9:20 a.m.
When trying to run crash 7.2 like this, the 'mod -s $mod' command loads
the modules from the running system instead of the specified directory:
d=${0%/*}
tee $t <<_EOF_
mod -s xen-kbdfront
mod -s xen_kbdfront
exit
_EOF_
strace -f -s 123 -tt -o /dev/shm/crash.trace.txt \
crash \
-i $t \
--mod "${d}" \
"${d}"/boot/vmlinux-*-default.gz \
"$1" \
"${d}"/usr/lib/debug/boot/vmlinux-*-default.debug
In the strace log I see a search for xen_kbdfront.ko, later for
xen-kbdfront.ko. Unfortunately /lib/modules/`uname -r` is searched before $d.
Now that I read the man page once more I noticed that --mod is just for
debug info?
It seems $d/boot is considered as well, likely because the kernel is
stored in that directory.
I wonder why there is no option to tell crash to only operate below a
certain directory when looking for module related things.
"--mod" seems to have an effect, if the directory exists it is searched.
How can I tell crash to only operate below $d?
Olaf
9:42 a.m.
----- Original Message -----
When trying to run crash 7.2 like this, the 'mod -s $mod'
command loads
the modules from the running system instead of the specified directory:
d=${0%/*}
tee $t <<_EOF_
mod -s xen-kbdfront
mod -s xen_kbdfront
exit
_EOF_
strace -f -s 123 -tt -o /dev/shm/crash.trace.txt \
crash \
-i $t \
--mod "${d}" \
"${d}"/boot/vmlinux-*-default.gz \
"$1" \
"${d}"/usr/lib/debug/boot/vmlinux-*-default.debug
In the strace log I see a search for xen_kbdfront.ko, later for
xen-kbdfront.ko. Unfortunately /lib/modules/`uname -r` is searched before $d.
Now that I read the man page once more I noticed that --mod is just for
debug info?
Correct.
It seems $d/boot is considered as well, likely because the kernel is
stored in that directory.
I wonder why there is no option to tell crash to only operate below a
certain directory when looking for module related things.
"--mod" seems to have an effect, if the directory exists it is searched.
How can I tell crash to only operate below $d?
For a single module whose debuginfo file is not located in one of the
default/standard locations, append the fully-qualified pathname to the
debuginfo file to "mod -s module":
crash> help mod
NAME
mod - module information and loading of symbols and debugging data
SYNOPSIS
mod -s module [objfile] | -d module | -S [directory] [-D|-t|-r|-R|-o|-g]
... [ cut ] ...
-s module [objfile] Loads symbolic and debugging data from the object file
for the module specified. If no objfile argument is
appended, a search will be made for an object file
consisting of the module name with a .o or .ko suffix,
starting at the /lib/modules/<release> directory on
the host system, or if not found there, starting at the
directory containing the kernel namelist file. If an
objfile argument is appended, then that file will be
used.
If the module.ko and module.ko.debug files are separate, you can use the debuginfo
file as the objfile argument.
Dave
5:43 a.m.
On Fri, Mar 16, Dave Anderson wrote:
For a single module whose debuginfo file is not located in one of the
default/standard locations, append the fully-qualified pathname to the
debuginfo file to "mod -s module":
That would be combersome I think. I rarely used crash in the last years,
so I probably do it all wrong.
If I have to use crash I get a core file and the used kernel version.
Then I dig the used kernel, kernel-debuginfo and perhaps the KMP rpm
packages from the archives. Everything gets copied to a workdir. The
binaries are extracted 'rpm2cpio|cpio -i' within that workdir. Thats why
I thought '--mod $dir' would let crash only operate below $dir.
So I have to rely on the fact that the uname of the running kernel never
matches the crashed kernel. And if it does happen to match I have to
hope all involved binaries match.
Olaf
2444
days inactive
2447
days old
devel@lists.crash-utility.osci.io
2 comments
2 participants
participants (2)
-
Dave Anderson -
Olaf Hering