[panda-users] Replay Failing

Danux danuxx at gmail.com
Mon Jul 10 21:56:24 EDT 2017


You were right Brendan, the loadvm command failed with same error, but I
installed Qemu through Panda installation script, I assumed it uses a
working version.

Can I upgrade Qemu to latest version separetely, or I need to compile Qemu
from scratch with Panda libraries?

Thanks.

On Sat, Jul 8, 2017 at 6:23 PM, Brendan Dolan-Gavitt <brendandg at nyu.edu>
wrote:

> It looks like the vexpress-a9 machine doesn't have a PCI bus, so you can't
> use IDE or SATA devices. However, if you need big endian I think you're out
> of luck anyway – I believe the vexpress-a9 board is also little endian.
> QEMU does support big endian arm, but I think it doesn't model any boards
> that use big-endian by default:
>
> https://stackoverflow.com/questions/41571643/emulatin-big-
> endian-arm-system-with-qemu
>
> (That answer is from one of the QEMU developers. Things may have changed
> since 2011 though.)
>
> Moving this back to the mailing list so that other people can benefit...
>
> -Brendan
>
> On Sat, Jul 8, 2017 at 9:02 PM, Danux <danuxx at gmail.com> wrote:
>
>> Hi Brendan,
>>
>> Thanks a lot for your help.
>> Happy to get in touch with you, really like Panda concept so hope to get
>> this solved.
>>
>> Is there a way to use ide or something else instead of sd? I used -sd
>> option just because I followed a blog.
>>
>> I need ARM big endian,  little endian does not help me.
>>
>> Will try all those suggestions and come back here with my feedback.
>>
>> On Jul 8, 2017 7:37 PM, "Brendan Dolan-Gavitt" <brendandg at nyu.edu> wrote:
>>
>> The ARM issue seems to be a bug in QEMU. Replaying a recording requires
>> loading a saved snapshot, and it appears that the SD card device in QEMU
>> isn't properly loading the snapshot (you can verify this by trying to take
>> a snapshot with savevm and loading it with loadvm afterward). One
>> workaround would be to use the "armel" image instead, which does not use an
>> SD card.
>>
>> As for the WinXP example, it's harder to tell what could be going wrong.
>> Are you able to do a successful recording using the i386 Linux image? (You
>> can test with panda/scripts/run_on_32bit_linux.py – it will download an
>> appropriate QCOW and run a command in a 32-bit x86 Debian image for you).
>>
>> Also, note that you should not need to use sudo to run PANDA (and it's
>> not recommended).
>>
>> Finally, I noticed you stopped by on IRC – unfortunately we can't be on
>> there 24 hours a day, so you may need to hang around a while before someone
>> can respond to your question. A persistent IRC client (e.g., irssi running
>> in a screen session) can help with this.
>>
>> On Sat, Jul 8, 2017 at 3:59 AM, Danux <danuxx at gmail.com> wrote:
>>
>>> Here is another try but now with WinXP, it is also failing :-(
>>>
>>> Running on:
>>>
>>> Linux panda 4.4.0-83-generic #106~14.04.1-Ubuntu SMP Mon Jun 26 18:10:19
>>> UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
>>>
>>> danux at panda:~/panda/build/i386-softmmu$ sudo ./qemu-system-i386 -drive
>>> file=/media/drive/winXP.qcow2,if=ide,cache=writeback -m 1024M --monitor
>>> stdio -vnc 192.168.1.74:1 -replay notepad
>>>
>>> [sudo] password for danux:
>>>
>>> QEMU 2.8.50 monitor - type 'help' for more information
>>>
>>> (qemu) loading snapshot
>>>
>>> ... done.
>>>
>>> opening nondet log for read : ./notepad-rr-nondet.log
>>>
>>> Ahead of log while looking for log entry of type RR_SKIPPED_CALL
>>>
>>> {guest_instr_count=7526}
>>>
>>> RR_INPUT_4 from RR_CALLSITE_READ_4
>>>
>>> FOUND DISAGREEMENT!
>>>
>>> Replay program point:
>>>
>>> {guest_instr_count=7528}
>>>
>>>
>>> Record program point:
>>>
>>> {guest_instr_count=7526}
>>>
>>>
>>> >>> guest instruction counts disagree
>>>
>>> notepad:        7528 (  0.00%) instrs.    0.09 sec.  0.30 GB ram.
>>>
>>> ERROR: replay failed!
>>>
>>> Time taken was: 0 seconds.
>>>
>>> Stats:
>>>
>>> RR_INPUT_1 number = 0, size = 0 bytes
>>>
>>> RR_INPUT_2 number = 0, size = 0 bytes
>>>
>>> RR_INPUT_4 number = 22, size = 308 bytes
>>>
>>> RR_INPUT_8 number = 25, size = 450 bytes
>>>
>>> RR_INTERRUPT_REQUEST number = 11, size = 154 bytes
>>>
>>> RR_EXIT_REQUEST number = 0, size = 0 bytes
>>>
>>> RR_SKIPPED_CALL number = 0, size = 0 bytes
>>>
>>> max_queue_len = 13
>>>
>>> ERROR: replay failed!
>>>
>>> On Fri, Jul 7, 2017 at 10:42 PM, Danux <danuxx at gmail.com> wrote:
>>>
>>>> Hi,
>>>>
>>>> When trying to run the replay option on a linux ARM system is failing
>>>> as follows, please help:
>>>>
>>>> (qemu) begin_record test
>>>>
>>>> (qemu) writing snapshot: ./test-rr-snp
>>>>
>>>> opening nondet log for write : ./test-rr-nondet.log
>>>>
>>>> (qemu) end_record
>>>>
>>>> (qemu) Time taken was: 75 seconds.
>>>>
>>>> (qemu) begin_replay test
>>>>
>>>> (qemu) loading snapshot
>>>>
>>>> qemu-system-arm: /home/danux/panda/hw/sd/sd.c:217: sd_ocr_powerup:
>>>> Assertion `!(sd->ocr & 0x80000000)' failed.
>>>>
>>>> Or by passing the replay in the command line:
>>>>
>>>>
>>>> danux at panda:~/panda/build/arm-softmmu$ ./qemu-system-arm  -M
>>>> vexpress-a9 -sd snapshot.img -kernel vmlinuz-3.2.0-4-vexpress -initrd
>>>> initrd.img-3.2.0-4-vexpress  -replay test -m 256M
>>>>
>>>> audio: Could not init `oss' audio driver
>>>>
>>>> loading snapshot
>>>>
>>>> qemu-system-arm: Missing section footer for sd-card
>>>>
>>>> Failed to load vmstate
>>>>
>>>> Failed to start replay
>>>>
>>>>
>>>> Please help me
>>>>
>>>>
>>>> Thanks!
>>>>
>>>>
>>>>
>>>
>>>
>>> --
>>> DanUx
>>>
>>> _______________________________________________
>>> panda-users mailing list
>>> panda-users at mit.edu
>>> http://mailman.mit.edu/mailman/listinfo/panda-users
>>>
>>>
>>
>>
>> --
>> Brendan Dolan-Gavitt
>> Assistant Professor, Department of Computer Science and Engineering
>> NYU Tandon School of Engineering
>>
>>
>>
>
>
> --
> Brendan Dolan-Gavitt
> Assistant Professor, Department of Computer Science and Engineering
> NYU Tandon School of Engineering
>



-- 
DanUx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/panda-users/attachments/20170710/45023484/attachment-0001.html


More information about the panda-users mailing list