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

add new case boot_order_check

changes from v1:
1) update time.sleep to wait_for()
2) add some comments
3) modify replace ascii char(backspace)

changes from v2:
1) fix check whether guest is booted

changes from v3:
1) move boot_fail_infos to the cfg files
2) update the check branch

changes from v4:
1) update the TestFail info

changes from v5:
1) check the boot order by boot_fail_infos

changes from v6:
1) get the nic_addr from 'info pci'

chages from v7:
1) add func to get nic_addr from nic_device_id
Signed-off-by: NShuping Cui <scui@redhat.com>
Acked-by: NFeng Yang <fyang@redhat.com>

Update the method for get pci address for nic deivce and fix some
minor problems.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

Hotplug one nonexistent cpus number
Signed-off-by: NXiaoqing Wei <xwei@redhat.com>

hotplug 159 vcpus and check vcpu number inside and reboot guest

Changelog from V3:
1) removed open/close session code.
2) use vm.monitor.send_args_cmd to auto choose monitor
3) use utils_misc.wait_for() to replace time.sleep()
Changelog from V2:
1) add all error.context with logging.info
Changelog from V1:
1) fixed the hardcode log
2) make the error.context with 'logging.info' argument
3) update the docstring
4) add try / except to the hotplug code
5) sync the log with test steps
6) added 10 Secs sleep time as Win guest needs it
Signed-off-by: NXiaoqing Wei <xwei@redhat.com>
Acked-by: NFeng Yang <fyang@redhat.com>

Merge the old scripts into cpu_hotplug

And cpu hotplug may support in qemu 1.5, so remove the manually
set cpu online inside guest related code.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

This patch make virtual_nic_sendbuffer test support windows guest.

For Windows guests:

1. Download udt.sdk.4.8.win32.zip.
   link: http://sourceforge.net/projects/udt/files/udt/4.8/udt.sdk.4.8.win32.zip
2. Unzip it, and put the file in your guest disk or cdrom.
3. Make sure the program can run normally in your guest.

Linux guest:
    make the udt_url is a valid git repo
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Measure the boot up time of RHEL guest when it run level is 1.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>
Acked-by: Golita Yue<gyue@redhat.com>

get_image_filename have been moved to virt_storage.
And we have moved image related script to virt_storage.

Changes from v2:
Delete duplicated module
Changes from v1:
Import missing kvm_storage in live_snapshot_chain.py
Signed-off-by: NFeng Yang <fyang@redhat.com>

All image related functions is moved to storage class. Update it
in boot_time.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

backup image before boot_time test and restore image after it.

Delete image snapshot part code and simplify the code.
Signed-off-by: NFeng Yang <fyang@redhat.com>

Update method for set and resotre run level. Also drop cache before
test the boot time.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

This patch adds a new test that verifies whether the
host allows tracing of exception injection in KVM.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

(1) Use qemu-img to create an unbootable image.
(2) Start vm with the image created by step 1.
(3) Use ps get qemu-kvm process %cpu, if greater
    than 90%, report fail.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>
Signed-off-by: NYiqiao Pu <ypu@redhat.com>
Signed-off-by: NSuqin Huang <shuang@redhat.com>
Signed-off-by: Nsshang <sshang@redhat.com>

Verify that qemu ROMS are loaded once, rather than multiple times.

changes from v1:
  1) Modified check function to set().

changes from v2:
  1) Add some comments.
  2) Delete the qmp_cmd parameter.

changes from v3:
  1) update some error contexts
  2) remove the for loop

changes from v4:
  1) make re string configure.
  2) update the Fail message

changes from v5:
  1) add check available for fw_filter and addr_filter parameter
Signed-off-by: NShuping Cui <scui@redhat.com>

This testcase which itself is doing nothing is used for dealing with VM(s)
via framework, i.e. create_image, destroy VM(s) etc.
Signed-off-by: NAmos Kong <akong@redhat.com>

This patch try to fix all the tab and trailing space problems.
Signed-off-by: NJason Wang <jasowang@redhat.com>

Use taskset to set the CPU affinity of tsc read program in both guest and host.
Make the code easy to read and drop the tsc list for both guest and host.
Signed-off-by: NAmos Kong <akong@redhat.com>

Use more reasonable varialbe name and optimize the code path.
Signed-off-by: NJason Wang <jasowang@redhat.com>

Convert tsc_drift ratio to absolute value
If tsc_drift ratio is a large negative, case could not raise exception.
Signed-off-by: NAmos Kong <akong@redhat.com>

Filter integer from the return value of get_tsc

If the return value of get_tsc() contains other string, it will fail to convert
the return value to float. Use re.findall('(\d+)',o)[0] to filter tsc value.
Signed-off-by: NAmos Kong <akong@redhat.com>

tsc_drift: Fix a bug of set CPU affinity

We need set process running on a single processor. There is a misunderstand of
mask and cpulist.
 "taskset 1 command": command will execute on CPU0 or CPU1, not only CPU1
 "taskset -c 1 command": command will only execute on CPU1
Signed-off-by: NAmos Kong <akong@redhat.com>
Reported-by: NShuxi Shang <sshang@redhat.com>

Update interface to fit the new framework
Signed-off-by: NFeng Yang <fyang@redhat.com>
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

Currently, qemu.convert.to_qed and qemu.convert.to_raw will
be skip, and the log is:

17:37:31 INFO | SKIP qcow2.virtio_blk.smp2.virtio_net.
Fedora.18.x86_64.qemu_img.convert.to_qed -> ParamNotFound:
Mandatory parameter 'compressed' is missing.
Check your cfg files for typos/mistakes

This is because in convert test, it will use params["compressed"]
and params["encrypted"], but for to_qed and to_raw, the two params
does not exist. In old version virttest, it will be OK, as it use
params.get("compressed"), and this will return 'None', the test can
run safely.

But in the new version, it will just SKIP due to no param 'compressed'.

This patch fixes problem with these two cases.
Signed-off-by: NMike Qiu <qiudayu@linux.vnet.ibm.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

No actual changes.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The main monitor name is overriden to hmp1 and causes test failures...
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Use VM.get_monitory_by_type() function to get QMP monitor(s).
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Now use test.bindir will not find the image file, just update it
to new interface data_dir.get_data_dir
Signed-off-by: NMike Qiu <qiudayu@linux.vnet.ibm.com>

This test checks groups when guest is started as
nobody [v2] using qemu -runas option to exec the qemu.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Add heavy disk write load test for windows guests. Using Crystal
Disk Mark [1] to generate the write load inside the guest. And set the
default cache to writeback in this test.

changes from v1:
  - Record steps in INFO logs
  - Close the session after test
  - Add command empty check

[1] Crystal Disk Mark is a BSD licensed program
    http://crystalmark.info/software/CrystalDiskMark/manual-en/License.html

    It can be downloaded here:
    http://crystalmark.info/software/CrystalDiskMark/index-e.html

    The file will be added to the winutils.iso
Signed-off-by: NYiqiao Pu <ypu@redhat.com>
Acked-by: NFeng Yang <fyang@redhat.com>

The previous patch introduced a bug in the qemu-ga.exe path.
This patch tries to fix this problem.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

The previous patches for adding windows support of qemu-guest-agent
depend on a guest image in which user has installed necessary DLL
files.

This patch tries to install these DLLs by test script, so that to
we can test this feature with a fresh windows system.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

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

This test will simply create a file system on disk, mount it and write
a file with some content to the disk. Check whether the write could be
succeeded.
Signed-off-by: Nsshang <sshang@redhat.com>

Checking guest os type with 'guest_name' parameter may not
always work for all guests, and not supported by upstream.
Thus update all these places to using 'os_type'.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

In the format disk test, the status of the harddisk is different in different
env in Windows guest. Should check if the disk plused is offline and writeable.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

This patch try to fix all the tab and trailing space problems.
Signed-off-by: NJason Wang <jasowang@redhat.com>

Make command timeout of format_disk.py scirpt longer and configurable
Signed-off-by: NFeng Yang <fyang@redhat.com>

Some test code in this script was deleted by accident in
commit 2b5e0f6151419dd403d7f959e0492510f790b8e0. This patch
brings them back.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

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

This case is used for performance testing.

Changes for V2
1. Only update gcc when testing on host.
2. Add record_result() function to record the result value to a file.
Changes for V3:
Clean the folder used to make kernel
Signed-off-by: NFeng Yang <fyang@redhat.com>
Acked-by: NQingtang Zhou <qzhou@redhat.com>
Acked-by: NAmos Kong <akong@redhat.com>

This case will check some properties bits for virtio devices set in
the command line. This script will support both block and nic devices.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

Add case for install cygwin env in windows guests.

change from v1:
   - Add cdrom letter replace to fit different system
   - typo fix and parameter adjust
change from v2:
   - cdrom check command change and the install command update
Signed-off-by: NYiqiao Pu <ypu@redhat.com>
Acked-by: NQingtang Zhou <qzhou@redhat.com>

Adapt to final version which was committed to qemu upstream.
Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>

Design this case according to manual case

[x2APIC] switch kernel with kexec

This case check that x2apic is still enabled after kexec.
Signed-off-by: NFeng Yang <fyang@redhat.com>

This case test QEMU Monitor Protocol Events.
It verifies if the appropriate qmp event
will be emitted when the required conditions
are met.

It tests the following events now:
RESET SHUTDOWN STOP RESUME WATCHDOG.

All tests follow the qmp commands:
system_reset, quit, system_powerdown.
Signed-off-by: NFeng Yang <fyang@redhat.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

This patch adds one more login step for qemu guest agent
reboot testing, to verify if the guest works well after
reboot.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

It is a fairly frequent task to pick up the current
qemu binary being used at a time, and as time passed,
we had a proliferation of mechanisms to do it, slightly
similar to each other. Introduce get_qemu_binary,
get_qemu_img_binary() and get_qemu_io_binary(), utility
functions that get the path of each of the QEMU binaries
for use in tests and utility functions. Use the new
functions across virt test.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

1). boot guest with cpu_model in the cluster one by one
2). check flags in guest if flag check is required, otherwise shutdown guest

Change from v1:
use cpu_model_%s instead of intel_model/amd_model
Signed-off-by: NSuqin Huang <shuang@redhat.com>
Acked-by: NFeng Yang <fyang@redhat.com>

    1. boot guest with -cpu model,+extra_flags (extra_flags is optional)
       a. no defined model_name in cfg file
          guest_model = host_model
       b. model_name defined in cfg file
          guest_model = params.get("cpu_model")
    2. get guest flags
    3. get expected model flags from dump file
       a. -cpu host: qemu_model = host_model
       b. guest_model > host_model --> expected_model = host_model
          e.g guest_model = Haswell, host_model = Sandybridge
              expected_model = Sandybridge
       c. guest_model < host_model --> expected_model = guest_model
    4. get extra flags
       a. add_flags = +flag
          1). flag is exposed to guest if it's supported in host
          2). flag is not supported to guest if it's unknown in host
          3). ignore "check", "enforce" which are params not flag
       b. del_flags = -flag
          flag is removed if it's supported in guest
       c. params check: check lack flag in host include unknow flag
    5. compare expected flag with flags in guest
       a. out_flags: not supported with some conf, this kinds of flag
          will be displayed in dump file, but not in guest.
          e.g tsc-dedline is not supported with -M rhel6.3.0
       b. option_flags: some flag is generated by kernel which is not
          defined in dump file. it's acceptable when display in guest.
          e.g rep_good
       expected_flags = expected_model_flags + add_flags - del_flags - out_flags
       miss_flag = expected_flags - guest_flags
       unexpect_flag = guest_flags - expected_flags - option_flags

Change v1:
1). add model_info check
2). improve for loop
3). put sub-func at the top

Change from v2:
1). move qemu_binary to function get_all_support_flags()
2). add flags value check

Change from v3:
Add one more blank line between functions

Change from v4:
change more_flags to unexpect_flag
improve get_extra_flags function

Change from V5:
1). remove unused model
2). add try, except for m.groups()[0]
3). result format of -cpu ? are diff between rhel6, and rhel7,
    set pattern in config file

Change from V6:
1). update flag.groups() to flag_list.groups()
2). add blank space between "" and ,
Signed-off-by: NSuqin Huang <shuang@redhat.com>
Acked-by: NXu Tian <xutian@redhat.com>

Update the dump_file dir to make it accept local file. And update
some flags should in option_flags.
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

Add five live snapshot cases have similar process:
- create live snapshot when guest paused
- restore to previous state when guest become crash
- restore live snapshot after guest update
- restore live snapshot chain 2-restore to base
- restore to parent snapshot node
Signed-off-by: NYiqiao Pu <ypu@redhat.com>

Update interface to date
Signed-off-by: NYiqiao Pu <ypu@redhat.com>
Signed-off-by: NFeng Yang <fyang@redhat.com>