[panda-users] Forking in callback

[Brendan Dolan-Gavitt]

Snapshots in QEMU take a long time because operations on QCOWs are
very slow, and full snapshots require creating a disk snapshot. You
may be able to make this faster by storing the QCOW on a ramdisk.

-Brendan



On Mon, Mar 23, 2015 at 3:16 PM, Igor R <boost.lists at gmail.com> wrote:
>> QEMU uses pthreads, and forking a multi-threaded program is very
>> tricky. If you're just looking for some parallelism, you may want to
>> consider just spawning another thread.
>>
>> What are you trying to accomplish?
>
>
>
>
> I'm trying to quickly get back to some previous state - to be able to change
> some input without re-running my process from the beginning.
> To accomplish this I'm experimenting with various approaches - without much
> success so far...
> Besides the experiments with fork, I tried to save the VM state and then
> restore it, but it turns out that vm_save/vm_load take a lot of time - much
> more than creating a snapshot during record initialization. Why does the
> latter snapshot get produced much faster? Is it just because it doesn't
> contain the state of the devices?
> In general, is there a fast way to do that in QEMU?
>
> Thanks!
>
> _______________________________________________
> panda-users mailing list
> panda-users at mit.edu
> http://mailman.mit.edu/mailman/listinfo/panda-users
>