Signed-off-by: NXiaoqing Wei <xwei@redhat.com>

This is an attempt to better fix the issue where system_output
doesn't have the verbose parameter in autotest 0.14. It is better
and less desruptive to simply lose the parameter, and we can
get it back once 0.15 rpms are available on Fedora.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

This reverts commit c080994ddf01914b0bb0162cbe97df9fef75adf3.

Long story short, pylint 1.0 drops some options and adds new ones.
This patch checks the pylint version and adds the formatting option
accordingly.

Unfortunately, pylint's documentation is really broken. The default
message format template is documented as:

--msg-template='{sigle}:{line:3d},{column}: {obj}: {msg}'

But the 'sigle' parameter simply does not exist. This patch uses
'{msg_id}' istead.

One thing this patch does not solve, though, is the conditional
formatting for the '{obj}' parameter, so you may notice a white
space followed by ':' in some of the messages (the ones that lack
reference to an object).

CC: Chris Evich <cevich@redhat.com>
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Use Monitor.info_block() which covers all versions of "info block"
command to get the information.

Also update the cases which use this function to provide info_block
output instead the Monitor.info('block') one.

Last but not least add unittest.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

As there is more and more versions of "info block" this patch
parses the output and creates dictionary of each disk params
(dict of dicts).

Currently supported methods are QMP, HMP old qemu and HMP qemu 1.5.
QMP representation contains the most informations, HMP qemu 1.5 the
least. The missing informations should have default value and it's
upto the user to set it. (use dict.get(key, default))
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

HMP monitor accepts debug for monitor.info() and monitor.query()
commands. QMP should support the same arguments.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

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

QMP returns "info qtree" with trailing '\r', which causes qtree_parser
to accept id = ". Strip those characters to avoid this problem.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

virtio-scsi-pci device was renamed to virtio-scsi-device in
qemu-1.5.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Instead of checking disk device (ide-hd), check the superior bus
(ich9-ahci) to verify autotest drive_format.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

As we use COUNTERS to manager VirshSession in VirshPersistent.
SESSION_COUNTER is no needed any more.
* remove SESSION_COUNTER from VirshPersistent.
* update virsh_unittest related with SESSION_COUNTER.
Signed-off-by: Nyangdongsheng <yangds.fnst@cn.fujitsu.com>

Follow the way used to add tapfds when creating vhostfd.
This can help us try to reuse vhostfd instead of getting a new
vhostfd every time.
Release vhostfd.
Support multi queue
Signed-off-by: NFeng Yang <fyang@redhat.com>

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

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

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

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

1.When a guest is shutoff, user can insert/update a source file
  without source path, but the guest cannot be started
2.When a guest is started, user cannot insert/update a source file
  without source path.
Signed-off-by: NLi Yang <liyang.fnst@cn.fujitsu.com>

Apperently drive_format doesn't have to be defined. Add default
value 'virtio_blk' to avoid using .startswith on a None object.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

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

use live_snapshot function function in test framework, drop useless
module and update test configuration;
Signed-off-by: NXu Tian <xutian@redhat.com>

My commit 756a1ff9a08ff2b1a16202a32e667b56204accc5 broke
unattended install with URL setting. This commit fixes
it in a very simplistic way, since I was tired to dig
into qemu devices code to try out adding unit=0/unit=1
suggested by Paolo, and also, this change would require
more testing with other qemu branches, such as RHEL5/RHEL6
before we know it is safe.

CC: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

run uses for setting vhost to network netdev_extra_params.
This param has to be used with ',' in-front of network param when
it is used for standard thinks. In future it should be replaced by
new vhost virt_net interface.
Signed-off-by: NJiří Župka <jzupka@redhat.com>

Signed-off-by: NLi Yang <liyang.fnst@cn.fujitsu.com>

rename variable id to plug_cpu_id to avoid overwriting python build-in
function.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Signed-off-by: NYu Mingfei <yumingfei@cn.fujitsu.com>

Signed-off-by: NYu Mingfei <yumingfei@cn.fujitsu.com>

We do not want to fail the case for exception in clean up command.
Signed-off-by: NFeng Yang <fyang@redhat.com>

Support ignore_all_errors parameter in cmd()
When run some clean up command in guest. We may do not want to fail the case
for any exception in the command.
So add ignore_all_errors parameter in cmd() to ignore exception.
Signed-off-by: NFeng Yang <fyang@redhat.com>

glibc-devel and glibc-static is requried by build libhugetlbfs source;
Signed-off-by: NXu Tian <xutian@redhat.com>

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

Add catch_serial_cmd for virtio_scsi, ide block device;
Signed-off-by: NXu Tian <xutian@redhat.com>

In some places of our code, we're calling utils.system_output
with a parameter verbose, introduced in autotest 0.15. Therefore,
for people using the Fedora/RHEL shipped RPMs, a bug will
happen since the API on their RPMs doesn't know about the verbose
param.

Therefore, remove such calls and get the equivalent construction
using the utils.run API, that does support the parameter.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

When last device should be appended to usb bus, qemu automatically
add usb-hub to keep free ports for additional devices. We don't want
this in autotest as all tests use strict amount of usb buses. By
enforcing the last item's port we avoid this behavior.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Correct bad English spelling and typos.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

1). update pre_command: used scsi_debug parm to add_host instead write
sys filesystem to avoid device create delay;

2). fix passthrough scsi device cache to writethrough to avoid fail to
open /dev/sg* when drive_cache=none; do this because /dev/sg* is
emulated by scsI_debug module, backend of these device is memery, open
with O_DRIECT is not supported;

3). removed useless import module 'qemu_monitor'
Signed-off-by: NXu Tian <xutian@redhat.com>

I don't know why 486 was being excluded, as the CPUID instruction really
works, and QEMU has CPUID data being set for 486, too. The tests keep
working after removing the "no 486" line from the config file.
Signed-off-by: NEduardo Habkost <ehabkost@redhat.com>

Older qemu versions (RHEL6) doesn't support ide-hd and ide-cd.
Use ide-drive instead.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Add cpu_hotplug test support in autotest.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

VM paused after install finished because of option '-no-shutdown', and
env.get_vm() got a paused VM, so case will fail at wait_for_login() when
need_install is 'yes';
Signed-off-by: NXu Tian <xutian@redhat.com>