- 16 Dec, 2018 4 commits
-
-
David Ahern authored
Similar to routes and rules, add protocol attribute to neighbor entries for easier tracking of how each was created. Signed-off-by:
David Ahern <dsahern@gmail.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Peter Oskolkov authored
This patch adds coverage of DCCP to reuseport_addr_any selftest. Signed-off-by:
Peter Oskolkov <posk@google.com> Signed-off-by:
-
Peter Oskolkov authored
Commit d9fbc7f6 "net: tcp: prefer listeners bound to an address" removes port-only listener lookups. This caused segfaults in DCCP lookups because DCCP did not initialize the (addr,port) hashtable. This patch adds said initialization. The only non-trivial issue here is the size of the new hashtable. It seemed reasonable to make it match the size of the port-only hashtable (= INET_LHTABLE_SIZE) that was used previously. Other parameters to inet_hashinfo2_init() match those used in TCP. Tested: syzcaller issues fixed; the second patch in the patchset tests that DCCP lookups work correctly. Fixes: d9fbc7f6 "net: tcp: prefer listeners bound to an address" Reported-by:
syzcaller <syzkaller@googlegroups.com> Signed-off-by:
-
Sam Protsenko authored
When Protocol Field Compression (PFC) is enabled, the "Protocol" field in PPP packet will be received without leading 0x00. See section 6.5 in RFC 1661 for details. So let's decompress protocol field if needed, the same way it's done in drivers/net/ppp/pptp.c. In case when "nopcomp" pppd option is not enabled, PFC (pcomp) can be negotiated during LCP handshake, and L2TP driver in kernel will receive PPP packets with compressed Protocol field, which in turn leads to next error: Protocol Rejected (unsupported protocol 0x2145) because instead of Protocol=0x0021 in PPP packet there will be Protocol=0x21. This patch unwraps it back to 0x0021, which fixes the issue. Sending the compressed Protocol field will be implemented in subsequent patch, this one is self-sufficient. Signed-off-by:Sam Protsenko <semen.protsenko@linaro.org> Signed-off-by:
-
- 15 Dec, 2018 22 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linuxDavid S. Miller authored
Saeed Mahameed says: ==================== mlx5e-updates-2018-12-14 (VF Lag) From Aviv Heller, Subsequent patches introduce VF LAG, which provdies load-balancing and high-availability capabilities for VFs associated with different physical ports of the same Connect-X card. This series consists of the following: - mlx5 devcom, driver infrastructure that facilitates operations that involve both core devices (physical functions) of the same card, to synchronize and communicate between two driver instances of the same card. - Infrastructure for TC rule duplication. - Changes to LAG logic to enable its use when SR-IOV is enabled - PFs in switchdev mode is the only mode currently supported. ==================== Signed-off-by: -
David S. Miller authored
Paolo Abeni says: ==================== net: mitigate retpoline overhead The spectre v2 counter-measures, aka retpolines, are a source of measurable overhead[1]. We can partially address that when the function pointer refers to a builtin symbol resorting to a list of tests vs well-known builtin function and direct calls. Experimental results show that replacing a single indirect call via retpoline with several branches and a direct call gives performance gains even when multiple branches are added - 5 or more, as reported in [2]. This may lead to some uglification around the indirect calls. In netconf 2018 Eric Dumazet described a technique to hide the most relevant part of the needed boilerplate with some macro help. This series is a [re-]implementation of such idea, exposing the introduced helpers in a new header file. They are later leveraged to avoid the indirect call overhead in the GRO path, when possible. Overall this gives > 10% performance improvement for UDP GRO benchmark and smaller but measurable for TCP syn flood. The added infra can be used in follow-up patches to cope with retpoline overhead in other points of the networking stack (e.g. at the qdisc layer) and possibly even in other subsystems. v2 -> v3: - fix build error with CONFIG_IPV6=m v1 -> v2: - list explicitly the builtin function names in INDIRECT_CALL_*(), as suggested by Ed Cree - expand the recipients list rfc -> v1: - use branch prediction hints, as suggested by Eric [1] http://vger.kernel.org/netconf2018_files/PaoloAbeni_netconf2018.pdf [2] https://linuxplumbersconf.org/event/2/contributions/99/attachments/98/117/lpc18_paper_af_xdp_perf-v2.pdf ==================== Signed-off-by:
-
Paolo Abeni authored
This avoids another indirect call for UDP GRO. Again, the test for the IPv6 variant is performed first. v1 -> v2: - adapted to INDIRECT_CALL_ changes Signed-off-by:
Paolo Abeni <pabeni@redhat.com> Signed-off-by:
-
Paolo Abeni authored
This avoids an indirect call in the receive path for TCP and UDP packets. TCP takes precedence on UDP, so that we have a single additional conditional in the common case. When IPV6 is build as module, all gro symbols except UDPv6 are builtin, while the latter belong to the ipv6 module, so we need some special care. v1 -> v2: - adapted to INDIRECT_CALL_ changes v2 -> v3: - fix build issue with CONFIG_IPV6=m Signed-off-by:
-
Paolo Abeni authored
This avoids an indirect calls for L3 GRO receive path, both for ipv4 and ipv6, if the latter is not compiled as a module. Note that when IPv6 is compiled as builtin, it will be checked first, so we have a single additional compare for the more common path. v1 -> v2: - adapted to INDIRECT_CALL_ changes Signed-off-by:
-
Paolo Abeni authored
This header define a bunch of helpers that allow avoiding the retpoline overhead when calling builtin functions via function pointers. It boils down to explicitly comparing the function pointers to known builtin functions and eventually invoke directly the latter. The macros defined here implement the boilerplate for the above schema and will be used by the next patches. rfc -> v1: - use branch prediction hint, as suggested by Eric v1 -> v2: - list explicitly the builtin function names in INDIRECT_CALL_*(), as suggested by Ed Cree Suggested-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Ilias Apalodimas authored
Currently the driver issues 2 mmio reads to figure out the number of transmitted packets and clean them. We can get rid of the expensive reads since BIT 31 of the Tx descriptor can be used for that. We can also remove the budget counting of Tx completions since all of the descriptors are not deliberately processed. Performance numbers using pktgen are: size pre-patch(pps) post-patch(pps) 64 362483 427916 128 358315 411686 256 352725 389683 512 215675 216464 1024 113812 114442 Signed-off-by:
Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by:
-
Ilias Apalodimas authored
Running pktgen with packets sizes > 512b ends up in the interface Txq getting stuck. "netsec 522d0000.ethernet eth0: netsec_netdev_start_xmit: TxQFull!" appears on dmesg but the interface never recovers. It requires an ifconfig down/up to make the interface usable again. The reason that triggers this, is a race condition between .ndo_start_xmit and the napi completion. The available budget is calculated first and indicates the queue is full. Due to a costly netif_err() the queue is not stopped in time while the napi completion runs, clears the irq and frees up descriptors, thus the queue never wakes up again. Fix this by moving the print after stopping the queue, make the print ratelimited, add barriers and check for cleaned descriptors.. Signed-off-by:
-
Fabrizio Castro authored
Document RZ/G2E (R8A774C0) SoC bindings. Signed-off-by:
Fabrizio Castro <fabrizio.castro@bp.renesas.com> Reviewed-by:
Geert Uytterhoeven <geert+renesas@glider.be> Reviewed-by:
Simon Horman <horms+renesas@verge.net.au> Reviewed-by:
Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by:
-
David Ahern authored
Move arp_queue_len_bytes ahead of arp_queue to remove two 4-byte holes. Ensure ha element is always 8-byte aligned. Signed-off-by:
-
Tonghao Zhang authored
Except for returning, the var leaf is not used in the qdisc_leaf(). For simplicity, remove it. Signed-off-by:
Tonghao Zhang <xiangxia.m.yue@gmail.com> Acked-by:
Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by:
-
David Ahern authored
A recent commit allows sockets bound to a VRF to receive ipv6 link local packets. However, it only works for UDP and worse TCP connection attempts to the LLA with the only listener bound to the VRF just hang where as before the client gets a reset and connection refused. Fix by adjusting ir_iif for LL addresses and packets received through a device enslaved to a VRF. Fixes: 6f12fa77 ("vrf: mark skb for multicast or link-local as enslaved to VRF") Reported-by:
Donald Sharp <sharpd@cumulusnetworks.com> Cc: Mike Manning <mmanning@vyatta.att-mail.com> Signed-off-by:
-
Heiner Kallweit authored
Improve detection of spurious interrupts by checking against the interrupt mask as currently set in the chip. Signed-off-by:
Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by:
-
Yangtao Li authored
We already have the DEFINE_SHOW_ATTRIBUTE. There is no need to define such a macro, so remove DEFINE_SIMPLE_DEBUGFS_FILE. Also use the DEFINE_SHOW_ATTRIBUTE macro to simplify some code. Signed-off-by:
Yangtao Li <tiny.windzz@gmail.com> Signed-off-by:
-
Yangtao Li authored
Use DEFINE_SHOW_ATTRIBUTE macro to simplify the code. Signed-off-by:
-
David S. Miller authored
Salil Mehta says: ==================== net: hns3: Add more commands to Debugfs in HNS3 driver This patch-set adds few more debugfs commands to HNS3 Ethernet Driver. Support has been added to query info related to below items: 1. Packet buffer descriptor ("echo bd info [queue no] [bd index] > cmd") 2. Manager table("echo dump mng tbl > cmd") 3. Dfx status register("echo dump reg ssu [prt id] > cmd") 4. Dcb status register("echo dump reg dcb [port id] > cmd") 5. Queue map ("echo queue map [queue no] > cmd") 6. Tm map ("echo tm map [queue no] > cmd") NOTE: Above commands are *read-only* and are only intended to query the information from the SoC(and dump inside the kernel, for now) and in no way tries to perform write operations for the purpose of configuration etc. Change Log: V1-->V2: 1. Addressed the GCC-8.2 compiler issue reported by David S. Miller. Link: https://lkml.org/lkml/2018/12/14/1298 ==================== Signed-off-by: -
liuzhongzhu authored
This patch prints dcb register status information by module. debugfs command: root@(none)# echo dump tm map 100 > cmd queue_id | qset_id | pri_id | tc_id 0100 | 0065 | 08 | 00 root@(none)# Signed-off-by:
liuzhongzhu <liuzhongzhu@huawei.com> Signed-off-by:
Salil Mehta <salil.mehta@huawei.com> Signed-off-by:
-
liuzhongzhu authored
This patch prints queue map information. debugfs command: echo dump queue map > cmd Sample Command: root@(none)# echo queue map > cmd local queue id | global queue id | vector id 0 32 769 1 33 770 2 34 771 3 35 772 4 36 773 5 37 774 6 38 775 7 39 776 8 40 777 9 41 778 10 42 779 11 43 780 12 44 781 13 45 782 14 46 783 15 47 784 root@(none)# Signed-off-by: -
liuzhongzhu authored
This patch prints dcb register status information by module. debugfs command: root@(none)# echo dump reg dcb > cmd roce_qset_mask: 0x0 nic_qs_mask: 0x0 qs_shaping_pass: 0x0 qs_bp_sts: 0x0 pri_mask: 0x0 pri_cshaping_pass: 0x0 pri_pshaping_pass: 0x0 root@(none)# Signed-off-by:
-
liuzhongzhu authored
This patch prints status register information by module. debugfs command: echo dump reg [mode name] > cmd Sample Command: root@(none)# echo dump reg bios common > cmd BP_CPU_STATE: 0x0 DFX_MSIX_INFO_NIC_0: 0xc000 DFX_MSIX_INFO_NIC_1: 0xf DFX_MSIX_INFO_NIC_2: 0x2 DFX_MSIX_INFO_NIC_3: 0x2 DFX_MSIX_INFO_ROC_0: 0xc000 DFX_MSIX_INFO_ROC_1: 0x0 DFX_MSIX_INFO_ROC_2: 0x0 DFX_MSIX_INFO_ROC_3: 0x0 root@(none)# Signed-off-by:
-
liuzhongzhu authored
This patch prints manager table information. debugfs command: echo dump mng tbl > cmd Sample Command: root@(none)# echo dump mng tbl > cmd entry|mac_addr |mask|ether|mask|vlan|mask|i_map|i_dir|e_type 00 |01:00:5e:00:00:01|0 |00000|0 |0000|0 |00 |00 |0 01 |c2:f1:c5:82:68:17|0 |00000|0 |0000|0 |00 |00 |0 root@(none)# Signed-off-by:
-
liuzhongzhu authored
This patch prints Sending and receiving package descriptor information. debugfs command: echo dump bd info 1 > cmd Sample Command: root@(none)# echo bd info 1 > cmd hns3 0000:7d:00.0: TX Queue Num: 0, BD Index: 0 hns3 0000:7d:00.0: (TX) addr: 0x0 hns3 0000:7d:00.0: (TX)vlan_tag: 0 hns3 0000:7d:00.0: (TX)send_size: 0 hns3 0000:7d:00.0: (TX)vlan_tso: 0 hns3 0000:7d:00.0: (TX)l2_len: 0 hns3 0000:7d:00.0: (TX)l3_len: 0 hns3 0000:7d:00.0: (TX)l4_len: 0 hns3 0000:7d:00.0: (TX)vlan_tag: 0 hns3 0000:7d:00.0: (TX)tv: 0 hns3 0000:7d:00.0: (TX)vlan_msec: 0 hns3 0000:7d:00.0: (TX)ol2_len: 0 hns3 0000:7d:00.0: (TX)ol3_len: 0 hns3 0000:7d:00.0: (TX)ol4_len: 0 hns3 0000:7d:00.0: (TX)paylen: 0 hns3 0000:7d:00.0: (TX)vld_ra_ri: 0 hns3 0000:7d:00.0: (TX)mss: 0 hns3 0000:7d:00.0: RX Queue Num: 0, BD Index: 120 hns3 0000:7d:00.0: (RX)addr: 0xffee7000 hns3 0000:7d:00.0: (RX)pkt_len: 0 hns3 0000:7d:00.0: (RX)size: 0 hns3 0000:7d:00.0: (RX)rss_hash: 0 hns3 0000:7d:00.0: (RX)fd_id: 0 hns3 0000:7d:00.0: (RX)vlan_tag: 0 hns3 0000:7d:00.0: (RX)o_dm_vlan_id_fb: 0 hns3 0000:7d:00.0: (RX)ot_vlan_tag: 0 hns3 0000:7d:00.0: (RX)bd_base_info: 0 Signed-off-by:
-
- 14 Dec, 2018 14 commits
-
-
David S. Miller authored
Peter Oskolkov says: ==================== net: prefer listeners bound to an address A relatively common use case is to have several IPs configured on a host, and have different listeners for each of them. We would like to add a "catch all" listener on addr_any, to match incoming connections not served by any of the listeners bound to a specific address. However, port-only lookups can match addr_any sockets when sockets listening on specific addresses are present if so_reuseport flag is set. This patchset eliminates lookups into port-only hashtable, as lookups by (addr,port) tuple are easily available. In a future patchset I plan to explore whether it is possible to remove port-only hashtables completely: additional refactoring will be required, as some non-lookup code uses the hashtables. ==================== Reviewed-by:
-
Peter Oskolkov authored
This patch adds a selftest that verifies that a socket listening on a specific address is chosen in preference over sockets that listen on any address. The test covers UDP/UDP6/TCP/TCP6. It is based on, and similar to, reuseport_dualstack.c selftest. Signed-off-by:
-
Peter Oskolkov authored
A relatively common use case is to have several IPs configured on a host, and have different listeners for each of them. We would like to add a "catch all" listener on addr_any, to match incoming connections not served by any of the listeners bound to a specific address. However, port-only lookups can match addr_any sockets when sockets listening on specific addresses are present if so_reuseport flag is set. This patch eliminates lookups into port-only hashtable, as lookups by (addr,port) tuple are easily available. In addition, compute_score() is tweaked to _not_ match addr_any sockets to specific addresses, as hash collisions could result in the unwanted behavior described above. Tested: the patch compiles; full test in the last patch in this patchset. Existing reuseport_* selftests also pass. Suggested-by:
-
Peter Oskolkov authored
A relatively common use case is to have several IPs configured on a host, and have different listeners for each of them. We would like to add a "catch all" listener on addr_any, to match incoming connections not served by any of the listeners bound to a specific address. However, port-only lookups can match addr_any sockets when sockets listening on specific addresses are present if so_reuseport flag is set. This patch eliminates lookups into port-only hashtable, as lookups by (addr,port) tuple are easily available. In addition, compute_score() is tweaked to _not_ match addr_any sockets to specific addresses, as hash collisions could result in the unwanted behavior described above. Tested: the patch compiles; full test in the last patch in this patchset. Existing reuseport_* selftests also pass. Suggested-by:
-
Peter Oskolkov authored
A relatively common use case is to have several IPs configured on a host, and have different listeners for each of them. We would like to add a "catch all" listener on addr_any, to match incoming connections not served by any of the listeners bound to a specific address. However, port-only lookups can match addr_any sockets when sockets listening on specific addresses are present if so_reuseport flag is set. This patch eliminates lookups into port-only hashtable, as lookups by (addr,port) tuple are easily available. In addition, compute_score() is tweaked to _not_ match addr_any sockets to specific addresses, as hash collisions could result in the unwanted behavior described above. Tested: the patch compiles; full test in the last patch in this patchset. Existing reuseport_* selftests also pass. Suggested-by:
-
Peter Oskolkov authored
A relatively common use case is to have several IPs configured on a host, and have different listeners for each of them. We would like to add a "catch all" listener on addr_any, to match incoming connections not served by any of the listeners bound to a specific address. However, port-only lookups can match addr_any sockets when sockets listening on specific addresses are present if so_reuseport flag is set. This patch eliminates lookups into port-only hashtable, as lookups by (addr,port) tuple are easily available. In addition, compute_score() is tweaked to _not_ match addr_any sockets to specific addresses, as hash collisions could result in the unwanted behavior described above. Tested: the patch compiles; full test in the last patch in this patchset. Existing reuseport_* selftests also pass. Suggested-by:
-
yupeng authored
Add explainations for some general IP counters, SACK and DSACK related counters Signed-off-by:
yupeng <yupeng0921@gmail.com> Signed-off-by:
-
David S. Miller authored
David Ahern says: ==================== neighbor: More gc_list changes More gc_list changes and cleanups. The first 2 patches are bug fixes from the first gc_list change. Specifically, fix the locking order to be consistent - table lock followed by neighbor lock, and then entries in the FAILED state should always be candidates for forced_gc without waiting for any time span (return to the eviction logic prior to the separate gc_list). Patch 3 removes 2 now unnecessary arguments to neigh_del. Patch 4 moves a helper from a header file to core code in preparation for Patch 5 which removes NTF_EXT_LEARNED entries from the gc_list. These entries are already exempt from forced_gc; patch 5 removes them from consideration and makes them on par with PERMANENT entries given that they are also managed by userspace. ==================== Signed-off-by: -
David Ahern authored
Externally learned entries are similar to PERMANENT entries in the sense they are managed by userspace and can not be garbage collected. As such remove them from the gc_list, remove the flags check from neigh_forced_gc and skip threshold checks in neigh_alloc. As with PERMANENT entries, this allows unlimited number of NTF_EXT_LEARNED entries. Signed-off-by:
-
David Ahern authored
neigh_update_ext_learned has one caller in neighbour.c so does not need to be defined in the header. Move it and in the process remove the intialization of ndm_flags and just set it based on the flags check. Signed-off-by:
-
David Ahern authored
neigh_del now only has 1 caller, and the state and flags arguments are both 0. Remove them and simplify neigh_del. Signed-off-by:
-
David Ahern authored
PERMANENT entries are not on the gc_list so the state check is now redundant. Also, the move to not purge entries until after 5 seconds should not apply to FAILED entries; those can be removed immediately to make way for newer ones. This restores the previous logic prior to the gc_list. Fixes: 58956317 ("neighbor: Improve garbage collection") Signed-off-by:
-
David Ahern authored
Lock checker noted an inverted lock order between neigh_change_state (neighbor lock then table lock) and neigh_periodic_work (table lock and then neighbor lock) resulting in: [ 121.057652] ====================================================== [ 121.058740] WARNING: possible circular locking dependency detected [ 121.059861] 4.20.0-rc6+ #43 Not tainted [ 121.060546] ------------------------------------------------------ [ 121.061630] kworker/0:2/65 is trying to acquire lock: [ 121.062519] (____ptrval____) (&n->lock){++--}, at: neigh_periodic_work+0x237/0x324 [ 121.063894] [ 121.063894] but task is already holding lock: [ 121.064920] (____ptrval____) (&tbl->lock){+.-.}, at: neigh_periodic_work+0x194/0x324 [ 121.066274] [ 121.066274] which lock already depends on the new lock. [ 121.066274] [ 121.067693] [ 121.067693] the existing dependency chain (in reverse order) is: ... Fix by renaming neigh_change_state to neigh_update_gc_list, changing it to only manage whether an entry should be on the gc_list and taking locks in the same order as neigh_periodic_work. Invoke at the end of neigh_update only if diff between old or new states has the PERMANENT flag set. Fixes: 8cc196d6 ("neighbor: gc_list changes should be protected by table lock") Signed-off-by: -
Cong Wang authored
After commit 69bd4840 ("net/sched: Remove egdev mechanism"), tc_setup_cb_call() is nearly identical to tcf_block_cb_call(), so we can just fold tcf_block_cb_call() into tc_setup_cb_call() and remove its unused parameter 'exts'. Fixes: 69bd4840 ("net/sched: Remove egdev mechanism") Cc: Oz Shlomo <ozsh@mellanox.com> Cc: Jiri Pirko <jiri@mellanox.com> Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Acked-by:
Oz Shlomo <ozsh@mellanox.com> Signed-off-by:
-
