[Crash-utility] Crash debugging vmcore, start error problem, crash seek error: kernel virtual address: fffffffffffffffa0 type: "kmem_cache buffer"

[Dave Anderson]

----- Original Message -----
> 
> hi:
> There is a problem that has been bothering me for several days. On the x86_64
> platform, I used crash to debug the vmcore file, but the crash cannot be
> started normally during startup, and the error is as follows:
> 
> # crash vmlinux vmcore
> 
> please wait...(gathering kmem slab cache data)
> crash seek error: kernel virtual address: fffffffffffffffa0 type: "kmem_cache
> buffer"
> crash: unable to initialize kmem slab cache subsystem
> crash: cannot determine idle task addresses fro m init_tasks[] or runqueues[]
> crash: cannot resolve "init_task_union"
> 
> At the beginning, I used crash7.0 version and reported the error: vmlinux is
> not SMP -- vmcore is SMP , so I have updated to version crash7.2. it was
> solved。But there are still another startup errors, crash: cannot determine
> idle task addresses from init_tasks[] or runqueues[] ------- crash: cannot
> resolve "init_task_union"
> 
> - -My hardware and software information:
> The arch: x86_64
> intel 8 the core CPU:
> Memroy: 8 gb
> 
> The kernel version: 4.4.39
> Kexec version: 2.0.14
> Crash version: 7.2.0
> 
> kernel config:
> CONFIG_CRASH_DUMP=y
> CONFIG_KEXEC=y
> CONFIG_SYSFS=y
> CONFIG_DEBUG_INFO=y
> CONFIG_PROC_VMCORE=y
> CONFIG_RELOCATABLE=y
> 
> 
> Ask for help:
> 1.Why did crash throw these errors? Is there a problem with my kernel
> configuration, or is there something else?
> 2.Can you give me some help according to my log?
> -- eager for your response

It's difficult to tell what the problem might be.  

 (1) Can you run crash successfully on the live system which generated the kdump?
 (2) Also, can you run crash with debugging turned on like this:

      $ crash -d4 vmlinux vmcore

     and then show or attach the full output.  That may yield some clues.

Dave