On Tue, 22 Oct 2019 at 17:23, Marc-André Lureau
<marcandre.lur...@redhat.com> wrote:
>
> Split IRQ and memory region initialization in a different function to
> be reused by callers that realize the serial device themself, and
> connect irq/mem after.
>
> Signed-off-by: Marc-André Lureau <marcandre.lur...@redhat.com>
> ---
> hw/char/serial.c | 20 +++++++++++++-------
> include/hw/char/serial.h | 3 +++
> 2 files changed, 16 insertions(+), 7 deletions(-)
>
> diff --git a/hw/char/serial.c b/hw/char/serial.c
> index 72d48b5cd8..41548e159c 100644
> --- a/hw/char/serial.c
> +++ b/hw/char/serial.c
> @@ -1078,27 +1078,33 @@ static const MemoryRegionOps serial_mm_ops[3] = {
> },
> };
>
> +void serial_mm_connect(SerialMMState *self, enum device_endian end,
> + MemoryRegion *address_space, qemu_irq irq)
> +{
> + SerialState *s = SERIAL(self);
> +
> + qdev_connect_gpio_out_named(DEVICE(self), "serial-irq", 0, irq);
> + memory_region_init_io(&s->io, NULL, &serial_mm_ops[end], self,
> + "serial", 8 << self->regshift);
> + memory_region_add_subregion(address_space, s->base, &s->io);
> +}
This function isn't needed if you make SerialMMState a sysbus
device which provides the MMIO region and the IRQ in the
usual way.
thanks
-- PMM
