qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [RFC v5 86/86] 440fx: fix PAM, PCI holes


From: Anthony Liguori
Subject: Re: [Qemu-devel] [RFC v5 86/86] 440fx: fix PAM, PCI holes
Date: Mon, 25 Jul 2011 09:08:17 -0500
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.17) Gecko/20110516 Lightning/1.0b2 Thunderbird/3.1.10

On 07/25/2011 09:05 AM, Avi Kivity wrote:
On 07/25/2011 04:47 PM, Anthony Liguori wrote:
On 07/25/2011 08:38 AM, Avi Kivity wrote:
On 07/25/2011 04:35 PM, Gleb Natapov wrote:
>
> That's the ISA TOM (15MB hole and friends).
>
Correct. What about:
3.2.19. DRB[0:7] DRAM ROW BOUNDARY REGISTERS

from 440fx spec?


Maybe. But we can't use that, since it ignores address line 31.

(440fx supports only 1GB RAM, and we're ignoring that)

What are we trying to do?

Can't we just register highest RAM address under 4G to 4G as PCI
memory and call it a day?

Do we really need a guest visible register to do this?

Why not use 3.5GB and call it a day? It's safer for memory hotplug, if
we ever get it.

The guest will never put a PCI BAR below that anyway.

My entire concern is that they will.

We're not just talking about Windows or Linux here, but any odd DOS application that's smart enough to switch into 32-bit mode, bootloaders, etc.

It's a significant behaviorial change that goes against the spec for no obviously good reason.

Even if we supported hot plug, we could easily just hot plug above the 4GB mark.

Regards,

Anthony Liguori






reply via email to

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