Re: [Qemu-devel] [RFC] Alpha system patchset updated for Memory API, batch 2: PCI devices

[Richard Henderson]

On 08/02/2011 02:23 AM, Avi Kivity wrote:
>> +const MemoryRegionOps alpha_pci_bw_io_ops = {
>> +    .read = bw_io_read,
>> +    .write = bw_io_write,
>> +    .endianness = DEVICE_LITTLE_ENDIAN,
>> +    .impl = {
>> +        .min_access_size = 1,
>> +        .max_access_size = 4,
>> +    },
>> +};
>> +
> 
> The plan for this (and all mmio-mapped PCI_ADDRESS_SPACE_IO regions)
> is to map the I/O address space as a subregion of the PCI host
> bridge, eliminating this thunk.

Yep, I figured that.  There's even a comment to that effect in
alpha_typhoon.c, mentioning that such a change is waiting on 
the ISA devices to be converted.


r~