"Nathan J. Williams" <nathanw@wasabisystems.com> writes:

> The driver for this chip does mess with the EEPROM. Quoting from the
> source:

Okay, having looked around a little more, let me remove my foot from
my mouth and explain what I think happened here, timewise.

The EEPROM tweak was necessary in certain Intel chips, as mentioned
before. There's a table in the driver that recognizes chip models (by
PCI ID) and determines whether they need the tweak. Unfortunately,
when a bunch of new models of chips were added to the table last
March, the tweak was turned on for them (with the comment, ironically,
"better safe than sorry"). That version of the table is what ended up
in 1.6.2.

More recently, the bug was reported (pr 22120) and the entries in the
table for the new chip were removed (last September). That corrected
version is in NetBSD-current and will be in 2.0. For some reason the
person who fixed it didn't update the bug report or request a pullup
to the 1.6 branch. I'll see about fixing those.

        - Nathan