----- "Adrien Kunysz" <adk(a)redhat.com> wrote:
Dear crash-utility list,
I played a bit with snap on RHEL 4 and besides the fact it doesn't build out of
the box (trivial Makefile compatibility problem), it seems it doesn't really
dump a full core properly (at least on the system I tested).
Since this is not really something that is supposed to be supported I am posting
this here rather than opening a Bugzilla. I haven't attempted to repro on other
distros.
# crash
crash> extend snap-rhel4-x86_64/snap.so
./snap-rhel4-x86_64/snap.so: shared object loaded
crash> snap mycore
mycore: [100%]
-rw-r--r-- 1 root root 4160684848 Sep 3 09:37 mycore
crash> quit
# free
total used free shared buffers
cached
Mem: 16319168 8065516 8253652 0 133060
6231128
-/+ buffers/cache: 1701328 14617840
Swap: 2097144 0 2097144
# uname -a
Linux *** 2.6.9-89.ELsmp #1 SMP Mon Apr 20 10:33:05 EDT 2009 x86_64
x86_64
x86_64 GNU/Linux
# crash /usr/lib/debug/lib/modules/2.6.9-89.ELsmp/vmlinux mycore
[...]
crash: read error: kernel virtual address: 1020009d024 type:
"tss_struct ist array"
Is that expected? Any idea what's going on?
It created at 4GB vmcore on a 16GB system, which is apparently the
problem. Try doing this (done on my 1GB RHEL4 box):
crash> !tty
/dev/pts/3
crash> set console /dev/pts/3
debug console [11505]: /dev/pts/3
console: /dev/pts/3
crash> extend snap.so
./extensions/snap.so: shared object loaded
crash> snap mycore
00000000-0009ffff : System RAM
00100000-3fe8abff : System RAM
ram_segments[0]: 0000000000000000 00000000000a0000 [00000000-0009ffff]
ram_segments[1]: 0000000000100000 000000003fe8a000 [00100000-3fe8abff]
mycore: [100%]
-rw-r--r-- 1 root root 1072210680 Sep 3 09:24 mycore
crash>
I'm guessing that the memory containing virtual address 0x1020009d024
(physical 0x20009d024) is not there for some reason.
Dave