Comment # 59 on bug 1188896 from LTC BugProxy

------- Comment From geraldsc@de.ibm.com 2022-02-15 09:05 EDT-------
(In reply to comment #54)
[...]
> - You can download latest Tumbleweed iso images (not released because of
> this bug) from here:
> https://build.opensuse.org/package/binaries/openSUSE:Factory:zSystems/
> 000product:openSUSE-dvd5-dvd-s390x/images
>
> The kernel is here:
> https://build.opensuse.org/package/binaries/openSUSE:Factory:zSystems/kernel-
> default/standard

Thanks, we are getting closer. It is not the matching kernel version (5.16.8-1
instead of 5.16.4-1), and therefore not usable to open the dump in crash. But
it showed me that I was wrong, and the debuginfo vmlinux is apparently not
enough, and it also takes the normal vmlinux in addition (don't ask me why...).

So, the following RPMs would be needed, or rather the included files which you
would find on your system when installed (Of course, in the very same version
as the kernel that was dumped):

kernel-default-debuginfo-5.16.8-1.1.s390x.rpm:
./usr/lib/debug/usr/lib/modules/5.16.8-1-default/vmlinux.debug

kernel-default-5.16.8-1.1.s390x.rpm:
./usr/lib/modules/5.16.8-1-default/vmlinux.xz

> well, the dump-guest-memory method was exactly what I used to create the
> dump above but comment 52 says it is not usable. (method used is basically:
> have qemu with debuginfo installed, attach gdb to hanging vm, load the
> dump-guest-memory.py from the qemu git to create the dump). but for the
> guest VM the runing kernel is taken from the project where the build
> happens, so this could be any kernel and the same debuginfo is not
> neccessarily available or even built at all.

The dump is not necessarily unusable for analysis in crash, at least if we
would also have matching vmlinux files (both with and w/o debuginfo, see
above).

What did not work with this dump, was "crash --log", which would not need any
vmlinux files, and only extract kernel message log. That might just "work as
designed", e.g. due to this special qemu dump, or maybe a bug in crash, who
knows. This is not really a problem, as long as we would be able to open the
dump in crash for analysis, which would require the vmlinux files.

As long as we do not have any vmlinux files, nobody can say if the dump us
usable or not.

Of course, this specific dump, w/o panic_on_taint, will most likely not give
useful insight, even if we had vmlinux files, other than maybe the kernel
message log. But at least it serves as exercise on how to submit dumps and
additional required files.