Re: [PATCH 0/7] hw/ide: Untangle ISA/PCI abuses of ide_init_ioport()

[Philippe Mathieu-Daudé]

On 8/2/23 01:07, Philippe Mathieu-Daudé wrote:
> Background thread:
> 5095dffc-309b-6c72-d255-8cdaa6fd3d52@ilande.co.uk/">https://lore.kernel.org/qemu-devel/5095dffc-309b-6c72-d255-8cdaa6fd3d52@ilande.co.uk/
>
> The ide_init_ioport() method expect an ISA device, but is
> massaged to accept NULL device (IOW, non-ISA devices...).
>
> A plausible explanation is QOM objects can only inherit one
> parent, and ide_init_ioport() was first used with ISA children.
> Later it was adapted to accept PCI children in a convoluted
> way. The PIIX IDE function abuse it.
>
> This series rename the current ide_init_ioport() as
> ide_init_ioport_isa(), then add a generic ide_init_ioport()
> which works with PCI devices.
>
> This is required to proceed with more PIIX cleanups.
>
> Philippe Mathieu-Daudé (7):
>    hw/isa: Un-inline isa_bus_from_device()
>    hw/isa: Use isa_address_space_io() to reduce access on global 'isabus'
>    hw/ide: Rename ISA specific ide_init_ioport() as ide_init_ioport_isa()
>    hw/ide: Introduce generic ide_init_ioport()
>    hw/ide/piix: Use generic ide_init_ioport()
>    hw/isa: Assert isa_register_portio_list() gets non-NULL ISA device
>    hw/ide/piix: Remove dead code in pci_piix_init_ports()

Forgot:
Based-on: <20230207234615.77300-1-philmd@linaro.org>
"exec/ioport: Factor portio_list_register[flush_coalesced]() out"
https://lore.kernel.org/qemu-devel/20230207234615.77300-1-philmd@linaro.org/