Re: [Qemu-devel] [PATCH 4/4] piix_pci: Implement reset for i440FX PAM configuration

[Laszlo Ersek]

On 02/21/13 12:02, David Woodhouse wrote:
> On Thu, 2013-02-21 at 09:17 +0000, David Woodhouse wrote:
>> Thanks. I suppose I'd better test that I haven't broken suspend/resume
>> first. I don't care about OS/2 or VDISK, but it's vaguely possible that
>> suspend/resume might be another "reset" user which actually *would* be
>> offended if the PAM configuration got reset, causing the BIOS to do a
>> complete reinit.
>>
>> If that breaks, this will have to wait for the 'soft-reset' bits that
>> Anthony is looking at.
> 
> Ah crap, it breaks. Suspend/resume now does a full PAM reset, so SeaBIOS
> doesn't see that it's a resume and goes through a full hardware init and
> reboot.

:(

But at least now I understand why the PAM regs were not reset before,
and why SeaBIOS has to jump through the qemu_prep_reset() hoop... I'm
afraid people had this entire discussion before :(

I'm sorry I couldn't think of this earlier... May be related to the fact
that I could never really appreciate the "suspend in a VM" thing.

Laszlo