[Qemu-devel] an error while loading checkpoint

[Qemu-devel] an error while loading checkpoint

[sparsh mittal]

[-- Attachment #1: Type: text/plain, Size: 287 bytes --]

I get this warning while using a checkpoint. I issued the command -m 4G
-icount 0 and other parameters.
Unknown savevm section type 32
qemu-system-x86_64: Error -22 while loading VM state

My computer has 48GB RAM, so I don't know what is the problem?


Thanks and Regards
Sparsh Mittal

[-- Attachment #2: Type: text/html, Size: 435 bytes --]

Re: [Qemu-devel] an error while loading checkpoint

[Stefan Hajnoczi]

On Wed, Dec 7, 2011 at 11:41 PM, sparsh mittal <sparsh0mittal@gmail.com> wrote:
> I get this warning while using a checkpoint. I issued the command -m 4G
> -icount 0 and other parameters.
> Unknown savevm section type 32
> qemu-system-x86_64: Error -22 while loading VM state
>
> My computer has 48GB RAM, so I don't know what is the problem?

Looks like the stream has gotten out of sync.

Please post the full QEMU command-line as well as any monitor commands
you issued so it is possible to reproduce this.

Stefan

Re: [Qemu-devel] an error while loading checkpoint

[sparsh mittal]

[-- Attachment #1: Type: text/plain, Size: 2502 bytes --]

I am using qemu as a part of Marss cycle-accurate simulator.
http://marss86.org/~marss86/index.php/Home

$ qemu/qemu-system-x86_64 -m 4G -snapshot -loadvm ffnn -vnc :3 -icount 0
-drive cache=unsafe,file=/local/sparsh/specimage_sftn3.qcow2 -simconfig
MyConfigs/ffnn_100M.cfg

Unknown savevm section type 32
qemu-system-x86_64: Error -22 while loading VM state
==================================
Another thing is that only in this checkpoint (ffnn), I get this error, and
not in any other checkpoint.
==================================
Here is the information about the image:

$ qemu/qemu-img info /local/sparsh/specimage_sftn3.qcow2
image: /local/sparsh/specimage_sftn3.qcow2
file format: qcow2
virtual size: 9.3G (10000000000 bytes)
disk size: 30G
cluster_size: 65536
Snapshot list:
ID        TAG                 VM SIZE                DATE       VM CLOCK
1         sfff1                  1.9G 2011-12-02 14:45:23   00:16:36.700
2         sfft1                  2.4G 2011-12-02 15:13:21   00:16:12.117
3         stnn                   2.5G 2011-12-02 15:42:51   00:16:46.837
4         ftnn                   1.5G 2011-12-02 16:06:46   00:16:45.460
6         sftt1                  2.3G 2011-12-02 17:06:53   00:16:13.254
7         ffnn                   1.0G 2011-12-02 17:30:37   00:16:20.771
8         snnn                   2.4G 2011-12-02 17:50:50   00:16:10.128
9         sfft2                  1.6G 2011-12-02 18:45:32   00:16:11.053
11        sftt2                  2.3G 2011-12-02 20:04:26   00:16:15.169
12        sffn                   1.8G 2011-12-06 10:46:26   00:16:31.435
13        ttff                   2.0G 2011-12-06 11:05:40   00:16:05.729
====================
I also have question whether -icount 0 or -icount auto can make a
simulation more deterministic and whether qemu is affected by the load on
the host (i.e. how many processes are running?)
Sparsh


On Thu, Dec 8, 2011 at 3:53 AM, Stefan Hajnoczi <stefanha@gmail.com> wrote:

> On Wed, Dec 7, 2011 at 11:41 PM, sparsh mittal <sparsh0mittal@gmail.com>
> wrote:
> > I get this warning while using a checkpoint. I issued the command -m 4G
> > -icount 0 and other parameters.
> > Unknown savevm section type 32
> > qemu-system-x86_64: Error -22 while loading VM state
> >
> > My computer has 48GB RAM, so I don't know what is the problem?
>
> Looks like the stream has gotten out of sync.
>
> Please post the full QEMU command-line as well as any monitor commands
> you issued so it is possible to reproduce this.
>
> Stefan
>

[-- Attachment #2: Type: text/html, Size: 3120 bytes --]

Re: [Qemu-devel] an error while loading checkpoint

[Stefan Hajnoczi]

On Thu, Dec 8, 2011 at 1:57 PM, sparsh mittal <sparsh0mittal@gmail.com> wrote:
> I am using qemu as a part of Marss cycle-accurate simulator.
> http://marss86.org/~marss86/index.php/Home
>
> $ qemu/qemu-system-x86_64 -m 4G -snapshot -loadvm ffnn -vnc :3 -icount 0
> -drive cache=unsafe,file=/local/sparsh/specimage_sftn3.qcow2 -simconfig
> MyConfigs/ffnn_100M.cfg

I wonder if the image loads in vanilla QEMU.  Try dropping the -icount
0 and -simconfig options just to see whether vanilla QEMU gets the
same error or not.

Also, have you asked on the Marss mailing list?  Please try that first
because the bug may be due to changes that were made to vanilla QEMU.
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Stefan

Re: [Qemu-devel] an error while loading checkpoint

[sparsh mittal]

[-- Attachment #1: Type: text/plain, Size: 1144 bytes --]

On Thu, Dec 8, 2011 at 8:18 AM, Stefan Hajnoczi <stefanha@gmail.com> wrote:

> On Thu, Dec 8, 2011 at 1:57 PM, sparsh mittal <sparsh0mittal@gmail.com>
> wrote:
> > I am using qemu as a part of Marss cycle-accurate simulator.
> > http://marss86.org/~marss86/index.php/Home
> >
> > $ qemu/qemu-system-x86_64 -m 4G -snapshot -loadvm ffnn -vnc :3 -icount 0
> > -drive cache=unsafe,file=/local/sparsh/specimage_sftn3.qcow2 -simconfig
> > MyConfigs/ffnn_100M.cfg
>
> I wonder if the image loads in vanilla QEMU.  Try dropping the -icount
> 0 and -simconfig options just to see whether vanilla QEMU gets the
> same error or not.
>
> I have tried dropping all these: snapshot, icount, vnc, cache=unsafe, but
still the error comes. I deleted this checkpoint and made another one (with
different name, but same benchmarks), still error comes.
I will also ask marss-forum.
Surprisingly, with other checkpoints, error does not come.


> Also, have you asked on the Marss mailing list?  Please try that first
> because the bug may be due to changes that were made to vanilla QEMU.
> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
>
> Stefan
>

[-- Attachment #2: Type: text/html, Size: 1925 bytes --]

Re: [Qemu-devel] an error while loading checkpoint

[sparsh mittal]

[-- Attachment #1: Type: text/plain, Size: 1525 bytes --]

I also have question whether -icount 0 or -icount auto can make a
simulation more deterministic and whether qemu is affected by the load on
the host (i.e. how many processes are running?) I would be grateful for a
reply.
Thanks and Regards
Sparsh Mittal




On Thu, Dec 8, 2011 at 8:36 AM, sparsh mittal <sparsh0mittal@gmail.com>wrote:

>
>
>
>
> On Thu, Dec 8, 2011 at 8:18 AM, Stefan Hajnoczi <stefanha@gmail.com>wrote:
>
>> On Thu, Dec 8, 2011 at 1:57 PM, sparsh mittal <sparsh0mittal@gmail.com>
>> wrote:
>> > I am using qemu as a part of Marss cycle-accurate simulator.
>> > http://marss86.org/~marss86/index.php/Home
>> >
>> > $ qemu/qemu-system-x86_64 -m 4G -snapshot -loadvm ffnn -vnc :3 -icount 0
>> > -drive cache=unsafe,file=/local/sparsh/specimage_sftn3.qcow2 -simconfig
>> > MyConfigs/ffnn_100M.cfg
>>
>> I wonder if the image loads in vanilla QEMU.  Try dropping the -icount
>> 0 and -simconfig options just to see whether vanilla QEMU gets the
>> same error or not.
>>
>> I have tried dropping all these: snapshot, icount, vnc, cache=unsafe, but
> still the error comes. I deleted this checkpoint and made another one (with
> different name, but same benchmarks), still error comes.
> I will also ask marss-forum.
> Surprisingly, with other checkpoints, error does not come.
>
>
>> Also, have you asked on the Marss mailing list?  Please try that first
>> because the bug may be due to changes that were made to vanilla QEMU.
>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
>>
>> Stefan
>>
>
>

[-- Attachment #2: Type: text/html, Size: 2606 bytes --]