Re: [PATCH 2/3] tests/acceptance/machine_s390_ccw_virtio: Test virtio-rng via /dev/hwrng

[Thomas Huth]

On 11/12/2020 21.30, Wainer dos Santos Moschetta wrote:
> Hi,
> 
> On 12/11/20 2:31 PM, Thomas Huth wrote:
>> /dev/hwrng is only functional if virtio-rng is working right, so let's
>> add a sanity check for this device node.
> 
> Good idea.
> 
>>
>> Signed-off-by: Thomas Huth <thuth@redhat.com>
>> ---
>>   tests/acceptance/machine_s390_ccw_virtio.py | 17 +++++++++++++++--
>>   1 file changed, 15 insertions(+), 2 deletions(-)
>>
>> diff --git a/tests/acceptance/machine_s390_ccw_virtio.py
>> b/tests/acceptance/machine_s390_ccw_virtio.py
>> index 733a7ca24a..7d0a78139b 100644
>> --- a/tests/acceptance/machine_s390_ccw_virtio.py
>> +++ b/tests/acceptance/machine_s390_ccw_virtio.py
>> @@ -64,9 +64,9 @@ class S390CCWVirtioMachine(Test):
>>                            '-append', kernel_command_line,
>>                            '-device', 'virtio-net-ccw,devno=fe.1.1111',
>>                            '-device',
>> -                         'virtio-rng-ccw,devno=fe.2.0000,max_revision=0',
>> +                        
>> 'virtio-rng-ccw,devno=fe.2.0000,max_revision=0,id=rn1',
>>                            '-device',
>> -                         'virtio-rng-ccw,devno=fe.3.1234,max_revision=2',
>> +                        
>> 'virtio-rng-ccw,devno=fe.3.1234,max_revision=2,id=rn2',
>>                            '-device', 'zpci,uid=5,target=zzz',
>>                            '-device', 'virtio-net-pci,id=zzz',
>>                            '-device', 'zpci,uid=0xa,fid=12,target=serial',
>> @@ -96,6 +96,19 @@ class S390CCWVirtioMachine(Test):
>>           exec_command_and_wait_for_pattern(self,
>>                           'cat
>> /sys/bus/ccw/devices/0.3.1234/virtio?/features',
>>                           virtio_rng_features)
>> +        # check that /dev/hwrng works - and that it's gone after ejecting
>> +        exec_command_and_wait_for_pattern(self,
>> +                        'dd if=/dev/hwrng of=/tmp/out.dat bs=1k count=10',
>> +                        '10+0 records out')
>> +        self.clear_guests_dmesg()
>> +        self.vm.command('device_del', id='rn1')
>> +        self.wait_for_crw_reports()
>> +        self.clear_guests_dmesg()
>> +        self.vm.command('device_del', id='rn2')
>> +        self.wait_for_crw_reports()
>> +        exec_command_and_wait_for_pattern(self,
>> +                        'dd if=/dev/hwrng of=/tmp/out.dat bs=1k count=10',
>> +                        'dd: /dev/hwrng: No such device')
> 
> Maybe the expected pattern is too fragile. On my Fedora 33 system, 'dd' will
> print a different message.

We are running this test with a well-defined kernel + initrd, so I don't
think we have to care of other versions of dd here.

> What if it checks for the presence of the device file, e.g:
> 
> ... self, 'test -c /dev/hwrng; echo $?', '1')

That doesn't work, the /dev/hwrng is still there (so test -c succeeds),
since this initrd uses static device nodes for this in /dev. /dev/hwrng just
can not be opened anymore after the device has been removed.

 Thomas