7:32 p.m.
On Wed, Jun 29, 2016 at 05:25:27PM -0400, Dave Anderson wrote:
Hi Takahiro,
Here is another thing that I would prefer not to change/omit.
In the current code, the raw exception frame data is dumped as
part of the "bt -[fF]" output, just prior to it being translated
as an exception frame:
crash> bt -F
PID: 1223 TASK: ffff800020ef5780 CPU: 3 COMMAND: "sh"
... [ cut ] ...
#5 [ffff800020b0bb70] do_mem_abort at ffff00000808128c
ffff800020b0bb70: ffff800020b0bd40 el1_da+24
ffff800020b0bb80: cpu_cgrp_subsys+152 0000000000000063
ffff800020b0bb90: ffff800020b0bd40 sysrq_handle_crash+32
ffff800020b0bba0: 0000000000000002 textbuf.34610
ffff800020b0bbb0: ffff800020b0bbd0 kallsyms_token_index+43128
ffff800020b0bbc0: 000000000000000f 0000000100000000
ffff800020b0bbd0: ffff800020b0bc70 vprintk_default+56
ffff800020b0bbe0: cpu_cgrp_subsys+152 0000000000000063
ffff800020b0bbf0: sysrq_crash_op 0000000000000009
ffff800020b0bc00: 0000000000000000 0000000000000015
ffff800020b0bc10: 0000000000000120 0000000000000040
ffff800020b0bc20: 0000000000000001 0000000000000000
ffff800020b0bc30: log_wait+8 0000000000000000
ffff800020b0bc40: 0000000000000000 00000000000047d4
ffff800020b0bc50: ffff800022f337a4 0000000000000000
ffff800020b0bc60: 0000000000000106 0000000000000001
ffff800020b0bc70: 0000000000000002 0000000000000106
ffff800020b0bc80: log_buf_len cont
ffff800020b0bc90: 0000ffff83cc28f0 text.34829+13
ffff800020b0bca0: sys_write 0000ffff83d266c0
ffff800020b0bcb0: 0000000000000006 cpu_cgrp_subsys+152
ffff800020b0bcc0: 0000000000000063 sysrq_crash_op
ffff800020b0bcd0: 0000000000000009 0000000000000000
ffff800020b0bce0: 0000000000000015 0000000000000120
ffff800020b0bcf0: 0000000000000040 sys_call_table
ffff800020b0bd00: ffff800020b08000 ffff800020b0bd40
ffff800020b0bd10: sysrq_handle_crash+12 ffff800020b0bd40
ffff800020b0bd20: sysrq_handle_crash+32 0000000060400149
ffff800020b0bd30: cpu_cgrp_subsys+152 [kmalloc-1024]
#6 [ffff800020b0bd40] el1_da at ffff000008084568
PC: ffff000008457fc8 [sysrq_handle_crash+32]
LR: ffff000008457fb4 [sysrq_handle_crash+12]
SP: ffff800020b0bd40 PSTATE: 60400149
X29: ffff800020b0bd40 X28: ffff800020b08000 X27: ffff0000087e2000
X26: 0000000000000040 X25: 0000000000000120 X24: 0000000000000015
X23: 0000000000000000 X22: 0000000000000009 X21: ffff000008e071b0
X20: 0000000000000063 X19: ffff000008dda000 X18: 0000000000000006
X17: 0000ffff83d266c0 X16: ffff0000081c68b8 X15: ffff000008e6cc95
X14: 0000ffff83cc28f0 X13: ffff000008e6c758 X12: ffff000008dda7a0
X11: 0000000000000106 X10: 0000000000000002 X9: 0000000000000001
X8: 0000000000000106 X7: 0000000000000000 X6: ffff800022f337a4
X5: 00000000000047d4 X4: 0000000000000000 X3: 0000000000000000
X2: ffff000008dda7b8 X1: 0000000000000000 X0: 0000000000000001
ORIG_X0: ffff000008dda000 SYSCALLNO: ffff80002104d418
...
whereas with the v5 patchset, the exception frame only gets translated,
but the actual raw memory never gets dumped:
I surely remember that you said that would not be an issue
when I submitted older version, maybe v1 or v2.
crash> bt -F
PID: 1223 TASK: ffff800020ef5780 CPU: 3 COMMAND: "sh"
... [ cut ] ...
#5 [ffff800020b0bb70] do_mem_abort at ffff000008081288
ffff800020b0bb70: ffff800020b0bd40 el1_da+24
ffff800020b0bb80: cpu_cgrp_subsys+152 0000000000000063
ffff800020b0bb90: ffff800020b0bd40 sysrq_handle_crash+32
ffff800020b0bba0: 0000000000000002 textbuf.34610
ffff800020b0bbb0: ffff800020b0bbd0 kallsyms_token_index+43128
ffff800020b0bbc0: 000000000000000f 0000000100000000
ffff800020b0bbd0: ffff800020b0bc70 vprintk_default+56
ffff800020b0bbe0: cpu_cgrp_subsys+152 0000000000000063
ffff800020b0bbf0: sysrq_crash_op 0000000000000009
ffff800020b0bc00: 0000000000000000 0000000000000015
ffff800020b0bc10: 0000000000000120 0000000000000040
#6 [ffff800020b0bc20] el1_da at ffff000008084564
Do you think that those symbolic display are still useful
though it is not quite easy to recognize which register has what value?
Even more, <ffff800020b0bb80-ffff800020b0bc10> is *not* a stack for
do_mem_abort(). It is just wrong and will confuse people.
So this is another example of improvement on my patches.
--- <Exception in kernel> ---
PC: ffff000008457fc8 [sysrq_handle_crash+32]
LR: ffff000008457fb4 [sysrq_handle_crash+12]
SP: ffff800020b0bd40 PSTATE: 60400149
X29: ffff800020b0bd40 X28: ffff800020b08000 X27: ffff0000087e2000
X26: 0000000000000040 X25: 0000000000000120 X24: 0000000000000015
X23: 0000000000000000 X22: 0000000000000009 X21: ffff000008e071b0
X20: 0000000000000063 X19: ffff000008dda000 X18: 0000000000000006
X17: 0000ffff83d266c0 X16: ffff0000081c68b8 X15: ffff000008e6cc95
X14: 0000ffff83cc28f0 X13: ffff000008e6c758 X12: ffff000008dda7a0
X11: 0000000000000106 X10: 0000000000000002 X9: 0000000000000001
X8: 0000000000000106 X7: 0000000000000000 X6: ffff800022f337a4
X5: 00000000000047d4 X4: 0000000000000000 X3: 0000000000000000
X2: ffff000008dda7b8 X1: 0000000000000000 X0: 0000000000000001
...
It's actually helpful being able to dump the exception frame data with
"bt -F" to show symbolic references that are found within the exception
frame itself.
IMO I think that you'd better improve the output of
arm64_print_exception_frame() for bt -F.
And this is an totally independent issue from other parts of my patches.
-Takahiro AKASHI
Dave
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility