Re: [Qemu-devel] RFC: ioapic polarity vs. qemu os-x guest

[Gabriel L. Somlo]

On Tue, Feb 11, 2014 at 09:54:44PM +0200, Michael S. Tsirkin wrote:
> On Tue, Feb 11, 2014 at 01:23:31PM -0500, Gabriel L. Somlo wrote:
> > I'm trying to get OS X to work as a QEMU guest, and one of the few
> > remaining "mysteries" I need to solve is that the OS X guest hangs
> > during boot, waiting for its boot disk to be available, unless the
> > following KVM patch is applied:
> > [...]
> > 2. Is there anything in QEMU (besides the ACPI DSDT .dsl files) which
> > has a hardcoded assumption re. "polarity == 0", or active-high, for
> > level-triggered interrupts? I tried to dig through hw/i386/kvm/ioapic.c
> > and a bunch of other files, but couldn't isolate anything that I could
> > "flip" to fix things in userspace.
> > 
> > 
> > Any ideas or suggestions about the appropriate way to move forward would
> > be much appreciated !!!
> > 
> > 
> > Thanks much,
> > --Gabriel
> 
> I think changing ACPI is the right thing to
> do really. But we'll need to fix some things
> first of course.
> 
> I think it's PC Q35 that has this assumption.
> hw/i386/pc_q35.c
> 
>         gsi = qemu_allocate_irqs(kvm_pc_gsi_handler, gsi_state,
>                                  GSI_NUM_PINS);
> 
> kvm_pc_gsi_handler simply forwards interrupts to kvm.
> 
> and
> 
> hw/isa/lpc_ich9.c
> static void ich9_lpc_update_pic(ICH9LPCState *lpc, int pic_irq)
> {
>     int i, pic_level;
> 
>     /* The pic level is the logical OR of all the PCI irqs mapped to it */
>     /* The pic level is the logical OR of all the PCI irqs mapped to it
>  * */
>     pic_level = 0;
>     for (i = 0; i < ICH9_LPC_NB_PIRQS; i++) {
>         int tmp_irq;
>         int tmp_dis;
>         ich9_lpc_pic_irq(lpc, i, &tmp_irq, &tmp_dis);
>         if (!tmp_dis && pic_irq == tmp_irq) {
>             pic_level |= pci_bus_get_irq_level(lpc->d.bus, i);
>         }
>     }
> 
> so somewhere we need to flip it, I am guessing in ich9
> along the lines of:
> 
> -    pic_level = 0;
> -            pic_level |= pci_bus_get_irq_level(lpc->d.bus, i);
> +    pic_level = 1;
> +            pic_level &= !pci_bus_get_irq_level(lpc->d.bus, i);

I think now we're on to something!

I managed to boot OS X on q35 with absolutely no kernel patches, but
Linux still hated it ("irqXX: nobody cared"). At least now I know what
I'm looking for, so I'll try to come up with a way to flip
level-triggered polarity to ActiveLow across all of i386, in a way
that works for Linux and Windows guests as well.

Thanks again for getting me unstuck!
--Gabriel