Semi-working patch for amd64 suspend/resume

[John Baldwin]

On Wednesday 22 October 2008 05:28:26 pm Nate Lawson wrote:
> Alexander Motin wrote:
> > Jung-uk Kim wrote:
> >> When you do 'sysctl debug.acpi.suspend_bounce=1' and 'acpiconf -s 3',
> >> does it bounce back?  If it does not, there are other problems, e.g.,
> >> device drivers.  On my desktop, for example, vga(4) tries to restore
> >> previous state while resuming but it hangs system.  In fact, I believe
> >> ISA-era EGA/VGA save/restore routines do not work for modern graphics
> >> cards at all. :-(
> > 
> > Test passed from both console and XOrg. I have integrated i965GM video.
> > 
> > Here is verbose messages from that trip "there and back again":
> > 
> > Oct 23 00:11:55 mavbook acpi: suspend at 20081023 00:11:55
> 
> > Oct 23 00:12:00 mavbook kernel: vga0: saving 68 bytes of video state
> > Oct 23 00:12:00 mavbook kernel: pci0:0:2:0: Transition from D0 to D3
> > Oct 23 00:12:00 mavbook kernel: pci0:0:2:1: Transition from D0 to D3
> > Oct 23 00:12:00 mavbook kernel: pci0:0:26:7: Transition from D0 to D3
> > Oct 23 00:12:00 mavbook kernel: acpi: bad write to port 0x080 (32), val
> > 0xbb
> > Oct 23 00:12:06 mavbook kernel: pci0:0:27:0: Transition from D0 to D3
> > Oct 23 00:12:06 mavbook kernel: pci0:0:31:2: Transition from D0 to D3
> 
> > Oct 23 00:12:06 mavbook kernel: pci0:0:2:0: Transition from D3 to D0
> > Oct 23 00:12:06 mavbook kernel: pci0:0:2:1: Transition from D3 to D0
> > Oct 23 00:12:06 mavbook kernel: acpi: bad write to port 0x080 (32), val
> > 0xaa
> > Oct 23 00:12:06 mavbook kernel: pci0:0:26:7: Transition from D3 to D0
> > Oct 23 00:12:06 mavbook kernel: pci0:0:27:0: Transition from D3 to D0
> > Oct 23 00:12:06 mavbook kernel: pci0:0:31:2: Transition from D3 to D0
> 
> That's kind of weird.  The above devices were turned off (D3) then back
> on again (D0).  I'm wondering about that blocked write also.

I think they were probably off before due to not having a driver and we turned 
them back on, noticed there wasn't a driver, and turned them back off again.  
If so, that should be fixable.

-- 
John Baldwin