)]}'
{
  "commit": "98b6ed0f2bf4abfb759206aa70690636372bdee7",
  "tree": "b9cf1610ae949db98916d0cedde5c409ecb6d341",
  "parents": [
    "1a506e473576cdcb922d339aea76b67d0fe344f7",
    "578a1310f2592ba90c5674bca21c1dbd1adf3f0a"
  ],
  "author": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Wed Jun 26 19:24:37 2013 -1000"
  },
  "committer": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Wed Jun 26 19:24:37 2013 -1000"
  },
  "message": "Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net\n\nPull networking fixes from David Miller:\n\n 1) Found via trinity:\n\n    If you connect up an ipv6 socket to an ipv4 mapped address then an\n    ipv6 one, sendmsg() can croak because ip6_sk_dst_check() assumes the\n    route cached in the socket is an ipv6 one.  In this case there is an\n    ipv4 route attached, so it gets stomped on.\n\n    Reported by Dave Jones and Hannes Frederic Sowa, fixed by Eric\n    Dumazet.\n\n 2) AF_KEY notifications leak some kernel memory to userspace, fix from\n    Mathias Krause.\n\n 3) DLCI calls __dev_get_by_name() without proper locking, and dlci_del\n    doesn\u0027t validate that the device being deleted is actually a DLCI\n    one.  Fixes from Li Zefan.\n\n 4) Length check on bluetooth l2cap information responses is wrong, each\n    response type has a different lenth, so we should make sure it\u0027s in\n    a given range rather than enforce one single valid length.  From\n    Jaganath Kanakkassery.\n\n 5) Receive FIFO overflow is really easy to trigger in stress scenerios\n    in the sh_eth driver, but the event isn\u0027t being handled properly at\n    all.  Specifically, the mask of error interrupts doesn\u0027t include the\n    event so we never clear it, resulting in the driver becomming wedged\n    processing an interrupt that never gets cleared.\n\n    Fix from Sergei Shtylyov.\n\n 6) qlcnic sleeps while holding a spinlock, use mdelay() instead of\n    msleep().  From Shahed Shaikh.\n\n 7) Missing curly braces causes SIP netfilter NAT module to always drop\n    packets.  Fix from Balazs Peter Odor.\n\n 8) ipt_ULOG in netfilter passes the wrong value to timer setup, causing\n    the timer to dereference crap when it fires.  Fix from Gao Feng.\n\n 9) Missing RCU protection around txq-\u003eaxq_acq traversal in\n    ath_txq_schedule().  Fix from Felix Fietkau.\n\n10) Idle state transition test in ath9k_htc_config() is reversed, fix\n    from Sujith Manoharan.\n\n11) IPV6 forwarding handles unicast Router Alert packets incorrectly.\n    It tests the wrong option state.  Previously opt-\u003era being non-zero\n    indicated a router alert marking in the SKB, but now it\u0027s indicated\n    by a bit in opt-\u003eflags.  Fix from YOSHIFUJI Hideaki.\n\n12) SKB leak in GRE tunnel GSO handling, from Eric Dumazet.\n\n13) get_user_pages_fast() error handling in TUN and MACVTAP use the same\n    local variable for the base index and the loop iterator for page\n    traversal, oops! Fix from Michael S Tsirkin.\n\n14) ipv6_get_lladdr() can fail, and we must therefore check it\u0027s return\n    value in inet6_set_iftoken().  For from Hannes Frederic Sowa.\n\n15) If you change an interface name and meanwhile can sneak in something\n    that looks up the name (like SO_BINDTODEVICE or SIOCGIFNAME) we can\n    deadlock with CONFIG_PREEMPT\u003dn.  Fix this by providing a helper\n    function that properly uses raw_seqcount_begin().  From Nicolas\n    Schichan.\n\n16) Chain noise calibration test is inverted in iwlwifi, fix from\n    Nikolay Martynov.\n\n17) Properly set TX iwlwifi descriptor flags for back requests.  Fix\n    from Emmanuel Grumbach.\n\n18) We can\u0027t assume skb_transport_header() is set in xt_TCPOPTSTRAP\n    module, fix from Pablo Neira Ayuso.\n\n19) Some crummy APs don\u0027t provide the proper High Throughput info in\n    association response frames.  Add a workaround by assume we\u0027ll use\n    whatever is in the beacon/probe.  Fix from Johannes Berg.\n\n20) mac80211 call to rate_idx_match_mask() swaps two arguments (mask and\n    channel width).  Fix from Simon Wunderlich.\n\n21) xt_TCPMSS (like xt_TCPOPTSTRAP) must not try to handle fragmented\n    frames.  Fix from Phil Oester.\n\n22) Fix rate control regression causing iwlwifi/iwlegacy chips to use\n    1Mbit/s on pre-11n networks.  From Moshe Benji and Stanslaw Gruszka.\n\n23) Disable brcmsmac power-save functions, they cause regressions.  From\n    Arend van Spriel.\n\n24) Enforce a sane minimum MTU in l2cap_build_cmd() otherwise we can\n    easily crash.  Fix from Anderson Lizardo.\n\n25) If a learning packet arrives during vxlan_stop() we crash, easily\n    fixed by checking netif_running().  From Stephen Hemminger.\n\n26) Static vxlan FDB entries should not be migrated, also from Stephen.\n\n27) skb_clone() failures not handled in vxlan_xmit(), oops.  Also from\n    Stephen.\n\n28) Add minimal driver for AR816x/AR817x ethernet chips, from Johannes\n    Berg.\n\n29) Fix regression in userspace VLAN acceleration control, added by the\n    802.1ad support changes.  Fix from Fernando Luis Vazquez Cao.\n\n30) Interval selection for MLD queries in the bridging code was\n    reversed.  Fix from Linus Lüssing.\n\n31) ipv6\u0027s ndisc_send_redirect() erroneously writes to the packet we\n    received not the packet we are building to send out.  Fix from\n    Matthias Schiffer.\n\n32) Don\u0027t free netdev before unregistering it, in usb_8dev can driver.\n    From Marc Kleine-Budde.\n\n33) Fix nl80211 attribute buffer races, from Johannes Berg.\n\n34) Although netlink_diag.h is under uapi/ it isn\u0027t present in Kbuild.\n    From Stephen Hemminger.\n\n35) Wrong address and family passed to MD5 key lookups in TCP, from\n    Aydin Arik.\n\n36) phy_type attribute created by SFC driver should not be writable.\n    From Ben Hutchings.\n\n37) Receive/Transmit queue allocations in pxa168_eth and mv643xx_eth\n    should use kzalloc().  Otherwise if setup fails half-way, we\u0027ll\n    dereference garbage when trying to teardown the rings.  From Lubomir\n    Rintel.\n\n38) Fix double-allocation of dst (resulting in unfreeable net device) in\n    ipv6\u0027s init_loopback().  From Gao Feng.\n\n39) Fix fragmentation handling SKB leak in netfilter conntrack, we were\n    freeing the wrong skb pointer.  From Phil Oester.\n\n40) Don\u0027t report \"-1\" (SPEED_UNKNOWN) in bond_miimon_commit(), from\n    Nikolay Aleksandrov.\n\n41) davinci_cpdma doesn\u0027t check for DMA mapping errors, letting the\n    device scribble to random addresses.  From Sebastian Siewior.\n\n* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (69 commits)\n  dlci: validate the net device in dlci_del()\n  dlci: acquire rtnl_lock before calling __dev_get_by_name()\n  af_key: fix info leaks in notify messages\n  ipv6: ip6_sk_dst_check() must not assume ipv6 dst\n  net: fix kernel deadlock with interface rename and netdev name retrieval.\n  net/tg3: Avoid delay during MMIO access\n  ipv6: check return value of ipv6_get_lladdr\n  macvtap: fix recovery from gup errors\n  tun: fix recovery from gup errors\n  gre: fix a possible skb leak\n  ipv6: Process unicast packet with Router Alert by checking flag in skb.\n  ath9k_htc: Handle IDLE state transition properly\n  ath9k: fix an RCU issue in calling ieee80211_get_tx_rates\n  netfilter: ipt_ULOG: fix incorrect setting of ulog timer\n  netfilter: ctnetlink: send event when conntrack label was modified\n  netfilter: nf_nat_sip: fix mangling\n  qlcnic: Do not sleep while holding spinlock\n  drivers: net: cpsw: fix compilation error with cpsw driver\n  tcp: doc : fix the syncookies default value\n  sh_eth: fix misreporting of transmit abort\n  ...\n",
  "tree_diff": []
}