Add support to detect if qemu becomes <defunct> (zombie)
after it was started. If qemu is detected defunct, it will
raise virt_vm.VMStartError().
Signed-off-by: NRuda Moura <rmoura@redhat.com>

Backports

They are not present in python 2.4. Replace them with
equivalent, py 2.4 compatible constructs.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

The methods hotplug and verify_hotplug are called
correctly in the inherited classes.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

The virttest.staging.backports module offers backport
implementations of the functions used, so they work under
python 2.4. Even if said tests won't be executed on older
distros, at least pylint checks will pass.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

We'll have to stick with python 2.4 for a little while
yet, so let's bring autotest's backport module to virt-test.

This module contains some functions people have been
using to write their tests. In order to use them, you
can just make

from virttest.backports import all, bin, next

The module takes care of figuring out if there's a
builtin present, or if the backported implementation
has to be used.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

try to catch windows bsod

Timedrift new case

This patch add bsod images catch when do crash check.
script will check whether the screendump image looks like a
bsod image by calculate the screendump image hamming distance
from the reference image(bsod image). if two images hamming
distance is less than the threshold(10), will mark this
screendump image as suspect, and raise guest BSOD error.
Configuration:
    check_guest_bsod = yes
    bsod_reference_img = XX.jpg (if not configure will use
    all the image under "share/deps/bsod_img")
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Signed-off-by: NXu Tian <xutian@redhat.com>

changes from v1:
    fix typo and add image_snapshot = no for s4
Signed-off-by: NXu Tian <xutian@redhat.com>

More pylint fixes

Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

According to the constructor of ConnPrivKeyError, such
exceptions should be passed a key file and a stderr.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

lvsb: Fix a bug of binary name for sandbox service.

Signed-off-by: NDongsheng Yang <yangds.fnst@cn.fujitsu.com>

Use older constructs that will make the tests possible
to work on older python versions.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Use a more generic method, that will work in older
python implementations.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Enable AHCI booting and sync Win2008-32 unattended file format

Pylint fixes

Here, the author called out to exit(), which is not
the intended effect. The actual exit condition is
`return`.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

For now, try/except/finally blocks shouldn't be used.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Rather than using '0b1', a representation not valid in
python 2.4, use 1, which should go for the same effect.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

 guest-os: add RHEL5.10 config

Periodically I was getting a failure when the first of the sequence
of virsh_edit's are run in one my long runs using the "--tests virsh"
option. The failing test is/was 'virsh.edit.normal_test.id_option'.
When run alone using "--tests virsh.edit" - the error didn't occur.
After looking at the test for a while I noticed that part of the
processing is to shutdown the guest prior to the last modify_vcpu()
call. Doing that means the 'id' is no longer valid - thus a failure
could occur that wasn't expected. Even after changing that I got a few
more failures and after an attempt to change the sleep() time from 1 to 2
I was directed to the remote.handle_prompts() call which should work.

Signed-off-by: NShuping Cui <scui@redhat.com>

virsh_migrate_stress: SKIP test if src and dest URI not defined

As we moved the check down the stack (given that we might
want to simply run run_pylint.py instead of the full blown
check_patch.py script).
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

tools/run_pylint.py: Handle unforseen exceptions running pylint

In some testing scenarios, such as daily pylint checking
on py 2.4, we noticed the attempt to inspect some pygtk
code generated problems such as:

RuntimeError: could not open display

Making the rest of the checks to be aborted. Let's handle
the exceptions and print the messages as we should.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Issue #1014.  Skip this test if we haven't redefined the source and
destination URI's.

qemu.tests.cpuinfo_query: Fix #1198

The test is wired up to check RHEL machine types, that
will not be present on anything that is not strictly
RHEL 6.4, so restrict the test to those hosts.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

run: Introduce new options, change virt-test default behavior

changes from V1:
    1. fix typo and update doc strings
Signed-off-by: NXu Tian <xutian@redhat.com>

In order to save time in virt-test:

1) The JeOS image is uncompressed once and reused
   every test, unless it is explicitly asked with
   --restore-image

2) After a test is done, the VMs won't be shut down
   unless it is explicitly asked with --restart-vm

3) After a test is done, the pristine image won't
   be restored, unless it is explicitly asked with
   --restore-image-between-tests

All of those behaviors, while might bring speed ups
to the test process, they are not safe. As users
frequently don't have the visibility of that when
using the test suite, we found out that it's in the
best interest of users and developers that the
defaults are changed to safer options.

With this, the options mentioned above changed to

--keep-guest-running
--keep-image
--keep-image-between-tests

The wording 'guest' was chose over 'vm' due to the
fact we already have --guest-os. So, mostly an attempt
to keep consistency.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

run: Add --console-level option, fix issue #179

With this we implement (admittedly very very late)
the functionality asked in issue #179. This introduces
a new --console-level option, that will allow people
using --verbose to adjust the level of the messages
printed to the console.

CC: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Spice tests updates/extensions (v3) (was #1076)