Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/22/2018 12:01 PM, Halil Pasic wrote:



On 04/22/2018 05:43 PM, Tony Krowiak wrote:

+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be 
more end-user

friendly?


It's a matter of opinion. I prefer facilities because AP is comprised 
of not only

instructions, but also AP processors.


Please elaborate. You mean processors like AP cards? If yes what if 
the matrix is
empty (e.g. the state we decided is the default when no further action 
is taken
(assign queues to the vfio-ap kernel driver, set up an mdev, -device 
vfio-ap on

qemu cmd line))?

I just wanted to point out that this plural is very vague. Not 
speaking about that
QCI should be an AP facility too, but is not included in this 
'features' and is not
covered by this cpu model feature. It has it's own cpu model feature 
and even a

dedicated STFLE bit.


FYI, I decided to take your suggestion and go with 'AP instructions 
installed'.





Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 04:59 AM, David Hildenbrand wrote:

On 18.04.2018 09:40, Cornelia Huck wrote:

On Tue, 17 Apr 2018 20:21:31 +0200
Christian Borntraeger  wrote:


On 04/16/2018 05:44 PM, David Hildenbrand wrote:
   
  
diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c

index 0cdbc15..0d5b0f7 100644
--- a/target/s390x/gen-features.c
+++ b/target/s390x/gen-features.c
@@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
  S390_FEAT_ADAPTER_INT_SUPPRESSION,
  S390_FEAT_EDAT_2,
  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
+S390_FEAT_AP_QUERY_CONFIG_INFO,
+S390_FEAT_AP_FACILITIES_TEST,
+S390_FEAT_AP,
  };


Now I have to ask a very stupid question:

I heard that the execution controls in the SIE block for AP are one of
the oldest ones we have around. How can it be that the AP feature cannot
be used before zEC12?

It was a suggestion from the crypto team due to testability. Nobody has a z196
with the older cards.


Might be worth adding a note to that respect?


We used to have the CPU model stick as close as possible to the real CPU
models.

Support statements should cover in specific products what is expected to
work and what not.

So is there any real (!support statement / !testability) reason to not
allow this feature on older CPU models that also had support for it?


This is a business decision based on the following factors:
1. We do not have access to the older cards/systems for testing of the 
older devices.
2. Pre-CEX2 cards are no longer supported in the kernel with others to 
follow.
3. Keep the code base as small as possible and ensure it can be 
adequately tested.

4. Business investment priorities - i.e., 's.

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/22/2018 12:01 PM, Halil Pasic wrote:



On 04/22/2018 05:43 PM, Tony Krowiak wrote:

+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be 
more end-user

friendly?


It's a matter of opinion. I prefer facilities because AP is comprised 
of not only

instructions, but also AP processors.


Please elaborate. You mean processors like AP cards? If yes what if 
the matrix is
empty (e.g. the state we decided is the default when no further action 
is taken
(assign queues to the vfio-ap kernel driver, set up an mdev, -device 
vfio-ap on

qemu cmd line))?


I don't want to get into a debate about this. If it means that much to 
you, I'll

go ahead and change it.




I just wanted to point out that this plural is very vague. Not 
speaking about that
QCI should be an AP facility too, but is not included in this 
'features' and is not
covered by this cpu model feature. It has it's own cpu model feature 
and even a

dedicated STFLE bit.


Uncle  I got it.




Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Halil Pasic]

On 04/22/2018 05:43 PM, Tony Krowiak wrote:

+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be more end-user
friendly?


It's a matter of opinion. I prefer facilities because AP is comprised of not 
only
instructions, but also AP processors.


Please elaborate. You mean processors like AP cards? If yes what if the matrix 
is
empty (e.g. the state we decided is the default when no further action is taken
(assign queues to the vfio-ap kernel driver, set up an mdev, -device vfio-ap on
qemu cmd line))?

I just wanted to point out that this plural is very vague. Not speaking about 
that
QCI should be an AP facility too, but is not included in this 'features' and is 
not
covered by this cpu model feature. It has it's own cpu model feature and even a
dedicated STFLE bit.

Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Halil Pasic]

On 04/22/2018 05:52 PM, Tony Krowiak wrote:

+    FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP Configuration 
facility"),


Why did you change this form qci to apqci. Too may people found the
qci good?


I changed it by request and I thought it made sense to do so. Maybe we should 
just
take a vote.


Sorry I've missed that request. Could you give me a link into the mail archive?
I would like to examine the discussion.

Thanks,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 07:03 AM, David Hildenbrand wrote:

On 18.04.2018 12:55, Halil Pasic wrote:


On 04/15/2018 09:07 PM, Tony Krowiak wrote:

A new CPU model feature and two new CPU model facilities are
introduced to support AP devices for a KVM guest.

[..]

diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
index 3b9e274..5ee3a2d 100644
--- a/target/s390x/cpu_features.c
+++ b/target/s390x/cpu_features.c
@@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
   FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status 
facility"),
   FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE facility"),
   FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, "Configuration-topology 
facility"),
+FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP Configuration 
facility"),

Why did you change this form qci to apqci. Too may people found the
qci good?

If the facility is called "Query AP Configuration facility" it should be
qapc

Where does the term "qci" come from ?


It comes from the name of the function Query AP Configuration 
Information which

is identified by the initials QCI.




   FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range facility"),
   FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing key-setting 
facility"),
+FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor Facilities Test 
facility"),
   FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, "Extended-translation facility 
2"),
   FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, "Message-security-assist 
facility (excluding subfunctions)"),
   FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement 
facility"),
@@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {

   FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
   FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
+FEAT_INIT_MISC("ap", "AP facilities installed"),

Why plural ('facilities')? Would not s/facilities/instructions be more end-user
friendly?

Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 06:55 AM, Halil Pasic wrote:



On 04/15/2018 09:07 PM, Tony Krowiak wrote:

A new CPU model feature and two new CPU model facilities are
introduced to support AP devices for a KVM guest.

[..]

diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
index 3b9e274..5ee3a2d 100644
--- a/target/s390x/cpu_features.c
+++ b/target/s390x/cpu_features.c
@@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
  FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status 
facility"),
  FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE 
facility"),
  FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, 
"Configuration-topology facility"),
+FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP 
Configuration facility"),


Why did you change this form qci to apqci. Too may people found the
qci good?


I changed it by request and I thought it made sense to do so. Maybe we 
should just

take a vote.




  FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range 
facility"),
  FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing 
key-setting facility"),
+FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor 
Facilities Test facility"),
  FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, 
"Extended-translation facility 2"),
  FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, 
"Message-security-assist facility (excluding subfunctions)"),
  FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement 
facility"),

@@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {

  FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
  FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be more 
end-user

friendly?

Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 06:55 AM, Halil Pasic wrote:



On 04/15/2018 09:07 PM, Tony Krowiak wrote:

A new CPU model feature and two new CPU model facilities are
introduced to support AP devices for a KVM guest.

[..]

diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
index 3b9e274..5ee3a2d 100644
--- a/target/s390x/cpu_features.c
+++ b/target/s390x/cpu_features.c
@@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
  FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status 
facility"),
  FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE 
facility"),
  FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, 
"Configuration-topology facility"),
+FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP 
Configuration facility"),


Why did you change this form qci to apqci. Too may people found the
qci good?

  FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range 
facility"),
  FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing 
key-setting facility"),
+FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor 
Facilities Test facility"),
  FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, 
"Extended-translation facility 2"),
  FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, 
"Message-security-assist facility (excluding subfunctions)"),
  FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement 
facility"),

@@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {

  FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
  FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be more 
end-user

friendly?


It's a matter of opinion. I prefer facilities because AP is comprised of 
not only

instructions, but also AP processors.




Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 04:59 AM, David Hildenbrand wrote:

On 18.04.2018 09:40, Cornelia Huck wrote:

On Tue, 17 Apr 2018 20:21:31 +0200
Christian Borntraeger  wrote:


On 04/16/2018 05:44 PM, David Hildenbrand wrote:
   
  
diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c

index 0cdbc15..0d5b0f7 100644
--- a/target/s390x/gen-features.c
+++ b/target/s390x/gen-features.c
@@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
  S390_FEAT_ADAPTER_INT_SUPPRESSION,
  S390_FEAT_EDAT_2,
  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
+S390_FEAT_AP_QUERY_CONFIG_INFO,
+S390_FEAT_AP_FACILITIES_TEST,
+S390_FEAT_AP,
  };


Now I have to ask a very stupid question:

I heard that the execution controls in the SIE block for AP are one of
the oldest ones we have around. How can it be that the AP feature cannot
be used before zEC12?

It was a suggestion from the crypto team due to testability. Nobody has a z196
with the older cards.


Might be worth adding a note to that respect?


We used to have the CPU model stick as close as possible to the real CPU
models.

Support statements should cover in specific products what is expected to
work and what not.

So is there any real (!support statement / !testability) reason to not
allow this feature on older CPU models that also had support for it?


I'll discuss it with the team and get back to you.

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/18/2018 03:40 AM, Cornelia Huck wrote:

On Tue, 17 Apr 2018 20:21:31 +0200
Christian Borntraeger  wrote:


On 04/16/2018 05:44 PM, David Hildenbrand wrote:
   
  
diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c

index 0cdbc15..0d5b0f7 100644
--- a/target/s390x/gen-features.c
+++ b/target/s390x/gen-features.c
@@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
  S390_FEAT_ADAPTER_INT_SUPPRESSION,
  S390_FEAT_EDAT_2,
  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
+S390_FEAT_AP_QUERY_CONFIG_INFO,
+S390_FEAT_AP_FACILITIES_TEST,
+S390_FEAT_AP,
  };


Now I have to ask a very stupid question:

I heard that the execution controls in the SIE block for AP are one of
the oldest ones we have around. How can it be that the AP feature cannot
be used before zEC12?

It was a suggestion from the crypto team due to testability. Nobody has a z196
with the older cards.


Might be worth adding a note to that respect?


I will.

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Pierre Morel]

On 18/04/2018 13:03, David Hildenbrand wrote:

On 18.04.2018 12:55, Halil Pasic wrote:


On 04/15/2018 09:07 PM, Tony Krowiak wrote:

A new CPU model feature and two new CPU model facilities are
introduced to support AP devices for a KVM guest.

[..]

diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
index 3b9e274..5ee3a2d 100644
--- a/target/s390x/cpu_features.c
+++ b/target/s390x/cpu_features.c
@@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
   FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status 
facility"),
   FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE facility"),
   FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, "Configuration-topology 
facility"),
+FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP Configuration 
facility"),

Why did you change this form qci to apqci. Too may people found the
qci good?

If the facility is called "Query AP Configuration facility" it should be
qapc

Where does the term "qci" come from ?


Query Configuration Information

:)





   FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range facility"),
   FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing key-setting 
facility"),
+FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor Facilities Test 
facility"),
   FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, "Extended-translation facility 
2"),
   FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, "Message-security-assist 
facility (excluding subfunctions)"),
   FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement 
facility"),
@@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {

   FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
   FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
+FEAT_INIT_MISC("ap", "AP facilities installed"),

Why plural ('facilities')? Would not s/facilities/instructions be more end-user
friendly?

Regards,
Halil





--
Pierre Morel
Linux/KVM/QEMU in Böblingen - Germany

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[David Hildenbrand]

On 18.04.2018 12:55, Halil Pasic wrote:
> 
> 
> On 04/15/2018 09:07 PM, Tony Krowiak wrote:
>> A new CPU model feature and two new CPU model facilities are
>> introduced to support AP devices for a KVM guest.
> [..]
>> diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
>> index 3b9e274..5ee3a2d 100644
>> --- a/target/s390x/cpu_features.c
>> +++ b/target/s390x/cpu_features.c
>> @@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
>>   FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status 
>> facility"),
>>   FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE 
>> facility"),
>>   FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, "Configuration-topology 
>> facility"),
>> +FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP Configuration 
>> facility"),
> 
> Why did you change this form qci to apqci. Too may people found the
> qci good?

If the facility is called "Query AP Configuration facility" it should be
qapc

Where does the term "qci" come from ?

> 
>>   FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range facility"),
>>   FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing key-setting 
>> facility"),
>> +FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor 
>> Facilities Test facility"),
>>   FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, "Extended-translation 
>> facility 2"),
>>   FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, 
>> "Message-security-assist facility (excluding subfunctions)"),
>>   FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement 
>> facility"),
>> @@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {
>>
>>   FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
>>   FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
>> +FEAT_INIT_MISC("ap", "AP facilities installed"),
> 
> Why plural ('facilities')? Would not s/facilities/instructions be more 
> end-user
> friendly?
> 
> Regards,
> Halil
> 


-- 

Thanks,

David / dhildenb

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Halil Pasic]

On 04/15/2018 09:07 PM, Tony Krowiak wrote:

A new CPU model feature and two new CPU model facilities are
introduced to support AP devices for a KVM guest.

[..]

diff --git a/target/s390x/cpu_features.c b/target/s390x/cpu_features.c
index 3b9e274..5ee3a2d 100644
--- a/target/s390x/cpu_features.c
+++ b/target/s390x/cpu_features.c
@@ -40,8 +40,10 @@ static const S390FeatDef s390_features[] = {
  FEAT_INIT("srs", S390_FEAT_TYPE_STFL, 9, "Sense-running-status facility"),
  FEAT_INIT("csske", S390_FEAT_TYPE_STFL, 10, "Conditional-SSKE facility"),
  FEAT_INIT("ctop", S390_FEAT_TYPE_STFL, 11, "Configuration-topology 
facility"),
+FEAT_INIT("apqci", S390_FEAT_TYPE_STFL, 12, "Query AP Configuration 
facility"),


Why did you change this form qci to apqci. Too may people found the
qci good?


  FEAT_INIT("ipter", S390_FEAT_TYPE_STFL, 13, "IPTE-range facility"),
  FEAT_INIT("nonqks", S390_FEAT_TYPE_STFL, 14, "Nonquiescing key-setting 
facility"),
+FEAT_INIT("apft", S390_FEAT_TYPE_STFL, 15, "Adjunct Processor Facilities Test 
facility"),
  FEAT_INIT("etf2", S390_FEAT_TYPE_STFL, 16, "Extended-translation facility 
2"),
  FEAT_INIT("msa-base", S390_FEAT_TYPE_STFL, 17, "Message-security-assist 
facility (excluding subfunctions)"),
  FEAT_INIT("ldisp", S390_FEAT_TYPE_STFL, 18, "Long-displacement facility"),
@@ -129,6 +131,7 @@ static const S390FeatDef s390_features[] = {

  FEAT_INIT_MISC("dateh2", "DAT-enhancement facility 2"),
  FEAT_INIT_MISC("cmm", "Collaborative-memory-management facility"),
+FEAT_INIT_MISC("ap", "AP facilities installed"),


Why plural ('facilities')? Would not s/facilities/instructions be more end-user
friendly?

Regards,
Halil

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[David Hildenbrand]

On 18.04.2018 09:40, Cornelia Huck wrote:
> On Tue, 17 Apr 2018 20:21:31 +0200
> Christian Borntraeger  wrote:
> 
>> On 04/16/2018 05:44 PM, David Hildenbrand wrote:
>>>   
  
 diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c
 index 0cdbc15..0d5b0f7 100644
 --- a/target/s390x/gen-features.c
 +++ b/target/s390x/gen-features.c
 @@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
  S390_FEAT_ADAPTER_INT_SUPPRESSION,
  S390_FEAT_EDAT_2,
  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
 +S390_FEAT_AP_QUERY_CONFIG_INFO,
 +S390_FEAT_AP_FACILITIES_TEST,
 +S390_FEAT_AP,
  };

>>>
>>> Now I have to ask a very stupid question:
>>>
>>> I heard that the execution controls in the SIE block for AP are one of
>>> the oldest ones we have around. How can it be that the AP feature cannot
>>> be used before zEC12?  
>>
>> It was a suggestion from the crypto team due to testability. Nobody has a 
>> z196
>> with the older cards.
>>
> 
> Might be worth adding a note to that respect?
> 

We used to have the CPU model stick as close as possible to the real CPU
models.

Support statements should cover in specific products what is expected to
work and what not.

So is there any real (!support statement / !testability) reason to not
allow this feature on older CPU models that also had support for it?

-- 

Thanks,

David / dhildenb

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Cornelia Huck]

On Tue, 17 Apr 2018 20:21:31 +0200
Christian Borntraeger  wrote:

> On 04/16/2018 05:44 PM, David Hildenbrand wrote:
> >   
> >>  
> >> diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c
> >> index 0cdbc15..0d5b0f7 100644
> >> --- a/target/s390x/gen-features.c
> >> +++ b/target/s390x/gen-features.c
> >> @@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
> >>  S390_FEAT_ADAPTER_INT_SUPPRESSION,
> >>  S390_FEAT_EDAT_2,
> >>  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
> >> +S390_FEAT_AP_QUERY_CONFIG_INFO,
> >> +S390_FEAT_AP_FACILITIES_TEST,
> >> +S390_FEAT_AP,
> >>  };
> >>
> > 
> > Now I have to ask a very stupid question:
> > 
> > I heard that the execution controls in the SIE block for AP are one of
> > the oldest ones we have around. How can it be that the AP feature cannot
> > be used before zEC12?  
> 
> It was a suggestion from the crypto team due to testability. Nobody has a z196
> with the older cards.
> 

Might be worth adding a note to that respect?

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Christian Borntraeger]

On 04/16/2018 05:44 PM, David Hildenbrand wrote:
> 
>>  
>> diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c
>> index 0cdbc15..0d5b0f7 100644
>> --- a/target/s390x/gen-features.c
>> +++ b/target/s390x/gen-features.c
>> @@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
>>  S390_FEAT_ADAPTER_INT_SUPPRESSION,
>>  S390_FEAT_EDAT_2,
>>  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
>> +S390_FEAT_AP_QUERY_CONFIG_INFO,
>> +S390_FEAT_AP_FACILITIES_TEST,
>> +S390_FEAT_AP,
>>  };
>>  
> 
> Now I have to ask a very stupid question:
> 
> I heard that the execution controls in the SIE block for AP are one of
> the oldest ones we have around. How can it be that the AP feature cannot
> be used before zEC12?

It was a suggestion from the crypto team due to testability. Nobody has a z196
with the older cards.

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[Tony Krowiak]

On 04/16/2018 11:44 AM, David Hildenbrand wrote:
  
diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c

index 0cdbc15..0d5b0f7 100644
--- a/target/s390x/gen-features.c
+++ b/target/s390x/gen-features.c
@@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
  S390_FEAT_ADAPTER_INT_SUPPRESSION,
  S390_FEAT_EDAT_2,
  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
+S390_FEAT_AP_QUERY_CONFIG_INFO,
+S390_FEAT_AP_FACILITIES_TEST,
+S390_FEAT_AP,
  };
  

Now I have to ask a very stupid question:

I heard that the execution controls in the SIE block for AP are one of
the oldest ones we have around. How can it be that the AP feature cannot
be used before zEC12?


Its not a stupid question, but I don't have an answer because this was
not a design decision I made. I will consult with the crypto team to
see if I can get you an answer.

Re: [Qemu-devel] [PATCH v4 3/5] s390x/cpumodel: Set up CPU model for AP device support

[David Hildenbrand]

>  
> diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c
> index 0cdbc15..0d5b0f7 100644
> --- a/target/s390x/gen-features.c
> +++ b/target/s390x/gen-features.c
> @@ -447,6 +447,9 @@ static uint16_t full_GEN12_GA1[] = {
>  S390_FEAT_ADAPTER_INT_SUPPRESSION,
>  S390_FEAT_EDAT_2,
>  S390_FEAT_SIDE_EFFECT_ACCESS_ESOP2,
> +S390_FEAT_AP_QUERY_CONFIG_INFO,
> +S390_FEAT_AP_FACILITIES_TEST,
> +S390_FEAT_AP,
>  };
>  

Now I have to ask a very stupid question:

I heard that the execution controls in the SIE block for AP are one of
the oldest ones we have around. How can it be that the AP feature cannot
be used before zEC12?


-- 

Thanks,

David / dhildenb