[v3,0/3] s390x/kvm: implement new hardware/firmware features

Message ID 20180118085628.40798-1-borntraeger@de.ibm.com
Headers show
Series
  • s390x/kvm: implement new hardware/firmware features
Related show

Message

Christian Borntraeger Jan. 18, 2018, 8:56 a.m.
We want to provide more hw features to guests, namely the new bpb
control as well as other transparent facilities that might be
introduced by firmware updates (e.g. the stfle facility 81).

See the kernel discussion for the KVM side
https://www.spinics.net/lists/kernel/msg2700551.html

v2->v3: - use bool for bpbc
	- sort cpu facilities
v1->v2: - style and comment fixes
	- drop transparent facility patch
	- add patch to introduce facility 81

Christian Borntraeger (3):
  header sync
  s390x/kvm: Handle bpb feature
  s390x/kvm: provide stfle.81

 linux-headers/asm-s390/kvm.h    |  9 ++++-----
 linux-headers/linux/kvm.h       |  5 +++--
 target/s390x/cpu.c              |  1 +
 target/s390x/cpu.h              |  1 +
 target/s390x/cpu_features.c     |  2 ++
 target/s390x/cpu_features_def.h |  2 ++
 target/s390x/gen-features.c     |  2 ++
 target/s390x/kvm.c              | 14 ++++++++++++++
 target/s390x/machine.c          | 17 +++++++++++++++++
 9 files changed, 46 insertions(+), 7 deletions(-)

Comments

Cornelia Huck Jan. 18, 2018, 11:53 a.m. | #1
On Thu, 18 Jan 2018 09:56:25 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> We want to provide more hw features to guests, namely the new bpb
> control as well as other transparent facilities that might be
> introduced by firmware updates (e.g. the stfle facility 81).
> 
> See the kernel discussion for the KVM side
> https://www.spinics.net/lists/kernel/msg2700551.html
> 
> v2->v3: - use bool for bpbc
> 	- sort cpu facilities
> v1->v2: - style and comment fixes
> 	- drop transparent facility patch
> 	- add patch to introduce facility 81
> 
> Christian Borntraeger (3):
>   header sync
>   s390x/kvm: Handle bpb feature
>   s390x/kvm: provide stfle.81
> 
>  linux-headers/asm-s390/kvm.h    |  9 ++++-----
>  linux-headers/linux/kvm.h       |  5 +++--
>  target/s390x/cpu.c              |  1 +
>  target/s390x/cpu.h              |  1 +
>  target/s390x/cpu_features.c     |  2 ++
>  target/s390x/cpu_features_def.h |  2 ++
>  target/s390x/gen-features.c     |  2 ++
>  target/s390x/kvm.c              | 14 ++++++++++++++
>  target/s390x/machine.c          | 17 +++++++++++++++++
>  9 files changed, 46 insertions(+), 7 deletions(-)
> 

Thanks, I've queued this. Once the bpb code lands in Linux, I'll
replace patch 1 with a proper header sync and push to s390-next.
Christian Borntraeger Jan. 18, 2018, 12:47 p.m. | #2
On 01/18/2018 12:53 PM, Cornelia Huck wrote:
> On Thu, 18 Jan 2018 09:56:25 +0100
> Christian Borntraeger <borntraeger@de.ibm.com> wrote:
> 
>> We want to provide more hw features to guests, namely the new bpb
>> control as well as other transparent facilities that might be
>> introduced by firmware updates (e.g. the stfle facility 81).
>>
>> See the kernel discussion for the KVM side
>> https://www.spinics.net/lists/kernel/msg2700551.html
>>
>> v2->v3: - use bool for bpbc
>> 	- sort cpu facilities
>> v1->v2: - style and comment fixes
>> 	- drop transparent facility patch
>> 	- add patch to introduce facility 81
>>
>> Christian Borntraeger (3):
>>   header sync
>>   s390x/kvm: Handle bpb feature
>>   s390x/kvm: provide stfle.81
>>
>>  linux-headers/asm-s390/kvm.h    |  9 ++++-----
>>  linux-headers/linux/kvm.h       |  5 +++--
>>  target/s390x/cpu.c              |  1 +
>>  target/s390x/cpu.h              |  1 +
>>  target/s390x/cpu_features.c     |  2 ++
>>  target/s390x/cpu_features_def.h |  2 ++
>>  target/s390x/gen-features.c     |  2 ++
>>  target/s390x/kvm.c              | 14 ++++++++++++++
>>  target/s390x/machine.c          | 17 +++++++++++++++++
>>  9 files changed, 46 insertions(+), 7 deletions(-)
>>
> 
> Thanks, I've queued this. Once the bpb code lands in Linux, I'll
> replace patch 1 with a proper header sync and push to s390-next.

Have the x86 features been marked as stable? If the answer is yes,
shall we mark these patches for stable as well?
Cornelia Huck Jan. 18, 2018, 1:15 p.m. | #3
On Thu, 18 Jan 2018 13:47:54 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> Have the x86 features been marked as stable? If the answer is yes,
> shall we mark these patches for stable as well?

Doesn't look like it.

TBH, I'm not quite sure whether this should go into stable as I'm a bit
unclear what our use case for stable is. It seems to be mostly "don't
let people run into known crashes" or something like that.

These patches are not very useful on their own unless you run on a z
with updated microcode and an updated host kernel. OTOH, the patches
should be low risk.
Christian Ehrhardt Jan. 18, 2018, 1:31 p.m. | #4
On Thu, Jan 18, 2018 at 2:15 PM, Cornelia Huck <cohuck@redhat.com> wrote:
> On Thu, 18 Jan 2018 13:47:54 +0100
> Christian Borntraeger <borntraeger@de.ibm.com> wrote:
>
>> Have the x86 features been marked as stable? If the answer is yes,
>> shall we mark these patches for stable as well?
>
> Doesn't look like it.
>
> TBH, I'm not quite sure whether this should go into stable as I'm a bit
> unclear what our use case for stable is. It seems to be mostly "don't
> let people run into known crashes" or something like that.

I read the public statement [1] as "... non-x86 processors ...
backported to recent stable releases." or did the changes in
discussion here end up structurally so different that this doesn't
apply anymore?

[1]: https://www.qemu.org/2018/01/04/spectre/
Cornelia Huck Jan. 18, 2018, 2:48 p.m. | #5
On Thu, 18 Jan 2018 14:31:50 +0100
Christian Ehrhardt <christian.ehrhardt@canonical.com> wrote:

> On Thu, Jan 18, 2018 at 2:15 PM, Cornelia Huck <cohuck@redhat.com> wrote:
> > On Thu, 18 Jan 2018 13:47:54 +0100
> > Christian Borntraeger <borntraeger@de.ibm.com> wrote:
> >  
> >> Have the x86 features been marked as stable? If the answer is yes,
> >> shall we mark these patches for stable as well?  
> >
> > Doesn't look like it.
> >
> > TBH, I'm not quite sure whether this should go into stable as I'm a bit
> > unclear what our use case for stable is. It seems to be mostly "don't
> > let people run into known crashes" or something like that.  
> 
> I read the public statement [1] as "... non-x86 processors ...
> backported to recent stable releases." or did the changes in
> discussion here end up structurally so different that this doesn't
> apply anymore?
> 
> [1]: https://www.qemu.org/2018/01/04/spectre/

OK, that would read as "do backport". (The x86 patches I saw didn't
carry an explicit cc:stable.)
Cornelia Huck Jan. 22, 2018, 10:20 a.m. | #6
On Thu, 18 Jan 2018 09:56:25 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> We want to provide more hw features to guests, namely the new bpb
> control as well as other transparent facilities that might be
> introduced by firmware updates (e.g. the stfle facility 81).
> 
> See the kernel discussion for the KVM side
> https://www.spinics.net/lists/kernel/msg2700551.html
> 
> v2->v3: - use bool for bpbc
> 	- sort cpu facilities
> v1->v2: - style and comment fixes
> 	- drop transparent facility patch
> 	- add patch to introduce facility 81
> 
> Christian Borntraeger (3):
>   header sync
>   s390x/kvm: Handle bpb feature
>   s390x/kvm: provide stfle.81
> 
>  linux-headers/asm-s390/kvm.h    |  9 ++++-----
>  linux-headers/linux/kvm.h       |  5 +++--
>  target/s390x/cpu.c              |  1 +
>  target/s390x/cpu.h              |  1 +
>  target/s390x/cpu_features.c     |  2 ++
>  target/s390x/cpu_features_def.h |  2 ++
>  target/s390x/gen-features.c     |  2 ++
>  target/s390x/kvm.c              | 14 ++++++++++++++
>  target/s390x/machine.c          | 17 +++++++++++++++++
>  9 files changed, 46 insertions(+), 7 deletions(-)
> 

Thanks, queued to s390-next (with a proper headers update).