Jeff,

I took your feedback, and cleaned up both the patch and more.  Attached are
two seperate diffs - one for adding the natsemi to pci_ids.h, and one
against natsemi.c

I'll do my best to narrate the dif hunk by hunk.  I hope this is sufficient
for you to send it on, now :)

* comments

* update release info
* add strings for 83816
 
* formatting for 8-space tab disply

* increase RX Queue to eliminate some Something Wicked messages
* Define the timer frequency
* formatting

* formatting

* strings for 83816
* use new PCI ID (depends on pci_ids.h patch

* formatting
* add constants for magic register's values
* rename an enum for consistancy
* rename all 'bit' enums so they are named the same as their 
  register, and are found in the same order
* Add a comment about the default interrupt state
* Add some missing but needed bits definitions (TXConfig...)

* more bits
* constants for Silicon Revisions

* more bits for descriptor status

* store SRR in private struct
* formatting
* add mdio_write function protoype

* formatting

* magic number removal

* call eeprom reload during probe - BEFORE chip reset

* use MII defines for mdio
* magic number removal
* save SRR

* moved bits

* cleanup mdio_read to look more like mdio_write
* add mdio_write
* add defines for bits to save across a reset
* expand natsemi_reset() to save and restore state that would be 
  nuked by a chip reset

* add a natsemi_reload_eeprom() func

* use a constant for netdev timer
* formatting

* magic number removal

* use the stored SRR
* poll for AnegDone for a bit during open()

* use magic register constants
* comments
* cleanup for WoL
* magic numbers
* no need to blast mac-address, natsemi_reset should be doing it now

* magic numbers

* print if we got a WoL event

* add the phy reset checker to catch spurious PHY resets
* use mod_timer

* formatting

* magic numbers

* increase a debug level (didn't check into netif_msg yet - later

* magic numbers

* use mdio_read where applicable

* magic

* get rid of "Something Wicked" and print what it actually is

* fixup bit definitions

* only do SOPASS for rev D or up

* remove FIXME comment

* More SOPASS for rev D

* Magic numbers

* use MII defines where appropriate

* use mdio_read()/mdio_write()

* WoL cleanup


Detailed enough?  Please don't make me break it down further :)	
-- 
Tim Hockin
Systems Software Engineer
Sun Microsystems, Cobalt Server Appliances
thockin@sun.com