- 10 Aug, 2015 15 commits
-
-
David S. Miller authored
Kaixu Xia says: ==================== bpf: Introduce the new ability of eBPF programs to access hardware PMU counter This patchset is base on the net-next: git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git commit 9dc20a64. Previous patch v6 url: https://lkml.org/lkml/2015/8/4/188 changes in V7: - rebase the whole patch set to net-next tree(9dc20a64); - split out the core perf APIs into Patch 1/5; - change the return value of function perf_event_attrs() from struct perf_event * to const struct perf_event * in Patch 1/5; - rename the function perf_event_read_internal() to perf_event_ read_local() and rewrite it in Patch 1/5; - rename the function check_func_limit() to check_map_func _compatibility() and remove the unnecessary pass pointer to a pointer in Patch 4/5; changes in V6: - make the Patch 1/4 commit message more meaning and readable; - remove the unnecessary comment in Patch 2/4 and make it clean; - declare the function perf_event_release_kernel() in include/ linux/perf_event.h to fix the build error when CONFIG_PERF_EVENTS isn't configured in Patch 2/4; - add function perf_event_attrs() to get the struct perf_event_attr in Patch 2/4. - move the related code from kernel/trace/bpf_trace.c to kernel/ events/core.c and add function perf_event_read_internal() to avoid poking inside of the event outside of perf code in Patch 3/4; - generial the func & map match-pair with an array in Patch 3/4; changes in V5: - move struct fd_array_map_ops* fd_ops to bpf_map; - move array perf event decrement refcnt function to map_free; - fix the NULL ptr of perf_event_get(); - move bpf_perf_event_read() to kernel/bpf/bpf_trace.c; - get rid of the remaining struct bpf_prog; - move the unnecessay cast on void *; changes in V4: - make the bpf_prog_array_map more generic; - fix the bug of event refcnt leak; - use more useful errno in bpf_perf_event_read(); changes in V3: - collapse V2 patches 1-3 into one; - drop the function map->ops->map_traverse_elem() and release the struct perf_event in map_free; - only allow to access bpf_perf_event_read() from programs; - update the perf_event_array_map elem via xchg(); - pass index directly to bpf_perf_event_read() instead of MAP_KEY; changes in V2: - put atomic_long_inc_not_zero() between fdget() and fdput(); - limit the event type to PERF_TYPE_RAW and PERF_TYPE_HARDWARE; - Only read the event counter on current CPU or on current process; - add new map type BPF_MAP_TYPE_PERF_EVENT_ARRAY to store the pointer to the struct perf_event; - according to the perf_event_map_fd and key, the function bpf_perf_event_read() can get the Hardware PMU counter value; Patch 5/5 is a simple example and shows how to use this new eBPF programs ability. The PMU counter data can be found in /sys/kernel/debug/tracing/trace(trace_pipe).(the cycles PMU value when 'kprobe/sys_write' sampling) $ cat /sys/kernel/debug/tracing/trace_pipe $ ./tracex6 ... syslog-ng-548 [000] d..1 76.905673: : CPU-0 681765271 syslog-ng-548 [000] d..1 76.905690: : CPU-0 681787855 syslog-ng-548 [000] d..1 76.905707: : CPU-0 681810504 syslog-ng-548 [000] d..1 76.905725: : CPU-0 681834771 syslog-ng-548 [000] d..1 76.905745: : CPU-0 681859519 syslog-ng-548 [000] d..1 76.905766: : CPU-0 681890419 syslog-ng-548 [000] d..1 76.905783: : CPU-0 681914045 syslog-ng-548 [000] d..1 76.905800: : CPU-0 681935950 syslog-ng-548 [000] d..1 76.905816: : CPU-0 681958299 ls-690 [005] d..1 82.241308: : CPU-5 3138451 sh-691 [004] d..1 82.244570: : CPU-4 7324988 <...>-699 [007] d..1 99.961387: : CPU-7 3194027 <...>-695 [003] d..1 99.961474: : CPU-3 288901 <...>-695 [003] d..1 99.961541: : CPU-3 383145 <...>-695 [003] d..1 99.961591: : CPU-3 450365 <...>-695 [003] d..1 99.961639: : CPU-3 515751 <...>-695 [003] d..1 99.961686: : CPU-3 579047 ... The detail of patches is as follow: Patch 1/5 add the necessary core perf APIs perf_event_attrs(), perf_event_get(),perf_event_read_local() when accessing events counters in eBPF programs Patch 2/5 rewrites part of the bpf_prog_array map code and make it more generic; Patch 3/5 introduces a new bpf map type. This map only stores the pointer to struct perf_event; Patch 4/5 implements function bpf_perf_event_read() that get the selected hardware PMU conuter; Patch 5/5 gives a simple example. ==================== Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Kaixu Xia authored
This is a simple example and shows how to use the new ability to get the selected Hardware PMU counter value. Signed-off-by:
Kaixu Xia <xiakaixu@huawei.com> Signed-off-by:
-
Kaixu Xia authored
According to the perf_event_map_fd and index, the function bpf_perf_event_read() can convert the corresponding map value to the pointer to struct perf_event and return the Hardware PMU counter value. Signed-off-by:
-
Kaixu Xia authored
Introduce a new bpf map type 'BPF_MAP_TYPE_PERF_EVENT_ARRAY'. This map only stores the pointer to struct perf_event. The user space event FDs from perf_event_open() syscall are converted to the pointer to struct perf_event and stored in map. Signed-off-by:
-
Wang Nan authored
All the map backends are of generic nature. In order to avoid adding much special code into the eBPF core, rewrite part of the bpf_prog_array map code and make it more generic. So the new perf_event_array map type can reuse most of code with bpf_prog_array map and add fewer lines of special code. Signed-off-by:
Wang Nan <wangnan0@huawei.com> Signed-off-by:
-
Kaixu Xia authored
This patch add three core perf APIs: - perf_event_attrs(): export the struct perf_event_attr from struct perf_event; - perf_event_get(): get the struct perf_event from the given fd; - perf_event_read_local(): read the events counters active on the current CPU; These APIs are needed when accessing events counters in eBPF programs. The API perf_event_read_local() comes from Peter and I add the corresponding SOB. Signed-off-by:
Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by:
-
David S. Miller authored
Vivien Didelot says: ==================== net: dsa: mv88e6xxx: support switchdev FDB objects This patchset refactors the DSA and mv88e6xxx code to use the switchdev FDB objects. The first two patches add minor but necessary changes to switchdev, the third one implements the switchdev glue in DSA for FDB routines, and the remaining ones refactor the FDB access functions in the mv88e6xxx code. Below is an usage example (ports 0-2 belongs to br0, ports 3-4 belongs to br1): # bridge fdb add 3c:97:0e:11:30:6e dev swp2 # bridge fdb add 3c:97:0e:11:40:78 dev swp3 # bridge fdb add 3c:97:0e:11:50:86 dev swp4 # bridge fdb del 3c:97:0e:11:40:78 dev swp3 # bridge fdb 01:00:5e:00:00:01 dev eth0 self permanent 01:00:5e:00:00:01 dev eth1 self permanent 00:50:d2:10:78:15 dev swp0 master br0 permanent 3c:97:0e:11:30:6e dev swp2 self static 00:50:d2:10:78:15 dev swp3 master br1 permanent 3c:97:0e:11:50:86 dev swp4 self static # cat /sys/kernel/debug/dsa0/atu # DB T/P Vec State Addr # 001 Port 004 e 3c:97:0e:11:30:6e # 004 Port 010 e 3c:97:0e:11:50:86 For the 88E6xxx switches, FIDs 1 to num_ports will be reserved for non-bridged ports and bridge groups, and the remaining will be later used by VLANs. This change is necessary to welcome the support for hardware VLANs (which will follow soon). Changes in v2: - remove ndo_bridge_{get,set,del}link from switchdev/DSA glue code - use ether_addr_copy instead of memcpy for MAC addresses - constify MAC address in port_fdb_{add,del} - split the mv88e6xxx code refactoring into several patches ==================== Signed-off-by: -
Vivien Didelot authored
Add a low level function for the ATU Load operation, and provide FDB add and delete wrappers functions. Signed-off-by:
Vivien Didelot <vivien.didelot@savoirfairelinux.com> Signed-off-by:
-
Vivien Didelot authored
This commit adds a low level _mv88e6xxx_atu_getnext function and helpers to rewrite the mv88e6xxx_port_fdb_getnext operation. A mv88e6xxx_atu_entry structure is added for convenient access to the hardware, and GLOBAL_ATU_FID is defined instead of the raw 0x01 value. The previous implementation did not handle the eventual trunk mapping. If the related bit is set, then the ATU data register would contain the trunk ID, and not the port vector. Check this in the FDB getnext operation and do not handle it (yet). Signed-off-by:
-
Vivien Didelot authored
Rename the __mv88e6xxx_{read,write}_addr functions to more explicit _mv88e6xxx_atu_mac_{read,write} functions, which also respect the single underscore convention used in the file (meaning SMI lock must be held). In the meantime, define their MAC address parameters as an array of ETH_ALEN bytes instead of a char pointer. Signed-off-by: -
Vivien Didelot authored
The driver currently manages one FID per port (or bridge group), with a mask of DSA_MAX_PORTS bits, where 0 means that the FID is in use. The Marvell 88E6xxx switches support up to 4094 FIDs (from 1 to 0xfff; FID 0 means that multiple address databases are not being used). This patch changes the fid_mask for an fid_bitmap of 4096 bits. >From now on, FIDs 1 to num_ports are reserved for non-bridged ports and bridge groups (a bridge group gets the FID of its first member). The remaining bits will be reserved for VLAN entries. Signed-off-by:
-
Vivien Didelot authored
Remove the fdb_{add,del,getnext} function pointer in favor of new port_fdb_{add,del,getnext}. Implement the switchdev_port_obj_{add,del,dump} functions in DSA to support the SWITCHDEV_OBJ_PORT_FDB objects. Signed-off-by: -
Vivien Didelot authored
This patch adds a is_static boolean to the switchdev_obj_fdb structure, in order to set the ndm_state to either NUD_NOARP or NUD_REACHABLE. Signed-off-by:
-
Vivien Didelot authored
The address in the switchdev_obj_fdb structure is currently represented as a pointer. Replacing it for a 6-byte array allows switchdev to carry addresses directly read from hardware registers, not stored by the switch chip driver (as in Rocker). Signed-off-by:
-
Masanari Iida authored
This patch fix a double word "the the" in Documentation/DocBook/networking.xml and Documentation/DocBook/networking/API-Wimax-report-rfkill-sw.html. These files are generated from comment in source, so I had to fix the typo in net/wimax/io-rfkill.c Signed-off-by:
Masanari Iida <standby24x7@gmail.com> Signed-off-by:
-
- 07 Aug, 2015 25 commits
-
-
Tom Herbert authored
There is a race condition in store_rps_map that allows jump label count in rps_needed to go below zero. This can happen when concurrently attempting to set and a clear map. Scenario: 1. rps_needed count is zero 2. New map is assigned by setting thread, but rps_needed count _not_ yet incremented (rps_needed count still zero) 2. Map is cleared by second thread, old_map set to that just assigned 3. Second thread performs static_key_slow_dec, rps_needed count now goes negative Fix is to increment or decrement rps_needed under the spinlock. Signed-off-by:
Tom Herbert <tom@herbertland.com> Signed-off-by:
-
Wenyu Zhang authored
When sampling rate is 1, the sampling probability is UINT32_MAX. The packet should be sampled even the prandom32() generate the number of UINT32_MAX. And none packet need be sampled when the probability is 0. Signed-off-by:
Wenyu Zhang <wenyuz@vmware.com> Acked-by:
Pravin B Shelar <pshelar@nicira.com> Signed-off-by:
-
Alexei Starovoitov authored
IFLA_VXLAN_FLOWBASED is useless without IFLA_VXLAN_COLLECT_METADATA, so combine them into single IFLA_VXLAN_COLLECT_METADATA flag. 'flowbased' doesn't convey real meaning of the vxlan tunnel mode. This mode can be used by routing, tc+bpf and ovs. Only ovs is strictly flow based, so 'collect metadata' is a better name for this tunnel mode. Signed-off-by:
Alexei Starovoitov <ast@plumgrid.com> Acked-by:
Thomas Graf <tgraf@suug.ch> Signed-off-by:
-
David S. Miller authored
Sowmini Varadhan says: ==================== RDS-TCP: Network namespace support This patch series contains the set of changes to correctly set up the infra for PF_RDS sockets that use TCP as the transport in multiple network namespaces. Patch 1 in the series is the minimal set of changes to allow a single instance of RDS-TCP to run in any (i.e init_net or other) net namespace. The changes in this patch set ensure that the execution of 'modprobe [-r] rds_tcp' sets up the kernel TCP sockets relative to the current netns, so that RDS applications can send/recv packets from that netns, and the netns can later be deleted cleanly. Patch 2 of the series further allows multiple RDS-TCP instances, one per network namespace. The changes in this patch allows dynamic creation/tear-down of RDS-TCP client and server sockets across all current and future namespaces. v2 changes from RFC sent out earlier: David Ahern comments in patch 1, net_device notifier in patch 2, patch 3 broken off and submitted separately. v3: Cong Wang review comments. ==================== Signed-off-by: -
Sowmini Varadhan authored
Register pernet subsys init/stop functions that will set up and tear down per-net RDS-TCP listen endpoints. Unregister pernet subusys functions on 'modprobe -r' to clean up these end points. Enable keepalive on both accept and connect socket endpoints. The keepalive timer expiration will ensure that client socket endpoints will be removed as appropriate from the netns when an interface is removed from a namespace. Register a device notifier callback that will clean up all sockets (and thus avoid the need to wait for keepalive timeout) when the loopback device is unregistered from the netns indicating that the netns is getting deleted. Signed-off-by:
Sowmini Varadhan <sowmini.varadhan@oracle.com> Signed-off-by:
-
Sowmini Varadhan authored
Open the sockets calling sock_create_kern() with the correct struct net pointer, and use that struct net pointer when verifying the address passed to rds_bind(). Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== Intel Wired LAN Driver Updates 2015-08-05 This series contains updates to i40e, i40evf and e1000e. Anjali adds support for x772 devices to i40e and i40evf. With the added support, x772 supports offloading of the outer UDP transmit and receive checksum for tunneled packets. Also supports evicting ATR filters in the hardware, so update the driver with this new feature set. Raanan provides several fixes for e1000e, first rectifies the Energy Efficient Ethernet in Sx code so that it only applies to parts that actually support EEE in Sx. Fix whitespace and moved ICH8 related define to the proper context. Fixed the ASPM locking which was reported by Bjorn Helgaas. Fix a workaround implementation for systime which could experience a large non-linear increment of the systime value when checking for overflow. ==================== Signed-off-by: -
Jason A. Donenfeld authored
The pr_debug family of functions turns into a no-op when -DDEBUG is not specified, opting instead to call "no_printk", which gets compiled to a no-op (but retains gcc's nice warnings about printf-style arguments). The problem with net_dbg_ratelimited is that it is defined to be a variant of net_ratelimited_function, which expands to essentially: if (net_ratelimit()) pr_debug(fmt, ...); When DEBUG is not defined, then this becomes, if (net_ratelimit()) ; This seems benign, except it isn't. Firstly, there's the obvious overhead of calling net_ratelimit needlessly, which does quite some book keeping for the rate limiting. Given that the pr_debug and net_dbg_ratelimited family of functions are sprinkled liberally through performance critical code, with developers assuming they'll be compiled out to a no-op most of the time, we certainly do not want this needless book keeping. Secondly, and most visibly, even though no debug message is printed when DEBUG is not defined, if there is a flood of invocations, dmesg winds up peppered with messages such as "net_ratelimit: 320 callbacks suppressed". This is because our aforementioned net_ratelimit() function actually prints this text in some circumstances. It's especially odd to see this when there isn't any other accompanying debug message. So, in sum, it doesn't make sense to have this function's current behavior, and instead it should match what every other debug family of functions in the kernel does with !DEBUG -- nothing. This patch replaces calls to net_dbg_ratelimited when !DEBUG with no_printk, keeping with the idiom of all the other debug print helpers. Also, though not strictly neccessary, it guards the call with an if (0) so that all evaluation of any arguments are sure to be compiled out. Signed-off-by:Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by:
-
Roopa Prabhu authored
This patch adds null dev check for the 'cfg->rc_via_table == NEIGH_LINK_TABLE or dev_get_by_index() failed' case Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
Roopa Prabhu <roopa@cumulusnetworks.com> Signed-off-by:
-
David S. Miller authored
Nicolas Schichan says: ==================== test_bpf improvements Please find below the patch series with my latest changes to test_bpf. The first patch checks for unexpected NULL generated skbs before running the filter. The second patch adds the possibility for tests to generate fragmented skbs. The third patch tests LD_ABS and LD_IND on fragmented skbs. The fourth patch adds the possibility to restrict the tests being run by specifying the name/id/range of the test(s) to run via module parameters. The fifth patch tests LD_ABS and LD_IND on non fragmented skbs with various sizes and alignments. The sixth and final patch checks that the interpreter or JIT correctly resets A and X to 0. This serie is against today's net-next tree. Changes in V2: * move declaration of 'ptr' in if() block in patch 2/6. * fix various typos in patch 4/6 * rework default init of test_range array and cleanup exclude_test() return condition in patch 4/6. ==================== Signed-off-by: -
Nicolas Schichan authored
It is mandatory for the JIT or interpreter to reset the A and X registers to 0 before running the filter. Check that it is the case on various ALU and JMP instructions. Signed-off-by:
Nicolas Schichan <nschichan@freebox.fr> Acked-by:
Daniel Borkmann <daniel@iogearbox.net> Signed-off-by:
-
Nicolas Schichan authored
This exerces the LD_ABS and LD_IND instructions for various sizes and alignments. This also checks that X when used as an offset to a BPF_IND instruction first in a filter is correctly set to 0. Signed-off-by:
-
Nicolas Schichan authored
When developping on the interpreter or a particular JIT, it can be interesting to restrict the tests list to a specific test or a particular range of tests. This patch adds the following module parameters to the test_bpf module: * test_name=<string>: only the specified named test will be run. * test_id=<number>: only the test with the specified id will be run (see the output of test_bpf without parameters to get the test id). * test_range=<number>,<number>: only the tests within IDs in the specified id range are run (see the output of test_bpf without parameters to get the test ids). Any invalid range, test id or test name will result in -EINVAL being returned and no tests being run. Signed-off-by:
-
Nicolas Schichan authored
These new tests exercise various load sizes and offsets crossing the head/fragment boundary. Signed-off-by:
-
Nicolas Schichan authored
This introduce a new test->aux flag (FLAG_SKB_FRAG) to tell the populate_skb() function to add a fragment to the test skb containing the data specified in test->frag_data). Signed-off-by:
-
Nicolas Schichan authored
Signed-off-by:
-
David S. Miller authored
Amir Vadai says: ==================== net/mlx5e: Driver updates 04-Aug-2015 This patchset introduces two features to the ConnectX-4 driver: Patch 8/8 ("Support physical port counters") exposes some hardware counters through ethtool. Rest of the patches are preparation and usage of what we call light-weight netdev open/close. Some flows that used to be in the ndo_open/stop are moved to the PCI probe/remove flows - i.e. we will make the netdev open/close operations more "light-weight". The benefits of this change are: 1) Reduce the execution time of the stop/open operations. 2) Avoid saving SW shadows of resource configurations that must persist through stop/open operations (e.g flow table steering rules), and avoid deleting/applying them from/to the device upon netdev stop/open. 3) Avoid synchronizing threads that access those resources with the netdev stop/open threads. Instead of create/destroy the resource during netdev open/stop, This patchset changes the behavior such that upon netdev stop, traffic is redirected to a "Drop RQ" (a RQ that silently drops, at the NIC HW level all incoming traffic). After redirecting the traffic, RX/TX software resources could be destroyed. During netdev open, the RX/TX rings are created and traffic is redirected to the RX rings. Patchset was applied and tested over commit ba7591d8 ("ebpf: add skb->hash to offset map for usage in {cls, act}_bpf or filters") ==================== Signed-off-by: -
Gal Pressman authored
Added physical port counters in the following standard formats to ethtool statistics: - IEEE 802.3 - RFC2863 - RFC2819 Signed-off-by:
Gal Pressman <galp@mellanox.com> Signed-off-by:
Saeed Mahameed <saeedm@mellanox.com> Signed-off-by:
Amir Vadai <amirv@mellanox.com> Signed-off-by:
-
Achiad Shochat authored
Now that TIRs, TISs and flow tables are kept alive while the netdev is stopped (after executing ndo_stop()) we can do the following improvements: - Obsolete the active_vlans SW shadow. - Do not delete/add flow table rules upon ndo_stop/open. In addition to simplifying the flow, this change also fastens the ndo_open/close operations. - Obsolete synchronization of threads accessing the flow tables with the netdev stop/open threads. Signed-off-by:
Achiad Shochat <achiad@mellanox.com> Signed-off-by:
-
Achiad Shochat authored
It does not make sense to allow events while the netdev is unregistered. Signed-off-by:
-
Achiad Shochat authored
Rename some functions that used to be invoked upon ndo_open/stop and are now invoked upon create/destroy_netdev() in order to better hint their place in the flow. Change some functions location in the file so that functions involved in ndo_open/stop flow will not be interleaved with other functions. This is a cosmetic change, no logical change here. Signed-off-by:
-
Achiad Shochat authored
Create/destroy TIRs, TISs and flow tables upon PCI probe/remove rather than upon the netdev ndo_open/stop. Upon ndo_stop(), redirect all RX traffic to the (lately introduced) "Drop RQ" and then close only the RX/TX rings, leaving the TIRs, TISs and flow tables alive. Signed-off-by:
-
Achiad Shochat authored
To be used by the mlx5 Eth driver in following commit. This is in preparation for netdev "light-weight" open/stop flow change described in previous commit. Signed-off-by:
-
Achiad Shochat authored
RX traffic routed to this RQ will be silently dropped, at the NIC HW level. This is in preparation for netdev "light-weight" open/stop flow change described in previous commit. Signed-off-by:
-
Achiad Shochat authored
Generally an RX packet flows through the following objects: Flow table --> TIR --> RQT --> RQ Where: - TIR stands for "Transport Interface Receive", defining the RSS and LRO paramaters. - RQT stands for "RQ Table", implementing the RSS indirection table. - RQ stands for "Receive Queue" For flows that do not need LRO, nor RSS, the driver made a shortcut to the above RX flow by pointing to the RQ directly from the TIR, yielding this flow: Flow table --> TIR --> RQ In this commit we remove this shortcut by "inserting" a single-RQ RQT between the TIR and the RQ, i.e RX packets will reach the same RQ but will go through an RQT of size 1, pointing to just a single RQ. This way the RX traffic re-direction to/from the "Drop RQ" will be more uniform (AKA "one flow"), as it will involve only RQTs re-direction and no TIRs re-direction. Signed-off-by:
-
