Re: [PATCH] hw/ide: Remove status register read side effect

[BALATON Zoltan]

Hello,

On Wed, 26 Feb 2020, address@hidden wrote:
> According to the CMD646U2 specification:
> "When an IDE port is in PCI IDE Legacy Mode, the PCI646U2 is compatible
> with standard ISA IDE. The IDE task file registers are mapped to the
> standard ISA port addresses, and IDE drive interrupts occur at IRQ14
> (primary) or IRQ15 (secondary)."
>
> In legacy mode, IRQ14 and IRQ15 mirror the state of INTRQ on each of
> the selected IDE devices. QEMU appears to emulate this correctly.

So CMD646 also seems to have a legacy mode. I've also seen a CMD PCI0640B 
spec which is proabably a similar chip which says for interrupt handling:

"When DSA1 is pulled low during reset, both IDE ports are in PCI IDE 
Legacy Mode. When DSA1 has no pull-down during reset, each IDE port may 
independently be set to PCI IDE Legacy Mode or Native Mode via the 
Programming Interface Byte (configuration register PROGIF, Index 9h). When 
an IDE port is in PCI IDE Legacy Mode, the PCI-0640B is compatible with 
standard ISA IDE. The IDE task file registers are mapped to the standard 
ISA port addresses, and IDE drive interrupts occur at IRQ14 (primary) or 
IRQ15 (secondary).

When an IDE port is in PCI IDE Native Mode, the IDE task file registers 
may be mapped to non-standard port addresses, and IDE drive interrupts 
occur at PCI INTA. Therefore, if both IDE ports are in PCI IDE Native 
Mode, drive interrupts from both IDE ports are multiplexed into PCI INTA. 
In this case, the interrupt status bits must be polled to determine which 
IDE port generated the interrupt, or whether the interrupt was generated 
by another PCI device sharing INTA on the bus."

This same explanation also appears in CMD646 doc. So what mode is the 
PROG_IF config reg set to and do the interrupts raised match that? 
cmd646_update_irq() only seems to raise PCI interrupt, should it also have 
an option to use INT 14 and 15 in legacy mode similar to what my patch 
does for via-ide?

Additionally Solaris may also get info from the OF device tree so that may 
also have to match the device config.

I'm not sure this helps but I don't have any better idea.

Regards,
BALATON Zoltan