Re: [Qemu-devel] [PATCH v4 1/8] fw_cfg: max access size and region size are the same for MMIO data reg

[Paolo Bonzini <pbonzini@redhat.com>]

On 16/12/2014 21:17, Laszlo Ersek wrote:
>> > I can't imagine how that would happen; fw_cfg_data_mem_read() ignores
>> > both "addr" and "size", and fw_cfg_read() simply advances the
>> > "cur_offset" member.
> Ah okay, I understand your point now; you're probably saying that
> access_with_adjusted_size() traverses the offsets in the wrong order.
> ... I don't see how; the only difference in the access() param list is
> the shift count. (I don't know how it should work by design.)

I think I have figured it out.

Guest endianness affects where those bytes are placed, but not the order
in which they are fetched; and the effects of guest endianness are
always cleaned up by adjust_endianness, so ultimately they do not matter.

Think of how you would implement the uint64_t read in fw_cfg:

File bytes         12 34 56 78 9a bc de f0

fw_cfg_data_mem_read should read

size==4 BE host    0x12345678
size==4 LE host    0x78563412
size==8 BE host    0x123456789abcdef0
size==8 LE host    0xf0debc9a78563412

So the implementation of fw_cfg_data_mem_read must depend on host
endianness.  Instead, memory.c always fills in bytes in the same order,
on the assumption that the reads are idempotent.

Paolo