summaryrefslogtreecommitdiff
path: root/include/net
AgeCommit message (Collapse)Author
2015-09-04mac80211: protect non-HT BSS when HT TDLS traffic existsAvri Altman
HT TDLS traffic should be protected in a non-HT BSS to avoid collisions. Therefore, when TDLS peers join/leave, check if protection is (now) needed and set the ht_operation_mode of the virtual interface according to the HT capabilities of the TDLS peer(s). This works because a non-HT BSS connection never sets (or otherwise uses) the ht_operation_mode; it just means that drivers must be aware that this field applies to all HT traffic for this virtual interface, not just the traffic within the BSS. Document that. Signed-off-by: Avri Altman <avri.altman@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2015-09-02netfilter: nf_conntrack: make nf_ct_zone_dflt built-inDaniel Borkmann
Fengguang reported, that some randconfig generated the following linker issue with nf_ct_zone_dflt object involved: [...] CC init/version.o LD init/built-in.o net/built-in.o: In function `ipv4_conntrack_defrag': nf_defrag_ipv4.c:(.text+0x93e95): undefined reference to `nf_ct_zone_dflt' net/built-in.o: In function `ipv6_defrag': nf_defrag_ipv6_hooks.c:(.text+0xe3ffe): undefined reference to `nf_ct_zone_dflt' make: *** [vmlinux] Error 1 Given that configurations exist where we have a built-in part, which is accessing nf_ct_zone_dflt such as the two handlers nf_ct_defrag_user() and nf_ct6_defrag_user(), and a part that configures nf_conntrack as a module, we must move nf_ct_zone_dflt into a fixed, guaranteed built-in area when netfilter is configured in general. Therefore, split the more generic parts into a common header under include/linux/netfilter/ and move nf_ct_zone_dflt into the built-in section that already holds parts related to CONFIG_NF_CONNTRACK in the netfilter core. This fixes the issue on my side. Fixes: 308ac9143ee2 ("netfilter: nf_conntrack: push zone object into functions") Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Use 'const' where possible.David S. Miller
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Don't use bit fields.David S. Miller
Just have a flags member instead. In file included from include/linux/linkage.h:4:0, from include/linux/kernel.h:6, from net/core/flow_dissector.c:1: In function 'flow_keys_hash_start', inlined from 'flow_hash_from_keys' at net/core/flow_dissector.c:553:34: >> include/linux/compiler.h:447:38: error: call to '__compiletime_assert_459' declared with attribute error: BUILD_BUG_ON failed: FLOW_KEYS_HASH_OFFSET % sizeof(u32) Reported-by: kbuild test robot <fengguang.wu@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Add control/reporting of encapsulationTom Herbert
Add an input flag to flow dissector on rather dissection should stop when encapsulation is detected (IP/IP or GRE). Also, add a key_control flag that indicates encapsulation was encountered during the dissection. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Add flag to stop parsing when an IPv6 flow label is seenTom Herbert
Add an input flag to flow dissector on rather dissection should be stopped when a flow label is encountered. Presumably, the flow label is derived from a sufficient hash of an inner transport packet so further dissection is not needed (that is ports are not included in the flow hash). Using the flow label instead of ports has the additional benefit that packet fragments should hash to same value as non-fragments for a flow (assuming that the same flow label is used). We set this flag by default in for skb_get_hash. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Add flag to stop parsing at L3Tom Herbert
Add an input flag to flow dissector on rather dissection should be stopped when an L3 packet is encountered. This would be useful if a caller just wanted to get IP addresses of the outermost header (e.g. to do an L3 hash). Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Add control/reporting of fragmentationTom Herbert
Add an input flag to flow dissector on rather dissection should be attempted on a first fragment. Also add key_control flags to indicate that a packet is a fragment or first fragment. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flowi: Abstract out functions to get flow hash based on flowiTom Herbert
Create __get_hash_from_flowi6 and __get_hash_from_flowi4 to get the flow keys and hash based on flowi structures. These are called by __skb_get_hash_flowi6 and __skb_get_hash_flowi4. Also, created get_hash_from_flowi6 and get_hash_from_flowi4 which can be called when just the hash value for a flowi is needed. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01skbuff: Make __skb_set_sw_hash a general functionTom Herbert
Move __skb_set_sw_hash to skbuff.h and add __skb_set_hash which is a common method (between __skb_set_sw_hash and skb_set_hash) to set the hash in an skbuff. Also, move skb_clear_hash to be closer to __skb_set_hash. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01flow_dissector: Move skb related functions to skbuff.hTom Herbert
Move the flow dissector functions that are specific to skbuffs into skbuff.h out of flow_dissector.h. This makes flow_dissector.h have no dependencies on skbuff.h. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01net: Make table id type u32David Ahern
A number of VRF patches used 'int' for table id. It should be u32 to be consistent with the rest of the stack. Fixes: 4e3c89920cd3a ("net: Introduce VRF related flags and helpers") 15be405eb2ea9 ("net: Add inet_addr lookup by table") 30bbaa1950055 ("net: Fix up inet_addr_type checks") 021dd3b8a142d ("net: Add routes to the table associated with the device") dc028da54ed35 ("inet: Move VRF table lookup to inlined function") f6d3c19274c74 ("net: FIB tracepoints") Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01netfilter: conntrack: use nf_ct_tmpl_free in CT/synproxy error pathsDaniel Borkmann
Commit 0838aa7fcfcd ("netfilter: fix netns dependencies with conntrack templates") migrated templates to the new allocator api, but forgot to update error paths for them in CT and synproxy to use nf_ct_tmpl_free() instead of nf_conntrack_free(). Due to that, memory is being freed into the wrong kmemcache, but also we drop the per net reference count of ct objects causing an imbalance. In Brad's case, this leads to a wrap-around of net->ct.count and thus lets __nf_conntrack_alloc() refuse to create a new ct object: [ 10.340913] xt_addrtype: ipv6 does not support BROADCAST matching [ 10.810168] nf_conntrack: table full, dropping packet [ 11.917416] r8169 0000:07:00.0 eth0: link up [ 11.917438] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready [ 12.815902] nf_conntrack: table full, dropping packet [ 15.688561] nf_conntrack: table full, dropping packet [ 15.689365] nf_conntrack: table full, dropping packet [ 15.690169] nf_conntrack: table full, dropping packet [ 15.690967] nf_conntrack: table full, dropping packet [...] With slab debugging, it also reports the wrong kmemcache (kmalloc-512 vs. nf_conntrack_ffffffff81ce75c0) and reports poison overwrites, etc. Thus, to fix the problem, export and use nf_ct_tmpl_free() instead. Fixes: 0838aa7fcfcd ("netfilter: fix netns dependencies with conntrack templates") Reported-by: Brad Jackson <bjackson0971@gmail.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2015-08-31tun_dst: Remove opts_sizePravin B Shelar
opts_size is only written and never read. Following patch removes this unused variable. Signed-off-by: Pravin B Shelar <pshelar@nicira.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-01ipvs: add schedule_icmp sysctlAlex Gartrell
This sysctl will be used to enable the scheduling of icmp packets. Signed-off-by: Alex Gartrell <agartrell@fb.com> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au>
2015-09-01ipvs: drop inverse argument to conn_{in,out}_getAlex Gartrell
No longer necessary since the information is included in the ip_vs_iphdr itself. Signed-off-by: Alex Gartrell <agartrell@fb.com> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au>
2015-09-01ipvs: Add hdr_flags to iphdrAlex Gartrell
These flags contain information like whether or not the addresses are inverted or from icmp. The first will allow us to drop an inverse param all over the place, and the second will later be useful in scheduling icmp. Signed-off-by: Alex Gartrell <agartrell@fb.com> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au>
2015-09-01ipvs: replace ip_vs_fill_ip4hdr with ip_vs_fill_iph_skb_offAlex Gartrell
This removes some duplicated code and makes the ICMPv6 path look more like the ICMP path. Signed-off-by: Alex Gartrell <agartrell@fb.com> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au>
2015-08-31gro_cells: remove spinlock protecting receive queuesEric Dumazet
As David pointed out, spinlock are no longer needed to protect the per cpu queues used in gro cells infrastructure. Also use new napi_complete_done() API so that gro_flush_timeout tweaks have an effect. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-31tcp: use dctcp if enabled on the route to the initiatorDaniel Borkmann
Currently, the following case doesn't use DCTCP, even if it should: A responder has f.e. Cubic as system wide default, but for a specific route to the initiating host, DCTCP is being set in RTAX_CC_ALGO. The initiating host then uses DCTCP as congestion control, but since the initiator sets ECT(0), tcp_ecn_create_request() doesn't set ecn_ok, and we have to fall back to Reno after 3WHS completes. We were thinking on how to solve this in a minimal, non-intrusive way without bloating tcp_ecn_create_request() needlessly: lets cache the CA ecn option flag in RTAX_FEATURES. In other words, when ECT(0) is set on the SYN packet, set ecn_ok=1 iff route RTAX_FEATURES contains the unexposed (internal-only) DST_FEATURE_ECN_CA. This allows to only do a single metric feature lookup inside tcp_ecn_create_request(). Joint work with Florian Westphal. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-31ip-tunnel: Use API to access tunnel metadata options.Pravin B Shelar
Currently tun-info options pointer is used in few cases to pass options around. But tunnel options can be accessed using ip_tunnel_info_opts() API without using the pointer. Following patch removes the redundant pointer and consistently make use of API. Signed-off-by: Pravin B Shelar <pshelar@nicira.com> Acked-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Jesse Gross <jesse@nicira.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-30net: Introduce helper functions to get the per cpu dataRaghavendra K T
Signed-off-by: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-30net/bonding: Export bond_option_active_slave_get_rcuMatan Barak
Some consumers of the netdev events API would like to know who is the active slave when a NETDEV_CHANGEUPPER or NETDEV_BONDING_FAILOVER events occur. For example, when managing RoCE GIDs, GIDs based on the bond's ips should only be set on the port which corresponds to active slave netdevice. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-30net/ipv6: Export addrconf_ifid_eui48Matan Barak
For loopback purposes, RoCE devices should have a default GID in the port GID table, even when the interface is down. In order to do so, we use the IPv6 link local address which would have been genenrated for the related Ethernet netdevice when it goes up as a default GID. addrconf_ifid_eui48 is used to gernerate this address, export it. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-29vxlan: do not receive IPv4 packets on IPv6 socketJiri Benc
By default (subject to the sysctl settings), IPv6 sockets listen also for IPv4 traffic. Vxlan is not prepared for that and expects IPv6 header in packets received through an IPv6 socket. In addition, it's currently not possible to have both IPv4 and IPv6 vxlan tunnel on the same port (unless bindv6only sysctl is enabled), as it's not possible to create and bind both IPv4 and IPv6 vxlan interfaces and there's no way to specify both IPv4 and IPv6 remote/group IP addresses. Set IPV6_V6ONLY on vxlan sockets to fix both of these issues. This is not done globally in udp_tunnel, as l2tp and tipc seems to work okay when receiving IPv4 packets on IPv6 socket and people may rely on this behavior. The other tunnels (geneve and fou) do not support IPv6. Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-29ip_tunnels: record IP version in tunnel infoJiri Benc
There's currently nothing preventing directing packets with IPv6 encapsulation data to IPv4 tunnels (and vice versa). If this happens, IPv6 addresses are incorrectly interpreted as IPv4 ones. Track whether the given ip_tunnel_key contains IPv4 or IPv6 data. Store this in ip_tunnel_info. Reject packets at appropriate places if they are supposed to be encapsulated into an incompatible protocol. Signed-off-by: Jiri Benc <jbenc@redhat.com> Acked-by: Alexei Starovoitov <ast@plumgrid.com> Acked-by: Thomas Graf <tgraf@suug.ch> Acked-by: Pravin B Shelar <pshelar@nicira.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-29ip_tunnels: convert the mode field of ip_tunnel_info to flagsJiri Benc
The mode field holds a single bit of information only (whether the ip_tunnel_info struct is for rx or tx). Change the mode field to bit flags. This allows more mode flags to be added. Signed-off-by: Jiri Benc <jbenc@redhat.com> Acked-by: Alexei Starovoitov <ast@plumgrid.com> Acked-by: Thomas Graf <tgraf@suug.ch> Acked-by: Pravin B Shelar <pshelar@nicira.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-nextDavid S. Miller
Pablo Neira Ayuso says: ==================== Netfilter updates for net-next The following patchset contains Netfilter/IPVS updates for your net-next tree. In sum, patches to address fallout from the previous round plus updates from the IPVS folks via Simon Horman, they are: 1) Add a new scheduler to IPVS: The weighted overflow scheduling algorithm directs network connections to the server with the highest weight that is currently available and overflows to the next when active connections exceed the node's weight. From Raducu Deaconu. 2) Fix locking ordering in IPVS, always take rtnl_lock in first place. Patch from Julian Anastasov. 3) Allow to indicate the MTU to the IPVS in-kernel state sync daemon. From Julian Anastasov. 4) Enhance multicast configuration for the IPVS state sync daemon. Also from Julian. 5) Resolve sparse warnings in the nf_dup modules. 6) Fix a linking problem when CONFIG_NF_DUP_IPV6 is not set. 7) Add ICMP codes 5 and 6 to IPv6 REJECT target, they are more informative subsets of code 1. From Andreas Herz. 8) Revert the jumpstack size calculation from mark_source_chains due to chain depth miscalculations, from Florian Westphal. 9) Calm down more sparse warning around the Netfilter tree, again from Florian Westphal. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28net: Add support for VRFs to inetpeer cacheDavid Ahern
inetpeer caches based on address only, so duplicate IP addresses within a namespace return the same cached entry. Enhance the ipv4 address key to contain both the IPv4 address and VRF device index. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28net: Refactor inetpeer address structDavid Ahern
Move the inetpeer_addr_base union to inetpeer_addr and drop inetpeer_addr_base. Both the a6 and in6_addr overlays are not needed; drop the __be32 version and rename in6 to a6 for consistency with ipv4. Add a new u32 array to the union which removes the need for the typecast in the compare function and the use of a consistent arg for both ipv4 and ipv6 addresses which makes the compare function more readable. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28net: Add helper function to compare inetpeer addressesDavid Ahern
tcp_metrics and inetpeer both have functions to compare inetpeer addresses. Consolidate into 1 version. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28net: Add set,get helpers for inetpeer addressesDavid Ahern
Use inetpeer set,get helpers in tcp_metrics rather than peeking into the inetpeer_addr struct. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-28net: Introduce ipv4_addr_hash and use it for tcp metricsDavid Ahern
Refactors a common line into helper function. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27net: sched: register noqueue qdiscPhil Sutter
This way users can attach noqueue just like any other qdisc using tc without having to mess with tx_queue_len first. Signed-off-by: Phil Sutter <phil@nwl.cc> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27geneve: Consolidate Geneve functionality in single module.Pravin B Shelar
geneve_core module handles send and receive functionality. This way OVS could use the Geneve API. Now with use of tunnel meatadata mode OVS can directly use Geneve netdevice. So there is no need for separate module for Geneve. Following patch consolidates Geneve protocol processing in single module. Signed-off-by: Pravin B Shelar <pshelar@nicira.com> Reviewed-by: Jesse Gross <jesse@nicira.com> Acked-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27geneve: Add support to collect tunnel metadata.Pravin B Shelar
Following patch create new tunnel flag which enable tunnel metadata collection on given device. These devices can be used by tunnel metadata based routing or by OVS. Geneve Consolidation patch get rid of collect_md_tun to simplify tunnel lookup further. Signed-off-by: Pravin B Shelar <pshelar@nicira.com> Reviewed-by: Jesse Gross <jesse@nicira.com> Acked-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27tunnel: introduce udp_tun_rx_dst()Pravin B Shelar
Introduce function udp_tun_rx_dst() to initialize tunnel dst on receive path. Signed-off-by: Pravin B Shelar <pshelar@nicira.com> Reviewed-by: Jesse Gross <jesse@nicira.com> Acked-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27net: sched: consolidate tc_classify{,_compat}Daniel Borkmann
For classifiers getting invoked via tc_classify(), we always need an extra function call into tc_classify_compat(), as both are being exported as symbols and tc_classify() itself doesn't do much except handling of reclassifications when tp->classify() returned with TC_ACT_RECLASSIFY. CBQ and ATM are the only qdiscs that directly call into tc_classify_compat(), all others use tc_classify(). When tc actions are being configured out in the kernel, tc_classify() effectively does nothing besides delegating. We could spare this layer and consolidate both functions. pktgen on single CPU constantly pushing skbs directly into the netif_receive_skb() path with a dummy classifier on ingress qdisc attached, improves slightly from 22.3Mpps to 23.1Mpps. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@plumgrid.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27netfilter: connlabels: Export setting connlabel lengthJoe Stringer
Add functions to change connlabel length into nf_conntrack_labels.c so they may be reused by other modules like OVS and nftables without needing to jump through xt_match_check() hoops. Suggested-by: Florian Westphal <fw@strlen.de> Signed-off-by: Joe Stringer <joestringer@nicira.com> Acked-by: Florian Westphal <fw@strlen.de> Acked-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-27dst: Add __skb_dst_copy() variationJoe Stringer
This variation on skb_dst_copy() doesn't require two skbs. Signed-off-by: Joe Stringer <joestringer@nicira.com> Acked-by: Pravin B Shelar <pshelar@nicira.com> Acked-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-26net_sched: act_bpf: remove spinlock in fast pathAlexei Starovoitov
Similar to act_gact/act_mirred, act_bpf can be lockless in packet processing with extra care taken to free bpf programs after rcu grace period. Replacement of existing act_bpf (very rare) is done with synchronize_rcu() and final destruction is done from tc_action_ops->cleanup() callback that is called from tcf_exts_destroy()->tcf_action_destroy()->__tcf_hash_release() when bind and refcnt reach zero which is only possible when classifier is destroyed. Previous two patches fixed the last two classifiers (tcindex and rsvp) to call tcf_exts_destroy() from rcu callback. Similar to gact/mirred there is a race between prog->filter and prog->tcf_action. Meaning that the program being replaced may use previous default action if it happened to return TC_ACT_UNSPEC. act_mirred race betwen tcf_action and tcfm_dev is similar. In all cases the race is harmless. Long term we may want to improve the situation by replacing the whole tc_action->priv as single pointer instead of updating inner fields one by one. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-26net_sched: make tcf_hash_destroy() staticAlexei Starovoitov
tcf_hash_destroy() used once. Make it static. Signed-off-by: Alexei Starovoitov <ast@plumgrid.com> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-25vxlan: fix multiple inclusion of vxlan.hJiri Benc
The vxlan_get_sk_family inline function was added after the last #endif, making multiple inclusion of net/vxlan.h fail. Move it to the proper place. Reported-by: Mark Rustad <mark.d.rustad@intel.com> Fixes: 705cc62f6728c ("vxlan: provide access function for vxlan socket address family") Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-25inetpeer: remove dead codeDavid Ahern
Remove various inlined functions not referenced in the kernel. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-25tcp: refine pacing rate determinationEric Dumazet
When TCP pacing was added back in linux-3.12, we chose to apply a fixed ratio of 200 % against current rate, to allow probing for optimal throughput even during slow start phase, where cwnd can be doubled every other gRTT. At Google, we found it was better applying a different ratio while in Congestion Avoidance phase. This ratio was set to 120 %. We've used the normal tcp_in_slow_start() helper for a while, then tuned the condition to select the conservative ratio as soon as cwnd >= ssthresh/2 : - After cwnd reduction, it is safer to ramp up more slowly, as we approach optimal cwnd. - Initial ramp up (ssthresh == INFINITY) still allows doubling cwnd every other RTT. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Neal Cardwell <ncardwell@google.com> Cc: Yuchung Cheng <ycheng@google.com> Acked-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-25tcp: fix slow start after idle vs TSO/GSOEric Dumazet
slow start after idle might reduce cwnd, but we perform this after first packet was cooked and sent. With TSO/GSO, it means that we might send a full TSO packet even if cwnd should have been reduced to IW10. Moving the SSAI check in skb_entail() makes sense, because we slightly reduce number of times this check is done, especially for large send() and TCP Small queue callbacks from softirq context. As Neal pointed out, we also need to perform the check if/when receive window opens. Tested: Following packetdrill test demonstrates the problem // Test of slow start after idle `sysctl -q net.ipv4.tcp_slow_start_after_idle=1` 0.000 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 65535 <mss 1000,sackOK,nop,nop,nop,wscale 7> +0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 6> +.100 < . 1:1(0) ack 1 win 511 +0 accept(3, ..., ...) = 4 +0 setsockopt(4, SOL_SOCKET, SO_SNDBUF, [200000], 4) = 0 +0 write(4, ..., 26000) = 26000 +0 > . 1:5001(5000) ack 1 +0 > . 5001:10001(5000) ack 1 +0 %{ assert tcpi_snd_cwnd == 10 }% +.100 < . 1:1(0) ack 10001 win 511 +0 %{ assert tcpi_snd_cwnd == 20, tcpi_snd_cwnd }% +0 > . 10001:20001(10000) ack 1 +0 > P. 20001:26001(6000) ack 1 +.100 < . 1:1(0) ack 26001 win 511 +0 %{ assert tcpi_snd_cwnd == 36, tcpi_snd_cwnd }% +4 write(4, ..., 20000) = 20000 // If slow start after idle works properly, we should send 5 MSS here (cwnd/2) +0 > . 26001:31001(5000) ack 1 +0 %{ assert tcpi_snd_cwnd == 10, tcpi_snd_cwnd }% +0 > . 31001:36001(5000) ack 1 Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Neal Cardwell <ncardwell@google.com> Cc: Yuchung Cheng <ycheng@google.com> Acked-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-24lwt: Add cfg argument to build_stateTom Herbert
Add cfg and family arguments to lwt build state functions. cfg is a void pointer and will either be a pointer to a fib_config or fib6_config structure. The family parameter indicates which one (either AF_INET or AF_INET6). LWT encpasulation implementation may use the fib configuration to build the LWT state. Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-23Merge tag 'nfc-next-4.3-1' of ↵David S. Miller
git://git.kernel.org/pub/scm/linux/kernel/git/sameo/nfc-next Samuel Ortiz says: ==================== NFC 4.3 pull request This is the NFC pull request for 4.3. With this one we have: - A new driver for Samsung's S3FWRN5 NFC chipset. In order to properly support this driver, a few NCI core routines needed to be exported. Future drivers like Intel's Fields Peak will benefit from this. - SPI support as a physical transport for STM st21nfcb. - An additional netlink API for sending replies back to userspace from vendor commands. - 2 small fixes for TI's trf7970a - A few st-nci fixes. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-23route: fix breakage after moving lwtunnel stateJiri Benc
__recnt and related fields need to be in its own cacheline for performance reasons. Commit 61adedf3e3f1 ("route: move lwtunnel state to dst_entry") broke that on 32bit archs, causing BUILD_BUG_ON in dst_hold to be triggered. This patch fixes the breakage by moving the lwtunnel state to the end of dst_entry on 32bit archs. Unfortunately, this makes it share the cacheline with __refcnt and may affect performance, thus further patches may be needed. Reported-by: kbuild test robot <fengguang.wu@intel.com> Fixes: 61adedf3e3f1 ("route: move lwtunnel state to dst_entry") Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-23vxlan: GRO support at tunnel layerTom Herbert
Add calls to gro_cells infrastructure to do GRO when receiving on a tunnel. Testing: Ran 200 netperf TCP_STREAM instance - With fix (GRO enabled on VXLAN interface) Verify GRO is happening. 9084 MBps tput 3.44% CPU utilization - Without fix (GRO disabled on VXLAN interface) Verified no GRO is happening. 9084 MBps tput 5.54% CPU utilization Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
generated by cgit (git 2.34.1) at 2025-09-21 04:51:29 +0000