Re: [Xen-devel] [PATCH v3 09/19] acpi/hvmloader: Include file/paths adjustments

[Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>]

On 09/08/2016 10:05 AM, Jan Beulich wrote:
>>
>> diff --git a/tools/firmware/hvmloader/acpi/libacpi.h 
>> b/tools/firmware/hvmloader/acpi/libacpi.h
>> index 3bcd226..d803139 100644
>> --- a/tools/firmware/hvmloader/acpi/libacpi.h
>> +++ b/tools/firmware/hvmloader/acpi/libacpi.h
>> @@ -84,6 +84,13 @@ struct acpi_config {
>>  
>>      /* RSDP address */
>>      unsigned int rsdp;
>> +
>> +    /* x86-specific parameters */
>> +    uint16_t (*lapic_id)(unsigned cpu);
> Why uint16_t? Legacy APIC IDs in MADT are in an 8-bit field, and
> x2APIC IDs require 32 bits. Depending on whether we want to be
> able to re-use the same function for x2APIC support (once that
> gets added) the return type should change accordingly. (But if we
> were to re-use it, I guess a second parameter would then also be
> needed.)

No good reason for uint16_t, should be uint8_t.

I think deciding on how interface will look for x2APIC should be
deferred until when we support it.

>
> And then - down the road you're not planning to build a shared
> library using this interface? Otherwise we'd need to consider some
> compatibility aspects.

No specific plans for now, the objects are expected to be built and
linked directly to whoever wants to use this (so libacpi is somewhat of
a misnomer, similar to libelf).


-boris


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