On Wed, Oct 26, 2011 at 9:14 AM, Ottavio
<address@hidden> wrote:
On 26 October 2011 16:40, avadh patel <
address@hidden> wrote:
> But the issue is that I want to keep writes in temporary file and also use
> 'loadvm' to load
> CPU and device states. When I invoke qemu with -loadvm and -snapshot both
> options,
> then qemu doesn't load the VM state and does a normal boot.
As far as I understand, what you want to achieve is a "named
snapshot", that is a temporary file without using -snapshot option.
According to the latest doc file you do this with:
No. Let me try to explain bit further. I use QEMU in a architectural simulation
For running multiple benchmarks we use 'savevm' to save CPU and device
states so in one qcow2 image we have about 10 to 12 snapshots. Every time we
run simulations we do a 'loadvm' to load CPU states and then run simulations
from that point. The simulator uses QEMU for full-system emulation part.
Now what I am trying to achieve is when I run 'qemu' (not qemu-img) I want
to use -loadvm to load CPU state and also apply -snapshot so qcow2 image is not
modified at all. So that we can share the same disk image across multiple
simulation runs without corrupting it.
Thanks for your quick response.
- Avadh
qemu-img snapshot [-l | -a snapshot | -c snapshot | -d snapshot] filename
http://qemu.weilnetz.de/qemu-doc.html#qemu_005fimg_005finvocation
I remember back in the day you would create a named snaphot like this:
qemu-img create -f qcow2 -b base-image.img snapshot.img
I am not sure if this syntax is still supported because I haven't used
it recently, maybe somebody else can help.