The ethtool api {get|set}_settings is deprecated.
We move this driver to new api {get|set}_link_ksettings.

As I don't have the hardware, I'd be very pleased if
someone may test this patch.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The ethtool api {get|set}_settings is deprecated.
We move this driver to new api {get|set}_link_ksettings.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Marcin Wojtas says:

====================
mvneta xmit_more and bql support

This is a delayed v2 of short patchset, which introduces xmit_more and BQL
to mvneta driver. The only one change was added in xmit_more support -
condition check preventing excessive descriptors concatenation before
flushing in HW.

Any comments or feedback would be welcome.

Changelog:
v1 -> v2:

* Add checking condition that ensures too much descriptors are not
  concatenated before flushing in HW.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tests showed that when whole bandwidth is consumed, the latency for
various kind of traffic can reach high values. With saturated
link (e.g. with iperf from target to host) simple ping could take
significant amount of time. BQL proved to improve this situation
when implemented in mvneta driver. Measurements of ping latency
for 3 link speeds:
Speed | Latency w/o BQL | Latency with BQL
10    |      7-14 ms    |     3.5 ms
100   |      2-12 ms    |     0.6 ms
1000  |   often timeout |   up to 2ms

Decreasing latency as above result in sligt performance cost - 4kpps
(-1.4%) when pushing 64B packets via two bridged interfaces of Armada 38x.
For 1500B packets in the same setup, the mpstat tool showed +8% of
CPU occupation (default affinity, second CPU idle). Even though this
cost seems reasonable to take, considering other improvements.

This commit adds byte queue limit mechanism for the mvneta driver.
Signed-off-by: NMarcin Wojtas <mw@semihalf.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Basing on xmit_more flag of the skb, TX descriptors can be concatenated
before flushing. This commit delay Tx descriptor flush if the queue is
running and if there is more skb's to send.

A maximum allowed number of descriptors for flushing at once due to
MVNETA_TXQ_UPDATE_REG(q) reqisters limitation, is 255. Because of that
a new macro was added (MVNETA_TXQ_DEC_SENT_MASK) in order to ensure that
concatenated amount of descriptor does not exceed that value.
Signed-off-by: NSimon Guinot <simon.guinot@sequanux.org>
Signed-off-by: NMarcin Wojtas <mw@semihalf.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Flower currently allows having the same filter twice with the same
priority. Actions (and statistics update) will always execute on the
first inserted rule leaving the second rule unused.
This patch disallows that.
Signed-off-by: NPaul Blakey <paulb@mellanox.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add missing IPv6-SR header files in include/uapi/linux/Kbuild.

Also, prevent seg6_lwt_headroom() from being exported and add
missing linux/types.h include.
Signed-off-by: NDavid Lebrun <david.lebrun@uclouvain.be>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make sure that ctx cannot potentially be accessed oob by asserting
explicitly that ctx access size into pt_regs for BPF_PROG_TYPE_KPROBE
programs must be within limits. In case some 32bit archs have pt_regs
not being a multiple of 8, then BPF_DW access could cause such access.

BPF_PROG_TYPE_KPROBE progs don't have a ctx conversion function since
there's no extra mapping needed. kprobe_prog_is_valid_access() didn't
enforce sizeof(long) as the only allowed access size, since LLVM can
generate non BPF_W/BPF_DW access to regs from time to time.

For BPF_PROG_TYPE_TRACEPOINT we don't have a ctx conversion either, so
add a BUILD_BUG_ON() check to make sure that BPF_DW access will not be
a similar issue in future (ctx works on event buffer as opposed to
pt_regs there).

Fixes: 2541517c ("tracing, perf: Implement BPF programs attached to kprobes")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the last patch da36e13c ("ipvlan: improvise dev_id generation
logic in IPvlan") I missed some part of Dave's suggestion and because
of that the dev_id creation could fail in a corner case scenario. This
would happen when more or less 64k devices have been already created and
several have been deleted. If the devices that are still sticking around
are the last n bits from the bitmap. So in this scenario even if lower
bits are available, the dev_id search is so narrow that it always fails.

Fixes: da36e13c ("ipvlan: improvise dev_id generation logic in IPvlan")
CC: David Miller <davem@davemloft.org>
CC: Eric Dumazet <edumazet@google.com>
Signed-off-by: NMahesh Bandewar <maheshb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The 8000 series SFC NICs have 4K PIO buffers, rather than the 2K of
 the 7000 series.  Rather than having a hard-coded PIO buffer size
 (ER_DZ_TX_PIOBUF_SIZE), read it from the GET_CAPABILITIES_V2 MCDI
 response.
Signed-off-by: NEdward Cree <ecree@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If an option descriptor has been sent on a queue but not followed by a
 packet, there will have been no completion event, so the read and write
 counts won't match and we'll think we can't do PIO.  This combines with
 the fact that we have two TX queues (for en/disable checksum offload),
 and that both must be empty for PIO to happen.
This patch adds a separate "packet_write_count" that tracks the most
 recent write_count we expect to see a completion event for; this excludes
 option descriptors but _includes_ PIO descriptors (even though they look
 like option descriptors).  This is then used, rather than write_count,
 in efx_nic_tx_is_empty().
We only bother to maintain packet_write_count on EF10, since on Siena
 (a) there are no option descriptors and it always equals write_count, and
 (b) there's no PIO, so we don't need it anyway.
Signed-off-by: NEdward Cree <ecree@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sctp_frag_point() doesn't store anything, and thus just calling it
cannot do anything useful.

sctp_apply_peer_addr_params is only called by
sctp_setsockopt_peer_addr_params. When operating on an asoc,
sctp_setsockopt_peer_addr_params will call sctp_apply_peer_addr_params
once for the asoc, and then once for each transport this asoc has,
meaning that the frag_point will be recomputed when updating the
transports and calling it when updating the asoc is not necessary.
IOW, no action is needed here and we can remove this call.
Signed-off-by: NMarcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Reviewed-by: NXin Long <lucien.xin@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Assigned but not used.
Signed-off-by: NMarcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Reviewed-by: NXin Long <lucien.xin@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

arp is being checked instead of arp_eth to see if the call to
__skb_header_pointer failed. Fix this by checking arp_eth is
null instead of arp.   Also fix to use length hlen rather than
hlen - sizeof(_arp); thanks to Eric Dumazet for spotting
this latter issue.

CoverityScan CID#1396428 ("Logically dead code") on 2nd
arp comparison (which should be arp_eth instead).

Fixes: commit 55733350 ("flow disector: ARP support")
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In commit d35c99ff ("netlink: do not enter direct reclaim from
netlink_dump()") we made sure to not trigger expensive memory reclaim.

Problem is that a bit later, netlink_trim() might be called and
trigger memory reclaim.

netlink_trim() should be best effort, and really as fast as possible.
Under memory pressure, it is fine to not trim this skb.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Perform an emergency shutdown of the adapter and stop it from
continuing any further communication on the ports or DMA to the
host. This is typically used when the adapter and/or firmware
have crashed and we want to prevent any further accidental
communication with the rest of the world. This will also force
the port Link Status to go down -- if register writes work --
which should help our peers figure out that we're down.
Signed-off-by: NHariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The bulk readpages support introduced a harmless warning:

fs/afs/file.c: In function 'afs_readpages_page_done':
fs/afs/file.c:270:20: error: unused variable 'vnode' [-Werror=unused-variable]

This adds an #ifdef to match the user of that variable.  The user of the
variable has to be conditional because it accesses a member of a struct
that is also conditional.

Fixes: 91b467e0 ("afs: Make afs_readpages() fetch data in bulk")
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use eth_zero_addr to assign zero address to the given address array
instead of memset when the second argument in memset is address
of zero which makes the code clearer and also add header
file linux/etherdevice.h
Signed-off-by: NShyam Saini <mayhs11saini@gmail.com>
Acked-by: NEdward Cree <ecree@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The break statement should be indented one more tab.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This driver is no longer necessary since it was merged into stmmac.
Acked-by: NLars Persson <larper@axis.com>
Signed-off-by: NJoao Pinto <jpinto@synopsys.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Merge tag 'mac80211-next-for-davem-2017-01-13' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211-next

Johannes Berg says:

====================
For 4.11, we seem to have more than in the past few releases:
 * socket owner support for connections, so when the wifi
   manager (e.g. wpa_supplicant) is killed, connections are
   torn down - wpa_supplicant is critical to managing certain
   operations, and can opt in to this where applicable
 * minstrel & minstrel_ht updates to be more efficient (time and space)
 * set wifi_acked/wifi_acked_valid for skb->destructor use in the
   kernel, which was already available to userspace
 * don't indicate new mesh peers that might be used if there's no
   room to add them
 * multicast-to-unicast support in mac80211, for better medium usage
   (since unicast frames can use *much* higher rates, by ~3 orders of
   magnitude)
 * add API to read channel (frequency) limitations from DT
 * add infrastructure to allow randomizing public action frames for
   MAC address privacy (still requires driver support)
 * many cleanups and small improvements/fixes across the board
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The region set by the call to memset, immediately overwritten by
the subsequent call to memcpy and thus makes the  memset redundant.

Also remove the memset((&info, 0, sizeof(info)) on line 398 because
info is memcpy()'ed to before being used in the loop and it isn't
used outside of the loop.
Signed-off-by: NShyam Saini <mayhs11saini@gmail.com>
Reviewed-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Do not count pause frames as part of general TX/RX frame
counters.

Based on the original work of Casey Leedom <leedom@chelsio.com>
Signed-off-by: NGanesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Michael Chan says:

====================
bnxt_en: Misc. updates for net-next.

Miscellaneous updates including firmware spec update, ethtool -p blinking
LED support, RDMA SRIOV config callback, and minor fixes.

v2: Dropped the DCBX RoCE app TLV patch until the ETH_P_IBOE RDMA patch
is merged.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add the ulp_sriov_cfg callbacks when the number of VFs is changing.  This
allows the RDMA driver to provision RDMA resources for the VFs.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add LED blinking code to support ethtool -p on the PF.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit bdbd1eb5 ("bnxt_en: Handle no aggregation ring gracefully.")
introduced the BNXT_FLAG_NO_AGG_RINGS flag.  For consistency,
bnxt_set_tpa_flags() should also clear TPA flags when there are no
aggregation rings.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

CC [M]  drivers/net/ethernet/broadcom/bnxt/bnxt.o
drivers/net/ethernet/broadcom/bnxt/bnxt.c:4947:21: warning: ‘bnxt_get_max_func_rss_ctxs’ defined but not used [-Wunused-function]
 static unsigned int bnxt_get_max_func_rss_ctxs(struct bnxt *bp)
                     ^
  CC [M]  drivers/net/ethernet/broadcom/bnxt/bnxt.o
drivers/net/ethernet/broadcom/bnxt/bnxt.c:4956:21: warning: ‘bnxt_get_max_func_vnics’ defined but not used [-Wunused-function]
 static unsigned int bnxt_get_max_func_vnics(struct bnxt *bp)
                     ^
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Yuchung Cheng says:

====================
tcp: RACK fast recovery

The patch set enables RACK loss detection (draft-ietf-tcpm-rack-01)
to trigger fast recovery with a reordering timer.

Previously RACK has been running in auxiliary mode where it is
used to detect packet losses once the recovery has triggered by
other algorithms (e.g., FACK). By inspecting packet timestamps,
RACK can start ACK-driven repairs timely. A few similar heuristics
are no longer needed and are either removed or disabled to reduce
the complexity of the Linux TCP loss recovery engine:

  1. FACK (Forward Acknowledgement)
  2. Early Retransmit (RFC5827)
  3. thin_dupack (fast recovery on single DUPACK for thin-streams)
  4. NCR (Non-Congestion Robustness RFC4653) (RFC4653)
  5. Forward Retransmit

After this change, Linux's loss recovery algorithms consist of
  1. Conventional DUPACK threshold approach (RFC6675)
  2. RACK and Tail Loss Probe (draft-ietf-tcpm-rack-01)
  3. RTO plus F-RTO extension (RFC5682)

The patch set has been tested on Google servers extensively and
presented in several IETF meetings. The data suggests that RACK
successfully improves recovery performance:
https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-draft-ietf-tcpm-rack-01.pdf
https://www.ietf.org/proceedings/96/slides/slides-96-tcpm-3.pdf
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch disables FACK by default as RACK is the successor of FACK
(inspired by the insights behind FACK).

FACK[1] in Linux works as follows: a packet P is deemed lost,
if packet Q of higher sequence is s/acked and P and Q are distant
by at least dupthresh number of packets in sequence space.

FACK is more aggressive than the IETF recommened recovery for SACK
(RFC3517 A Conservative Selective Acknowledgment (SACK)-based Loss
 Recovery Algorithm for TCP), because a single SACK may trigger
fast recovery. This obviously won't work well with reordering so
FACK is dynamically disabled upon detecting reordering.

RACK supersedes FACK by using time distance instead of sequence
distance. On reordering, RACK waits for a quarter of RTT receiving
a single SACK before starting recovery. (the timer can be made more
adaptive in the future by measuring reordering distance in time,
but currently RTT/4 seem to work well.) Once the recovery starts,
RACK behaves almost like FACK because it reduces the reodering
window to 1ms, so it fast retransmits quickly. In addition RACK
can detect loss retransmission as it does not care about the packet
sequences (being repeated or not), which is extremely useful when
the connection is going through a traffic policer.

Google server experiments indicate that disabling FACK after enabling
RACK has negligible impact on the overall loss recovery performance
with more reordering events detected.  But we still keep the FACK
implementation for backup if RACK has bugs that needs to be disabled.

[1] M. Mathis, J. Mahdavi, "Forward Acknowledgment: Refining
TCP Congestion Control," In Proceedings of SIGCOMM '96, August 1996.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Thin stream DUPACK is to start fast recovery on only one DUPACK
provided the connection is a thin stream (i.e., low inflight).  But
this older feature is now subsumed with RACK. If a connection
receives only a single DUPACK, RACK would arm a reordering timer
and soon starts fast recovery instead of timeout if no further
ACKs are received.

The socket option (THIN_DUPACK) is kept as a nop for compatibility.
Note that this patch does not change another thin-stream feature
which enables linear RTO. Although it might be good to generalize
that in the future (i.e., linear RTO for the first say 3 retries).
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch removes the (partial) implementation of the aggressive
limited transmit in RFC4653 TCP Non-Congestion Robustness (NCR).

NCR is a mitigation to the problem created by the dynamic
DUPACK threshold.  With the current adaptive DUPACK threshold
(tp->reordering) could cause timeouts by preventing fast recovery.
For example, if the last packet of a cwnd burst was reordered, the
threshold will be set to the size of cwnd. But if next application
burst is smaller than threshold and has drops instead of reorderings,
the sender would not trigger fast recovery but instead resorts to a
timeout recovery.

NCR mitigates this issue by checking the number of DUPACKs against
the current flight size additionally. The techniqueue is similar to
the early retransmit RFC.

With RACK loss detection, this mitigation is not needed, because RACK
does not use DUPACK threshold to detect losses. RACK arms a reordering
timer to fire at most a quarter RTT later to start fast recovery.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch removes the support of RFC5827 early retransmit (i.e.,
fast recovery on small inflight with <3 dupacks) because it is
subsumed by the new RACK loss detection. More specifically when
RACK receives DUPACKs, it'll arm a reordering timer to start fast
recovery after a quarter of (min)RTT, hence it covers the early
retransmit except RACK does not limit itself to specific inflight
or dupack numbers.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Forward retransmit is an esoteric feature in RFC3517 (condition(3)
in the NextSeg()). Basically if a packet is not considered lost by
the current criteria (# of dupacks etc), but the congestion window
has room for more packets, then retransmit this packet.

However it actually conflicts with the rest of recovery design. For
example, when reordering is detected we want to be conservative
in retransmitting packets but forward-retransmit feature would
break that to force more retransmission. Also the implementation is
fairly complicated inside the retransmission logic inducing extra
iterations in the write queue. With RACK losses are being detected
timely and this heuristic is no longer necessary. There this patch
removes the feature.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Current F-RTO reverts cwnd reset whenever a never-retransmitted
packet was (s)acked. The timeout can be declared spurious because
the packets acknoledged with this ACK was transmitted before the
timeout, so clearly not all the packets are lost to reset the cwnd.

This nice detection does not really depend F-RTO internals. This
patch applies the detection universally. On Google servers this
change detected 20% more spurious timeouts.
Suggested-by: NNeal Cardwell <ncardwell@google.com>
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch changes two things:

1. Start fast recovery with RACK in addition to other heuristics
   (e.g., DUPACK threshold, FACK). Prior to this change RACK
   is enabled to detect losses only after the recovery has
   started by other algorithms.

2. Disable TCP early retransmit. RACK subsumes the early retransmit
   with the new reordering timer feature. A latter patch in this
   series removes the early retransmit code.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently RACK would mark loss before the undo operations in TCP
loss recovery. This could incorrectly identify real losses as
spurious. For example a sender first experiences a delay spike and
then eventually some packets were lost due to buffer overrun.
In this case, the sender should perform fast recovery b/c not all
the packets were lost.

But the sender may first trigger a (spurious) RTO and reset
cwnd to 1. The following ACKs may used to mark real losses by
tcp_rack_mark_lost. Then in tcp_process_loss this ACK could trigger
F-RTO undo condition and unmark real losses and revert the cwnd
reduction. If there are no more ACKs coming back, eventually the
sender would timeout again instead of performing fast recovery.

The patch fixes this incorrect process by always performing
the undo checks before detecting losses.

Fixes: 4f41b1c5 ("tcp: use RACK to detect losses")
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The packets inside a jumbo skb (e.g., TSO) share the same skb
timestamp, even though they are sent sequentially on the wire. Since
RACK is based on time, it can not detect some packets inside the
same skb are lost.  However, we can leverage the packet sequence
numbers as extended timestamps to detect losses. Therefore, when
RACK timestamp is identical to skb's timestamp (i.e., one of the
packets of the skb is acked or sacked), we use the sequence numbers
of the acked and unacked packets to break ties.

We can use the same sequence logic to advance RACK xmit time as
well to detect more losses and avoid timeout.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch makes RACK install a reordering timer when it suspects
some packets might be lost, but wants to delay the decision
a little bit to accomodate reordering.

It does not create a new timer but instead repurposes the existing
RTO timer, because both are meant to retransmit packets.
Specifically it arms a timer ICSK_TIME_REO_TIMEOUT when
the RACK timing check fails. The wait time is set to

  RACK.RTT + RACK.reo_wnd - (NOW - Packet.xmit_time) + fudge

This translates to expecting a packet (Packet) should take
(RACK.RTT + RACK.reo_wnd + fudge) to deliver after it was sent.

When there are multiple packets that need a timer, we use one timer
with the maximum timeout. Therefore the timer conservatively uses
the maximum window to expire N packets by one timeout, instead of
N timeouts to expire N packets sent at different times.

The fudge factor is 2 jiffies to ensure when the timer fires, all
the suspected packets would exceed the deadline and be marked lost
by tcp_rack_detect_loss(). It has to be at least 1 jiffy because the
clock may tick between calling icsk_reset_xmit_timer(timeout) and
actually hang the timer. The next jiffy is to lower-bound the timeout
to 2 jiffies when reo_wnd is < 1ms.

When the reordering timer fires (tcp_rack_reo_timeout): If we aren't
in Recovery we'll enter fast recovery and force fast retransmit.
This is very similar to the early retransmit (RFC5827) except RACK
is not constrained to only enter recovery for small outstanding
flights.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>