- 06 Feb, 2016 19 commits
-
-
David S. Miller authored
Alexei Starovoitov says: ==================== bpf: introduce per-cpu maps We've started to use bpf to trace every packet and atomic add instruction (event JITed) started to show up in perf profile. The solution is to do per-cpu counters. For PERCPU_(HASH|ARRAY) map the existing bpf_map_lookup() helper returns per-cpu area which bpf programs can use to store and increment the counters. The BPF_MAP_LOOKUP_ELEM syscall command returns areas from all cpus and user process aggregates the counters. The usage example is in patch 6. The api turned out to be very easy to use from bpf program and from user space. Long term we were discussing to add 'bounded loop' instruction, so bpf programs can do aggregation within the program which may help some use cases. Right now user space aggregation of per-cpu counters fits the best. This patch set is new approach for per-cpu hash and array maps. I've reused the map tests written by Martin and Ming, but implementation and api is new. Old discussion here: http://thread.gmane.org/gmane.linux.kernel/2123800/focus=2126435 ==================== Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Alexei Starovoitov authored
Signed-off-by:
Alexei Starovoitov <ast@kernel.org> Signed-off-by:
-
tom.leiming@gmail.com authored
A sanity test for BPF_MAP_TYPE_PERCPU_ARRAY Signed-off-by:
Ming Lei <tom.leiming@gmail.com> Signed-off-by:
-
Martin KaFai Lau authored
A sanity test for BPF_MAP_TYPE_PERCPU_HASH. Signed-off-by:
Martin KaFai Lau <kafai@fb.com> Signed-off-by:
-
Alexei Starovoitov authored
The functions bpf_map_lookup_elem(map, key, value) and bpf_map_update_elem(map, key, value, flags) need to get/set values from all-cpus for per-cpu hash and array maps, so that user space can aggregate/update them as necessary. Example of single counter aggregation in user space: unsigned int nr_cpus = sysconf(_SC_NPROCESSORS_CONF); long values[nr_cpus]; long value = 0; bpf_lookup_elem(fd, key, values); for (i = 0; i < nr_cpus; i++) value += values[i]; The user space must provide round_up(value_size, 8) * nr_cpus array to get/set values, since kernel will use 'long' copy of per-cpu values to try to copy good counters atomically. It's a best-effort, since bpf programs and user space are racing to access the same memory. Signed-off-by: -
Alexei Starovoitov authored
Primary use case is a histogram array of latency where bpf program computes the latency of block requests or other events and stores histogram of latency into array of 64 elements. All cpus are constantly running, so normal increment is not accurate, bpf_xadd causes cache ping-pong and this per-cpu approach allows fastest collision-free counters. Signed-off-by:
-
Alexei Starovoitov authored
Introduce BPF_MAP_TYPE_PERCPU_HASH map type which is used to do accurate counters without need to use BPF_XADD instruction which turned out to be too costly for high-performance network monitoring. In the typical use case the 'key' is the flow tuple or other long living object that sees a lot of events per second. bpf_map_lookup_elem() returns per-cpu area. Example: struct { u32 packets; u32 bytes; } * ptr = bpf_map_lookup_elem(&map, &key); /* ptr points to this_cpu area of the value, so the following * increments will not collide with other cpus */ ptr->packets ++; ptr->bytes += skb->len; bpf_update_elem() atomically creates a new element where all per-cpu values are zero initialized and this_cpu value is populated with given 'value'. Note that non-per-cpu hash map always allocates new element and then deletes old after rcu grace period to maintain atomicity of update. Per-cpu hash map updates element values in-place. Signed-off-by: -
Kim Jones authored
netdev_rss_key is written to once and thereafter is read by drivers when they are initialising. The fact that it is mostly read and not written to makes it a candidate for a __read_mostly declaration. Signed-off-by:
Kim Jones <kim-marie.jones@intel.com> Signed-off-by:
Alan Carey <alan.carey@intel.com> Acked-by:
Rami Rosen <rami.rosen@intel.com> Signed-off-by:
-
David S. Miller authored
Eric Dumazet says: ==================== tcp: fastopen: accept data/FIN present in SYNACK Implements RFC 7413 (TCP Fast Open) 4.2.2, accepting payload and/or FIN in SYNACK messages, and prepare removal of SYN flag in tcp_recvmsg() ==================== Signed-off-by: -
Eric Dumazet authored
If we remove the SYN flag from the skbs that tcp_fastopen_add_skb() places in socket receive queue, then we can remove the test that tcp_recvmsg() has to perform in fast path. All we have to do is to adjust SEQ in the slow path. For the moment, we place an unlikely() and output a message if we find an skb having SYN flag set. Goal would be to get rid of the test completely. Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
Yuchung Cheng <ycheng@google.com> Signed-off-by:
Neal Cardwell <ncardwell@google.com> Signed-off-by:
-
Eric Dumazet authored
RFC 7413 (TCP Fast Open) 4.2.2 states that the SYNACK message MAY include data and/or FIN This patch adds support for the client side : If we receive a SYNACK with payload or FIN, queue the skb instead of ignoring it. Since we already support the same for SYN, we refactor the existing code and reuse it. Note we need to clone the skb, so this operation might fail under memory pressure. Sara Dickinson pointed out FreeBSD server Fast Open implementation was planned to generate such SYNACK in the future. The server side might be implemented on linux later. Reported-by:
Sara Dickinson <sara@sinodun.com> Signed-off-by:
-
David S. Miller authored
Jarod Wilson says: ==================== net: add and use rx_nohandler stat counter The network core tries to keep track of dropped packets, but some packets you wouldn't really call dropped, so much as intentionally ignored, under certain circumstances. One such case is that of bonding and team device slaves that are currently inactive. Their respective rx_handler functions return RX_HANDLER_EXACT (the only places in the kernel that return that), which ends up tracking into the network core's __netif_receive_skb_core() function's drop path, with no pt_prev set. On a noisy network, this can result in a very rapidly incrementing rx_dropped counter, not only on the inactive slave(s), but also on the master device, such as the following: $ cat /proc/net/dev Inter-| Receive | Transmit face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed p7p1: 14783346 140430 0 140428 0 0 0 2040 680 8 0 0 0 0 0 0 p7p2: 14805198 140648 0 0 0 0 0 2034 0 0 0 0 0 0 0 0 bond0: 53365248 532798 0 421160 0 0 0 115151 2040 24 0 0 0 0 0 0 lo: 5420 54 0 0 0 0 0 0 5420 54 0 0 0 0 0 0 p5p1: 19292195 196197 0 140368 0 0 0 56564 680 8 0 0 0 0 0 0 p5p2: 19289707 196171 0 140364 0 0 0 56547 680 8 0 0 0 0 0 0 em3: 20996626 158214 0 0 0 0 0 383 0 0 0 0 0 0 0 0 em2: 14065122 138462 0 0 0 0 0 310 0 0 0 0 0 0 0 0 em1: 14063162 138440 0 0 0 0 0 308 0 0 0 0 0 0 0 0 em4: 21050830 158729 0 0 0 0 0 385 71662 469 0 0 0 0 0 0 ib0: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 In this scenario, p5p1, p5p2 and p7p1 are all inactive slaves in an active-backup bond0, and you can see that all three have high drop counts, with the master bond0 showing a tally of all three. I know that this was previously discussed some here: http://www.spinics.net/lists/netdev/msg226341.html It seems additional counters never came to fruition, so this is a first attempt at creating one of them, so that we stop calling these drops, which for users monitoring rx_dropped, causes great alarm, and renders the counter much less useful for them. This adds a sysfs statistics node and makes the counter available via netlink. Additionally, I'm not certain if this set qualifies for net, or if it should be put aside and resubmitted for net-next after 4.5 is put to bed, but I do have users who consider this an important bugfix. This has been tested quite a bit on x86_64, and now lightly on i686 as well, to verify functionality of updates to netdev_stats_to_stats64() on 32-bit arches. ==================== Signed-off-by:Jarod Wilson <jarod@redhat.com> Signed-off-by:
-
Jarod Wilson authored
Sample output with this set applied for an active-backup bond: $ cat /sys/devices/virtual/net/bond0/lower_p7p1/statistics/rx_nohandler 16568 $ cat /sys/devices/virtual/net/bond0/lower_p5p2/statistics/rx_nohandler 16583 $ cat /sys/devices/virtual/net/bond0/statistics/rx_nohandler 33151 CC: Jay Vosburgh <j.vosburgh@gmail.com> CC: Veaceslav Falico <vfalico@gmail.com> CC: Andy Gospodarek <gospo@cumulusnetworks.com> CC: netdev@vger.kernel.org Signed-off-by:
-
Jarod Wilson authored
CC: Jiri Pirko <jiri@resnulli.us> CC: netdev@vger.kernel.org Signed-off-by:
-
Jarod Wilson authored
This adds an rx_nohandler stat counter, along with a sysfs statistics node, and copies the counter out via netlink as well. CC: "David S. Miller" <davem@davemloft.net> CC: Eric Dumazet <edumazet@google.com> CC: Jiri Pirko <jiri@mellanox.com> CC: Daniel Borkmann <daniel@iogearbox.net> CC: Tom Herbert <tom@herbertland.com> CC: Jay Vosburgh <j.vosburgh@gmail.com> CC: Veaceslav Falico <vfalico@gmail.com> CC: Andy Gospodarek <gospo@cumulusnetworks.com> CC: netdev@vger.kernel.org Signed-off-by:
-
Jarod Wilson authored
The netdev_stats_to_stats64 function copies the deprecated net_device_stats format stats into rtnl_link_stats64 for legacy support purposes, but with the BUILD_BUG_ON as it was, it wasn't possible to extend rtnl_link_stats64 without also extending net_device_stats. Relax the BUILD_BUG_ON to only require that rtnl_link_stats64 is larger, and zero out all the stat counters that aren't present in net_device_stats. CC: Eric Dumazet <edumazet@google.com> CC: netdev@vger.kernel.org Signed-off-by:
-
Richard Alpe authored
Currently link priority changes isn't handled for active links. In this patch we resolve this by changing our priority if the peer passes a valid priority in a state message. Reviewed-by:
Jon Maloy <jon.maloy@ericsson.com> Signed-off-by:
Richard Alpe <richard.alpe@ericsson.com> Signed-off-by:
-
Richard Alpe authored
Changing certain link attributes (link tolerance and link priority) from the TIPC management tool is supposed to automatically take effect at both endpoints of the affected link. Currently the media address is not instantiated for the link and is used uninstantiated when crafting protocol messages designated for the peer endpoint. This means that changing a link property currently results in the property being changed on the local machine but the protocol message designated for the peer gets lost. Resulting in property discrepancy between the endpoints. In this patch we resolve this by using the media address from the link entry and using the bearer transmit function to send it. Hence, we can now eliminate the redundant function tipc_link_prot_xmit() and the redundant field tipc_link::media_addr. Fixes: 2af5ae37 (tipc: clean up unused code and structures) Reviewed-by:
Jason Hu <huzhijiang@gmail.com> Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== 40GbE Intel Wired LAN Driver Updates 2016-02-03 This series contains updates to i40e and i40evf only. Kiran adds the MAC filter element to the end of the list instead of HEAD just in case there are ever any ordering issues in the future. Anjali fixes several RSS issues, first fixes the hash PCTYPE enable for X722 since it supports a broader selection of PCTYPES for TCP and UDP. Then fixes a bug in XL710, X710, and X722 support for RSS since we cannot reduce the 4-tuple for RSS for TCP/IPv4/IPv6 or UDP/IPv4/IPv6 packets since this requires a product feature change coming in a later release. Cleans up the reset code where the restart-autoneg workaround is applied, since X722 does not need the workaround, add a flag to indicate which MAC and firmware version require the workaround to be applied. Adds new device id's for X722 and code to add their support. Also adds another way to access the RSS keys and lookup table using the admin queue for X722 devices. Catherine updates the driver to replace the MAC check with a feature flag check for 100M SGMII, since it is only support on X722 devices currently. Mitch reworks the VF driver to allow channel bonding, which was not possible before this patch due to the asynchronous nature of the admin queue mechanism. Also fixes a rare case which causes a panic if the VF driver is removed during reset recovery, resolve this by setting the ring pointers to NULL after freeing them. Shannon cleans up the driver where device capabilities were defined in two different places, and neither had all the definitions, so he consolidates the definitions in the admin queue API. Also adds the new proxy-wake-on-lan capability bit available with the new X722 device. Lastly, added the new External Device Power Ability field to the get_link_status data structure by using a reserved field at the end of the structure. Jesse mimics the ixgbe driver's use of a private work queue in the i40e and i40evf drivers to avoid blocking the system work queue. Greg cleans up the driver to limit the firmware revision checks to properly handle DCB configurations from the firmware to the older devices which need these checks (specifically X710 and XL710 devices only). ==================== Signed-off-by:
-
- 05 Feb, 2016 1 commit
-
-
Mahesh Bandewar authored
When we create IPvlan slave; we use ether_setup() and that sets up default MTU to 1500 while the master device may have lower / different MTU. Any subsequent changes to the masters' MTU are reflected into the slaves' MTU setting. However if those don't happen (most likely scenario), the slaves' MTU stays at 1500 which could be bad. This change adds code to inherit MTU from the master device instead of using the default value during the link initialization phase. Signed-off-by:
Mahesh Bandewar <maheshb@google.com> CC: Eric Dumazet <eric.dumazet@gmail.com> CC: Tim Hockins <thockins@google.com> Acked-by:
-
- 04 Feb, 2016 20 commits
-
-
Catherine Sullivan authored
Add some missing reporting/advertisement of 100Mb capability for adapters that support it. Change-ID: I8b8523fbdc99517bec29d90c71b3744db11542ac Signed-off-by:
Catherine Sullivan <catherine.sullivan@intel.com> Tested-by:
Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
Shannon Nelson authored
Add the new External Device Power Ability field to the get_link_status data structure, using space from the reserved field at the end of the struct. Signed-off-by:
Shannon Nelson <shannon.nelson@intel.com> Acked-by:
Kevin Scott <kevin.c.scott@intel.com> Tested-by:
-
Shannon Nelson authored
Fix the name of the new cloud tunnel type from the place-holder NGE name to the official Geneve. Also fix the spelling of the VXLAN type. Signed-off-by:
-
Shannon Nelson authored
Add the AQ opcode and struct definitions for the Run PHY Activity command Signed-off-by:
-
Greg Bowers authored
X710/XL710 devices require FW version checks to properly handle DCB configurations from the FW. Newer devices do not, so limit these checks to X710/XL710. Signed-off-by:
Greg Bowers <gregory.j.bowers@intel.com> Acked-by:
-
Shannon Nelson authored
Add the new proxy-wake-on-lan capability bit available with the new X722 device. Signed-off-by:
Jesse Brandeburg <jesse.brandeburg@intel.com> Tested-by:
-
Jesse Brandeburg authored
As done per ixgbe, use a private workqueue to avoid blocking the system workqueue. This avoids some strange side effects when some other entity is depending on the system work queue. Change-ID: Ic8ba08f5b03696cf638b21afd25fbae7738d55ee Signed-off-by:
-
Anjali Singhai Jain authored
Add write-back on interrupt throttle rate timer expiration support for the i40evf driver, when running on X722 devices. Signed-off-by:
Anjali Singhai Jain <anjali.singhai@intel.com> Tested-by:
-
Anjali Singhai Jain authored
The PCTYPES for the X710 and X722 families are different. This patch makes adjustments for that. Signed-off-by:
-
Anjali Singhai Jain authored
This patch adds another way to access the RSS keys and lut using the AQ for X722 devices. Signed-off-by:
-
Anjali Singhai Jain authored
Add the KX and QSFP device IDs for X722. Signed-off-by:
-
Catherine Sullivan authored
Bump. Change-ID: Ic9a495feb9ab0606f953c3848b0acf67169d3930 Signed-off-by:
-
Jesse Brandeburg authored
Synchronize code bases and add SCTP offload support. Change-ID: I9f99071f7176225479026930c387bf681a47494e Signed-off-by:
-
Anjali Singhai Jain authored
The restart-autoneg work around does not apply to X722. Added a flag to set it only for the right MAC and right FW version where the work around should be applied. Signed-off-by:
-
Mitch Williams authored
Since we check these ring pointers to make sure we don't double-allocate or double-free the rings, we had better null them out after we free them. In very rare cases this can cause a panic if the driver is removed during reset recovery. Change-ID: Ib06eb4910a3058275c8f7ec5ef7f45baa4674f96 Signed-off-by:
Mitch Williams <mitch.a.williams@intel.com> Tested-by:
-
Shannon Nelson authored
The device capabilities were defined in two places, and neither had all the definitions. It really belongs with the AQ API definition, so this patch removes the other set of definitions and fills out the missing item. Change-ID: I273ba7d79a476cd11d2e0ca5825fec1716740de2 Signed-off-by:
-
Mitch Williams authored
In some modes, bonding would not enslave VF interfaces. This is due to bonding calling change_mtu and the immediately calling open. Because of the asynchronous nature of the admin queue mechanism, the VF returns -EBUSY to the open call, because it knows the previous operation hasn't finished yet. This causes bonding to fail with a less-than-useful error message. To fix this, remove the check for pending operations at the beginning of open. But this introduces a new bug where the driver will panic on a quick close/open cycle. To fix that, we add a new driver state, __I40EVF_DOWN_PENDING, that the driver enters when down is called. The driver finally transitions to a fully DOWN state when it receives confirmation from the PF driver that all the queues are disabled. This allows open to complete even if there is a pending mtu change, and bonding is finally happy. Change-ID: I06f4c7e435d5bacbfceaa7c3f209e0ff04be21cc Signed-off-by:
-
Catherine Sullivan authored
100M SGMII is only supported on X722. Replace the mac check with a feature flag check that is only set for the X722 device. Change-ID: I53452d9af6af8cd9dca8500215fbc6ce93418f52 Signed-off-by:
-
Anjali Singhai Jain authored
This patch fixes the Hash PCTYPE enable for X722 since it supports a broader selection of PCTYPES for TCP and UDP. This patch also fixes a bug in XL710, X710, X722 support for RSS, as of now we cannot reduce the (4)tuple for RSS for TCP/IPv4/IPV6 or UDP/IPv4/IPv6 packets since this requires a product feature change that comes in a later release. A VF should never be allowed to change the tuples for RSS for any PCTYPE since that's a global setting for the device in case of i40e devices. Change-ID: I0ee7203c9b24813260f58f3220798bc9d9ac4a12 Signed-off-by:
-
Kiran Patil authored
Add MAC filter element to the end of the list in the given order, just to be tidy, and just in case there are ever any ordering issues in the future. Change-ID: Idc15276147593ea9393ac72c861f9c7905a791b4 Signed-off-by:
Kiran Patil <kiran.patil@intel.com> Tested-by:
-
