- 17 Nov, 2008 18 commits
-
-
Gerrit Renker authored
This patch deprecates the Ack Ratio sysctl, since * Ack Ratio is entirely ignored by CCID-3 and CCID-4, * Ack Ratio currently doesn't work in CCID-2 (i.e. is always set to 1); * even if it would work in CCID-2, there is no point for a user to change it: - Ack Ratio is constrained by cwnd (RFC 4341, 6.1.2), - if Ack Ratio > cwnd, the system resorts to spurious RTO timeouts (since waiting for Acks which will never arrive in this window), - cwnd is not a user-configurable value. The only reasonable place for Ack Ratio is to print it for debugging. It is planned to do this later on, as part of e.g. dccp_probe. With this patch Ack Ratio is now under full control of feature negotiation: * Ack Ratio is resolved as a dependency of the selected CCID; * if the chosen CCID supports it (i.e. CCID == CCID-2), Ack Ratio is set to the default of 2, following RFC 4340, 11.3 - "New connections start with Ack Ratio 2 for both endpoints"; * what happens then is part of another patch set, since it concerns the dynamic update of Ack Ratio while the connection is in full flight. Thanks to Tomasz Grobelny for discussion leading up to this patch. Signed-off-by:Gerrit Renker <gerrit@erg.abdn.ac.uk> Acked-by:
Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Gerrit Renker authored
This provides feature negotiation for server minimum checksum coverage which so far has been missing. Since sender/receiver coverage values range only from 0...15, their type has also been reduced in size from u16 to u4. Feature-negotiation options are now generated for both sender and receiver coverage, i.e. when the peer has `forgotten' to enable partial coverage then feature negotiation will automatically enable (negotiate) the partial coverage value for this connection. Signed-off-by:
Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by:
-
Gerrit Renker authored
The previous setsockopt interface, which passed socket options via struct dccp_so_feat, is complicated/difficult to use. Continuing to support it leads to ugly code since the old approach did not distinguish between NN and SP values. This patch removes the old setsockopt interface and replaces it with two new functions to register NN/SP values for feature negotiation. These are essentially wrappers around the internal __feat_register functions, with checking added to avoid * wrong usage (type); * changing values while the connection is in progress. Signed-off-by:
-
Gerrit Renker authored
This adds a hook to resolve features whose value depends on the choice of CCID. It is done at the server since it can only be done after the CCID values have been negotiated; i.e. the client will add its CCID preference list on the Change options sent in the Request, which will be reconciled with the local preference list of the server. The concept is documented on http://www.erg.abdn.ac.uk/users/gerrit/dccp/notes/feature_negotiation/\ implementation_notes.html#ccid_dependencies Signed-off-by:
-
Mark McLoughlin authored
If segmentation offload is enabled by the host, we currently allocate maximum sized packet buffers and pass them to the host. This uses up 20 ring entries, allowing us to supply only 20 packet buffers to the host with a 256 entry ring. This is a huge overhead when receiving small packets, and is most keenly felt when receiving MTU sized packets from off-host. The VIRTIO_NET_F_MRG_RXBUF feature flag is set by hosts which support using receive buffers which are smaller than the maximum packet size. In order to transfer large packets to the guest, the host merges together multiple receive buffers to form a larger logical buffer. The number of merged buffers is returned to the guest via a field in the virtio_net_hdr. Make use of this support by supplying single page receive buffers to the host. On receive, we extract the virtio_net_hdr, copy 128 bytes of the payload to the skb's linear data buffer and adjust the fragment offset to point to the remaining data. This ensures proper alignment and allows us to not use any paged data for small packets. If the payload occupies multiple pages, we simply append those pages as fragments and free the associated skbs. This scheme allows us to be efficient in our use of ring entries while still supporting large packets. Benchmarking using netperf from an external machine to a guest over a 10Gb/s network shows a 100% improvement from ~1Gb/s to ~2Gb/s. With a local host->guest benchmark with GSO disabled on the host side, throughput was seen to increase from 700Mb/s to 1.7Gb/s. Based on a patch from Herbert Xu. Signed-off-by:
Mark McLoughlin <markmc@redhat.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (use netdev_priv) Signed-off-by:
-
Mark McLoughlin authored
Seems like an oversight that we have set-tx-csum and set-sg hooked up, but not set-tso. Also leads to the strange situation that if you e.g. disable tx-csum, then tso doesn't get disabled. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au> Signed-off-by:
-
Mark McLoughlin authored
Each time we re-fill the recv queue with buffers, we allocate one too many skbs and free it again when adding fails. We should recycle the pages allocated in this case. A previous version of this patch made trim_pages() trim trailing unused pages from skbs with some paged data, but this actually caused a barely measurable slowdown. Signed-off-by:
-
Alexey Dobriyan authored
Technically, patch changes format for modules, but I think nobody cares. -86dd :ipv6:ipv6_rcv+0x0 +86dd ipv6_rcv+0x0/0x400 [ipv6] Signed-off-by:
Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by:
-
Eric Dumazet authored
As found in the past (commit f1dd9c37 [NET]: Fix tbench regression in 2.6.25-rc1), it is really important that struct dst_entry refcount is aligned on a cache line. We cannot use __atribute((aligned)), so manually pad the structure for 32 and 64 bit arches. for 32bit : offsetof(truct dst_entry, __refcnt) is 0x80 for 64bit : offsetof(truct dst_entry, __refcnt) is 0xc0 As it is not possible to guess at compile time cache line size, we use a generic value of 64 bytes, that satisfies many current arches. (Using 128 bytes alignment on 64bit arches would waste 64 bytes) Add a BUILD_BUG_ON to catch future updates to "struct dst_entry" dont break this alignment. "tbench 8" is 4.4 % faster on a dual quad core (HP BL460c G1), Intel E5450 @3.00GHz (2350 MB/s instead of 2250 MB/s) Signed-off-by:
Eric Dumazet <dada1@cosmosbay.com> Signed-off-by:
-
Eric Dumazet authored
Adds Documentation/RCU/rculist_nulls.txt file to describe how 'nulls' end-of-list can help in some RCU algos. Signed-off-by:
Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by:
-
Eric Dumazet authored
RCU was added to UDP lookups, using a fast infrastructure : - sockets kmem_cache use SLAB_DESTROY_BY_RCU and dont pay the price of call_rcu() at freeing time. - hlist_nulls permits to use few memory barriers. This patch uses same infrastructure for TCP/DCCP established and timewait sockets. Thanks to SLAB_DESTROY_BY_RCU, no slowdown for applications using short lived TCP connections. A followup patch, converting rwlocks to spinlocks will even speedup this case. __inet_lookup_established() is pretty fast now we dont have to dirty a contended cache line (read_lock/read_unlock) Only established and timewait hashtable are converted to RCU (bind table and listen table are still using traditional locking) Signed-off-by:
-
Eric Dumazet authored
This is a straightforward patch, using hlist_nulls infrastructure. RCUification already done on UDP two weeks ago. Using hlist_nulls permits us to avoid some memory barriers, both at lookup time and delete time. Patch is large because it adds new macros to include/net/sock.h. These macros will be used by TCP & DCCP in next patch. Signed-off-by:
-
Eric Dumazet authored
hlist uses NULL value to finish a chain. hlist_nulls variant use the low order bit set to 1 to signal an end-of-list marker. This allows to store many different end markers, so that some RCU lockless algos (used in TCP/UDP stack for example) can save some memory barriers in fast paths. Two new files are added : include/linux/list_nulls.h - mimics hlist part of include/linux/list.h, derived to hlist_nulls variant include/linux/rculist_nulls.h - mimics hlist part of include/linux/rculist.h, derived to hlist_nulls variant Only four helpers are declared for the moment : hlist_nulls_del_init_rcu(), hlist_nulls_del_rcu(), hlist_nulls_add_head_rcu() and hlist_nulls_for_each_entry_rcu() prefetches() were removed, since an end of list is not anymore NULL value. prefetches() could trigger useless (and possibly dangerous) memory transactions. Example of use (extracted from __udp4_lib_lookup()) struct sock *sk, *result; struct hlist_nulls_node *node; unsigned short hnum = ntohs(dport); unsigned int hash = udp_hashfn(net, hnum); struct udp_hslot *hslot = &udptable->hash[hash]; int score, badness; rcu_read_lock(); begin: result = NULL; badness = -1; sk_nulls_for_each_rcu(sk, node, &hslot->head) { score = compute_score(sk, net, saddr, hnum, sport, daddr, dport, dif); if (score > badness) { result = sk; badness = score; } } /* * if the nulls value we got at the end of this lookup is * not the expected one, we must restart lookup. * We probably met an item that was moved to another chain. */ if (get_nulls_value(node) != hash) goto begin; if (result) { if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt))) result = NULL; else if (unlikely(compute_score(result, net, saddr, hnum, sport, daddr, dport, dif) < badness)) { sock_put(result); goto begin; } } rcu_read_unlock(); return result; Signed-off-by: -
Balazs Scheidler authored
In case UDP traffic is redirected to a local UDP socket, the originally addressed destination address/port cannot be recovered with the in-kernel tproxy. This patch adds an IP_RECVORIGDSTADDR sockopt that enables a IP_ORIGDSTADDR ancillary message in recvmsg(). This ancillary message contains the original destination address/port of the packet being received. Signed-off-by:
Balazs Scheidler <bazsi@balabit.hu> Signed-off-by:
-
Ben Greear authored
Ben Greear wrote: > I have 500 mac-vlans on a system talking to 500 other > mac-vlans. My problem is that the arp-table gets extremely > huge because every time an arp-request comes in on all mac-vlans, > a stale arp entry is added for each mac-vlan. I have filtering > turned on, but that doesn't help because the neigh_event_ns call > below will cause a stale neighbor entry to be created regardless > of whether a replay will be sent or not. > Maybe the neigh_event code should be below the checks for dont_send, > and only create check neigh_event_ns if we are !dont_send? The attached patch makes it work much better for me. The patch will cause the code to NOT create a stale neighbor entry if we are not going to respond to the ARP request. The old code *would* create a stale entry even if we are not going to respond. Signed-off-by:
Ben Greear <greearb@candelatech.com> Signed-off-by:
-
Alexander Duyck authored
This change enables ECC correction for the packet buffer on all 82571 silicon. Signed-off-by:
Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
Paulius Zaleckas authored
make mdio-gpio work with non OpenFirmware gpio implementation. Aditional changes to mdio-gpio: - use gpio_request() and gpio_free() - place irq[] array in struct mdio_gpio_info - add module description, author and license - add note about compiling this driver as module - rename mdc and mdio function (were ugly names) - change MII to MDIO in bus name - add __init __exit to module (un)loading functions - probe fails if no phys added to the bus - kzalloc bitbang with sizeof(*bitbang) Changes since v3: - keep bus naming "%x" to be compatible with existing drivers. Changes since v2: - more #ifdefs reduction - platform driver will be registered on OF platforms also - unified platform and OF bus_id to phy%i Changes since v1: - removed NO_IRQ - reduced #idefs Laurent, please test this driver under OF. Signed-off-by:
Paulius Zaleckas <paulius.zaleckas@teltonika.lt> Signed-off-by:
-
Paulius Zaleckas authored
Signed-off-by:
-
- 16 Nov, 2008 3 commits
-
-
David S. Miller authored
Reported by Stephen Rothwell: drivers/net/dm9000.c:1450: error: expected ')' before ';' token drivers/net/dm9000.c:1455: error: expected ';' before '}' token Signed-off-by: -
David Brownell authored
Make pegasus driver not allocate a workqueue until the driver is bound to some device, which will need that workqueue if the device is brought up. This conserves resources when the driver is linked but there's no pegasus device connected. Also shrink the runtime footprint a smidgeon by moving some init-only code into its proper section, and move an obnoxious (frequent and meaningless) message to be debug-only. Signed-off-by:
David Brownell <dbrownell@users.sourceforge.net> Signed-off-by:
-
PJ Waskiewicz authored
netif_carrier_off() is sufficient to stop Tx into the driver. Stopping the Tx queues is redundant and unnecessary. By the same token, netif_carrier_on() will be sufficient to re-enable Tx, so waking the queues is unnecessary. Signed-off-by:
Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com> Signed-off-by:
-
- 14 Nov, 2008 4 commits
-
-
Eric Dumazet authored
During tbench/oprofile sessions, I found that dst_release() was in third position. CPU: Core 2, speed 2999.68 MHz (estimated) Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (Unhalted core cycles) count 100000 samples % symbol name 483726 9.0185 __copy_user_zeroing_intel 191466 3.5697 __copy_user_intel 185475 3.4580 dst_release 175114 3.2648 ip_queue_xmit 153447 2.8608 tcp_sendmsg 108775 2.0280 tcp_recvmsg 102659 1.9140 sysenter_past_esp 101450 1.8914 tcp_current_mss 95067 1.7724 __copy_from_user_ll 86531 1.6133 tcp_transmit_skb Of course, all CPUS fight on the dst_entry associated with 127.0.0.1 Instead of first checking the refcount value, then decrement it, we use atomic_dec_return() to help CPU to make the right memory transaction (ie getting the cache line in exclusive mode) dst_release() is now at the fifth position, and tbench a litle bit faster ;) CPU: Core 2, speed 3000.1 MHz (estimated) Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (Unhalted core cycles) count 100000 samples % symbol name 647107 8.8072 __copy_user_zeroing_intel 258840 3.5229 ip_queue_xmit 258302 3.5155 __copy_user_intel 209629 2.8531 tcp_sendmsg 165632 2.2543 dst_release 149232 2.0311 tcp_current_mss 147821 2.0119 tcp_recvmsg 137893 1.8767 sysenter_past_esp 127473 1.7349 __copy_from_user_ll 121308 1.6510 ip_finish_output 118510 1.6129 tcp_transmit_skb 109295 1.4875 tcp_v4_rcv Signed-off-by:
-
Jarek Poplawski authored
After implementing qdisc->ops->peek() and changing sch_netem into classless qdisc there are no more qdisc->ops->requeue() users. This patch removes this method with its wrappers (qdisc_requeue()), and also unused qdisc->requeue structure. There are a few minor fixes of warnings (htb_enqueue()) and comments btw. The idea to kill ->requeue() and a similar patch were first developed by David S. Miller. Signed-off-by:
Jarek Poplawski <jarkao2@gmail.com> Signed-off-by:
-
Petr Tesarik authored
The urg_ptr field is not used anywhere and is merely confusing. Signed-off-by:
Petr Tesarik <ptesarik@suse.cz> Signed-off-by:
-
Harvey Harrison authored
They would have been better named as get_be16, put_be16, etc. as they were hiding an endian shift inside. They don't add much over explicitly coding the byteshifting and gcc sometimes has a problem with builtin_constant_p inside inline functions, so it may do a better job of byteswapping at compile time rather than runtime. Signed-off-by:
Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by:
-
- 13 Nov, 2008 11 commits
-
-
Wang Chen authored
We have some reasons to kill netdev->priv: 1. netdev->priv is equal to netdev_priv(). 2. netdev_priv() wraps the calculation of netdev->priv's offset, obviously netdev_priv() is more flexible than netdev->priv. But we cann't kill netdev->priv, because so many drivers reference to it directly. This patch is a safe convert for netdev->priv to netdev_priv(netdev). Since all of the netdev->priv is only for read. But it is too big to be sent in one mail. I split it to 4 parts and make every part smaller than 100,000 bytes, which is max size allowed by vger. Signed-off-by:
Wang Chen <wangchen@cn.fujitsu.com> Signed-off-by:
-
Wang Chen authored
We have some reasons to kill netdev->priv: 1. netdev->priv is equal to netdev_priv(). 2. netdev_priv() wraps the calculation of netdev->priv's offset, obviously netdev_priv() is more flexible than netdev->priv. But we cann't kill netdev->priv, because so many drivers reference to it directly. This patch is a safe convert for netdev->priv to netdev_priv(netdev). Since all of the netdev->priv is only for read. But it is too big to be sent in one mail. I split it to 4 parts and make every part smaller than 100,000 bytes, which is max size allowed by vger. Signed-off-by:
-
Wang Chen authored
We have some reasons to kill netdev->priv: 1. netdev->priv is equal to netdev_priv(). 2. netdev_priv() wraps the calculation of netdev->priv's offset, obviously netdev_priv() is more flexible than netdev->priv. But we cann't kill netdev->priv, because so many drivers reference to it directly. This patch is a safe convert for netdev->priv to netdev_priv(netdev). Since all of the netdev->priv is only for read. But it is too big to be sent in one mail. I split it to 4 parts and make every part smaller than 100,000 bytes, which is max size allowed by vger. Signed-off-by:
-
Wang Chen authored
We have some reasons to kill netdev->priv: 1. netdev->priv is equal to netdev_priv(). 2. netdev_priv() wraps the calculation of netdev->priv's offset, obviously netdev_priv() is more flexible than netdev->priv. But we cann't kill netdev->priv, because so many drivers reference to it directly. This patch is a safe convert for netdev->priv to netdev_priv(netdev). Since all of the netdev->priv is only for read. But it is too big to be sent in one mail. I split it to 4 parts and make every part smaller than 100,000 bytes, which is max size allowed by vger. Signed-off-by:
-
Arnaud Ebalard authored
In commit 2518c7c2 ("[XFRM]: Hash policies when non-prefixed."), the last use of xfrm_gen_policy() first argument was removed, but the argument was left behind in the prototype. Signed-off-by:
Arnaud Ebalard <arno@natisbad.org> Acked-by:
Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Alexey Dobriyan authored
Every user is under CONFIG_NET_DMA already, so ifdef field as well. Signed-off-by:
-
Michael Chan authored
Signed-off-by:
Michael Chan <mchan@broadcom.com> Signed-off-by:
Matt Carlson <mcarlson@broadcom.com> Signed-off-by:
Benjamin Li <benli@broadcom.com> Signed-off-by:
-
Michael Chan authored
Move all related timeout constants to the same location. BNX2 prefix is also added to make them more consistent. Signed-off-by:
-
Michael Chan authored
The default rx buffer water marks for XOFF/XON are for 1500 MTU. At larger MTUs, these water marks need to be adjusted for effective flow control. Signed-off-by:
-
Michael Chan authored
On some quad-port cards that cannot support WoL on all ports due to excessive power consumption, the driver needs to restrict WoL on some ports by checking VAUX_PRESET bit. Signed-off-by:
-
Michael Chan authored
Signed-off-by:
-
- 12 Nov, 2008 4 commits
-
-
Eric Dumazet authored
Using read_pnet() and write_pnet() in neighbour code ease the reading of code. Signed-off-by:
-
Eric Dumazet authored
We can shrink size of "struct inet_bind_bucket" by 50%, using read_pnet() and write_pnet() Signed-off-by:
-
Eric Dumazet authored
This patch introduces two helpers that deal with reading and writing struct net pointers in various network structures. Their implementation depends on CONFIG_NET_NS For symmetry, both functions work with "struct net **pnet". Their usage should reduce the number of #ifdef CONFIG_NET_NS, without adding many helpers for each network structure that hold a "struct net *pointer" Signed-off-by:
-
Gerrit Renker authored
This provides a missing link in the code chain, as several features implicitly depend and/or rely on the choice of CCID. Most notably, this is the Send Ack Vector feature, but also Ack Ratio and Send Loss Event Rate (also taken care of). For Send Ack Vector, the situation is as follows: * since CCID2 mandates the use of Ack Vectors, there is no point in allowing endpoints which use CCID2 to disable Ack Vector features such a connection; * a peer with a TX CCID of CCID2 will always expect Ack Vectors, and a peer with a RX CCID of CCID2 must always send Ack Vectors (RFC 4341, sec. 4); * for all other CCIDs, the use of (Send) Ack Vector is optional and thus negotiable. However, this implies that the code negotiating the use of Ack Vectors also supports it (i.e. is able to supply and to either parse or ignore received Ack Vectors). Since this is not the case (CCID-3 has no Ack Vector support), the use of Ack Vectors is here disabled, with a comment in the source code. An analogous consideration arises for the Send Loss Event Rate feature, since the CCID-3 implementation does not support the loss interval options of RFC 4342. To make such use explicit, corresponding feature-negotiation options are inserted which signal the use of the loss event rate option, as it is used by the CCID3 code. Lastly, the values of the Ack Ratio feature are matched to the choice of CCID. The patch implements this as a function which is called after the user has made all other registrations for changing default values of features. The table is variable-length, the reserved (and hence for feature-negotiation invalid, confirmed by considering section 19.4 of RFC 4340) feature number `0' is used to mark the end of the table. Signed-off-by:
-
