path: root/include
diff options
authorLinus Torvalds <torvalds@linux-foundation.org>2012-03-05 14:30:54 -0800
committerLinus Torvalds <torvalds@linux-foundation.org>2012-03-05 14:30:54 -0800
commitaa139092de3bb7d8b81f9b8433e12e87ac1eec3f (patch)
treef59a16c5fe7f1cece34b7d2e0b58a3c7438f45c1 /include
parent4f0449e26f7ce72933af99e80b381a4dce792d7b (diff)
parent9d1dfc06b14df0818d86d6ba6aaec0f641e94990 (diff)
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Pull networking fixes from David Miller: 1) TCP SACK processing can calculate an incorrect reordering value in some cases, fix from Neal Cardwell. 2) tcp_mark_head_lost() can split SKBs in situations where it should not, violating send queue invariants expected by other pieces of code and thus resulting (eventually) in corrupted retransmit state counters. Also from Neal Cardwell. 3) qla3xxx erroneously calls spin_lock_irqrestore() with constant hw_flags of zero. Fix from Santosh Nayak. 4) Fix NULL deref in rt2x00, from Gabor Juhos. 5) pch_gbe passes address of wrong typed object to pch_gbe_validate_option thus corrupting part of the value. From Dan Carpenter. 6) We must check the return value of nlmsg_parse() before trying to use the results. From Eric Dumazet. 7) Bridging code fails to check return value of ipv6_dev_get_saddr() thus potentially leaving uninitialized garbage in the outgoing ipv6 header. From Ulrich Weber. 8) Due to rounding and a reversed operation on jiffies, bridge message ages can go backwards instead of forwards, thus breaking STP. Fixes from Joakim Tjernlund. 9) r8169 modifies Config* registers without properly holding the Config9346 lock, resulting in corrupted IP fragments on some chips. Fix from Francois Romieu. 10) NET_PACKET_ENGINE default wan't set properly during the network driver mega-move. Fix from Stephen Hemminger. 11) vmxnet3 uses TCP header size where it actually should use the UDP header size, fix from Shreyas Bhatewara. 12) Netfilter bridge module autoload is busted in the compat case, fix from Florian Westphal. 13) Wireless Key removal was not setting multicast bits correctly thus accidently killing the unicast key 0 and thus all traffic stops. Fix from Johannes Berg. 14) Fix endless retries of A-MPDU transmissions in brcm80211 driver. * git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (22 commits) qla3xxx: ethernet: Fix bogus interrupt state flag. bridge: check return value of ipv6_dev_get_saddr() rtnetlink: fix rtnl_calcit() and rtnl_dump_ifinfo() bridge: message age needs to increase, not decrease. bridge: Adjust min age inc for HZ > 256 tcp: don't fragment SACKed skbs in tcp_mark_head_lost() r8169: corrupted IP fragments fix for large mtu. packetengines: fix config default vmxnet3: Fix transport header size enic: fix an endian bug in enic_probe() pch_gbe: memory corruption calling pch_gbe_validate_option() tg3: Fix tg3_get_stats64 for 5700 / 5701 devs tcp: fix false reordering signal in tcp_shifted_skb tcp: fix comment for tp->highest_sack netfilter: bridge: fix module autoload in compat case brcm80211: smac: only print block-ack timeout message at trace level brcm80211: smac: fix endless retry of A-MPDU transmissions mac80211: Fix a warning on changing to monitor mode from STA mac80211: zero initialize count field in ieee80211_tx_rate iwlwifi: fix key removal ...
Diffstat (limited to 'include')
2 files changed, 5 insertions, 3 deletions
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 46a85c9e1f25..3c7ffdb40dc6 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -412,7 +412,8 @@ struct tcp_sock {
struct tcp_sack_block recv_sack_cache[4];
- struct sk_buff *highest_sack; /* highest skb with SACK received
+ struct sk_buff *highest_sack; /* skb just after the highest
+ * skb with SACKed bit set
* (validity guaranteed only if
* sacked_out > 0)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 42c29bfbcee3..2d80c291fffb 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1364,8 +1364,9 @@ static inline void tcp_push_pending_frames(struct sock *sk)
-/* Start sequence of the highest skb with SACKed bit, valid only if
- * sacked > 0 or when the caller has ensured validity by itself.
+/* Start sequence of the skb just after the highest skb with SACKed
+ * bit, valid only if sacked_out > 0 or when the caller has ensured
+ * validity by itself.
static inline u32 tcp_highest_sack_seq(struct tcp_sock *tp)