Xen project Mailing List

Thanks for explanation. It makes sense then.

From: Corneliu ZUZU [mailto:czuzu@xxxxxxxxxxxxxxx]
Sent: Monday, July 11, 2016 1:32 PM
To: Tian, Kevin; xen-devel@xxxxxxxxxxxxx
Cc: Andrew Cooper; George Dunlap; Jan Beulich; Julien Grall; Nakajima, Jun; Paul Durrant; Razvan Cojocaru; Stefano Stabellini; Tamas K Lengyel
Subject: Re: [PATCH 00/16] x86/vm-event: numerous adjustments & fixes

On 7/11/2016 5:54 AM, Tian, Kevin wrote:

what's the difference between this series and earlier one?

[PATCH v3 0/8] x86/vm-event: Adjustments & fixes

looks you have some patches cross-posted (e.g. 1/16)...

Corneliu ZUZU (16):

  x86/vmx_update_guest_cr: minor optimization

  x86: fix: make atomic_read() param const

  x86/monitor: mechanical renames

  x86/monitor: relocate vm_event_register_write_resume() function to

    monitor code

  x86/monitor: relocate code more appropriately

  x86/monitor: fix: set msr_bitmap to NULL after xfree

  x86/vm-event: fix: call cleanup when init fails, to free partial

    allocs

  x86/vm-event: call monitor init & cleanup funcs from respective

    vm_event funcs

  arm/monitor: move d->monitor cleanup to monitor_cleanup_domain()

  x86/vm-event: centralize vcpu-destroy cleanup in vm-events code

  x86/monitor: fix: treat -monitor- properly, as a subsys of the

    vm-event subsys

  x86/vm-event: fix: move cleanup of mem_access_emulate_each_rep to

    monitor stub

  x86/monitor: introduce writes_pending field in monitor_write_data

  x86/monitor: clarify separation between monitor subsys and vm-event as

    a whole

  x86/monitor: fix: don't compromise a monitor_write_data with pending

    CR writes

  x86/monitor: fix: xc_monitor _write_ctrlreg w/o previous _enable must

    fail

--

2.5.0

Hi Kevin,

I'm sorry about the confusion, the _older v3 series is superseded by this 16-patches one_. I thought the latter was too complex to consider as a v4 over the former, but I should have at least stated something to lead you guys to that conclusion (and to specify changes..). The patches that are still in this series from the older one, the changes and the acks are as following:

- P 1/16, was 1/8 in v3
    * changed since v3: nothing
    * acked-by in old v3: Acked-by: Kevin Tian <kevin.tian@xxxxxxxxx>

- P 5/16, was 2/8 in v3
    * changed since v3:
        - title slightly changed
        - renames due to 3/16: arch_monitor_write_data()->monitor_ctrlreg_write_data(), write_ctrlreg_adjust_traps()->monitor_ctrlreg_adjust_traps(), write_ctrlreg_disable_traps()->monitor_ctrlreg_disable_traps()
        - added 'const' attribute to "struct *domain" param of vmx_vm_event_update_cr3_traps
        - 'index' param of monitor_ctrlreg_adjust_traps now of type 'unsigned int'
    * acked-by in old v3: no-one

- P 15/16, was 3/8 in v3:
     * changed since v3 (IMO changes are too many and depends on previous patches in this 16-p-series and it would be better if the old v3 one is simply forgotten-about):
        - separated fields in arch_vm_event in another structure and arch_vm_event only holds a pointer to that struct (arch_vm_event->monitor->...) - see 14/16 for details
        - arch_vm_event.monitor freed entirely when the pending writes are committed (monitor_ctrlreg_write_data()) or when the domain/vcpu is destroyed
     * acked-by in old v3: no-one

- P 16/16, was 4/8 in v3:
     * changed since v3: use ENODEV error code instead of ENOSYS
     * acked-by in old v3: no-one

The rest of patches in v3 (P5-P8) already made it to staging.

Sorry again and thanks,
Zuzu C.

Re: [Xen-devel] [PATCH 00/16] x86/vm-event: numerous adjustments & fixes