Re: [PATCH v3 1/2] s390/setup: diag318: remove bit check and refactor struct

[Collin Walling <walling@linux.ibm.com>]

On 4/3/19 8:33 AM, Cornelia Huck wrote:
> On Wed, 3 Apr 2019 14:03:21 +0200
> David Hildenbrand <david@redhat.com> wrote:
> 
>> On 02.04.19 19:46, Collin Walling wrote:
>>> Execution of DIAGNOSE 0x318 is fenced by checking an SCLP bit
>>> for the availability of hardware support for the instruction.
>>>
>>> In order to support this instruction for a KVM/QEMU guest, we
>>> would need to provide modifications to the SCLP Read SCP Info
>>> data, which will in turn reduce the maximum number of CPUs that
>>> may be provided to the guest. This issue introduces compatability
>>> and legacy concerns.
>>>
>>> Let's circumvent this issue by removing the bit check and blindly
>>> executing the instruction. An exception table rule is in place to
>>> catch the case where hardware does not support this instruction.
>>>
>>> While we're at it, let's condense the version code fields in the
>>> diag318_info struct until we can determine how it will be used.
>>>
>>> This modifies commit 4ad78b8651aacf26b3ab6d1e784952eb70469c43
>>>
>>> Signed-off-by: Collin Walling <walling@linux.ibm.com>
>>> ---
>>>   arch/s390/include/asm/diag.h |  6 ++----
>>>   arch/s390/kernel/setup.c     | 12 ++++++------
>>>   2 files changed, 8 insertions(+), 10 deletions(-)
>>>
>>> diff --git a/arch/s390/include/asm/diag.h b/arch/s390/include/asm/diag.h
>>> index 19562be22b7e..215516284175 100644
>>> --- a/arch/s390/include/asm/diag.h
>>> +++ b/arch/s390/include/asm/diag.h
>>> @@ -298,10 +298,8 @@ struct diag26c_mac_resp {
>>>   union diag318_info {
>>>   	unsigned long val;
>>>   	struct {
>>> -		unsigned int cpnc : 8;
>>> -		unsigned int cpvc_linux : 24;
>>> -		unsigned char cpvc_distro[3];
>>> -		unsigned char zero;
>>> +		unsigned long cpnc : 8;
>>> +		unsigned long cpvc : 56;
> 
> That part looks reasonable (we don't have a proper convention yet, have
> we?)
> 
>>>   	};
>>>   };
>>>   
>>> diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
>>> index 2c642af526ce..fe70201f8b5d 100644
>>> --- a/arch/s390/kernel/setup.c
>>> +++ b/arch/s390/kernel/setup.c
>>> @@ -1011,15 +1011,15 @@ static void __init setup_control_program_code(void)
>>>   {
>>>   	union diag318_info diag318_info = {
>>>   		.cpnc = CPNC_LINUX,
>>> -		.cpvc_linux = 0,
>>> -		.cpvc_distro = {0},
>>> +		.cpvc = 0,
>>>   	};
>>>   
>>> -	if (!sclp.has_diag318)
>>> -		return;
>>> -
>>>   	diag_stat_inc(DIAG_STAT_X318);
>>> -	asm volatile("diag %0,0,0x318\n" : : "d" (diag318_info.val));
>>> +	asm volatile(
>>> +		"	diag	%0,0,0x318\n"
>>> +		"0:	nopr	%%r7\n"
>>> +		EX_TABLE(0b,0b)
>>> +		: : "d" (diag318_info.val));
>>>   }
>>>   
>>>   /*
>>>    
>>
>> That smells like a nasty hack to not expose new features in QEMU and
>> deal with the issue of handling CPU limits. No, I don't like this.
>>
>> Fix QEMU, not the kernel.
>>
> 
> I agree. The compat handling is a bit annoying, but I don't think we
> can get around it.
> 

Thanks for the feedback, everyone.

The consensus here is to keep the bit check, so I'll throw that back in.
I'll squeeze in a "clean-up" patch for the diag318_info struct in v4.

I'll see that QEMU 4.1 has a max cpu limit of at most 247 (one less than
the current max). If anyone has a suggestion on a better limit
(Christian mentioned 240), please let me know. Otherwise we can discuss
that value in the next version.