In order to distinguish better between a legitimate
TestFail (the test did fail) and a TestError
(unexpected condition, test bug or misconfiguration),
print a different ERROR state when error.TestError
is raised.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Autotest upstream renamed one particular API:

global_config -> settings

In order to make it cleaner. As we are still in the process
to make autotest 0.15, let's handle gracefully both
libraries installed in the system. This way, we can resume
regression testing for the next branches of both projects.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

The guest might not get ip address sometimes, thus let's record
tcpdump output as well, so that we can debug network problem.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

The list of CPU models is the same on RHEL-6 and QEMU-1.3.
Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>

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

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

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

The duplication of settings makes it very hard to spot mistakes. This is
a new version of the config file that uses nesting and cartesian
products of variants to avoid duplication.

To make it easier to test wheter the changes are working, this generates
exactly the same dictionaries as the original config, in exactly the
same sequence, including all the bugs. Look for the #FIXME comments for
places that will be cleaned up or fixed later.
Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * s/TestFail/TestNAError/
    * s/custom_model_id/custom.model_id/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * s/TestFail/TestNAError/
    * s/custom_xlevel/custom.xlevel/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
   * s/TestFail/TestNAError/
   * s/custom_stepping/custom.stepping/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
v2:
   * s/TestFail/TestNAError/
   * s/custom_model/custom.model/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * s/TestFail/TestNAError/
    * s/custom_family/custom.family/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * s/TestFail/TestNAError/
    * s/custom_level/custom.level/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * s/TestFail/TestNAError/
    * s/custom_vendor/custom.vendor/ in config

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
v2:
    * remove rhelcpu models form qemu13 test
    * if vendor is missing use  host's vendor
    * remove no needed cpu_models list from kvm test
    * replace cpu_models with shorter ignore_cpu_models in rhel64.kvm.cpuid.default.vendor
    * fix typo
    * s/default_vendor/default.vendor/ in config so we could use
      cpuid.default selector for all default tests

default.vendor fix

... and test utilities to parse its output

CPUID leaves are capped by reported level, xlevel, x2level values.
Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
---
  v3:
    * dump CPUID(EAX=3), CPUID(EAX=4,ECX=3), CPUID(EAX={0xB, 0xC})
    * dump CPUID(EAX=0x80000007) and known paravirt cpuid leaves if
      detected
    * make target name specific to function cpuid_dump_kernel.bin,
      so we could later add another targets for other tests
  v2:
    * log cpuid dump on parsing failure
    * cap by reported level, xlevel, x2level

Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>
---
v2:
  * rebased on top of tree where there is no subtests.cfg.example
  * s/kvm/qemu/ in pathnames and test config
  * Change some errors types to a more suitable ones

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

Signed-off-by: NChris Evich <cevich@redhat.com>

Signed-off-by: NChris Evich <cevich@redhat.com>

Signed-off-by: NChris Evich <cevich@redhat.com>

* data_dir.SubdirList class returns a flat-list of all non-hidden
  subdirectories beneath a base directory (inclusive).

* data_dir.SubdirGlobList class returns flat-list of all files
  matching a glob pattern, within all non-hidden subdirectories
  beneath a base directory (inclusive).
Signed-off-by: NChris Evich <cevich@redhat.com>

Iterate over all allowed cpu_models and in case none is supported,
force the last one.

This should suit negative testing and also it should be safer when
using auto_cpu_model.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Currently, it will rase and error when '/dev/sdaa' match the pattern
of the blacklist. This should not be happen as we just want sda sda1
and so on to be in the blacklist.

add one more check to avoid this error.

Signed-off-by: Mike Qiu<qiudayu@linux.vnet.ibm.com>

Currently, autotest use such as /dev/sda.* to get the disk of
the blacklist, but this will raise an error if the disk 'sdaa'
or some name like this, it's not correct as we just want to add
sda or name like sda1-9 to the blacklist.

change the pattern to /dev/sda[\d]* can avoid this.

Signed-off-by: Mike Qiu<qiudayu@linux.vnet.ibm.com>

Currently, function sorted inside the python was not suitable for
storage, for example, if ls -1d /dev/sd* gets result like this
after split('\n'):
['/dev/sda', '/dev/sda1', '/dev/sda2', '/dev/sda3',
 '/dev/sdaa', '/dev/sdab','/dev/sdac', '/dev/sdad', '/dev/sdae',
 '/dev/sdaf', '/dev/sdag', '/dev/sdah', '/dev/sdai', '/dev/sdaj',
 '/dev/sdak', '/dev/sdal', '/dev/sdam', '/dev/sdan', '/dev/sdao',
 '/dev/sdap', '/dev/sdaq', '/dev/sdar', '/dev/sdas', '/dev/sdb',
 '/dev/sdb1', '/dev/sdb2', '/dev/sdb3', '/dev/sdc', '/dev/sdd',
 '/dev/sde', '/dev/sdf', '/dev/sdg', '/dev/sdg1', '/dev/sdg2',
 '/dev/sdg3', '/dev/sdh', '/dev/sdh1', '/dev/sdh2', '/dev/sdh3',
 '/dev/sdi', '/dev/sdj', '/dev/sdk', '/dev/sdl', '/dev/sdm',
 '/dev/sdn', '/dev/sdo', '/dev/sdp', '/dev/sdq', '/dev/sdr',
 '/dev/sds', '/dev/sdt', '/dev/sdu', '/dev/sdv', '/dev/sdw',
 '/dev/sdx', '/dev/sdy', '/dev/sdz']

it will make no any change if just use sorted. As the default,
it will choose last 9 disk:
['/dev/sdr', '/dev/sds', '/dev/sdt', '/dev/sdu', '/dev/sdv', '/dev/sdw',
 '/dev/sdx', '/dev/sdy', '/dev/sdz']

Sometimes this is very dangerous, for lots of them may be the disk contains
lots of important data. Actually, maybe you just use scsi_debug just create:
['/dev/sdak','/dev/sdal', '/dev/sdam', '/dev/sdan', '/dev/sdao',
 '/dev/sdap', '/dev/sdaq', '/dev/sdar', '/dev/sdas']

But select the wrong disks as mentioned above.

To solve this problem, add param cmp to sorted and get the right order.
Signed-off-by: NMike Qiu <qiudayu@linux.vnet.ibm.com>

maxcpus is a sub-option of -smp, so need a Regex to match
Signed-off-by: NXiaoqing Wei <xwei@redhat.com>

Allow calling vm.get_cpu_count() before guest ssh
running. In this case, calling simply vm.login()
would fail.
Signed-off-by: NXiaoqing Wei <xwei@redhat.com>

qemu_img can be found on a different location than the
images, in case qemu was compiled from source, so let's
fix the reference to the right path when setting the
binary.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Client can suggest IP address which he wants to have.
Autotest check if IP address is accessible, thanks to
it there is no problem with changing of guest network.
Additionally it helps with windows DHCP compatibility.
Signed-off-by: NJiří Župka <jzupka@redhat.com>

* max_disk must not have any usbs, otherwise there are not enough
  slots in PCI
* default image_boot = yes which generates in multi_disk test
  multiple disks with index = 1 and test failure. By forcing default
  value stg_image_boot = 'no' we can keep the default value (and index
  of image1) and let the other disk to have their own indexes.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When we set disable_kvm to 'yes', we obviously don't want
-enable-kvm set regardless of the value of enable_kvm.

So let's fix that. Now you can just set

disable_kvm = yes

And everything else will be taken care of.

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

The implementation of the devices needed for KVM unit
testing changed a bit on qemu:

1) There's no need for a testlog anymore (as kvm-unit-tests
   now writes to the standard serial port)
2) Now there's a shared new device, called isa-debug-exit,
   that when written to, causes qemu to exit
3) Due to the new behavior of the isa-debug-exit device,
   the new known good behavior for qemu at the end of the
   unit test is to return 1.

Take that all into account, while still keeping backwards
compatibility. Also, document how one can run the unittests
with the new virt-tests runner both in the config file
and the wiki.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

For qemu upstream, in order to run the kvm-unit-tests suite,
there are new test devices, that differ in syntax from the
ones present in qemu-kvm:

* pc-testdev
* isa-debug-exit

Add them both to the qemu VM class, and make things
backwards compatible (ie, if an older qemu-kvm is being
tested, it'll handle the correct syntaxes as well).
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

When we introduced qemu CPU model configuration, we opted
for robustness, so we wouldn't add a cpu_model to the
command line in case it was not showing on qemu -cpu help
text. This is bad because:

1) It was all done very silently
2) This makes negative cpu model testing unfeasible

So let's honor the setting, and print a error message,
so people have a better idea of what's going on. In
such cases, the bad param will be passed to qemu, it
will fail and a VMCreateError will be thrown.

CC: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Signed-off-by: NJiří Župka <jzupka@redhat.com>

This is a solution for issue #184.

Create the utility function _restore_stdout, so we
don't need to resort to standalone_utils.print_stdout
on error paths, changing all error paths to use it.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Based on experience and advice we are increasing the default
timeouts. We have seen non intended failures with the timeouts
as they are. Increasing them makes legit tests (migration ones
in our experience) not fail anymore.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Do a better job at logging which VM is being killed,
and which method to kill it was used. Other parts of
the VM classes need cleanup, but let's start with this
method.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>