On Thu, Aug 15, 2019 at 02:25:07AM +0200, BALATON Zoltan wrote:
> The MacOS driver exits if the card does not have an interrupt. If we
> set PCI_INTERRUPT_PIN to 1 then it enables VBlank interrupts and it
> boots but the mouse poniter can not be moved. This patch implements a
> dummy VBlank interrupt by a timer triggered at 60 Hz to test if it
> helps. Unfortunately it doesn't: MacOS with this patch hangs during
> boot just polling interrupts and acknowledging them so maybe it needs
> something else or there may be some other problem with this
> implementation.
>
> This is posted for comments and to let others experiment with it but
> probably should not be committed upstream yet.
>
> Signed-off-by: BALATON Zoltan <bala...@eik.bme.hu>
> ---
> hw/display/ati.c | 41 +++++++++++++++++++++++++++++++++++++++++
> hw/display/ati_dbg.c | 1 +
> hw/display/ati_int.h | 4 ++++
> hw/display/ati_regs.h | 1 +
> 4 files changed, 47 insertions(+)
>
> diff --git a/hw/display/ati.c b/hw/display/ati.c
> index a365e2455d..e06cbf3e91 100644
> --- a/hw/display/ati.c
> +++ b/hw/display/ati.c
> @@ -243,6 +243,21 @@ static uint64_t ati_i2c(bitbang_i2c_interface *i2c,
> uint64_t data, int base)
> return data;
> }
>
> +static void ati_vga_update_irq(ATIVGAState *s)
> +{
> + pci_set_irq(&s->dev, s->regs.gen_int_status & 1);
This should be "s->regs.gen_int_status & s->regs.gen_int_cntl" I guess?
> +static void ati_vga_vblank_irq(void *opaque)
> +{
> + ATIVGAState *s = opaque;
> +
> + timer_mod(&s->vblank_timer, qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) +
> + NANOSECONDS_PER_SECOND / 60);
> + s->regs.gen_int_status |= 1;
#defines for the irq status bits would be nice.
> + case GEN_INT_CNTL:
> + s->regs.gen_int_cntl = data;
> + if (data & 1) {
> + ati_vga_vblank_irq(s);
> + } else {
> + timer_del(&s->vblank_timer);
> + }
ati_vga_update_irq() needed here.
> + break;
> + case GEN_INT_STATUS:
> + data &= (s->dev_id == PCI_DEVICE_ID_ATI_RAGE128_PF ?
> + 0x000f040fUL : 0xfc080effUL);
Add IRQ_MASK #define ?
> + s->regs.gen_int_status &= ~data;
ati_vga_update_irq() needed here too.
cheers,
Gerd
