On 08/04/2011 11:17 AM, Avi Kivity wrote:
On 08/04/2011 07:14 PM, Avi Kivity wrote:
Yes, I'm just talking about the function names, not about the
implementation.
qemu_chr_fe_write(chr,...) -> qemu_chr_write(chr, ...)
qemu_chr_be_write(chr, ...) -> qemu_chr_write(&chr->backend, ...)
And, if you want an internal pipe:
QemuPipeEndpoint pipe1, pipe2;
qemu_chr_pipe_init(&pipe0, &pipe1);
so clients don't have to choose between frontend and backend. Finally
you can connect COM1: to COM2: with a qemu-provided null modem cable!
Yes, that's exactly the goal :-)
The only problem left to be resolved is how to handle ioctl().
qemu_chr_ioctl() forces asymmetry today.
I think if we merge ioctl into qemu_chr_event() by adding a return and
data payload, it'll make the semantics be synchronous messaging and
it'll be easy to make ioctl work.
Regards,
Anthony Liguori