PATCH: power down acpi and pci devices in suspend/resume

[M. Warner Losh]

In message: <41A43787.7070809 at root.org>
            Nate Lawson <nate at root.org> writes:
: M. Warner Losh wrote:
: > In message: <419EF7AD.8050007 at root.org>
: >             Nate Lawson <nate at root.org> writes:
: > : cbb0: bad Vcc request. ctrl=0xffffff88, status=0xffffffff
: > : cbb_power: 0V
: > : cbb1: bad Vcc request. ctrl=0xffffff88, status=0xffffffff
: > : cbb_power: 0V
: > 
: > This looks like a problem to me.
: > 
: > Warner
: 
: I did a suspend with dhclient active and an an0 card present and got this:
: 
: cbb0: bad Vcc request. ctrl=0xffffff88, status=0xffffffff
: cbb_power: 0V
: cbb1: bad Vcc request. ctrl=0xffffff88, status=0xffffffff
: cbb_power: 0V
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: DMA timeout
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: fxp0: SCB timeout: 0xff 0xff 0xff 0xffff
: 
: Note that the policy is to set all PCI and ACPI devices to D3 when 
: suspending unless an _SxD method says to use a different power state. 
: fxp0 isn't present in the ACPI namespace but both cardbus ports are and 
: both have _S3D methods that say to use D3 for S3!  So this is not an 
: error unless it's in cbb needing to do more before I power it down.  I 
: don't get any of the above messages on suspend without my patch.
: 
: Suggestions?  What does the "bad Vcc request" message mean?

At a guess, I'd say that we're accessing the hardware after the power
state has been set to d3, likely via an interrupt.

'bad Vcc request' is just a bit that's set in the pccbb hardware.  It
is almost never right, especially with the status of 0xffffffff.

Warner