Re: [Qemu-devel] [PATCH v3 11/14] ioport: Switch dispatching to memory core layer

[Peter Maydell]

On 14 July 2013 16:18, Anthony Liguori <address@hidden> wrote:
> On Sun, Jul 14, 2013 at 9:58 AM, Peter Maydell <address@hidden> wrote:
>> Devices themselves do have a byte order, though, right? Specifically,
>> if you do a 32 bit read of address 0 on a device and an 8 bit read,
>
> It depends on the bus and device.  Busses don't necessary pass the I/O
> size down to the device like that.  If it does, the device may do any
> number of things (including ignoring the request entirely.

Agreed.

> What's most common AFAIK is that the access is treated as a word
> access and then truncated.  IOW, the device sees the 32-bit word read
> but somewhere along the way, the top 24 bits are discarded.
>
> The real interesting question is what happens when you do a byte
> access at address 1.  I think most devices simply don't allow that.

Mmm, and where it is permitted it's going to be device dependent
(for instance the ARM GIC has most registers be 32-bit access only,
with a few which allow 8-bit access; and the GIC spec states that
those byte accesses use a little endian memory order model).

It looks like most of the ARM PLxxx devices on the APB bus simply
don't wire up address lines 0 and 1, so an access to address 1
will behave like one to address 0. (Of course we don't model this.)

-- PMM