Re: [kvm-unit-tests PATCH v3 7/8] s390x: Add diag318 intercept test

[Thomas Huth <thuth@redhat.com>]

On 17/12/2020 10.59, Christian Borntraeger wrote:
> 
> 
> On 17.12.20 10:53, Thomas Huth wrote:
>> On 11/12/2020 11.00, Janosch Frank wrote:
>>> Not much to test except for the privilege and specification
>>> exceptions.
>>>
>>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>>> Reviewed-by: Thomas Huth <thuth@redhat.com>
>>> ---
>>>  lib/s390x/sclp.c  |  2 ++
>>>  lib/s390x/sclp.h  |  6 +++++-
>>>  s390x/intercept.c | 19 +++++++++++++++++++
>>>  3 files changed, 26 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/lib/s390x/sclp.c b/lib/s390x/sclp.c
>>> index cf6ea7c..0001993 100644
>>> --- a/lib/s390x/sclp.c
>>> +++ b/lib/s390x/sclp.c
>>> @@ -138,6 +138,8 @@ void sclp_facilities_setup(void)
>>>  
>>>  	assert(read_info);
>>>  
>>> +	sclp_facilities.has_diag318 = read_info->byte_134_diag318;
>>> +
>>>  	cpu = (void *)read_info + read_info->offset_cpu;
>>>  	for (i = 0; i < read_info->entries_cpu; i++, cpu++) {
>>>  		if (cpu->address == cpu0_addr) {
>>> diff --git a/lib/s390x/sclp.h b/lib/s390x/sclp.h
>>> index 6c86037..58f8e54 100644
>>> --- a/lib/s390x/sclp.h
>>> +++ b/lib/s390x/sclp.h
>>> @@ -105,7 +105,8 @@ extern struct sclp_facilities sclp_facilities;
>>>  
>>>  struct sclp_facilities {
>>>  	uint64_t has_sief2 : 1;
>>> -	uint64_t : 63;
>>> +	uint64_t has_diag318 : 1;
>>> +	uint64_t : 62;
>>>  };
>>>  
>>>  typedef struct ReadInfo {
>>> @@ -130,6 +131,9 @@ typedef struct ReadInfo {
>>>      uint16_t highest_cpu;
>>>      uint8_t  _reserved5[124 - 122];     /* 122-123 */
>>>      uint32_t hmfai;
>>> +    uint8_t reserved7[134 - 128];
>>> +    uint8_t byte_134_diag318 : 1;
>>> +    uint8_t : 7;
>>>      struct CPUEntry entries[0];
>>
>> ... the entries[] array can be moved around here without any further ado?
>> Looks confusing to me. Should there be a CPUEntry array here at all, or only
>> in ReadCpuInfo?
> 
> there is offset_cpu for the cpu entries at the beginning of the structure.

Ah, thanks, right, this was used earlier in the patch series, now I
remember. But I think the "struct CPUEntry entries[0]" here is rather
confusing, since there is no guarantee that the entries are really at this
location ... I think this line should rather be replaced by a comment saying
that offset_cpu should be used instead.

 Thomas