Re: acpi_vbios_reset hangs while vbetool works (Was: sporadic failure to wakeup from S3)

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Tue, Sep 06, 2011 at 06:25:32PM +0100, David Brownlee wrote:
> Does anyone know if acpi_vbios_reset is doing anything intrinsically
> different to vbetool reset, or possibly is it just doing it at a
> different point in the resume?

Depends on which value you use. With acpi_vbios_reset=1, it is done very
early in real mode. Most importantly, PCI bridges etc are not powered up
by NetBSD. This fails if the BIOS doesn't do the work itself.

With acpi_vbios_reset=2, the POST is done as part of the wake up of
vga0, using an emulated 8086 and emulated timer. It means the PCI
bridges etc are handled by NetBSD and in a consistent state. Different
to acpi_vbios_reset=0 is that it doesn't allow reprogramming the real
AT timer and that only part of the real mode memory are mapped.

Depending on whether you are running i386 or AMD64, vbetool will use
VM86 mode or CPU emulation, acpi_vbios_reset=2 always does the latter.
There is some additional mode setting magic in vbetool, too. I don't
have any hardware that needs it, so it isn't hooked up.

Joerg