Re: [Qemu-devel] [PATCH RFC 0/5] Xen: introduce Xen PV target

[Peter Maydell <peter.maydell@linaro.org>]

On 24 January 2014 14:30, Paolo Bonzini <pbonzini@redhat.com> wrote:
> Il 23/01/2014 23:30, Peter Maydell ha scritto:
>> I'm afraid I still think this is a terrible idea. "Xen" isn't a CPU, and
>> "the binary is smaller" isn't IMHO sufficient justification for breaking
>> QEMU's basic structure of "target-* define target CPUs and we have
>> a lot of compile time constants which are specific to a CPU which
>> get defined there". How would you support a bigendian Xen CPU,
>> just to pick one example of where this falls down?
>
>
> (1) decide that the Xen ring buffers are little-endian even on big-endian
> CPUs
>
> (2) communicate the endianness of the Xen ring buffers via Xenstore, just
> like we do for sizeof(long), and let the guest use either endianness on any
> architecture.

You still have to make a choice about what you think
TARGET_WORDS_BIGENDIAN should be, and it's still going
to be wrong half the time and horribly confusing.
I just think this is completely the wrong solution to
the problem.

If Xen really wants a totally standalone binary of the
smallest possible size with just paravirtualized hardware
and minimal to no dependency on guest CPU architecture then
they should write one, along the lines of kvmtool :-)

thanks
-- PMM