1. 15 Jun, 2019 15 commits
    • David S. Miller's avatar
      Merge branch 'tcp-add-three-static-keys' · 35fc07ae
      David S. Miller authored
      Eric Dumazet says:
      
      ====================
      tcp: add three static keys
      
      Recent addition of per TCP socket rx/tx cache brought
      regressions for some workloads, as reported by Feng Tang.
      
      It seems better to make them opt-in, before we adopt better
      heuristics.
      
      The last patch adds high_order_alloc_disable sysctl
      to ask TCP sendmsg() to exclusively use order-0 allocations,
      as mm layer has specific optimizations.
      ====================
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      35fc07ae
    • Eric Dumazet's avatar
      net: add high_order_alloc_disable sysctl/static key · ce27ec60
      Eric Dumazet authored
      >From linux-3.7, (commit 5640f768 "net: use a per task frag
      allocator") TCP sendmsg() has preferred using order-3 allocations.
      
      While it gives good results for most cases, we had reports
      that heavy uses of TCP over loopback were hitting a spinlock
      contention in page allocations/freeing.
      
      This commits adds a sysctl so that admins can opt-in
      for order-0 allocations. Hopefully mm layer might optimize
      order-3 allocations in the future since it could give us
      a nice boost  (see 8 lines of following benchmark)
      
      The following benchmark shows a win when more than 8 TCP_STREAM
      threads are running (56 x86 cores server in my tests)
      
      for thr in {1..30}
      do
       sysctl -wq net.core.high_order_alloc_disable=0
       T0=`./super_netperf $thr -H 127.0.0.1 -l 15`
       sysctl -wq net.core.high_order_alloc_disable=1
       T1=`./super_netperf $thr -H 127.0.0.1 -l 15`
       echo $thr:$T0:$T1
      done
      
      1: 49979: 37267
      2: 98745: 76286
      3: 141088: 110051
      4: 177414: 144772
      5: 197587: 173563
      6: 215377: 208448
      7: 241061: 234087
      8: 267155: 263373
      9: 295069: 297402
      10: 312393: 335213
      11: 340462: 368778
      12: 371366: 403954
      13: 412344: 443713
      14: 426617: 473580
      15: 474418: 507861
      16: 503261: 538539
      17: 522331: 563096
      18: 532409: 567084
      19: 550824: 605240
      20: 525493: 641988
      21: 564574: 665843
      22: 567349: 690868
      23: 583846: 710917
      24: 588715: 736306
      25: 603212: 763494
      26: 604083: 792654
      27: 602241: 796450
      28: 604291: 797993
      29: 611610: 833249
      30: 577356: 841062
      Signed-off-by: default avatarEric Dumazet <edumazet@google.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ce27ec60
    • Eric Dumazet's avatar
      tcp: add tcp_tx_skb_cache sysctl · 0b7d7f6b
      Eric Dumazet authored
      Feng Tang reported a performance regression after introduction
      of per TCP socket tx/rx caches, for TCP over loopback (netperf)
      
      There is high chance the regression is caused by a change on
      how well the 32 KB per-thread page (current->task_frag) can
      be recycled, and lack of pcp caches for order-3 pages.
      
      I could not reproduce the regression myself, cpus all being
      spinning on the mm spinlocks for page allocs/freeing, regardless
      of enabling or disabling the per tcp socket caches.
      
      It seems best to disable the feature by default, and let
      admins enabling it.
      
      MM layer either needs to provide scalable order-3 pages
      allocations, or could attempt a trylock on zone->lock if
      the caller only attempts to get a high-order page and is
      able to fallback to order-0 ones in case of pressure.
      
      Tests run on a 56 cores host (112 hyper threads)
      
      -	35.49%	netperf 		 [kernel.vmlinux]	  [k] queued_spin_lock_slowpath
         - 35.49% queued_spin_lock_slowpath
      	  - 18.18% get_page_from_freelist
      		 - __alloc_pages_nodemask
      			- 18.18% alloc_pages_current
      				 skb_page_frag_refill
      				 sk_page_frag_refill
      				 tcp_sendmsg_locked
      				 tcp_sendmsg
      				 inet_sendmsg
      				 sock_sendmsg
      				 __sys_sendto
      				 __x64_sys_sendto
      				 do_syscall_64
      				 entry_SYSCALL_64_after_hwframe
      				 __libc_send
      	  + 17.31% __free_pages_ok
      +	31.43%	swapper 		 [kernel.vmlinux]	  [k] intel_idle
      +	 9.12%	netperf 		 [kernel.vmlinux]	  [k] copy_user_enhanced_fast_string
      +	 6.53%	netserver		 [kernel.vmlinux]	  [k] copy_user_enhanced_fast_string
      +	 0.69%	netserver		 [kernel.vmlinux]	  [k] queued_spin_lock_slowpath
      +	 0.68%	netperf 		 [kernel.vmlinux]	  [k] skb_release_data
      +	 0.52%	netperf 		 [kernel.vmlinux]	  [k] tcp_sendmsg_locked
      	 0.46%	netperf 		 [kernel.vmlinux]	  [k] _raw_spin_lock_irqsave
      
      Fixes: 472c2e07 ("tcp: add one skb cache for tx")
      Signed-off-by: default avatarEric Dumazet <edumazet@google.com>
      Reported-by: default avatarFeng Tang <feng.tang@intel.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      0b7d7f6b
    • Eric Dumazet's avatar
      tcp: add tcp_rx_skb_cache sysctl · ede61ca4
      Eric Dumazet authored
      Instead of relying on rps_needed, it is safer to use a separate
      static key, since we do not want to enable TCP rx_skb_cache
      by default. This feature can cause huge increase of memory
      usage on hosts with millions of sockets.
      Signed-off-by: default avatarEric Dumazet <edumazet@google.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ede61ca4
    • Eric Dumazet's avatar
      sysctl: define proc_do_static_key() · a8e11e5c
      Eric Dumazet authored
      Convert proc_dointvec_minmax_bpf_stats() into a more generic
      helper, since we are going to use jump labels more often.
      
      Note that sysctl_bpf_stats_enabled is removed, since
      it is no longer needed/used.
      Signed-off-by: default avatarEric Dumazet <edumazet@google.com>
      Acked-by: default avatarAlexei Starovoitov <ast@kernel.org>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      a8e11e5c
    • Haiyang Zhang's avatar
      hv_netvsc: Set probe mode to sync · 9a33629b
      Haiyang Zhang authored
      For better consistency of synthetic NIC names, we set the probe mode to
      PROBE_FORCE_SYNCHRONOUS. So the names can be aligned with the vmbus
      channel offer sequence.
      
      Fixes: af0a5646 ("use the new async probing feature for the hyperv drivers")
      Signed-off-by: default avatarHaiyang Zhang <haiyangz@microsoft.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      9a33629b
    • Vlad Buslov's avatar
      net: sched: flower: don't call synchronize_rcu() on mask creation · 99815f50
      Vlad Buslov authored
      Current flower mask creating code assumes that temporary mask that is used
      when inserting new filter is stack allocated. To prevent race condition
      with data patch synchronize_rcu() is called every time fl_create_new_mask()
      replaces temporary stack allocated mask. As reported by Jiri, this
      increases runtime of creating 20000 flower classifiers from 4 seconds to
      163 seconds. However, this design is no longer necessary since temporary
      mask was converted to be dynamically allocated by commit 2cddd201
      ("net/sched: cls_flower: allocate mask dynamically in fl_change()").
      
      Remove synchronize_rcu() calls from mask creation code. Instead, refactor
      fl_change() to always deallocate temporary mask with rcu grace period.
      
      Fixes: 195c234d ("net: sched: flower: handle concurrent mask insertion")
      Reported-by: default avatarJiri Pirko <jiri@mellanox.com>
      Signed-off-by: default avatarVlad Buslov <vladbu@mellanox.com>
      Tested-by: default avatarJiri Pirko <jiri@mellanox.com>
      Acked-by: default avatarJiri Pirko <jiri@mellanox.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      99815f50
    • Anders Roxell's avatar
      net: dsa: fix warning same module names · f0c03ee0
      Anders Roxell authored
      When building with CONFIG_NET_DSA_REALTEK_SMI and CONFIG_REALTEK_PHY
      enabled as loadable modules, we see the following warning:
      
      warning: same module names found:
        drivers/net/phy/realtek.ko
        drivers/net/dsa/realtek.ko
      
      Rework so the driver name is realtek-smi instead of realtek.
      Reviewed-by: default avatarLinus Walleij <linus.walleij@linaro.org>
      Reviewed-by: default avatarAndrew Lunn <andrew@lunn.ch>
      Signed-off-by: default avatarAnders Roxell <anders.roxell@linaro.org>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      f0c03ee0
    • Neil Horman's avatar
      sctp: Free cookie before we memdup a new one · ce950f10
      Neil Horman authored
      Based on comments from Xin, even after fixes for our recent syzbot
      report of cookie memory leaks, its possible to get a resend of an INIT
      chunk which would lead to us leaking cookie memory.
      
      To ensure that we don't leak cookie memory, free any previously
      allocated cookie first.
      
      Change notes
      v1->v2
      update subsystem tag in subject (davem)
      repeat kfree check for peer_random and peer_hmacs (xin)
      
      v2->v3
      net->sctp
      also free peer_chunks
      
      v3->v4
      fix subject tags
      
      v4->v5
      remove cut line
      Signed-off-by: default avatarNeil Horman <nhorman@tuxdriver.com>
      Reported-by: syzbot+f7e9153b037eac9b1df8@syzkaller.appspotmail.com
      CC: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
      CC: Xin Long <lucien.xin@gmail.com>
      CC: "David S. Miller" <davem@davemloft.net>
      CC: netdev@vger.kernel.org
      Acked-by: default avatarMarcelo Ricardo Leitner <marcelo.leitner@gmail.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ce950f10
    • Robert Hancock's avatar
      net: dsa: microchip: Don't try to read stats for unused ports · 6bb9e376
      Robert Hancock authored
      If some of the switch ports were not listed in the device tree, due to
      being unused, the ksz_mib_read_work function ended up accessing a NULL
      dp->slave pointer and causing an oops. Skip checking statistics for any
      unused ports.
      
      Fixes: 7c6ff470 ("net: dsa: microchip: add MIB counter reading support")
      Signed-off-by: default avatarRobert Hancock <hancock@sedsystems.ca>
      Reviewed-by: default avatarVivien Didelot <vivien.didelot@gmail.com>
      Reviewed-by: default avatarAndrew Lunn <andrew@lunn.ch>
      Reviewed-by: default avatarFlorian Fainelli <f.fainelli@gmail.com>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      6bb9e376
    • David S. Miller's avatar
      Merge branch 'qmi_wwan-fix-QMAP-handling' · 2309f517
      David S. Miller authored
      Reinhard Speyerer says:
      
      ====================
      qmi_wwan: fix QMAP handling
      
      This series addresses the following issues observed when using the
      QMAP support of the qmi_wwan driver:
      
      1. The QMAP code in the qmi_wwan driver is based on the CodeAurora
         GobiNet driver ([1], [2]) which does not process QMAP padding
         in the RX path correctly. This causes qmimux_rx_fixup() to pass
         incorrect data to the IP stack when padding is used.
      
      2. qmimux devices currently lack proper network device usage statistics.
      
      3. RCU stalls on device disconnect with QMAP activated like this
      
         # echo Y > /sys/class/net/wwan0/qmi/raw_ip
         # echo 1 > /sys/class/net/wwan0/qmi/add_mux
         # echo 2 > /sys/class/net/wwan0/qmi/add_mux
         # echo 3 > /sys/class/net/wwan0/qmi/add_mux
      
         have been observed in certain setups:
      
         [ 2273.676593] option1 ttyUSB16: GSM modem (1-port) converter now disconnected from ttyUSB16
         [ 2273.676617] option 6-1.2:1.0: device disconnected
         [ 2273.676774] WARNING: CPU: 1 PID: 141 at kernel/rcu/tree_plugin.h:342 rcu_note_context_switch+0x2a/0x3d0
         [ 2273.676776] Modules linked in: option qmi_wwan cdc_mbim cdc_ncm qcserial cdc_wdm usb_wwan sierra sierra_net usbnet mii edd coretemp iptable_mangle ip6_tables iptable_filter ip_tables cdc_acm dm_mod dax iTCO_wdt evdev iTCO_vendor_support sg ftdi_sio usbserial e1000e ptp pps_core i2c_i801 ehci_pci button lpc_ich i2c_core mfd_core uhci_hcd ehci_hcd rtc_cmos usbcore usb_common sd_mod fan ata_piix thermal
         [ 2273.676817] CPU: 1 PID: 141 Comm: kworker/1:1 Not tainted 4.19.38-rsp-1 #1
         [ 2273.676819] Hardware name: Not Applicable   Not Applicable  /CX-GS/GM45-GL40             , BIOS V1.11      03/23/2011
         [ 2273.676828] Workqueue: usb_hub_wq hub_event [usbcore]
         [ 2273.676832] EIP: rcu_note_context_switch+0x2a/0x3d0
         [ 2273.676834] Code: 55 89 e5 57 56 89 c6 53 83 ec 14 89 45 f0 e8 5d ff ff ff 89 f0 64 8b 3d 24 a6 86 c0 84 c0 8b 87 04 02 00 00 75 7a 85 c0 7e 7a <0f> 0b 80 bf 08 02 00 00 00 0f 84 87 00 00 00 e8 b2 e2 ff ff bb dc
         [ 2273.676836] EAX: 00000001 EBX: f614bc00 ECX: 00000001 EDX: c0715b81
         [ 2273.676838] ESI: 00000000 EDI: f18beb40 EBP: f1a3dc20 ESP: f1a3dc00
         [ 2273.676840] DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068 EFLAGS: 00010002
         [ 2273.676842] CR0: 80050033 CR2: b7e97230 CR3: 2f9c4000 CR4: 000406b0
         [ 2273.676843] Call Trace:
         [ 2273.676847]  ? preempt_count_add+0xa5/0xc0
         [ 2273.676852]  __schedule+0x4e/0x4f0
         [ 2273.676855]  ? __queue_work+0xf1/0x2a0
         [ 2273.676858]  ? _raw_spin_lock_irqsave+0x14/0x40
         [ 2273.676860]  ? preempt_count_add+0x52/0xc0
         [ 2273.676862]  schedule+0x33/0x80
         [ 2273.676865]  _synchronize_rcu_expedited+0x24e/0x280
         [ 2273.676867]  ? rcu_accelerate_cbs_unlocked+0x70/0x70
         [ 2273.676871]  ? wait_woken+0x70/0x70
         [ 2273.676873]  ? rcu_accelerate_cbs_unlocked+0x70/0x70
         [ 2273.676875]  ? _synchronize_rcu_expedited+0x280/0x280
         [ 2273.676877]  synchronize_rcu_expedited+0x22/0x30
         [ 2273.676881]  synchronize_net+0x25/0x30
         [ 2273.676885]  dev_deactivate_many+0x133/0x230
         [ 2273.676887]  ? preempt_count_add+0xa5/0xc0
         [ 2273.676890]  __dev_close_many+0x4d/0xc0
         [ 2273.676892]  ? skb_dequeue+0x40/0x50
         [ 2273.676895]  dev_close_many+0x5d/0xd0
         [ 2273.676898]  rollback_registered_many+0xbf/0x4c0
         [ 2273.676901]  ? raw_notifier_call_chain+0x1a/0x20
         [ 2273.676904]  ? call_netdevice_notifiers_info+0x23/0x60
         [ 2273.676906]  ? netdev_master_upper_dev_get+0xe/0x70
         [ 2273.676908]  rollback_registered+0x1f/0x30
         [ 2273.676911]  unregister_netdevice_queue+0x47/0xb0
         [ 2273.676915]  qmimux_unregister_device+0x1f/0x30 [qmi_wwan]
         [ 2273.676917]  qmi_wwan_disconnect+0x5d/0x90 [qmi_wwan]
         ...
         [ 2273.677001] ---[ end trace 0fcc5f88496b485a ]---
         [ 2294.679136] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
         [ 2294.679140] rcu: 	Tasks blocked on level-0 rcu_node (CPUs 0-1): P141
         [ 2294.679144] rcu: 	(detected by 0, t=21002 jiffies, g=265857, q=8446)
         [ 2294.679148] kworker/1:1     D    0   141      2 0x80000000
      
      In addition the permitted QMAP mux_id value range is extended for
      compatibility with ip(8) and the rmnet driver.
      
      Reinhard
      
      [1]: https://portland.source.codeaurora.org/patches/quic/gobi
      [2]: https://portland.source.codeaurora.org/quic/qsdk/oss/lklm/gobinet/
      ====================
      Tested-by: default avatarDaniele Palmas <dnlplm@gmail.com>
      Acked-by: default avatarBjørn Mork <bjorn@mork.no>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      2309f517
    • Reinhard Speyerer's avatar
      qmi_wwan: extend permitted QMAP mux_id value range · 36815b41
      Reinhard Speyerer authored
      Permit mux_id values up to 254 to be used in qmimux_register_device()
      for compatibility with ip(8) and the rmnet driver.
      
      Fixes: c6adf779 ("net: usb: qmi_wwan: add qmap mux protocol support")
      Cc: Daniele Palmas <dnlplm@gmail.com>
      Signed-off-by: default avatarReinhard Speyerer <rspmn@arcor.de>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      36815b41
    • Reinhard Speyerer's avatar
      qmi_wwan: avoid RCU stalls on device disconnect when in QMAP mode · a8fdde1c
      Reinhard Speyerer authored
      Switch qmimux_unregister_device() and qmi_wwan_disconnect() to
      use unregister_netdevice_queue() and unregister_netdevice_many()
      instead of unregister_netdevice(). This avoids RCU stalls which
      have been observed on device disconnect in certain setups otherwise.
      
      Fixes: c6adf779 ("net: usb: qmi_wwan: add qmap mux protocol support")
      Cc: Daniele Palmas <dnlplm@gmail.com>
      Signed-off-by: default avatarReinhard Speyerer <rspmn@arcor.de>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      a8fdde1c
    • Reinhard Speyerer's avatar
      qmi_wwan: add network device usage statistics for qmimux devices · 44f82312
      Reinhard Speyerer authored
      Add proper network device usage statistics for qmimux devices
      instead of reporting all-zero values for them.
      
      Fixes: c6adf779 ("net: usb: qmi_wwan: add qmap mux protocol support")
      Cc: Daniele Palmas <dnlplm@gmail.com>
      Signed-off-by: default avatarReinhard Speyerer <rspmn@arcor.de>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      44f82312
    • Reinhard Speyerer's avatar
      qmi_wwan: add support for QMAP padding in the RX path · 61356088
      Reinhard Speyerer authored
      The QMAP code in the qmi_wwan driver is based on the CodeAurora GobiNet
      driver which does not process QMAP padding in the RX path correctly.
      Add support for QMAP padding to qmimux_rx_fixup() according to the
      description of the rmnet driver.
      
      Fixes: c6adf779 ("net: usb: qmi_wwan: add qmap mux protocol support")
      Cc: Daniele Palmas <dnlplm@gmail.com>
      Signed-off-by: default avatarReinhard Speyerer <rspmn@arcor.de>
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      61356088
  2. 14 Jun, 2019 10 commits
  3. 12 Jun, 2019 15 commits