized
that way. So move the initialization to fix this.
introduced by:
commit e89e9cf539 "[IPv4/IPv6]: UFO Scatter-gather approach"
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
