[panda-users] about scissors

Wed Apr 22 15:21:05 EDT 2015

Ending a replay with rr_do_end_replay() from within the CPU thread
does bad things (don't know why precisely). The solution is to instead
just set the rr_end_replay_requested flag to 1. Fix pushed.

I think there are still problems with the scissors though. I just
tried to snip the win7iessl replay at start=181577170 end=318079730,
and only got 24 instructions in the log...

-Brendan

On Wed, Apr 22, 2015 at 2:53 PM, Manolis Stamatogiannakis
<mstamat at gmail.com> wrote:
> I also had a segfault with scissors. The trimmed trace looks alright,
> despite the segfault.
>
> My command line was:  ./i386-softmmu/qemu-system-i386 -panda
> scissors:start=17401969,end=47488576,name=vlassis_vi -replay vlassis
>
> I added -g to the plugin flags, and here is the stack trace from the core:
> http://pastebin.com/y6Dzr6YH
> I may have sometime to further look this next week.
>
> M.
>
> 2015-04-22 8:36 GMT-07:00 Brendan Dolan-Gavitt <brendandg at gatech.edu>:
>>
>> Hi,
>>
>> >From the output you posted it looks like you also have the
>> stringsearch plugin enabled ("stringsearch: added string of length 14
>> to search set"). Could you try again with that turned off?
>>
>> -Brendan
>>
>> On Wed, Apr 22, 2015 at 2:01 AM, xiaojuan Li <xiaotan6666 at gmail.com>
>> wrote:
>> > Hi,
>> > i want to reflect a problem about scissor plugin:
>> > I found that scissor plugin did not work, cannot extract a part from the
>> > original snp.
>> > i use the command line:
>> > scissors:start=4335499535,end=5244538335,name=api21-256-scissor...
>> >
>> > Adding PANDA arg scissors:start=4335499535.
>> > Adding PANDA arg scissors:end=5244538335.
>> > Adding PANDA arg scissors:name=api21-256-scissor.
>> > adding
>> >
>> > /home/shentanli/pandanew/qemu/arm-softmmu/panda_plugins/panda_scissors.so to
>> > panda_plugin_files 0
>> > Adding PANDA arg stringsearch:name=test.
>> > adding
>> >
>> > /home/shentanli/pandanew/qemu/arm-softmmu/panda_plugins/panda_stringsearch.so
>> > to panda_plugin_files 1
>> > emulator: registered 'boot-properties' qemud service
>> > emulator: Adding boot property: 'dalvik.vm.heapsize' = '48m'
>> > emulator: Adding boot property: 'qemu.sf.fake_camera' = 'both'
>> > emulator: Adding boot property: 'qemu.hw.mainkeys' = '0'
>> > loading
>> >
>> > /home/shentanli/pandanew/qemu/arm-softmmu/panda_plugins/panda_scissors.so
>> > Success
>> > loading
>> >
>> > /home/shentanli/pandanew/qemu/arm-softmmu/panda_plugins/panda_stringsearch.so
>> > Initializing plugin stringsearch
>> > panda_require: callstack_instr
>> > loading
>> >
>> > /home/shentanli/pandanew/qemu/arm-softmmu/panda_plugins/panda_callstack_instr.so
>> > Initializing plugin callstack_instr
>> > Success
>> > stringsearch: added string of length 14 to search set
>> > Success
>> >
>> >
>> >  when it reaches the 4335499535,it prints:
>> >
>> > api214-20-256:  4335499535 ( 62.00%) instrs.  296.73 sec.  0.89 GB ram.
>> > Original ending prog point: {guest_instr_count=6992494441 pc=0xc00158c4,
>> > secondary=0x00000000}
>> > Saving snapshot at instr count 4335499535...
>> > Beginning cut-and-paste process at prog point:
>> > {guest_instr_count=4335499535 pc=0xc000db0c, secondary=0x00000000}
>> > Writing entries to api21-256-scissor-rr-nondet.log...
>> > Reached end of old nondet log.
>> > Continuing with replay.
>> >
>> > then when it reaches the end, it shows:
>> >
>> > api214-20-256:  5244538335 ( 75.00%) instrs.  377.58 sec.  1.09 GB ram.
>> > Ending cut-and-paste on prog point:
>> > {guest_instr_count=5244538336 pc=0xffff0008, secondary=0x00000000}
>> > api214-20-256:  5244538336 ( 75.00%) instrs.  377.59 sec.  1.09 GB ram.
>> > Replay completed successfully.
>> > Time taken was: 380 seconds.
>> > Stats:
>> > RR_INPUT_1 number = 634, size = 17118 bytes
>> > RR_INPUT_2 number = 237, size = 6636 bytes
>> > RR_INPUT_4 number = 659406, size = 19782180 bytes
>> > RR_INPUT_8 number = 0, size = 0 bytes
>> > RR_INTERRUPT_REQUEST number = 1343219, size = 37610132 bytes
>> > RR_EXIT_REQUEST number = 0, size = 0 bytes
>> > RR_SKIPPED_CALL number = 424454, size = 238156678 bytes
>> > RR_DEBUG number = 0, size = 0 bytes
>> > max_queue_len = 747
>> > 746 items on recycle list, 65648 bytes total
>> > Replay terminated at user request.
>> >
>> > Logging all cpu states
>> >
>> >
>> > it exits.and when i check the scissor file, it is the same size as the
>> > original one.
>> >
>> >
>> >
>> > Please correct me!
>> > Thanks a lot!
>> >
>> >
>> > 2015-04-21 7:48 GMT-04:00 xiaojuan Li <xiaotan6666 at gmail.com>:
>> >
>> >> excuse me,  scissors failed again:
>> >> api214-20-256:  4268808000 ( 61.05%) instrs.  329.84 sec.  0.88 GB ram.
>> >> api214-20-256:  4335499535 ( 62.00%) instrs.  335.77 sec.  0.89 GB ram.
>> >> Original ending prog point: {guest_instr_count=6992494441
>> >> pc=0xc00158c4,
>> >> secondary=0x00000000}
>> >> Saving snapshot at instr count 4335499535...
>> >> Beginning cut-and-paste process at prog point:
>> >> {guest_instr_count=4335499535 pc=0xc000db0c, secondary=0x00000000}
>> >> Writing entries to api4-21-rr-nondet.log...
>> >> Assertion failure @ count 2656994906!
>> >> api214-20-256:  4335499535 ( 62.00%) instrs.  336.32 sec.  1.08 GB ram.
>> >> ERROR: replay failed!
>> >> Time taken was: 337 seconds.
>> >> Stats:
>> >>
>> >>
>> >> now i use taint encounter 2 errors:aborted and killed.
>> >>
>> >> the first one may caused malloc(),but i donnot know why killed.
>> >>
>> >> so i want to scissor it to a smaller one (just use the stringseach) and
>> >> then use the smaller to taint, but scissor failed too.
>> >>
>> >>
>> >> 2015-04-20 22:06 GMT-04:00 Brendan Dolan-Gavitt <mooyix at gmail.com>:
>> >>>
>> >>> This might be caused by a bug that was just fixed today:
>> >>>
>> >>> https://github.com/moyix/panda/issues/58
>> >>>
>> >>> Could you do a git pull and try again?
>> >>>
>> >>> -Brendan
>> >>>
>> >>> On Sun, Apr 19, 2015 at 8:29 AM, xiaojuan Li <xiaotan6666 at gmail.com>
>> >>> wrote:
>> >>> > HI,Brendan,
>> >>> > I am the one sending emails asking for help about the taint
>> >>> > segfault.
>> >>> > Till
>> >>> > now, it haven't been fixed.(have tried your suggestions,but seems
>> >>> > does
>> >>> > not
>> >>> > work)
>> >>> > during my solving this problem,i find another problem about scissors
>> >>> > plugin:
>> >>> > using qemu-system-arm  *** -replay ***** -panda
>> >>> > "scissor:start=*,end=**,name=name" failed to get the cut one. does
>> >>> > it
>> >>> > have
>> >>> > something wrong?
>> >>> >
>> >>> > Thanks for your patience.
>> >>> > Best wishes.
>> >>> >
>> >>> >
>> >>> > --
>> >>> > wait and hope~~
>> >>
>> >>
>> >>
>> >>
>> >> --
>> >> wait and hope~~
>> >
>> >
>> >
>> >
>> > --
>> > wait and hope~~
>> _______________________________________________
>> panda-users mailing list
>> panda-users at mit.edu
>> http://mailman.mit.edu/mailman/listinfo/panda-users
>
>