[Crash-utility] unwind not working on x86?

[Dave Anderson]

----- Original Message -----
> 
> Hello
> 
> I am trying to use both the unwind function on x86 in crash, and the fp
> extension. Both are not working for me.
> 
> I did the following tests: (crash 7.0.9)
> 
> crash/crash vmcore.201412161409 linux-3.2.64/vmlinux
> 
> GNU gdb (GDB) 7.6
> 
>       KERNEL: linux-3.2.64/vmlinux
>     DUMPFILE: vmcore.201412161409
>         CPUS: 1
>         DATE: Tue Dec 16 15:09:46 2014
>       UPTIME: 00:00:46
> LOAD AVERAGE: 0.05, 0.01, 0.01
>        TASKS: 55
>     NODENAME: debian
>      RELEASE: 3.2.64
>      VERSION: #2 SMP Tue Dec 16 15:08:10 CET 2014
>      MACHINE: x86_64  (2392 Mhz)
>       MEMORY: 383.5 MB
>        PANIC: "[   46.736164] Oops: 0002 [#1] SMP " (check log for details)
>          PID: 1962
>      COMMAND: "tee"
>         TASK: ffff88000e5e0000  [THREAD_INFO: ffff88000c8e8000]
>          CPU: 0
>        STATE: TASK_RUNNING (PANIC)
> 
> crash> bt
> PID: 1962   TASK: ffff88000e5e0000  CPU: 0   COMMAND: "tee"
>  #0 [ffff88000c8e99f0] machine_kexec at ffffffff81038e0a
>  #1 [ffff88000c8e9a60] crash_kexec at ffffffff810b3a92
>  #2 [ffff88000c8e9b30] oops_end at ffffffff816427e8
>  #3 [ffff88000c8e9b60] no_context at ffffffff81635d6f
>  #4 [ffff88000c8e9bc0] __bad_area_nosemaphore at ffffffff81635f49
>  #5 [ffff88000c8e9c20] bad_area at ffffffff81635fc2
>  #6 [ffff88000c8e9c50] do_page_fault at ffffffff81645454
>  #7 [ffff88000c8e9d60] do_async_page_fault at ffffffff81644b75
>  #8 [ffff88000c8e9d80] async_page_fault at ffffffff81641de5
>     [exception RIP: sysrq_handle_crash+22]
>     RIP: ffffffff813cad46  RSP: ffff88000c8e9e38  RFLAGS: 00010092
>     RAX: 0000000000000010  RBX: 0000000000000063  RCX: 00000000ffffffff
>     RDX: 0000000000000000  RSI: 0000000000000082  RDI: 0000000000000063
>     RBP: ffff88000c8e9e38   R8: 0000000000000000   R9: 0000000000000000
>     R10: 0000000000000000  R11: 0000000000000000  R12: ffffffff81c725e0
>     R13: 0000000000000286  R14: 0000000000000007  R15: 0000000000000000
>     ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018
>  #9 [ffff88000c8e9e40] __handle_sysrq at ffffffff813cb461
> #10 [ffff88000c8e9e80] write_sysrq_trigger at ffffffff813cb51a
> #11 [ffff88000c8e9eb0] proc_reg_write at ffffffff811d60e2
> #12 [ffff88000c8e9f00] vfs_write at ffffffff81176053
> #13 [ffff88000c8e9f30] sys_write at ffffffff8117637a
> #14 [ffff88000c8e9f80] sysenter_dispatch at ffffffff8164bc70
>     RIP: 00000000f775e430  RSP: 00000000ffe6a720  RFLAGS: 00000296
>     RAX: 0000000000000004  RBX: ffffffff8164bc70  RCX: 00000000ffe6a81c
>     RDX: 0000000000000002  RSI: 0000000000000002  RDI: 00000000ffe6a81c
>     RBP: 00000000ffe6a758   R8: 0000000000000000   R9: 0000000000000000
>     R10: 0000000000000000  R11: 0000000000000000  R12: 0000000000000000
>     R13: 0000000000000000  R14: 0000000000000003  R15: 0000000000000000
>     ORIG_RAX: 0000000000000004  CS: 0023  SS: 002b
> 
> crash> set unwind on
> unwind: on
> 
> crash> bt
> PID: 1962   TASK: ffff88000e5e0000  CPU: 0   COMMAND: "tee"
>  #0 [ffff88000c8e99f0] machine_kexec at ffffffff81038e0a
>     RIP: 00000000f775e430  RSP: 00000000ffe6a720  RFLAGS: 00000296
>     RAX: 0000000000000004  RBX: ffffffff8164bc70  RCX: 00000000ffe6a81c
>     RDX: 0000000000000002  RSI: 0000000000000002  RDI: 00000000ffe6a81c
>     RBP: 00000000ffe6a758   R8: 0000000000000000   R9: 0000000000000000
>     R10: 0000000000000000  R11: 0000000000000000  R12: 0000000000000000
>     R13: 0000000000000000  R14: 0000000000000003  R15: 0000000000000000
>     ORIG_RAX: 0000000000000004  CS: 0023  SS: 002b
> crash>
> 
> -> no functions no parameters
> 
> -------------------------------------------------------------------------------------------------------------
> 
> crash> extend crash/extensions/fp.so
> ./crash/extensions/fp.so: shared object loaded
> crash> fp
> .................
> 
> -> no functions no parameters
> 
> Did I anything wrong?
> 

Probably not...

I have never done any support of the x86_64 optional unwind support, nor do I
ever use it.  It was put into place by IBM back in 2006, and nothing substantive
has been applied to it since then.  The original author does not seem to be a 
member of this list any more, but perhaps there are other users on the list.

The author of the "fp" extension module is still a member of this list.  Perhaps
he can be of assistance.

Dave