Re: [PATCH] s390: vfio-ap: trace the update of APCB masks

[Christian Borntraeger]

On 10/05/2018 09:54 AM, David Hildenbrand wrote:
> On 05/10/2018 09:50, Cornelia Huck wrote:
>> On Fri,  5 Oct 2018 09:33:01 +0200
>> Pierre Morel <pmorel@xxxxxxxxxxxxx> wrote:
>>
>>> Define a tracing function to trace in the KVM trace buffer
>>> and trace the changes of the APCB masks.
>>
>> In general, trace events are good :)
>>
>>>
>>> Signed-off-by: Pierre Morel <pmorel@xxxxxxxxxxxxx>
>>> ---
>>>  drivers/s390/crypto/vfio_ap_ops.c     | 22 ++++++++++++++++++++++
>>>  drivers/s390/crypto/vfio_ap_private.h |  4 ++++
>>>  2 files changed, 26 insertions(+)
>>>
>>> diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c
>>> index a8a9984..00e632a 100644
>>> --- a/drivers/s390/crypto/vfio_ap_ops.c
>>> +++ b/drivers/s390/crypto/vfio_ap_ops.c
>>> @@ -753,6 +753,28 @@ static void vfio_ap_mdev_copy_masks(struct ap_matrix_mdev *matrix_mdev)
>>>  	memcpy(aqm, matrix_mdev->matrix.aqm, nbytes);
>>>  	nbytes = DIV_ROUND_UP(matrix_mdev->matrix.adm_max + 1, BITS_PER_BYTE);
>>>  	memcpy(adm, matrix_mdev->matrix.adm, nbytes);
>>> +
>>> +	switch (matrix_mdev->kvm->arch.crypto.crycbd & CRYCB_FORMAT_MASK) {
>>> +	case CRYCB_FORMAT2:
>>> +		VM_AP_EVENT(matrix_mdev->kvm, 3,
>>> +			    "SET CRYCB: apm %016lx %016lx %016lx %016lx",
>>> +			    apm[0], apm[1], apm[2], apm[3]);
>>> +		VM_AP_EVENT(matrix_mdev->kvm, 3,
>>> +			    "SET CRYCB: aqm %016lx %016lx %016lx %016lx",
>>> +			    aqm[0], aqm[1], aqm[2], aqm[3]);
>>> +		VM_AP_EVENT(matrix_mdev->kvm, 3,
>>> +			    "SET CRYCB: adm %016lx %016lx %016lx %016lx",
>>> +			    adm[0], adm[1], adm[2], adm[3]);
>>> +		break;
>>> +	case CRYCB_FORMAT1:
>>> +	case CRYCB_FORMAT0: /* Both CRYCB format uses APCB format 0 */
>>> +	default: /* Can not happen */
>>
>> If it can't happen, trace "SET CRYCB: invalid format '%x'" instead?
>>
>>> +		VM_AP_EVENT(matrix_mdev->kvm, 3,
>>> +			    "SET CRYCB: apm %016lx aqm %04x adm %04x", apm[0],
>>> +			    *((unsigned short *)aqm), *((unsigned short *)adm));
>>> +		break;
>>> +	}
>>> +
>>>  }
>>>  
>>>  /**
>>> diff --git a/drivers/s390/crypto/vfio_ap_private.h b/drivers/s390/crypto/vfio_ap_private.h
>>> index 5675492..4cd779d 100644
>>> --- a/drivers/s390/crypto/vfio_ap_private.h
>>> +++ b/drivers/s390/crypto/vfio_ap_private.h
>>> @@ -19,6 +19,10 @@
>>>  
>>>  #include "ap_bus.h"
>>>  
>>> +#define VM_AP_EVENT(d_kvm, d_loglevel, d_string, d_args...)\
>>> +	debug_sprintf_event(d_kvm->arch.dbf, d_loglevel, d_string "\n", \
>>> +			    d_args)
>>> +
>>>  #define VFIO_AP_MODULE_NAME "vfio_ap"
>>>  #define VFIO_AP_DRV_NAME "vfio_ap"
>>>  
>>
>> One thing I don't like about this is that you're using the kvm dbf from
>> driver code, which looks odd. OTOH, these are kvm-specific masks...
> 
> Maybe factor out the setting of the actual masks to KVM code and do
> tracing there?

We had both versions internally, and this version looked shorter. But Pierre,
can you send your v1 as well and then we can compare both variants.
> 
>>
>> Is there anything else that you may want to trace in the vfio-ap
>> driver, so that you could add a dbf for it?
>>
> 
>