Il 06/05/2013 17:02, Jan Kiszka ha scritto: > On 2013-05-06 16:59, Paolo Bonzini wrote: >> Il 06/05/2013 16:55, Andreas Färber ha scritto: >>> Am 06.05.2013 16:26, schrieb Jan Kiszka: >>>> Move isa_is_ioport_assigned to the ISA core and implement it via a >>>> memory region lookup. As all IO ports are now directly or indirectly >>>> registered via the memory API, this becomes possible and will finally >>>> allow us to drop the ioport tables. >>>> >>>> Signed-off-by: Jan Kiszka <jan.kis...@siemens.com> >>>> --- >>>> hw/acpi/piix4.c | 6 +++--- >>>> hw/isa/isa-bus.c | 11 +++++++++++ >>>> hw/isa/lpc_ich9.c | 8 ++++---- >>>> include/exec/ioport.h | 1 - >>>> include/hw/isa/isa.h | 2 ++ >>>> ioport.c | 7 ------- >>>> 6 files changed, 20 insertions(+), 15 deletions(-) >>>> >>>> diff --git a/hw/acpi/piix4.c b/hw/acpi/piix4.c >>>> index c4af1cc..5955217 100644 >>>> --- a/hw/acpi/piix4.c >>>> +++ b/hw/acpi/piix4.c >>>> @@ -386,10 +386,10 @@ static void piix4_pm_machine_ready(Notifier *n, void >>>> *opaque) >>>> uint8_t *pci_conf; >>>> >>>> pci_conf = s->dev.config; >>>> - pci_conf[0x5f] = (isa_is_ioport_assigned(0x378) ? 0x80 : 0) | 0x10; >>>> + pci_conf[0x5f] = (isa_is_ioport_assigned(NULL, 0x378) ? 0x80 : 0) | >>>> 0x10; >>>> pci_conf[0x63] = 0x60; >>>> - pci_conf[0x67] = (isa_is_ioport_assigned(0x3f8) ? 0x08 : 0) | >>>> - (isa_is_ioport_assigned(0x2f8) ? 0x90 : 0); >>>> + pci_conf[0x67] = (isa_is_ioport_assigned(NULL, 0x3f8) ? 0x08 : 0) | >>>> + (isa_is_ioport_assigned(NULL, 0x2f8) ? 0x90 : 0); >>>> >>>> } >>>> >>> >>> Is there really no way to access the ISABus from this device? Would be >>> nice to get rid of global ISA variables and not introduce more >>> dependencies. :) >> >> There's always a way to find the ISABus via QOM: >> >> ISABus *isa_bus = (ISABus *) object_resolve_path_type("", TYPE_ISA_BUS, >> NULL); > > Err, in what way is this better? It also assumes that there is only one.
I didn't say it is better. :) Unfortunately, the PIIX4 has these register on the "wrong" function, ICH9 fixed it. You could make this take an AddressSpace instead of an ISABus, and use pci_address_space_io. The assumption then becomes that the ISA and PM devices have the same address spaces, which is somewhat hackish but reasonable. In fact, with the other new API I have added (address_space_valid), this would become address_space_valid(pci_address_space_io(dev), 0x2f8, 1) and similar. I need to check, but you could remove isa_is_ioport_assigned completely. Paolo