This series applies miscellaneous fixes and enhancements to Hyper-V
emulation code in QEMU, and lays out the ground for VMBus devices.

v1 -> v2:
 - drop the already merged patch
 - split and rework SINTx and SVERSION msrs init
 - factor out hyperv vcpu init to a function
 - rework vp_index management
 - distinguish kvm-only (== legacy) mode for SynIC
 - use new capabilities recently submitted to KVM
 - add compat logic for SynIC
 - drop workaround for KVM zeroing SynIC pages
 - minor fixes according to comments

Evgeny Yakovlev (1):
  hyperv: set partition-wide MSRs only on first vcpu

Roman Kagan (22):
  hyperv: add header with protocol definitions
  update-linux-headers: prepare for hyperv.h removal
  hyperv: ensure SINTx msrs are reset properly
  hyperv: make SynIC version msr constant
  [not to commit] add new hyperv-related caps
  hyperv: ensure VP index equal to QEMU cpu_index
  hyperv_testdev: refactor for readability
  hyperv: cosmetic: g_malloc -> g_new
  hyperv: synic: only setup ack notifier if there's a callback
  hyperv: allow passing arbitrary data to sint ack callback
  hyperv: address HvSintRoute by X86CPU pointer
  hyperv: make HvSintRoute reference-counted
  hyperv: qom-ify SynIC
  hyperv: block SynIC use in QEMU in incompatible configurations
  hyperv: make overlay pages for SynIC
  hyperv: add synic message delivery
  hyperv: add synic event flag signaling
  hyperv: process SIGNAL_EVENT hypercall
  hyperv: process POST_MESSAGE hypercall
  hyperv_testdev: add SynIC message and event testmodes
  MAINTAINERS: add myself and eyakovlev@ for hyperv*
  hyperv: update copyright notices

 include/hw/i386/pc.h            |   5 +
 linux-headers/linux/kvm.h       |   2 +
 target/i386/cpu.h               |  16 +-
 target/i386/hyperv.h            |  40 ++-
 target/i386/hyperv_proto.h      | 257 ++++++++++++++++
 hw/misc/hyperv_testdev.c        | 267 +++++++++++++----
 target/i386/cpu.c               |   4 +-
 target/i386/hyperv.c            | 634 ++++++++++++++++++++++++++++++++++++++--
 target/i386/kvm.c               | 167 +++++++----
 target/i386/machine.c           |  24 +-
 MAINTAINERS                     |   7 +
 scripts/update-linux-headers.sh |   4 +-
 12 files changed, 1256 insertions(+), 171 deletions(-)
 create mode 100644 target/i386/hyperv_proto.h

-- 
2.9.4

On Wed, 21 Jun 2017 19:24:01 +0300
Roman Kagan <rkagan@virtuozzo.com> wrote:

> This series applies miscellaneous fixes and enhancements to Hyper-V
> emulation code in QEMU, and lays out the ground for VMBus devices.

Series is a big random mix of cleanup/refactoring/new features
which is hard to review due to mixed content of above
within patches as well inter them.

I suggest to split series in at least in 2
 1: make cleanups/refactoring first
and on top of this
 2: add new features

* hyperv_find_vcpu() with expensive lookup and related renaming
  might be not needed at all if local_apics[12/23] is used.

*pls add in cover letter a link to public git repo where
 it would be possible to pull from for testing/review.

PS:
if renaming, do it consistently.
ex:
 vcpu_id -> vp_index touches only hyperv.c
 but leaves old naming in user of kvm_hv_sint_route_create(): hyperv_testdev.c

> v1 -> v2:
>  - drop the already merged patch
>  - split and rework SINTx and SVERSION msrs init
>  - factor out hyperv vcpu init to a function
>  - rework vp_index management
>  - distinguish kvm-only (== legacy) mode for SynIC
>  - use new capabilities recently submitted to KVM
>  - add compat logic for SynIC
>  - drop workaround for KVM zeroing SynIC pages
>  - minor fixes according to comments
> 
> Evgeny Yakovlev (1):
>   hyperv: set partition-wide MSRs only on first vcpu
> 
> Roman Kagan (22):
>   hyperv: add header with protocol definitions
>   update-linux-headers: prepare for hyperv.h removal
>   hyperv: ensure SINTx msrs are reset properly
>   hyperv: make SynIC version msr constant
>   [not to commit] add new hyperv-related caps
>   hyperv: ensure VP index equal to QEMU cpu_index
>   hyperv_testdev: refactor for readability
>   hyperv: cosmetic: g_malloc -> g_new
>   hyperv: synic: only setup ack notifier if there's a callback
>   hyperv: allow passing arbitrary data to sint ack callback
>   hyperv: address HvSintRoute by X86CPU pointer
>   hyperv: make HvSintRoute reference-counted
>   hyperv: qom-ify SynIC
>   hyperv: block SynIC use in QEMU in incompatible configurations
>   hyperv: make overlay pages for SynIC
>   hyperv: add synic message delivery
>   hyperv: add synic event flag signaling
>   hyperv: process SIGNAL_EVENT hypercall
>   hyperv: process POST_MESSAGE hypercall
>   hyperv_testdev: add SynIC message and event testmodes
>   MAINTAINERS: add myself and eyakovlev@ for hyperv*
>   hyperv: update copyright notices
> 
>  include/hw/i386/pc.h            |   5 +
>  linux-headers/linux/kvm.h       |   2 +
>  target/i386/cpu.h               |  16 +-
>  target/i386/hyperv.h            |  40 ++-
>  target/i386/hyperv_proto.h      | 257 ++++++++++++++++
>  hw/misc/hyperv_testdev.c        | 267 +++++++++++++----
>  target/i386/cpu.c               |   4 +-
>  target/i386/hyperv.c            | 634 ++++++++++++++++++++++++++++++++++++++--
>  target/i386/kvm.c               | 167 +++++++----
>  target/i386/machine.c           |  24 +-
>  MAINTAINERS                     |   7 +
>  scripts/update-linux-headers.sh |   4 +-
>  12 files changed, 1256 insertions(+), 171 deletions(-)
>  create mode 100644 target/i386/hyperv_proto.h
> 

On Thu, Jun 29, 2017 at 05:20:30PM +0200, Igor Mammedov wrote:
> On Wed, 21 Jun 2017 19:24:01 +0300
> Roman Kagan <rkagan@virtuozzo.com> wrote:
> 
> > This series applies miscellaneous fixes and enhancements to Hyper-V
> > emulation code in QEMU, and lays out the ground for VMBus devices.
> 
> Series is a big random mix of cleanup/refactoring/new features
> which is hard to review due to mixed content of above
> within patches as well inter them.
> 
> I suggest to split series in at least in 2
>  1: make cleanups/refactoring first
> and on top of this
>  2: add new features

Indeed, I started to realize this.  I'll try to re-split better when
respinning.

> * hyperv_find_vcpu() with expensive lookup and related renaming
>   might be not needed at all if local_apics[12/23] is used.
> 
> *pls add in cover letter a link to public git repo where
>  it would be possible to pull from for testing/review.

I'm in the process of setting it up, will include it next time.

> PS:
> if renaming, do it consistently.
> ex:
>  vcpu_id -> vp_index touches only hyperv.c
>  but leaves old naming in user of kvm_hv_sint_route_create(): hyperv_testdev.c

I strived for eventual consistency: no vcpu_id is left after the last
patch in the series.  But it's probably harder to review patch by patch.
So I'll redo the way you suggest.

Thanks,
Roman.