B3 GPIO changes: new EC -> Host IRQ

[Paul Fox]

hi wad -- this sounds fine, and makes sense.  given the extra
complication this will entail, current plan is to attempt
to do everything via ACPI, but this backup is a good thing.

paul

john wrote:
 > 
 > Paul,
 >     Having discovered problems with our previously proposed
 > method of co-generating interrupts on the host, tonight's installment
 > details a work around restoring a second interrupt line from the EC
 > to the VX855 (eliminated earlier in the design, and incorrectly  
 > implemented
 > then anyway.).
 > 
 > In a previous episode:
 > The embedded controller (EC) in the laptop communicates a bunch of stuff
 > to the "host" processor.  It does this through some ports on the I/O  
 > bus and
 > and interrupt wire to the host.  Our intrepid code hacker pgf  
 > proposed that
 > it would be much easier to support many of the cool features we support
 > through this EC communciations channel (lid switches, EB mode switches,
 > gamepad and rotate button presses, etc.) would be easier to support if
 > we had a separate virtual communications channel with the EC.   This
 > channel would have an interrupt handler which wasn't in the ACPI code.
 > 
 > Fine, we'll just wire the EC interrupt to another convenient  
 > interrupt input
 > on the VX855.   That input turned out to have complicated issues with  
 > it's
 > parents, and couldn't be used.   The only "available" inputs are also  
 > used
 > to set chip options at "power-on", which is every resume.   As the EC  
 > interrupt
 > is frequently asserted at power-up, it would screw up these key  
 > initializations.
 > 
 > Current proposal:
 > We will provide a second interrupt from the EC to the host.   This  
 > interrupt
 > will go through an input which isn't "owned" by ACPI.   This  
 > interrupt will
 > not be asserted unless the system is running (S0).
 > This interrupt will not be able to "wake" the system, although the EC
 > may preceed it with an SCI interrupt to wake the system.
 > 
 > This allows this second interrupt to now use one of the available  
 > inputs.
 > One slight hitch is that the host control of the WLAN LED will be  
 > bumped.
 > This will now be driven solely from the WLAN module as in XO-1.
 > 
 > New GPIO mappings are attached.
 > 
 > Cheers,
 > wad
 > 
 > part 5     text/plain                   2
 > 
 > 

=---------------------
 paul fox, pgf at laptop.org