The transmit window values must be configured for streaming mode, even
though streaming mode does not have a window.  This enables use of
extended headers when the transmit window socket option is set to 64
or larger.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

Let the compiler decide if inlining is appropriate.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

Since l2cap_send_ack can trigger extra actions like sending iframes,
don't call it.  Just send an RR or RNR frame if an ack needs sending.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

Different states have different actions for retransmit and monitor
timeouts, so remove the logic for those actions from the timer handlers.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

The ERTM specification requires the retransmit timer to be cancelled
when the monitor timer is set.  The retransmit timer cannot be set
again while the monitor timer is pending.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

struct l2cap_ctrl is now used, and the sframe is now sent directly
rather than depending on a separate call.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

REJ frames are sent by the remote device to request that all frames
after a given sequence number be retransmitted.  These are also an
implicit indication that the remote device is not in a busy state and
can receive new iframes.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

When a remote device sends an SREJ, retransmit the frame with the
corresponding sequence number (subject to special cases with poll and
final flags).  An SREJ is also an implicit indication the the remote
device is not in a busy state.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

As retransmitted packets arrive, attempt to reassemble SDUs.  If all
requested retransmissions have been received, acknowledge them and
transition back to the RECV state.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

The ERTM specification lays out three scenarios for sending SREJ
frames to request retransmission of specific frames.  l2cap_send_srej
requests all frames up to a given txseq that are not already queued
for reassembly.  l2cap_send_srej_tail only requests the most recent
missing frame.  l2cap_send_srej_list resends SREJ frames for data that
was requested for resend but never received.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This adds l2cap_ertm_resend to retransmit frames based on the sequence
numbers in chan->retrans_list.  If the retransmit limit is reached for
any individual frame is reached, the connection is dropped.  skbs that
are cloned already are copied to avoid modifying shared data (this is
uncommon).  To retransmit all frames, l2cap_retransmit_all now builds
a list of all unacked sequence numbers and then calls
l2cap_ertm_resend.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This adds a top-level state machine with handlers for two receive
states defined in the ERTM spec, RECV and SREJ_SENT.  The reqseq value
of the incoming frame is also validated at the top level and a
disconnection is forced if it is invalid.  The actions for the RECV
and SREJ_SENT states are implemented according to the state tables in
the ERTM specification.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This action now exactly matches what is defined in the ERTM
specification, including clearing the remote busy flag and setting the
retransmit timer rather than retransmitting frames directly.  The spec
does not retransmit frames as part of this action, since
retransmission is only triggered by REJ, SREJ, or an RR with the final
bit set.  struct l2cap_ctrl is also used to set up header values.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This lets the transmit state machine handle local busy state changes,
since different actions are taken in the different transmit states.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

The function now encapsulates more of the logic to either immediately
send an ack if the transmit window is over 75% full, or wait for the
ack timer to expire if the transmit window is not full enough.  It is
also able to push out waiting iframes that can carry an
acknowledgement.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This deletes the receive code that had handlers for each frame type at
the top level, and then had logic to determine the receive state
within each handler.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

Streaming mode reception is fairly simple, with in-sequence frames
being reassembled as they arrive.  Out-of-sequence frames are dropped,
and also clear any partially-assembled SDUs that may exist.

The packet classifier determines if the txseq value of the incoming
packet is expected, invalid (resulting in a disconnection), invalid
(ignorable), duplicate, or having to do with an SREJ request that was
previously sent.  The rules for each classification are defined in the
ERTM specification, and consolidating these rules in one place helps
clarify the receive state machine.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

Creates a new l2cap_data_rcv function that combines previous code from
l2cap_ertm_data_rcv and l2cap_data_channel.  This reduces duplicate
code for streaming mode, and sets up a framework for the ERTM receive
state machine.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

The new implementation uses struct l2cap_ctrl to set up the sframe
fields, and also reduces duplicate acks by canceling the ack timer
whenever an RR or RNR frame is sent.  sframe PDU generation is also
split in to a separate function to separate it from the logic related
to the connection state and sframe type.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

The new implementation is aware of the new transmit state machine, and
uses struct l2cap_ctrl to compose ERTM headers.  It also has improved
error handling for allocation failures, and does not send the packet
until after all skb and channel data structures are updated.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This new implementation uses struct l2cap_ctrl to compose the
streaming mode headers.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This implements a top-level transmit state machine with handlers for
the two ERTM states defined in the specification: XMIT and WAIT_F.

The state machine accepts an event and, optionally, a list of skbs to
transmit.  In addition to data transmission, the local busy state can
be modified, acks are processed, and monitor and retransmit timeouts
are handled.  This mirrors the structure of the state tables in the
spec.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This is to allow for ERTM state machine replacement in the patches
that follow.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>

This fixes a regression from commit
2ead70b8 that is present in all
kernels starting at v3.0.

When L2CAP information was moved to struct l2cap_chan, a check was
added to l2cap_chan_del to avoid certain cleanup operations when ERTM
or streaming mode had not yet been initialized.  The logic in the
check did not take in to account that chan->conf_state is set to 0 in
l2cap_chan_ready, so l2cap_chan_del failed to cancel timers and leaked
memory any time the ERTM queues or lists were not empty.

This change makes sure that l2cap_chan_del only returns early if
ERTM initialization was not performed.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

If the ERTM SREJ list is properly allocated but the retransmit list
allocation fails, the SREJ list must be freed before returning from
l2cap_ertm_init.  l2cap_chan_del will not clean up the SREJ list
if l2cap_ertm_init returns a failure code.
Signed-off-by: Mat Martineau <mathewm@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow the net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow the net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow the net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow the net subsystem rules.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Follow the coding style of the net subsystem.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>

Pull device-mapper updates from Alasdair G Kergon:
 "Improve multipath's retrying mechanism in some defined circumstances
  and provide a simple reserve/release mechanism for userspace tools to
  access thin provisioning metadata while the pool is in use."

* tag 'dm-3.5-changes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm:
  dm thin: provide userspace access to pool metadata
  dm thin: use slab mempools
  dm mpath: allow ioctls to trigger pg init
  dm mpath: delay retry of bypassed pg
  dm mpath: reduce size of struct multipath

This patch implements two new messages that can be sent to the thin
pool target allowing it to take a snapshot of the _metadata_.  This,
read-only snapshot can be accessed by userland, concurrently with the
live target.

Only one metadata snapshot can be held at a time.  The pool's status
line will give the block location for the current msnap.

Since version 0.1.5 of the userland thin provisioning tools, the
thin_dump program displays the msnap as follows:

    thin_dump -m <msnap root> <metadata dev>

Available here: https://github.com/jthornber/thin-provisioning-tools

Now that userland can access the metadata we can do various things
that have traditionally been kernel side tasks:

     i) Incremental backups.

     By using metadata snapshots we can work out what blocks have
     changed over time.  Combined with data snapshots we can ensure
     the data doesn't change while we back it up.

     A short proof of concept script can be found here:

     https://github.com/jthornber/thinp-test-suite/blob/master/incremental_backup_example.rb

     ii) Migration of thin devices from one pool to another.

     iii) Merging snapshots back into an external origin.

     iv) Asyncronous replication.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

Use dedicated caches prefixed with a "dm_" name rather than relying on
kmalloc mempools backed by generic slab caches so the memory usage of
thin provisioning (and any leaks) can be accounted for independently.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

After the failure of a group of paths, any alternative paths that
need initialising do not become available until further I/O is sent to
the device.  Until this has happened, ioctls return -EAGAIN.

With this patch, new paths are made available in response to an ioctl
too.  The processing of the ioctl gets delayed until this has happened.

Instead of returning an error, we submit a work item to kmultipathd
(that will potentially activate the new path) and retry in ten
milliseconds.

Note that the patch doesn't retry an ioctl if the ioctl itself fails due
to a path failure.  Such retries should be handled intelligently by the
code that generated the ioctl in the first place, noting that some SCSI
commands should not be retried because they are not idempotent (XOR write
commands).  For commands that could be retried, there is a danger that
if the device rejected the SCSI command, the path could be errorneously
marked as failed, and the request would be retried on another path which
might fail too.  It can be determined if the failure happens on the
device or on the SCSI controller, but there is no guarantee that all
SCSI drivers set these flags correctly.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

If I/O needs retrying and only bypassed priority groups are available,
set the pg_init_delay_retry flag to wait before retrying.

If, for example, the reason for the bypass is that the controller is
getting reset or there is a firmware upgrade happening, retrying right
away would cause a flood of log messages and retries for what could be a
few seconds or even several minutes.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

Move multipath structure's 'lock' and 'queue_size' members to eliminate
two 4-byte holes.  Also use a bit within a single unsigned int for each
existing flag (saves 8-bytes).  This allows future flags to be added
without each consuming an unsigned int.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Acked-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

Pull networking updates from David Miller:

 1) Make syn floods consume significantly less resources by

    a) Not pre-COW'ing routing metrics for SYN/ACKs
    b) Mirroring the device queue mapping of the SYN for the SYN/ACK
       reply.

    Both from Eric Dumazet.

 2) Fix calculation errors in Byte Queue Limiting, from Hiroaki SHIMODA.

 3) Validate the length requested when building a paged SKB for a
    socket, so we don't overrun the page vector accidently.  From Jason
    Wang.

 4) When netlabel is disabled, we abort all IP option processing when we
    see a CIPSO option.  This isn't the right thing to do, we should
    simply skip over it and continue processing the remaining options
    (if any).  Fix from Paul Moore.

 5) SRIOV fixes for the mellanox driver from Jack orgenstein and Marcel
    Apfelbaum.

 6) 8139cp enables the receiver before the ring address is properly
    programmed, which potentially lets the device crap over random
    memory.  Fix from Jason Wang.

 7) e1000/e1000e fixes for i217 RST handling, and an improper buffer
    address reference in jumbo RX frame processing from Bruce Allan and
    Sebastian Andrzej Siewior, respectively.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  fec_mpc52xx: fix timestamp filtering
  mcs7830: Implement link state detection
  e1000e: fix Rapid Start Technology support for i217
  e1000: look into the page instead of skb->data for e1000_tbi_adjust_stats()
  r8169: call netif_napi_del at errpaths and at driver unload
  tcp: reflect SYN queue_mapping into SYNACK packets
  tcp: do not create inetpeer on SYNACK message
  8139cp/8139too: terminate the eeprom access with the right opmode
  8139cp: set ring address before enabling receiver
  cipso: handle CIPSO options correctly when NetLabel is disabled
  net: sock: validate data_len before allocating skb in sock_alloc_send_pskb()
  bql: Avoid possible inconsistent calculation.
  bql: Avoid unneeded limit decrement.
  bql: Fix POSDIFF() to integer overflow aware.
  net/mlx4_core: Fix obscure mlx4_cmd_box parameter in QUERY_DEV_CAP
  net/mlx4_core: Check port out-of-range before using in mlx4_slave_cap
  net/mlx4_core: Fixes for VF / Guest startup flow
  net/mlx4_en: Fix improper use of "port" parameter in mlx4_en_event
  net/mlx4_core: Fix number of EQs used in ICM initialisation
  net/mlx4_core: Fix the slave_id out-of-range test in mlx4_eq_int