Re: [Xen-devel] [PATCH v4 01/15] x86/pmtimer: Move ACPI registers from PMTState to hvm_domain

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 01/12/16 16:28, Boris Ostrovsky wrote:
> On 12/01/2016 10:52 AM, Jan Beulich wrote:
>>> --- a/xen/include/public/arch-x86/hvm/save.h
>>> +++ b/xen/include/public/arch-x86/hvm/save.h
>>> @@ -525,16 +525,16 @@ DECLARE_HVM_SAVE_TYPE(HPET, 12, struct hvm_hw_hpet);
>>>  
>>>  
>>>  /*
>>> - * PM timer
>>> + * ACPI registers
>>>   */
>>>  
>>> -struct hvm_hw_pmtimer {
>>> +struct hvm_hw_acpi {
>>>      uint32_t tmr_val;   /* PM_TMR_BLK.TMR_VAL: 32bit free-running counter 
>>> */
>>>      uint16_t pm1a_sts;  /* PM1a_EVT_BLK.PM1a_STS: status register */
>>>      uint16_t pm1a_en;   /* PM1a_EVT_BLK.PM1a_EN: enable register */
>>>  };
>>>  
>>> -DECLARE_HVM_SAVE_TYPE(PMTIMER, 13, struct hvm_hw_pmtimer);
>>> +DECLARE_HVM_SAVE_TYPE(ACPI, 13, struct hvm_hw_acpi);
>> However much I appreciate this switch to a better name, I'm not
>> convinced we can actually do this as easily: There's no
>> __XEN_TOOLS__ guard anywhere in this file, and hence everything
>> here is part of the stable ABI. I'm afraid you minimally will have to
>> add interface version guards, retaining the old naming for old
>> consumers.
>
> Right, I haven't though about out-of-tree users. Should new fields
> (added in patch 7) also be guarded?

Be aware that my Hypervisor migration v2 plans (which follow the CPUID
plans) will remove all of this (as it should never have gotten into the
ABI to start with), and replace it with something looking suspiciously
like the other migration v2 stream formats.

If you can get away without changing names for now, probably best to
just leave comment.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel