Re: [Xen-devel] [XEN][RFC PATCH 01/15] hvm: Modify interface to support multiple ioreq server

[Julien Grall <julien.grall@citrix.com>]

On 03/23/2012 11:33 AM, Ian Campbell wrote:
> On Thu, 2012-03-22 at 15:59 +0000, Julien Grall wrote:
>    
>> Add structure to handle ioreq server. It's server which can
>> handle a range of IO (MMIO and/or PIO) and emulate a PCI.
>> Each server as its own shared page to receive ioreq. So
>> we have introduced to HVM PARAM to set/get the first and
>> the last shared used for ioreq.
>> With it's id, the server knows which page it must use.
>>      
> So id is always the page offset with the range? Why not just call it
> iobuf_offset then? Is the additional layer of abstraction from calling
> it "id" useful if we are just going to peek around it?
>    
Indeed, but this parameter is also used to register bdf/io range.
So can we call it server_id or server_offset ?

>> diff --git a/xen/include/public/hvm/hvm_op.h b/xen/include/public/hvm/hvm_op.h
>> index 6a78f75..1e0e27b 100644
>> --- a/xen/include/public/hvm/hvm_op.h
>> +++ b/xen/include/public/hvm/hvm_op.h
>> @@ -24,6 +24,8 @@
>>   #include "../xen.h"
>>   #include "../trace.h"
>>
>> +#include "hvm_info_table.h" /* HVM_MAX_VCPUS */
>>      
> You don't appear to use HVM_MAX_VCPUS anywhere in your additions?
>
>    
I use it in xen-all.c in QEMU. It permits to check that smp_cpus is lower
that the maximum of vpcus. It avoids page overflow.