Re: New version of iwi(4) - Call for testers [regression!]

On Tuesday 21 March 2006 00:38, Max Laier wrote:
On Monday 20 March 2006 23:06, Darren Pilgrim wrote:
dima wrote:
I had a chance to test the driver more throughly today. It can't
connect to some access points. Sometimes reboot helps, sometimes not. I
didn't experience this with the driver currently present in the tree;
well, I used /usr/ports/net/iwi-firmware (not the kmod version). This
can be the case.

> When I run the 2 following commands in the script it sometimes

fails to load the firmware properly:

# ifconfig iwi0 inet 192.168.1.2 netmask 255.255.255.0 up

> # iwiconfig -i iwi0 -d /boot/firmware -m bss
>
> It seems that if_iwi module needs some time to initialize the card
> properly (I have a 2915 one); and it's really ready to boot a firmware
> in several milliseconds.

What is the "iwiconfig" program and why are you using it with Max's
driver? The first command is all that is needed to get the interface up
and configured, assuming you're using static IPs on an open AP.

Let me clear up some things here:
1) This is not "mine" driver. It's Damien's with a lot of patches that I
hope to test by putting out this version.
2) iwiconfig is no longer needed to load firmware. iwi-firmware-kmod from
ports is the weapon of choice for both, the driver version from CVS and the
one from my page.
3) The aim is to merge the changes in "my" driver to CVS modulo
regressions. That is why I am mostly interested in side-by-side testing. I
will put out a new tarball later this week that will help with that. If
you already did some such testing, I'd be happy to hear your results.

Okay, here is the newest version:
http://people.freebsd.org/~mlaier/new_iwi/20060325.both_nofw.tgz

This includes the following changes towards the last version:

1) Different locking. This should shut up several WITNESS warnings and panics
when doing long ifconfig commands or tcpdumping on a downed interface. It
will not get rid of WITNESS warnings from slowcopyout when doing ifconfig
scan or the like.

2) Allow wpa_supplicant to recover from occasionally stuck scans - I haven't
seen any lately, but some reports exist. If you do see them, please let me
know - maybe we can find a pattern and cause.

3) Get rid of the "cmd 0x19 not send..."

This version installs from modules/iwiNG to make it easier to test as we move
patches to CVS. The module internally still identifies as "iwi" so that you
don't load two instances concurrently. The tarball also doesn't include any
firmware. net/iwi-firmware-kmod is all you need!

Make sure that you test against a recent kernel. This means especially one
with the net80211 fixes. You can verify this by looking at:
src/sys/net80211/ieee80211_output.c
You want either 1.40 (HEAD) or 1.26.2.7 (RELENG_6)

--
/"\ Best regards, | mlaier@xxxxxxxxxxx
\ / Max Laier | ICQ #67774661
X http://pf4freebsd.love2party.net/ | mlaier@EFnet
/ \ ASCII Ribbon Campaign | Against HTML Mail and News

Attachment: pgpUSFdJWimJm.pgp
Description: PGP signature

Relevant Pages