Re: kern/39864: iwn network interface driver broken in current & -5 (i386 - others??)

[Robert Elz <kre%munnari.OZ.AU@localhost>]

The following reply was made to PR kern/39864; it has been noted by GNATS.

From: Robert Elz <kre%munnari.OZ.AU@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/39864: iwn network interface driver broken in current & -5 
(i386 - others??) 
Date: Wed, 05 Nov 2008 03:41:40 +0700

 Well, that has to be the most bogus PR I've ever sent ... aside from
 the iwn driver not working in -current (which remains correct) almost
 nothing in the PR was accurate.   I really should remember never to
 rely upon my memory for anything -- but somehow I just keep on
 forgetting that...
 
 OK, to start again, kernels compiled on 2008-10-16 and 2008-10-29
 both contain iwn drivers that fail.   A kernel from 2008-09-23 worked.
 (See how I messed up the dates...)
 
 Given that, the if_iwn.c commits I suspected in the PR cannot
 possibly be relevant (more idiotic, I wasted time testing them
 only to find that no combination of those, including rolling back
 both of them, made any difference at all...)
 
 Further, in this case "fails" is just about all that I can say - the
 driver just doesn't associate with the AP (not with either of two I
 have available at home where this system lives - I sent the PR from
 work earlier.)   There are no kernel diagnostics, errors, crashes,
 or anything else interesting to report (on that one I think I was
 remembering what happens with the pcmcia wi innterface which was
 reported in PR kern/39838...)
 
 Given all of this, the only if_iwn.c change that falls into the
 relevant date range is this one ...
 
 ----------------------------
 revision 1.20
 date: 2008/10/13 12:41:45;  author: blymn;  state: Exp;  lines: +90 -25
 
 DMA fixes thanks to rtr & skrll
 Don't set tsf in auth, this stops firmware errors on WPA renegotiation
 Various other tweaks that the linux driver seems to have that we did not
 ----------------------------
 
 That, and changes to if_iwnreg.h (rev 1.4) on the same day that's
 clearly a part of the same change.
 
 I rolled back both of those files to the state they were on Sep 23,
 while leaving all the rest of the kernel as current on Oct 29, and with
 that change, iwn started working in an (almost) up to date current again
 (still using Oct 29 sources.)
 
 Now I just need to see if I can spot which of the (fairly large set of)
 changes is causing things to fail.
 
 kre