thunder: faulty TSO padding

[dann frazier]

[Impact]
Certain packet sizes will trigger spurious TCP retransmissions on Cavium Thunder-X systems when TSO is enabled (the default). This can result in excess traffic and additional latency.

[Test Case]
Connect 2 systems to a 10G switch:
 - 1 ThunderX node (the client)
 - 1 "other" node - doesn't have to be ThunderX (the server)
On the server, run the "receiver" command from the attached tcptest tarball:
  $ ./receiver
Also on the server, run tcpdump to capture the traffic:
  $ sudo tcpdump -i <iface> -w tcpdump.out
On the client, run the "sender" command from the tcptest tarball, with a 1461 packet size:
  $ ./sender <ip of receiver> 1461 1461
Kill the tcpdump process, and open it in wireshark.

On failure, you'll see several "[TCP Spurious Retransmission]" packets - on success you should see none.

[Regression Risk]
Fix is upstream, so regressions will have upstream support. Fix is limited to a driver that is only applicable to the Cavium Thunder-X SoC, so risk is negligible to other platforms.

[Brad Figg (brad-figg)]

This bug is awaiting verification that the kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-xenial' to 'verification-done-xenial'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 4.4.0-42.62

---------------
linux (4.4.0-42.62) xenial; urgency=low

  * Fix GRO recursion overflow for tunneling protocols (LP: #1631287)
    - tunnels: Don't apply GRO to multiple layers of encapsulation.
    - gro: Allow tunnel stacking in the case of FOU/GUE

  * CVE-2016-7039
    - SAUCE: net: add recursion limit to GRO

linux (4.4.0-41.61) xenial; urgency=low

  [ Kamal Mostafa ]

  * Release Tracking Bug
    - LP: #1628204

  * nvme drive probe failure (LP: #1626894)
    - (fix) NVMe: Don't unmap controller registers on reset

linux (4.4.0-40.60) xenial; urgency=low

  [ Kamal Mostafa ]

  * Release Tracking Bug
    - LP: #1627074

  * Permission denied in CIFS with kernel 4.4.0-38 (LP: #1626112)
    - Fix memory leaks in cifs_do_mount()
    - Compare prepaths when comparing superblocks
    - SAUCE: Fix regression which breaks DFS mounting

  * Backlight does not change when adjust it higher than 50% after S3
    (LP: #1625932)
    - SAUCE: i915_bpo: drm/i915/backlight: setup and cache pwm alternate
      increment value
    - SAUCE: i915_bpo: drm/i915/backlight: setup backlight pwm alternate
      increment on backlight enable

linux (4.4.0-39.59) xenial; urgency=low

  [ Joseph Salisbury ]

  * Release Tracking Bug
    - LP: #1625303

  * thunder: chip errata w/ multiple CQEs for a TSO packet (LP: #1624569)
    - net: thunderx: Fix for issues with multiple CQEs posted for a TSO packet

  * thunder: faulty TSO padding (LP: #1623627)
    - net: thunderx: Fix for HW issue while padding TSO packet

  * CVE-2016-6828
    - tcp: fix use after free in tcp_xmit_retransmit_queue()

  * Sennheiser Officerunner - cannot get freq at ep 0x83 (LP: #1622763)
    - SAUCE: (no-up) ALSA: usb-audio: Add quirk for sennheiser officerunner

  * Backport E3 Skylake Support in ie31200_edac to Xenial (LP: #1619766)
    - EDAC, ie31200_edac: Add Skylake support

  * Ubuntu 16.04 - Full EEH Recovery Support for NVMe devices (LP: #1602724)
    - SAUCE: nvme: Don't suspend admin queue that wasn't created

  * ISST-LTE:pNV: system ben is hung during ST (nvme) (LP: #1620317)
    - blk-mq: Allow timeouts to run while queue is freezing
    - blk-mq: improve warning for running a queue on the wrong CPU
    - blk-mq: don't overwrite rq->mq_ctx

  * lsattr 32bit does not work on 64bit kernel (Inappropriate ioctl error)
    (LP: #1619918)
    - btrfs: bugfix: handle FS_IOC32_{GETFLAGS, SETFLAGS, GETVERSION} in
      btrfs_ioctl

  * radeon: monitor connected to onboard VGA doesn't work with Xenial
    (LP: #1600092)
    - drm/radeon/dp: add back special handling for NUTMEG

  * initramfs includes qle driver, but not firmware (LP: #1623187)
    - qed: add MODULE_FIRMWARE()

  * [Hyper-V] Rebase Hyper-V to 4.7.2 (stable) (LP: #1616677)
    - hv_netvsc: Implement support for VF drivers on Hyper-V
    - hv_netvsc: Fix the list processing for network change event
    - Drivers: hv: vmbus: Introduce functions for estimating room in the ring
      buffer
    - Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile
    - Drivers: hv: vmbus: Export the vmbus_set_event() API
    - lcoking/barriers, arch: Use smp barriers...