Re: [Autotest] [PATCH] Add a subtest pci_hotplug in kvm test

[Yolkfull Chow <yzhou@redhat.com>]

On 07/08/2009 09:51 AM, Lucas Meneghel Rodrigues wrote:
> I've spent some time doing a second review and test of the code.
> During my tests:
>
>   * I found some problems with PCI hotplug itself and would like help
> to figure out why things are not working as expected.
>   * Made suggestions regarding the phrasing of the error messages
> thrown by the test. Mostly nipticking. Let me know if you think the
> new messages make sense.
>   * The order of the final test steps looks a bit weird to me
>
> Comments follow.
>
> On Fri, Jul 3, 2009 at 3:00 AM, Yolkfull Chow<yzhou@redhat.com>  wrote:
>    
>> On 07/03/2009 01:57 PM, Yolkfull Chow wrote:
>>      
>>> Signed-off-by: Yolkfull Chow<yzhou@redhat.com>
>>> ---
>>>    client/tests/kvm/kvm.py               |    1 +
>>>    client/tests/kvm/kvm_tests.cfg.sample |   65 ++++++++++++++++++++++-
>>>    client/tests/kvm/kvm_tests.py         |   94 +++++++++++++++++++++++++++++++++
>>>    client/tests/kvm/kvm_vm.py            |    2 +
>>>    4 files changed, 161 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/client/tests/kvm/kvm.py b/client/tests/kvm/kvm.py
>>> index b18b643..fc92e10 100644
>>> --- a/client/tests/kvm/kvm.py
>>> +++ b/client/tests/kvm/kvm.py
>>> @@ -55,6 +55,7 @@ class kvm(test.test):
>>>                    "kvm_install":  test_routine("kvm_install", "run_kvm_install"),
>>>                    "linux_s3":     test_routine("kvm_tests", "run_linux_s3"),
>>>                    "stress_boot":  test_routine("kvm_tests", "run_stress_boot"),
>>> +                "pci_hotplug":  test_routine("kvm_tests", "run_pci_hotplug"),
>>>                    }
>>>
>>>            # Make it possible to import modules from the test's bindir
>>> diff --git a/client/tests/kvm/kvm_tests.cfg.sample b/client/tests/kvm/kvm_tests.cfg.sample
>>> index 2f864de..a9e16d6 100644
>>> --- a/client/tests/kvm/kvm_tests.cfg.sample
>>> +++ b/client/tests/kvm/kvm_tests.cfg.sample
>>> @@ -94,6 +94,53 @@ variants:
>>>            max_vms = 5
>>>            alive_test_cmd = ps aux
>>>
>>> +
>>> +    - nic_hotplug:
>>> +        type = pci_hotplug
>>> +        pci_type = nic
>>> +        modprobe_acpiphp = yes
>>> +        reference_cmd = lspci
>>> +        find_pci_cmd = 'lspci | tail -n1'
>>>        
> I tried block device hotplug, lspci doesn't show up the newly added
> devices. Already tried with F8 and F9. Any idea why?
>    
   It doesn't need to modprobe acpiphp on Fedora, also, both F8 and F9 
don't have virtio & virtio_ring modules.  I ran it on guest Fedora-11 
and Win2008,  both ran successfully:

# ./scan_results.py
test                                                    status    
seconds    info
----                                                    ------    
-------    ----
Fedora.11.32.nic_hotplug.nic_rtl8139                            GOOD    
68    completed successfully
Fedora.11.32.nic_hotplug.nic_virtio                     GOOD    46    
completed successfully
Fedora.11.32.block_hotplug.fmt_qcow2.block_virtio        GOOD    46    
completed successfully
Fedora.11.32.block_hotplug.fmt_qcow2.block_scsi         GOOD    44    
completed successfully
Fedora.11.32.block_hotplug.fmt_raw.block_virtio         GOOD    45    
completed successfully
Fedora.11.32.block_hotplug.fmt_raw.block_scsi           GOOD    46    
completed successfully
Win2008.32.nic_hotplug.nic_rtl8139                              GOOD    
66    completed successfully
Win2008.32.block_hotplug.fmt_qcow2.block_scsi           GOOD    186    
completed successfully
Win2008.32.block_hotplug.fmt_raw.block_scsi             GOOD    71    
completed successfully

>    
>>> +        pci_test_cmd = 'nslookupwww.redhat.com'
>>> +        seconds_wait_for_device_install = 3
>>> +        variants:
>>> +            - @nic_8139:
>>> +                pci_model = rtl8139
>>> +                match_string = "8139"
>>> +            - nic_virtio:
>>> +                pci_model = virtio
>>> +                match_string = "Virtio network device"
>>> +            - nic_e1000:
>>> +                pci_model = e1000
>>> +                match_string = "Gigabit Ethernet Controller"
>>>        
> Pretty much all block hotplug 'guest side check' is failing during the
> stage where the output  of lspci | tail -n1 is being compared with the
> match strings. Hypervisor is qemu 0.10.5 (kvm-87 upstream).
>
>    
>>> +    - block_hotplug:
>>> +        type = pci_hotplug
>>> +        pci_type = block
>>> +        modprobe_acpiphp = yes
>>> +        reference_cmd = lspci
>>> +        find_pci_cmd = 'lspci | tail -n1'
>>> +        images += " stg"
>>> +        boot_drive_stg = no
>>> +        image_name_stg = storage
>>> +        image_size = 1G
>>> +        force_create_image_stg = yes
>>> +        pci_test_cmd = 'dir'
>>> +        seconds_wait_for_device_install = 3
>>> +        variants:
>>> +            - block_virtio:
>>> +                pci_model = virtio
>>> +                match_string = "Virtio block device"
>>> +            - block_scsi:
>>> +                pci_model = scsi
>>> +                match_string = "SCSI storage controller"
>>> +        variants:
>>> +            - fmt_qcow2:
>>> +                image_format_stg = qcow2
>>> +            - fmt_raw:
>>> +                image_format_stg = raw
>>> +
>>> +
>>>    # NICs
>>>    variants:
>>>        - @rtl8139:
>>> @@ -306,6 +353,22 @@ variants:
>>>                migration_test_command = ver&&    vol
>>>            stress_boot:
>>>                alive_test_cmd = systeminfo
>>> +        nic_hotplug:
>>> +            modprobe_acpiphp = no
>>> +            reference_cmd = systeminfo
>>> +            seconds_wait_for_device_install = 10
>>> +            find_pci_cmd = ipconfig /all | find "Description"
>>> +            nic_e1000:
>>> +                match_string = "Intel(R) PRO/1000 MT Network Connection"
>>> +        block_hotplug:
>>> +            use_telnet = yes
>>> +            modprobe_acpiphp = no
>>> +            reference_cmd = wmic diskdrive
>>> +            find_pci_cmd = wmic diskdrive | find "disk drives"
>>> +            seconds_wait_for_device_install = 10
>>> +            only block_scsi
>>> +            block_scsi:
>>> +                match_string = "SCSI"
>>>
>>>        
>> It supports Windows block_hotplug now. But we need use_telnet to login
>> Windows OS since command 'wmic' could only run on telnet session.
>>      
>>>            variants:
>>>                - Win2000:
>>> @@ -571,4 +634,4 @@ variants:
>>>            only rtl8139
>>>
>>>    # Choose your test list
>>> -only fc8_quick
>>> +#only fc8_quick
>>> diff --git a/client/tests/kvm/kvm_tests.py b/client/tests/kvm/kvm_tests.py
>>> index 2d11fed..230385a 100644
>>> --- a/client/tests/kvm/kvm_tests.py
>>> +++ b/client/tests/kvm/kvm_tests.py
>>> @@ -585,3 +585,97 @@ def run_stress_boot(tests, params, env):
>>>            for se in sessions:
>>>                se.close()
>>>            logging.info("Total number booted: %d" % (num -1))
>>> +
>>> +
>>> +def run_pci_hotplug(test, params, env):
>>> +    """
>>> +    Test pci devices' hotplug
>>> +    1) pci_add a deivce (nic or storage)
>>> +    2) Compare 'info pci' output
>>> +    3) Compare $reference_cmd output
>>> +    4) Verify whether pci_model is shown in $pci_find_cmd
>>> +    5) pci_del the device, verify whether could remove the pci device
>>>        
> What about if the last steps were
>
> 5) Verify if the newly added devices are working properly
>    
Had worked out how to verify the newly added block device. Will send it 
here right now. Any suggestion about how to verify newly added nic card 
? So far I just keep
  'nslookup www.redhat.com' to check network simply.
> 6) pci_del the device, verify whether could remove the pci device
>
> 6) is already happening in the code, but in reverse order than mentioned above.
>
>    
>>> +    @param test:   kvm test object
>>> +    @param params: Dictionary with the test parameters
>>> +    @param env:    Dictionary with test environment.
>>> +    """
>>> +    vm = kvm_utils.env_get_vm(env, params.get("main_vm"))
>>> +    if not vm:
>>> +        raise error.TestError("VM object not found in environment")
>>> +    if not vm.is_alive():
>>> +        raise error.TestError("VM seems to be dead; Test requires a living VM")
>>> +
>>> +    logging.info("Waiting for guest to be up...")
>>> +
>>> +    session = kvm_utils.wait_for(vm.ssh_login, 240, 0, 2)
>>> +    if not session:
>>> +        raise error.TestFail("Could not log into guest")
>>> +
>>> +    logging.info("Logged in")
>>> +
>>> +    # modprobe the module that enable hotplug
>>> +    if params.get("modprobe_acpiphp") == "yes":
>>> +        if session.get_command_status("modprobe acpiphp"):
>>> +            raise error.TestError("Modprobe module 'acpiphp' failed")
>>> +
>>> +    # get reference output
>>> +    s, info_pci_ref = vm.send_monitor_cmd("info pci")
>>> +
>>> +    # compare the output of `reference_cmd`
>>> +    ref_cmd = params.get("reference_cmd")
>>> +    reference = session.get_command_output(ref_cmd)
>>> +
>>> +    # implement pci hotplug
>>> +    tested_model = params.get("pci_model")
>>> +    logging.info("Testing hotplug pci device:%s" % tested_model)
>>> +
>>> +    test_type = params.get("pci_type")
>>> +    if test_type == "nic":
>>> +        pci_add_cmd = "pci_add pci_addr=auto nic model=%s" % tested_model
>>> +        s, add_output = vm.send_monitor_cmd(pci_add_cmd)
>>>        
> The command
>
> s, add_output = vm.send_monitor_cmd(pci_add_cmd)
>
> Can be moved out the if blocks since it will be executed in both
> paths, we save one line of code with this (and it's cleaner anyway).
>
>    
>>> +    elif test_type == "block":
>>> +        image_name = params.get("image_name_stg")
>>> +        image_format = params.get("image_format_stg", "qcow2")
>>> +        image_filename = "%s.%s" % (image_name, image_format)
>>> +        image_dir = os.path.join(test.bindir, "images")
>>> +        storage_name = os.path.join(image_dir, image_filename)
>>> +        pci_add_cmd = "pci_add pci_addr=auto storage file=%s,if=%s" % \
>>> +                                              (storage_name, tested_model)
>>>        
> The above line would be better with implicit line continuation
>
> pci_add_cmd = ("pci_add pci_addr=auto storage file=%s,if=%s" %
>                          (storage_name, tested_model))
>
>    
>>> +        s, add_output = vm.send_monitor_cmd(pci_add_cmd)
>>> +
>>> +    if not "OK domain" in add_output:
>>> +        raise error.TestFail("Command failed: %s" % pci_add_cmd)
>>>        
> Since we're executing a command on the hypervisor to add the device, a
> better way to put the above message would be
>
>      if not "OK domain" in add_output:
>          raise error.TestFail("Hypervisor command to add device '%s'failed."
>                               "Output: %s" % (pci_add_cmd, add_output))
>
>    
>>> +    # compare the output of 'info pci'
>>> +    s, after_add = vm.send_monitor_cmd("info pci")
>>> +    if after_add == info_pci_ref:
>>> +        raise error.TestFail("No new pci device shown in 'info pci'")
>>>        
> Another suggestion:
>
>          raise error.TestFail("No new PCI device shown after executing "
>                               "hypervisor command 'info pci'")
>
>
>    
>>> +    time.sleep(int(params.get("seconds_wait_for_device_install")))
>>> +
>>> +    o = session.get_command_output(ref_cmd)
>>> +    if reference == o:
>>> +        raise error.TestFail("No new device shown in cmd: %s" % ref_cmd)
>>>        
> I found useful to explain on which side of the check the test fail,
> 'hypervisor', or 'guest'
>
>          raise error.TestFail("No new device shown on guest command %s output" %
>                               ref_cmd)
>
>
>    
>>> +    cmd = params.get("find_pci_cmd")
>>> +    output = session.get_command_output(cmd)
>>> +    if not params.get("match_string") in output:
>>> +        raise error.TestFail("Not found pci model: %s; Command is: %s" %
>>> +                                                    (tested_model, cmd))
>>>        
>          raise error.TestFail("PCI device model '%s' not found on host "
>                               "command '%s' ouput" % (tested_model, cmd))
>
>    
>>> +    # del pci device
>>> +    slot_id = "0" + add_output.split(",")[2].split()[1]
>>> +    cmd = "pci_del pci_addr=%s" % slot_id
>>> +    s, after_del = vm.send_monitor_cmd(cmd)
>>> +    if after_del == after_add:
>>> +        raise error.TestFail("Failed to hot remove pci device:%s; Command:%s" %
>>> +                                                     (tested_model, cmd))
>>>        
>          raise error.TestFail("Failed to hot remove pci device '%s'. "
>                               "Hypervisor command: '%s'" % (tested_model, cmd))
>
> Hot remove doesn't work for me either... Since we are here, I still
> didn't understand why we do a hot remove of a device and right below
> it we try to verify if the newly added devices are working properly.
> Shouldn't be the other way around (ie checking and then hot removing)?
>    

It did actually not verify the newly added device but simply verify if 
pci_del bring in any problem to the guest (disk or network). Since 
sometimes I found pci_del a nic card can break the guest's network as 
tested in our internal kvm version.

>    
>>> +    # check whether VM's network&    disk work fine
>>> +    if session.get_command_status(params.get("pci_test_cmd")):
>>> +        raise error.TestFail("Check for %s device failed after PCI hotplug" %
>>> +                                                                 test_type)
>>> +
>>> +    session.close()
>>> diff --git a/client/tests/kvm/kvm_vm.py b/client/tests/kvm/kvm_vm.py
>>> index 503f636..95b55eb 100644
>>> --- a/client/tests/kvm/kvm_vm.py
>>> +++ b/client/tests/kvm/kvm_vm.py
>>> @@ -239,6 +239,8 @@ class VM:
>>>
>>>            for image_name in kvm_utils.get_sub_dict_names(params, "images"):
>>>                image_params = kvm_utils.get_sub_dict(params, image_name)
>>> +            if image_params.get("boot_drive") == "no":
>>> +                continue
>>>                qemu_cmd += " -drive file=%s" % get_image_filename(image_params,
>>>                                                                   image_dir)
>>>                if image_params.get("drive_format"):
>>>        


-- 
Yolkfull
Regards,