qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 02/52] ide: split ioport registration to a separ


From: BALATON Zoltan
Subject: Re: [Qemu-devel] [PATCH 02/52] ide: split ioport registration to a separate file
Date: Wed, 30 Jan 2019 16:55:27 +0100 (CET)
User-agent: Alpine 2.21.9999 (BSF 287 2018-06-16)

On Wed, 30 Jan 2019, Yang Zhong wrote:
On Wed, Jan 30, 2019 at 01:20:42PM +0100, Paolo Bonzini wrote:
On 30/01/19 13:07, Thomas Huth wrote:
I think adding ioport.o to IDE_PCI and IDE_VIA is wrong. The
ide_init_ioport() function is only used by isa.c and piix.c, so it
should only be added for IDE_ISA and IDE_PIIX here.

Good idea, another one for Yang. :)

 Yes, for ioport.o, i will add as below:

 +common-obj-$(CONFIG_IDE_ISA) += isa.o ioport.o
 +common-obj-$(CONFIG_IDE_PIIX) += piix.o ioport.o

 The newest hw/ide/via.c has removed ide_init_ioport() call.

This was my change. I've changed via.c to implement PCI IDE instead of ISA legacy mode that it used to do before (as that's what clients use more). However it still needs isa_get_irq() and thus ISA. This IDE controller is usually part of VIA superio chips (such as VT82C686 and similar) that have ISA and the IRQ seems to be wired to that. This could probably be separated further but since device has a register to set IRQ number it probably does not worth the added complexity. So it's still not completely independent of ISA. (But don't make it dependent on vt82c686 either because I'll add another such chip which will also use it later. The vt82c686 should depend on via-ide.)

Regards,
BALATON Zoltan



reply via email to

[Prev in Thread] Current Thread [Next in Thread]