On 22 November 2011 11:06, Juan Quintela <quint...@redhat.com> wrote:
> Peter Chubb <pet...@gelato.unsw.edu.au> wrote:
>> Hi Peter,
>> Please find appended a patch containing initial support for the
>> FreeScale i.MX31 and the KZM Arm11 evaluation board.
>>
>> The implementation was originally written by Hans Jang and Adam
>> Clench of OK-Labs; I've updated it to the current qdev and memory
>> region paradigms and implemented enough extra that Linux will boot
>> on the patched QEMU using a ram disk.
>>
>> The i.MX 31 Serial controller is found in most of the i.MX SoCs;
>> the AVIC and timer implementations can also be shared, albeit with
>> fewer chips.
>>
>> Signed-off-by: Peter Chubb <peter.ch...@nicta.com.au>
>> Signed-off-by: Hans Jang <hsj...@ok-labs.com>
>> Signed-off-by: Adam Clench <ad...@ok-labs.com>
>
>> +static const VMStateDescription vmstate_imxg_timer = {
>> + .name = "imxg-timer",
>> + .version_id = 1,
>> + .minimum_version_id = 1,
>> + .minimum_version_id_old = 1,
>> + .fields = (VMStateField[]) {
>> + VMSTATE_UINT32(cr, imxg_timer_state),
>> + VMSTATE_UINT32(sr, imxg_timer_state),
>> + VMSTATE_UINT32(ir, imxg_timer_state),
>> + VMSTATE_UINT32(cnt, imxg_timer_state),
>> + VMSTATE_UINT32(ocr1, imxg_timer_state),
>> + VMSTATE_TIMER(timer, imxg_timer_state),
>> + VMSTATE_END_OF_LIST()
>> + }
>> +};
>
> This is not used. Shouldn't a call like this be needed?
>
>> +static int imxg_timer_init(SysBusDevice *dev)
>> +{
>> + imxg_timer_state *s = FROM_SYSBUS(imxg_timer_state, dev);
>> +
>> + sysbus_init_irq(dev, &s->irq);
>> + memory_region_init_io(&s->iomem, &imxg_timer_ops,
>> + s, "imxg-timer",
>> + 0x00001000);
>> + sysbus_init_mmio_region(dev, &s->iomem);
>> +
>> + s->timer = qemu_new_timer_ns(vm_clock, imxg_timer_timeout, s);
>> + s->cr = 0;
>> + s->ir = 0;
>> + s->pr = 0;
>> + s->ocr1 = 0;
>> + imxg_timer_update_count(s);
>
> vmstate_register(&dev->qdev, -1, &vmstate_imxg_timer, s);
> ???
Any particular reason to prefer a vmstate_register() call
over just having a SysBusDeviceInfo struct with the .qdev.vmsd
field set? I tend to prefer the latter myself.
-- PMM
