[lkcd-devel] Cannot save dump...

[lanas <la...@se...>]

Hi folks,

  I'm trying to configure a 2.4.21 real-time Timesys kernel for use
with lkcd and after much going around trying to get things going I run
into a problem where a dump is seemingly made but cannot be saved
after reboot.  If someone can help, it'd be appreciated.  

  First, the lkcd-2.4.21-20030825.patch cannot be applied to the
2.4.21 Timesys kernel:

[...]
patching file Makefile
Hunk #1 FAILED at 278.
Hunk #2 succeeded at 332 (offset 36 lines).
1 out of 2 hunks FAILED -- saving rejects to file Makefile.rej
patching file arch/alpha/config.in
Hunk #1 succeeded at 435 with fuzz 2 (offset 6 lines).
patching file arch/alpha/kernel/setup.c
Hunk #1 FAILED at 435.
[...]

  Maybe some parts are OK, but others certainly aren't.  So I applied
the patch to a vanilla 2.4.21 kernel without any problems and manually
diffed the concerned files using emacs' ediff.  Obviously some Timesys
kernel files were enormously different fromt eh vanilla kernel and I
left these alone.  I paid more attention around the dump files and
where the CONFIG_DUMP condition was used.

  The system I'm using is a devfs system with the following partitions
being used for testing lkcd:

/dev/discs/disc0/part3 25752       29567     3907584   82  Linux swap
/dev/discs/disc0/part4 29568       38154     8793088   83  Linux

  This system has swap disabled.  It uses no swap file, thus no swap
partition in normal use.  Here I made part3 type swap for testing
purposes.

  The /etc/sysconfig/dump file options are:

DUMP_ACTIVE="1"
DUMPDEV=/dev/vmdump
DUMPDIR="/var/log/dump"
DUMP_LEVEL="2"
DUMP_COMPRESS="2"
DUMP_FLAGS="0x80000000"
DUMP_SAVE="1"
PANIC_TIMEOUT="5"
BOUNDS_LIMIT=10
KEXEC_IMAGE=/boot/vmlinuz
KEXEC_CMDLINE="root console=tty0"
TARGET_HOST=""
TARGET_PORT=6688
SOURCE_PORT=6688
ETH_ADDRESS=ff:ff:ff:ff:ff:ff

  So I make a link in dev like this:

  cd /dev
  ln -s /dev/discs/disc0/part3 vmdump

  The kernel is configured with the lkcd options, as well as SysRq.

  lkcd_config is called from the lkcd script like this:

/sbin/lkcd_config -d /dev/vmdump -l 2 -f 0x80000000 -c 2 -t -p 6688 
                  -P 6688 -e ff:ff:ff:ff:ff:ff

  I press the Alt-SysRq-c key and the kernel crashes while these two
lines are last ones to be seen at the console:

dump: Dumping to device 0x303 [ide0(3,3)] on CPU 0...
dump: Compression value is 0x02.  Writing dump header.

  Should there be any other msgs about 'dump completed' or somesuch ?

  I'm not familiar with this drive IDE notation.  Does [ide0(3,3)]
looks OK for part3 ?

  When the system reboots I issue the lkcd save command and I get this:

lcrash 0.10.2 (xlcrash) build at Nov 7 2006 13:37:57 
Error: could not save crash dump: lcrash returned with 1. Maybe there
was no dump?
Reconfiguring LKCD....

  Is there a way to probe and test further to see that a dump occured
or not at the SysRq sequence ?  Is it possible to check the part3
partition ?  Same error happens when using part4.

Any help would be greatly appreciated.  Please do not hesistate to ask
me for further details.

Cheers,
Al