On 05/23/2010 03:34 PM, Blue Swirl wrote:
Signed-off-by: Blue Swirl<blauwir...@gmail.com> --- hw/cirrus_vga.c | 12 ++++++------ hw/e1000.c | 2 +- hw/eepro100.c | 2 +- hw/isa.h | 1 + hw/isa_mmio.c | 18 ++++++++++++++++-- hw/lsi53c895a.c | 4 ++-- hw/macio.c | 24 ++++++++++++------------ hw/msix.c | 4 ++-- hw/openpic.c | 6 +++--- hw/pcnet.c | 3 ++- hw/rtl8139.c | 2 +- hw/sun4u.c | 7 +++++-- hw/usb-ohci.c | 2 +- hw/vga-pci.c | 4 ++-- hw/vmware_vga.c | 7 +++---- hw/wdt_i6300esb.c | 2 +- 16 files changed, 59 insertions(+), 41 deletions(-)diff --git a/hw/cirrus_vga.c b/hw/cirrus_vga.c index ba48289..52e51e0 100644 --- a/hw/cirrus_vga.c +++ b/hw/cirrus_vga.c @@ -3145,10 +3145,10 @@ static void cirrus_pci_lfb_map(PCIDevice *d, int region_num, CirrusVGAState *s =&DO_UPCAST(PCICirrusVGAState, dev, d)->cirrus_vga; /* XXX: add byte swapping apertures */ - cpu_register_physical_memory(addr, s->vga.vram_size, - s->cirrus_linear_io_addr); - cpu_register_physical_memory(addr + 0x1000000, 0x400000, - s->cirrus_linear_bitblt_io_addr); + pci_register_memory(d->bus, addr, s->vga.vram_size, + s->cirrus_linear_io_addr); + pci_register_memory(d->bus, addr + 0x1000000, 0x400000, + s->cirrus_linear_bitblt_io_addr);
It would probably be better to pass the PCIDevice instead of the bus. That would allow a per-device PCI IOMMU to be used (such as emulated VT-d).
Ultimately, I don't think devices should be handling their IO region mappings but I think this could be a reasonable intermediate solution.
Regards, Anthony Liguori
