parameter 'md5sum' in test configuration file will conflict with
cdrom iso md5sum parameter, and it will caused md5sum mismatch error
when re-create(env_process.preprocess_vm) VM in test script, so rename
it;
Signed-off-by: NXu Tian <xutian@redhat.com>

Replace the method which used only disk names to match autotest
image format with virttest.qemu_qtree one. It additionally verifies
all kind of other params of "info qtree" and "info block".
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

some fix for block stream tests

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

add_drive and add_cdrom are obsoleted by
virttest.qemu_devices.images_define_by_variables()
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

q35 doesn't have ide bus. People should use ahci instead.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Libvirt doesn't use the full cmdline, but only -device $driver,$params
format. This patch makes autotest follow this syntax too.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When we force insert device into bus where it doesn't belongs to
we should probably set bus and addr as it's not the default behavior.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Error reason was shown only in force insertion mode.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Add support for QDrive, which supports drive_add (QHPDrive) or
__com.redhat_drive_add (QRHDrive).
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This patch adds the possibility to safely hotplug/unplug devices from
running VM.

Additionally it adds another VM state and VM param. When the test
requires clean VM, you might specify "allow_hotplugged_vm = no" and
when VM used hotplug it will be recreated.

For hotplugging you might specify monitor, which should be used and
whether to verify results. This is not supported by all devices, but
when there is a method to automatically verify hotplug results, it'll
do it for you and cleans the qdev state for you. Otherwise it's upto the
user to do it by qdev.hotplug_verified().
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

To simplify and unify usage of QBaseDevice.child_bus I added
add/rm_child_bus functions.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Fix issues of DevContainer.remove() and add support for
non-recursive removal.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Currently during vm_preprocess params of unused running VMs are updated,
but VM is not recreated (it's not used, so there is no need for it).
If we create VM in 1st test, in 2nd test we use different params but set
start_vm = no, params are updated so when we use the same params from
2nd test in 3rd test, preprocess would match the VM from 1st test as
the params were updated. Anyway as the VM is not recreated, the running
VM doesn't represents the params from 2nd test.

With my modification the preprocess would always handle VM properly. The
only problem might be, when someone uses vm.create() directly without
preprocess in the 2nd test. Anyway it's a bad practise and I haven't found
it in upstream tests. We always use preprocess to prepare the VM.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Original autotest ignored some params.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This patch changes the way of handling images, cdroms and floppies
params. Instead of custom handling in make_create_command it uses
wrappers in self.devices. It keeps track of all image buses according
to autotest.

It shouldn't change the way how autotest used to create disks. Anyway
few params which were omited for some devices were added. Also all
buses are tracked so in case there is no free bus it should create one.

This feature works 100% with strict_mode. For non-strict_mode there
might be few differences as qemu doesn't differentiate between ide and
ahci buses. Handling of this would require deeper modifications to
the way of images handling and will be discussed and resolved later.

Images should be added with the correct version of device thus we should
be able to unplug/hotplug the device in test.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Use qemu_devices to create usb controllers and usb devices.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

this patch adds support to create USB controllers using qemu_devices.
This version supports all current options including usb-hubs plus
adds couple of new options.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

We were assembling the network device cmd line for RHEL5
QEMU incorrectly, post the qemu_devices work.

 bad cmdline:
     -net id=idprh29y,macaddr=9a:49:4a:4b:4c:4d,model=virtio,type=nic,vlan=0  \
     -net tap,vlan=0,fd=24  \

 good cmdline:
     -net nic,vlan=0,model=virtio,macaddr='9a:21:22:23:24:25',id='idxlzqAk' \
     -net tap,vlan=0,fd=24 \

Fix this by making the 'nic' param of the custom QDevice
NO_STRING_EQUAL, and making sure such params come first before others.

CC: Lukas Doktor <ldoktor@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

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

add new qemu block format support in cdrom, like tray, lock,
block file, etc.
Signed-off-by: NMike Qiu <qiudayu@linux.vnet.ibm.com>

add info block support of get_block() and
check_block_locked() in new qemu
Signed-of-by: NMike Qiu <qiudayu@linux.vnet.ibm.com>

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

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

Only Win8 and above wmic will give the results in the right format.
So disable it for lower version of Windows guests
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

When fail to get vm from env file, we will meet following issue:
Traceback (most recent call last):
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 426, in _exec
    _call_test_function(self.execute, *p_args, **p_dargs)
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 853, in _call_test_function
    raise error.UnhandledTestFail(e)
UnhandledTestFail: Unhandled AttributeError: 'NoneType' object has no attribute 'create'    [context: Boot up guest with a floppy]
Traceback (most recent call last):
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 846, in _call_test_function
    return func(*args, **dargs)
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 299, in execute
    postprocess_profiled_run, args, dargs)
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 216, in _call_run_once
    *args, **dargs)
  File "/root/ypu/autotest_upstream_master/client/shared/test.py", line 322, in run_once_profiling
    self.run_once(*args, **dargs)
  File "/root/ypu/autotest_upstream_master/client/tests/virt/virt.py", line 133, in run_once
    run_func(self, params, env)
  File "/root/ypu/autotest_upstream_master/client/shared/error.py", line 138, in new_fn
    return fn(*args, **kwargs)
  File "/root/ypu/autotest_upstream_master/client/tests/virt/qemu/tests/floppy.py", line 33, in run_floppy
    vm.create()
AttributeError: 'NoneType' object has no attribute 'create'

Call env_process.preprocess_vm(test, params, env, vm_name) to create vm could skip this issue.
Signed-off-by: NFeng Yang <fyang@redhat.com>

changes from v1:
    remove non-related changes

enable serial console in preseed for ubuntu guest
Signed-off-by: NXu Tian <xutian@redhat.com>

This adds a new test function to the cpuid.py test module, where the
CPUID data exposed by QEMU is compared with a full CPUID dump extracted
previously.

The new test code ensures that there are no ABI changes for a given
machine-type + CPU model combination, between QEMU versions.

The reference dumps were generated using the build-qemu-and-dump-cpus,
and copied using the copy-ref-dumps script.

There are some known QEMU bugs and past ABI changes that require
ignoring some differences. Those cases handled as rules that set
ignore_cpuid_leaves on qemu_cpu.cfg.
Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>

Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>

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

The messages assume images will be copied from an
NFS mount in case of unattended install failure,
when said mount may not even exist. Let's fix the
message and notify the image copy only if the grid
owners actually configured it to do so.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

The namespace autotest.virt.tests is not available
when running virt-tests from outside autotest. So
move image_copy to utils_test, and use it from there.

This commit moves the code to the right places and
changes the imports accordingly.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

The param image_copy_on_error tells virt-test that
when an unattended install fails, copy an already
installed guest image from a NFS mount. However,
not all test grids will have such a NFS mount configured,
so just assuming by default that such a mount is configured
is wrong.

This param was introduced on a previous commit, and it
is not available anywhere in the default config files.
Add it and set it by default to "no".
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

remove useless script block_stream_with_stress, since
block_stream_stress in used by block_stream.with_stress tests;
Signed-off-by: NXu Tian <xutian@redhat.com>

updated test script since case steps changed;
Signed-off-by: NXu Tian <xutian@redhat.com>

 virttest.utils_net : modify the method of get_linux_ifname

virt.test.nic_promisc : send command safely

This patch add script for multiple nics stress test, both support linux
and windows
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>