summaryrefslogtreecommitdiff
path: root/net
AgeCommit message (Collapse)AuthorFilesLines
2012-01-03net_sched: qdisc_alloc_handle() can be too slowEric Dumazet1-5/+9
When trying to allocate ~32768 qdiscs using autohandle mechanism, we can fill the space managed by kernel (handles in [8000-FFFF]:0000 range) But O(N^2) qdisc_alloc_handle() loops 0x10000 times instead of 0x8000 time tc add qdisc add dev eth0 parent 10:7fff pfifo limit 10 RTNETLINK answers: Cannot allocate memory real 1m54.826s user 0m0.000s sys 0m0.004s INFO: rcu_sched_state detected stall on CPU 0 (t=60000 jiffies) Half number of loops, and add a cond_resched() call. We hold rtnl at this point. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Dave Taht <dave.taht@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-01-03sch_qfq: accurate wsum handlingEric Dumazet1-8/+9
We can underestimate q->wsum in case of "tc class replace ... qfq" and/or qdisc_create_dflt() error. wsum is not really used in fast path, only at qfq qdisc/class setup, to catch user error. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-01-03sch_sfq: dont put new flow at the end of flowsEric Dumazet1-1/+1
SFQ enqueue algo puts a new flow _behind_ all pre-existing flows in the circular list. In fact this is probably an old SFQ implementation bug. 100 Mbits = ~8333 full frames per second, or ~8 frames per ms. With 50 flows, it means your "new flow" will have to wait 50 packets being sent before its own packet. Thats the ~6ms. We certainly can change SFQ to give a priority advantage to new flows, so that next dequeued packet is taken from a new flow, not an old one. Reported-by: Dave Taht <dave.taht@gmail.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-01-03Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/netDavid S. Miller4-6/+12
2012-01-01netfilter: nfnetlink_acct: fix nfnl_acct_get operationPablo Neira Ayuso1-4/+13
The get operation was not sending the message that was built to user-space. This patch also includes the appropriate handling for the return value of netlink_unicast(). Moreover, fix error codes on error (for example, for non-existing entry was uncorrect). Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-31netfilter: ctnetlink: fix timeout calculationXi Wang1-2/+2
The sanity check (timeout < 0) never works; the dividend is unsigned and so is the division, which should have been a signed division. long timeout = (ct->timeout.expires - jiffies) / HZ; if (timeout < 0) timeout = 0; This patch converts the time values to signed for the division. Signed-off-by: Xi Wang <xi.wang@gmail.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-31ipvs: try also real server with port 0 in backup serverJulian Anastasov3-4/+10
We should not forget to try for real server with port 0 in the backup server when processing the sync message. We should do it in all cases because the backup server can use different forwarding method. Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-31netem: fix classful handlingEric Dumazet1-121/+81
Commit 10f6dfcfde (Revert "sch_netem: Remove classful functionality") reintroduced classful functionality to netem, but broke basic netem behavior : netem uses an t(ime)fifo queue, and store timestamps in skb->cb[] If qdisc is changed, time constraints are not respected and other qdisc can destroy skb->cb[] and block netem at dequeue time. Fix this by always using internal tfifo, and optionally attach a child qdisc to netem (or a tree of qdiscs) Example of use : DEV=eth3 tc qdisc del dev $DEV root tc qdisc add dev $DEV root handle 30: est 1sec 8sec netem delay 20ms 10ms tc qdisc add dev $DEV handle 40:0 parent 30:0 tbf \ burst 20480 limit 20480 mtu 1514 rate 32000bps qdisc netem 30: root refcnt 18 limit 1000 delay 20.0ms 10.0ms Sent 190792 bytes 413 pkt (dropped 0, overlimits 0 requeues 0) rate 18416bit 3pps backlog 0b 0p requeues 0 qdisc tbf 40: parent 30: rate 256000bit burst 20Kb/8 mpu 0b lat 0us Sent 190792 bytes 413 pkt (dropped 6, overlimits 10 requeues 0) backlog 0b 5p requeues 0 Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31IPv6: Avoid taking write lock for /proc/net/ipv6_routeJosh Hunt2-1/+21
During some debugging I needed to look into how /proc/net/ipv6_route operated and in my digging I found its calling fib6_clean_all() which uses "write_lock_bh(&table->tb6_lock)" before doing the walk of the table. I found this on 2.6.32, but reading the code I believe the same basic idea exists currently. Looking at the rtnetlink code they are only calling "read_lock_bh(&table->tb6_lock);" via fib6_dump_table(). While I realize reading from proc isn't the recommended way of fetching the ipv6 route table; taking a write lock seems unnecessary and would probably cause network performance issues. To verify this I loaded up the ipv6 route table and then ran iperf in 3 cases: * doing nothing * reading ipv6 route table via proc (while :; do cat /proc/net/ipv6_route > /dev/null; done) * reading ipv6 route table via rtnetlink (while :; do ip -6 route show table all > /dev/null; done) * Load the ipv6 route table up with: * for ((i = 0;i < 4000;i++)); do ip route add unreachable 2000::$i; done * iperf commands: * client: iperf -i 1 -V -c <ipv6 addr> * server: iperf -V -s * iperf results - 3 runs each (in Mbits/sec) * nothing: client: 927,927,927 server: 927,927,927 * proc: client: 179,97,96,113 server: 142,112,133 * iproute: client: 928,927,928 server: 927,927,927 lock_stat shows taking the write lock is causing the slowdown. Using this info I decided to write a version of fib6_clean_all() which replaces write_lock_bh(&table->tb6_lock) with read_lock_bh(&table->tb6_lock). With this new function I see the same results as with my rtnetlink iperf test. Signed-off-by: Josh Hunt <joshhunt00@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31unix_diag: Fixup RQLEN extension reportPavel Emelyanov1-1/+12
While it's not too late fix the recently added RQLEN diag extension to report rqlen and wqlen in the same way as TCP does. I.e. for listening sockets the ack backlog length (which is the input queue length for socket) in rqlen and the max ack backlog length in wqlen, and what the CINQ/OUTQ ioctls do for established. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31af_unix: Move CINQ/COUTQ code to helpersPavel Emelyanov1-23/+36
Currently tcp diag reports rqlen and wqlen values similar to how the CINQ/COUTQ iotcls do. To make unix diag report these values in the same way move the respective code into helpers. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31unix_diag: Add the MEMINFO extensionPavel Emelyanov1-8/+12
[ Fix indentation of sock_diag*() calls. -DaveM ] Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31inet_diag: Add the SKMEMINFO extensionPavel Emelyanov1-0/+4
Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-31sock_diag: Introduce the meminfo nla core (v2)Pavel Emelyanov1-0/+23
Add a routine that dumps memory-related values of a socket. It's made as an array to make it possible to add more stuff here later without breaking compatibility. Since v1: The SK_MEMINFO_ constants are in userspace visible part of sock_diag.h, the rest is under __KERNEL__. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-30Merge branch 'tipc-Dec29-2011' of ↵David S. Miller18-237/+248
git://git.kernel.org/pub/scm/linux/kernel/git/paulg/linux
2011-12-30Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/netDavid S. Miller3-9/+22
2011-12-30tipc: rename struct bearer_name to struct tipc_bearer_namesPaul Gortmaker2-6/+6
The addition of the "s" to indicate pluralization is intentional, since the struct actually contains two name variants. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct link* to struct tipc_link*Paul Gortmaker10-146/+156
This converts the following: struct link -> struct tipc_link struct link_req -> struct tipc_link_req struct link_name -> struct tipc_link_name Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct bcbearer* to tipc_bcbearer*Paul Gortmaker1-10/+10
This changes both the struct bcbearer and struct bcbearer_pair to have the "tipc_" prefix. Runtime behaviour is unchanged. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct bclink to struct tipc_bclinkPaul Gortmaker1-4/+4
Make this rename so that it is consistent with the majority of the other tipc structs and to assist in removing any ambiguity with other similar names in other subsystems. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct subscriber to struct tipc_subscriberPaul Gortmaker1-11/+11
Make this rename so that it is consistent with the majority of the other tipc structs and to assist in removing any ambiguity with other similar names in other subsystems. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct subscription to struct tipc_subscriptionPaul Gortmaker4-26/+27
Make this rename so that it is consistent with the majority of the other tipc structs and to assist in removing any ambiguity with other similar names in other subsystems. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct port_list to struct tipc_port_listPaul Gortmaker6-19/+19
Make this rename so that it is consistent with the majority of the other tipc structs and to assist in removing any ambiguity with other similar names in other subsystems. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30tipc: rename struct media to struct tipc_mediaPaul Gortmaker5-15/+15
Give it a meaningful prefix, as suggested by DaveM, so that it is consistent with things like struct tipc_bearer, and so it isn't confused with anything else. This has no impact on the actual runtime code behaviour. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-30ipv6: Fix neigh lookup using NULL device.David S. Miller1-7/+10
In some of the rt6_bind_neighbour() call sites, it hasn't hooked up the rt->dst.dev pointer yet, so we'd deref a NULL pointer when obtaining dev->ifindex for the neighbour hash function computation. Just pass the netdevice explicitly in to fix this problem. Reported-by: Bjarke Istrup Pedersen <gurligebis@gentoo.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-30ipv6: Report TCP timetstamp info in cacheinfo just like ipv4 does.David S. Miller1-1/+10
I missed this while adding ipv6 support to inet_peer. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-30sch_tbf: report backlog informationEric Dumazet1-0/+1
Provide child qdisc backlog (byte count) information so that "tc -s qdisc" can report it to user. qdisc netem 30: root refcnt 18 limit 1000 delay 20.0ms 10.0ms Sent 948517 bytes 898 pkt (dropped 0, overlimits 0 requeues 1) rate 175056bit 16pps backlog 114b 1p requeues 1 qdisc tbf 40: parent 30: rate 256000bit burst 20Kb/8 mpu 0b lat 0us Sent 948517 bytes 898 pkt (dropped 15, overlimits 611 requeues 0) backlog 18168b 12p requeues 0 Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-29netfilter: Kconfig: fix unmet xt_nfacct dependenciesPablo Neira Ayuso1-1/+1
warning: (NETFILTER_XT_MATCH_NFACCT) selects NETFILTER_NETLINK_ACCT which has unmet direct dependencies (NET && INET && NETFILTER && NETFILTER_ADVANCED) and then ERROR: "nfnetlink_subsys_unregister" [net/netfilter/nfnetlink_acct.ko] undefined! ERROR: "nfnetlink_subsys_register" [net/netfilter/nfnetlink_acct.ko] undefined! Reported-by: Randy Dunlap <rdunlap@xenotime.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Acked-by: Randy Dunlap <rdunlap@xenotime.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-29ipv6: Kill rt6i_dev and rt6i_expires defines.David S. Miller9-49/+49
It just obscures that the netdevice pointer and the expires value are implemented in the dst_entry sub-object of the ipv6 route. And it makes grepping for dst_entry member uses much harder too. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-29ipv6: Create fast inline ipv6 neigh lookup just like ipv4.David S. Miller1-19/+22
Also, create and use an rt6_bind_neighbour() in net/ipv6/route.c to consolidate some common logic. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-29ipv6: Use universal hash for NDISC.David S. Miller4-18/+18
In order to perform a proper universal hash on a vector of integers, we have to use different universal hashes on each vector element. Which means we need 4 different hash randoms for ipv6. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-28netrom: avoid overflows in nr_setsockopt()Xi Wang1-7/+7
Check setsockopt arguments to avoid overflows and return -EINVAL for too large arguments. Signed-off-by: Xi Wang <xi.wang@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-28ax25: avoid overflows in ax25_setsockopt()Xi Wang1-5/+6
Commit be639ac6 ("NET: AX.25: Check ioctl arguments to avoid overflows further down the road") rejects very large arguments, but doesn't completely fix overflows on 64-bit systems. Consider the AX25_T2 case. int opt; ... if (opt < 1 || opt > ULONG_MAX / HZ) { res = -EINVAL; break; } ax25->t2 = opt * HZ; The 32-bit multiplication opt * HZ would overflow before being assigned to 64-bit ax25->t2. This patch changes "opt" to unsigned long. Signed-off-by: Xi Wang <xi.wang@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-28genetlink: add auto module loadingStephen Hemminger1-0/+9
When testing L2TP support, I discovered that the l2tp module is not autoloaded as are other netlink interfaces. There is because of lack of hook in genetlink to call request_module and load the module. Signed-off-by: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-28ipv6: Remove optimistic DAD flag test in ipv6_add_addr()David Miller1-10/+0
The route we have here is for the address being added to the interface, ie. for input packet processing. Therefore using that route to determine whether an output nexthop gateway is known and resolved doesn't make any sense. So, simply remove this test, it never triggered anyways. Signed-off-by: David S. Miller <davem@davemloft.net> Acked-By: Neil Horman <nhorman@tuxdriver.com>
2011-12-28packet: fix possible dev refcnt leak when bind failWei Yongjun1-1/+5
If bind is fail when bind is called after set PACKET_FANOUT sock option, the dev refcnt will leak. Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-27netfilter: provide config option to disable ancient procfs partsJan Engelhardt4-9/+19
Using /proc/net/nf_conntrack has been deprecated in favour of the conntrack(8) tool. Signed-off-by: Jan Engelhardt <jengelh@medozas.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-27netfilter: xtables: collapse conditions in xt_ecnJan Engelhardt1-9/+6
One simplification of an if clause. Signed-off-by: Jan Engelhardt <jengelh@medozas.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-27netfilter: xtables: add an IPv6 capable version of the ECN matchPatrick McHardy1-26/+80
References: http://www.spinics.net/lists/netfilter-devel/msg18875.html Augment xt_ecn by facilities to match on IPv6 packets' DSCP/TOS field similar to how it is already done for the IPv4 packet field. Signed-off-by: Jan Engelhardt <jengelh@medozas.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-27netfilter: xtables: give xt_ecn its own nameJan Engelhardt1-17/+17
Use the new macro and struct names in xt_ecn.h, and put the old definitions into a definition-forwarding ipt_ecn.h. Signed-off-by: Jan Engelhardt <jengelh@medozas.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-27netfilter: xtables: move ipt_ecn to xt_ecnJan Engelhardt5-6/+16
Prepare the ECN match for augmentation by an IPv6 counterpart. Since no symbol dependencies to ipv6.ko are added, having a single ecn match module is the more so welcome. Signed-off-by: Jan Engelhardt <jengelh@medozas.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2011-12-27tipc: Allow use of buf_seqno() helper routine by unicast linksAllan Stephens3-25/+24
Migrates the buf_seqno() helper routine from broadcast link level to unicast link level so that it can be used both types of TIPC links. This is a cosmetic change only, and does not affect the operation of TIPC. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Ignore broadcast acknowledgements that are out-of-rangeAllan Stephens4-13/+37
Adds checks to TIPC's broadcast link so that it ignores any acknowledgement message containing a sequence number that does not correspond to an unacknowledged message currently in the broadcast link's transmit queue. This change prevents the broadcast link from becoming stalled if a newly booted node receives stale broadcast link acknowledgement information from another node that has not yet fully synchronized its end of the broadcast link to reflect the current state of the new node's end. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Flush unsent broadcast messages when contact with last node is lostAllan Stephens1-1/+11
Adds code to release any unsent broadcast messages in the broadcast link transmit queue if TIPC loses contact with its only neighboring node. Previously, a broadcast link that was in the congested state would hold on to the unsent messages, even though the messages were now undeliverable. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Minor optimization of broadcast link transmit queue statisticAllan Stephens1-4/+4
The two broadcast link statistics fields that are used to derive the average length of that link's transmit queue are now updated only after a successful attempt to send a broadcast message, since there is no need to update these values when an unsuccessful send attempt leaves the queue unchanged. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Handle broadcast attempt when no neighboring nodes existAllan Stephens1-1/+7
Adds a check to detect when an attempt is made to send a message via the broadcast link and no neighboring nodes are currently available to receive it. Rather than wasting effort passing the message to the broadcast link and broadcast bearer, who will only throw it away, TIPC now frees the message immediately and reports success (i.e. the message has been delivered to all available destinations). Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Ensure broadcast link spinlock is held when updating node mapAllan Stephens3-10/+22
Fixes oversight that allowed broadcast link node map to be updated without first taking the broadcast link spinlock that protects the map. As part of this fix the node map has been incorporated into the broadcast link structure to make the need for such protection more evident. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Eliminate dynamic allocation of broadcast link data structuresAllan Stephens3-31/+13
Creates global variables to hold the broadcast link's pseudo-bearer and pseudo-link structures, rather than allocating them dynamically. There is only a single instance of each structure, and changing over to static allocation allows elimination of code to handle the cases where dynamic allocation was unsuccessful. The memset in the teardown code may look like they aren't used, but the same teardown code is run when there is a non-fatal error at init-time, so that stale data isn't present when the user fixes the cause of the soft error. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Eliminate useless check when network address is assignedAllan Stephens1-7/+5
Gets rid of an unnecessary check in the routine that updates the port id of a node's name publications when the node is assigned a network address, since the routine is only invoked if the new address is different from the existing one. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-27tipc: Minor correction to TIPC module unloadingAllan Stephens1-1/+1
Modifies TIPC's module unloading logic to switch itself into "single node" mode before starting to terminate networking support. This helps to ensure that no operations that require TIPC to be in "networking" mode can initiate once unloading starts. Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
generated by cgit v1.2.3 (git 2.39.0) at 2024-11-16 22:09:59 +0300