Add support for saving VM state when a test fails.

Add support for saving VM state when a test fails.

VM state information is stored in a copy-on-write QCOW2 file that is
created based on the chromiumos_qemu_image.bin file.  Only changes are
written to this file.

The guest VM triggers the save VM command by writing "savevm" to its
/dev/ttyUSB0.  This USB serial device is a virtual USB serial device
created by QEMU and is redirected to the QEMU monitor.

Since QEMU USB serial device emulates an FTDI UART, image_to_vm.sh
modifies the kernel command-line parameter to force the
usbserial_generic driver to claim the QEMU USB serial device.

BUG=9577
TEST=Run cros_run_vm_test with various parameter combinations, force a
test failure in the middle of smoke_Suite to ensure a snapshot is
created and verify snapshot is correct by loading it, run
cros_run_vm_test without new parameters to ensure existing
functionalities

Change-Id: I776968a61e25efffca2ab761aa5f8a24b80a7369

[thieule, 2010-11-20 00:30:02]

Test infrastructure changes for saving a VM state after a test failure is in a
separate CL.

PTAL

[Chris Masone, 2010-11-20 01:16:26]

lgtm

On 2010/11/20 00:30:02, thieule wrote:
> Test infrastructure changes for saving a VM state after a test failure is in a
> separate CL.
> 
> PTAL

[sosa, 2010-11-20 08:55:08]

LGTM w/ nit

http://codereview.chromium.org/5174009/diff/1/bin/cros_run_vm_test
File bin/cros_run_vm_test (right):

http://codereview.chromium.org/5174009/diff/1/bin/cros_run_vm_test#newcode26
bin/cros_run_vm_test:26: DEFINE_boolean save_vm_state ${FLAGS_FALSE} \
Sort in alphabetical order

[thieule, 2010-11-23 22:25:15]

Had to change from TCP socket to unix socket for the redirection to make this
work with cros_run_parrallel_vm_test.

PTAL

[sosa, 2010-11-24 00:44:04]

LGTM

[sosa, 2010-11-24 00:44:30]

Actually don't forget about the nit :)

On 2010/11/24 00:44:04, sosa wrote:
> LGTM

[thieule, 2010-11-24 00:51:34]

http://codereview.chromium.org/5174009/diff/1/bin/cros_run_vm_test
File bin/cros_run_vm_test (right):

http://codereview.chromium.org/5174009/diff/1/bin/cros_run_vm_test#newcode26
bin/cros_run_vm_test:26: DEFINE_boolean save_vm_state ${FLAGS_FALSE} \
On 2010/11/20 08:55:08, sosa wrote:
> Sort in alphabetical order

Done.

[sosa, 2011-01-04 20:27:00]

Was this ever committed?

[thieule, 2011-01-04 23:38:01]

It's been checked in but it hasn't been turned on.

We need to start uploading the logs from the VMs to google storage first
since this process also uploads the QEMU disk image.  Eventually, it should
also upload the state files.

On Tue, Jan 4, 2011 at 12:27 PM, <sosa@chromium.org> wrote:

> Was this ever committed?
>
> http://codereview.chromium.org/5174009/
>

[sosa, 2011-01-04 23:39:39]

Is there any status on the uploading of logs to google storage?  I
remember seeing the CL for cbuildbot but it doesn't look like it got
committed.

On Tue, Jan 4, 2011 at 3:37 PM, Thieu Le <thieule@chromium.org> wrote:
> It's been checked in but it hasn't been turned on.
> We need to start uploading the logs from the VMs to google storage first
> since this process also uploads the QEMU disk image.  Eventually, it should
> also upload the state files.
>
> On Tue, Jan 4, 2011 at 12:27 PM, <sosa@chromium.org> wrote:
>>
>> Was this ever committed?
>>
>> http://codereview.chromium.org/5174009/
>
>

[thieule, 2011-01-04 23:57:05]

+scottz

The CL was committed.  Scottz is looking to integrate it into the buildbots.

On Tue, Jan 4, 2011 at 3:39 PM, Chris Sosa <sosa@chromium.org> wrote:

> Is there any status on the uploading of logs to google storage?  I
> remember seeing the CL for cbuildbot but it doesn't look like it got
> committed.
>
> On Tue, Jan 4, 2011 at 3:37 PM, Thieu Le <thieule@chromium.org> wrote:
> > It's been checked in but it hasn't been turned on.
> > We need to start uploading the logs from the VMs to google storage first
> > since this process also uploads the QEMU disk image.  Eventually, it
> should
> > also upload the state files.
> >
> > On Tue, Jan 4, 2011 at 12:27 PM, <sosa@chromium.org> wrote:
> >>
> >> Was this ever committed?
> >>
> >> http://codereview.chromium.org/5174009/
> >
> >
>