← Back to team overview

kernel-packages team mailing list archive

[Bug 1461078] Re: kdump service fails to start on Power8

 

** Tags removed: targetmilestone-inin--- verification-needed
** Tags added: targetmilestone-inin14043 verification-done

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to kexec-tools in Ubuntu.
https://bugs.launchpad.net/bugs/1461078

Title:
  kdump service fails to start on Power8

Status in kexec-tools package in Ubuntu:
  Fix Released
Status in kexec-tools source package in Trusty:
  Fix Committed

Bug description:
  [SRU justification]
  Needed to avoid load failure on Power8

  [Impact]
  Without this fix, kexec will not load on Power8

  [Fix]
  Apply upstream fix present in Dev release

  [Test Case]
  1) sudo apt-get install linux-crashdump
  2) reboot server
  3) kdump-config load

  As outlined in the initial description, kdump will fail to load
  without the fix.

  [Regression]
  None expected, fix is already present in the development version (Wily)

  [Original description of the problem]
  == Comment: #0 - SACHIN P. SANT <ssant@xxxxxxxxxx> - 2015-06-01 06:39:56 ==
  ---Problem Description---
  kdump service fails to start

  Contact Information = Sachin Sant / ssant@xxxxxxxxxx

  ---uname output---
  3.19.0-18-generic

  Machine Type = POWER8

  ---Debugger---
  A debugger is not configured

  ---Steps to Reproduce---
   1) Using latest daily ISO install 14.04.02 as a Power VM guest
  2) Upgrade the kernel to 3.19 level (3.19.0-18-generic)
  3) Configure kdump

  kdump-config load command fails with following error:

  root@alp9:~# cat /proc/cmdline
  BOOT_IMAGE=/boot/vmlinux-3.19.0-18-generic root=UUID=3ea23bcf-7269-432f-bacc-f82c6cdd774e ro splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M vt.handoff=7
  root@alp9:~# kdump-config show
  USE_KDUMP:        1
  KDUMP_SYSCTL:     kernel.panic_on_oops=1
  KDUMP_COREDIR:    /var/crash
  crashkernel addr:
  current state:    Not ready to kdump

  kexec command:
    no kexec command recorded
  root@alp9:~# kdump-config load
  unrecoverable error: short read from"/proc/device-tree//ibm,drc-names"
   * failed to load kdump kernel
  root@alp9:~#

  Userspace tool common name: kexec-tools

  The userspace tool has the following bit modes: 64-bit

  Userspace rpm: kdump-tools(1.5.5-2ubuntu1.2)

  Userspace tool obtained from project website:  na

  *Additional Instructions for Sachin Sant / ssant@xxxxxxxxxx:
  -Post a private note with access information to the machine that the bug is occuring on.
  -Attach ltrace and strace of userspace application.

  == Comment: #2 - MAMATHA INAMDAR <mainamdar@xxxxxxxxxx> - 2015-06-02 02:23:51 ==
  root@alp9:~# /etc/init.d/kdump-tools start
  Starting kdump-tools: unrecoverable error: short read from"/proc/device-tree//ibm,drc-names"
   * failed to load kdump kernel

  root@alp9:~# dpkg -l | grep kexec
  ii  kexec-tools                         1:2.0.6-0ubuntu2.1               ppc64el      tools to support fast kexec reboots

  root@alp9:~# dpkg -l | grep kdump
  ii  kdump-tools                         1.5.5-2ubuntu1.2                 all          scripts and tools for automating kdump (Linux crash dumps)

  root@alp9:~# cat /sys/kernel/kexec_crash_loaded
  0
  root@alp9:~# kdump-config status
  current state   : Not ready to kdump

  root@alp9:~# dmesg | grep -i crash
  [    0.000000] Reserving 2048MB of memory at 128MB for crashkernel (System RAM: 102400MB)
  [    0.000000] Kernel command line: BOOT_IMAGE=/boot/vmlinux-3.19.0-18-generic root=UUID=3ea23bcf-7269-432f-bacc-f82c6cdd774e ro splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M vt.handoff=7
  root@alp9:~#

  == Comment: #3 - Hari Krishna Bathini <hbathini@xxxxxxxxxx> - 2015-06-02 07:29:23 ==
  The following upstream commit fixes this issue:

  ---
  commit d1932cd592e2a6aaf50ed22cfa2d7562b583854a
  Author: Anton Blanchard <anton@xxxxxxxxx>
  Date:   Tue Dec 2 10:59:40 2014 +1100

      kexec/fs2dt: Use slurp_file_len to avoid partial read of files

      The OPAL firmware is going to embed its symbol map in the device tree.
      The size is large enough to be more than a page, and it takes
      multiple reads to get the whole file. This is because sysfs uses
      the seq_file helpers which do a page at a time.

      Unfortunately fs2dt has no handling for short reads and we die
  with:

      unrecoverable error: short read from"/proc/device-
  tree//ibm,opal/firmware/symbol-map"

      This patch uses the slurp_file_len helper which does the right thing.
      It moves the explicit open of the file further down for
      add_usable_mem_property and add_dyn_reconf_usable_mem_property.
      We should convert both of these to use the buffer provided by
      slurp_file_len at some stage.

      Signed-off-by: Anton Blanchard <anton@xxxxxxxxx>
      Signed-off-by: Simon Horman <horms@xxxxxxxxxxxx>
  ---

  This patch applies cleanly on the kexec-tools version
  1:2.0.6-0ubuntu2.1 on Ubuntu 14.04.2.

  Thanks
  Hari

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/kexec-tools/+bug/1461078/+subscriptions