Re: Cannot use iwi(4): "could not load firmware iwi_bss"

> Date: Fri, 10 Aug 2007 08:48:35 +0200
> From: Jeremie Le Hen <jeremie_at_le-hen.org>
> 
> Hi Kevin,
> 
> Sorry for this late reply.
> 
> On Mon, Jul 23, 2007 at 01:13:26PM -0700, Kevin Oberman wrote:
> > > Date: Mon, 23 Jul 2007 22:00:21 +0200
> > > From: Jeremie Le Hen <jeremie_at_le-hen.org>
> > > Sender: owner-freebsd-current_at_freebsd.org
> > > 
> > > On Mon, Jul 23, 2007 at 11:16:00AM -0400, John Baldwin wrote:
> > > > On Friday 20 July 2007 07:21:00 pm Jeremie Le Hen wrote:
> > > > > Ok, I've tried with ACPI enabled and I confirm that iwi(4) can
> > > > > successfully load its firmware in this case.  Unfortunately psm(4)
> > > > > doesn't work so this is not an option for me.
> > > > 
> > > > Can you get verbose dmesg's both with and without ACPI?
> > > 
> > > I've not attached them as verbose dmesgs are quite huge, you can get
> > > them here.
> > > 
> > > With ACPI (psm(4) doesn't work, iwi(4) does):
> > > http://tataz.chchile.org/~tataz/tmp/dmesg.with_acpi.gz
> > > 
> > > Without ACPI (hint.acpi.0.disabled="1", psm(4) works, iwi(4) doesn't):
> > > http://tataz.chchile.org/~tataz/tmp/dmesg.without_acpi.gz
> > > 
> > > Note that I've manually loaded iwi_bss and if_iwi modules from root
> > > prompt.
> > 
> > I have a very similar problem when I load drivers at the boot prompt (or
> > did you mean "root"?)
> 
> I meant "root" prompt :-).
> 
> > If you look at your dmesg (verbose not needed). I suspect that you will
> > se that there is no interrupt assigned to psm. If you are loading them
> > before the probe, try booting with ACPI and boot to single user (boot
> > -s). Hit Enter to get a shell prompt and load the needed modules
> > (kldload). Then exit to move to multi-user. I bet everything works.
> > 
> > I found that my system fails to assign an interrupt to psm if I have the
> > module loaded before I probe I/O devices.
> 
> I wished to try this but unfortunately psm(4) doesn't seem to be
> available as a module.
> 
> # cd /sys/modules
> # grep -rl psm\\.c .
> #
> 
> > jhb suggested an approach to fixing this, but I have not gotten around
> > to trying it.
> 
> Could you point out this please?
> 
> Thank you.
> Best regards,
> -- 
> Jeremie Le Hen
> < jeremie at le-hen dot org >< ttz at chchile dot org >
> 

Jeremie,

I guess I was not clear. As you noted, psm is not available as a
module. Let me run through this from the top.

Boot the system with drivers that attach to interrupts (I can only
confirm this for umass and if_ath, but I suspect it applies to other
drivers including the iwi driver) and the psm fails to attach to its IRQ
and is non-functional. No pointer. No X. Ho joy.

The only work-around I have found is to load any module that causes this
problem after the device probe has completed. This can be done
by:
1. Booting to single-user and using kldload, load the iwi module, or
2. Loading the module after booting. At least most network driver
   modules will be loaded automatically when the first ifconfig(8) is
   issues for that device. This is the mechanism I use to load if_ath, or
3. If the device is not required by the rc scripts, use kldload(8) after
   the system has booted.

A new issue has shown up of late that might also impact you if you use
the wpa_supplicant. It seems to hang up and never associate with the AP
when using WEP or open connections. I always need to do a
'/etc/rc.d/wpa_supplicant reload ath0' after I log in. The
wpa_supplicant them behaves normally. I hope to have time to track this
down some day. I don't know whether you will have this problem.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman_at_es.net			Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4  EADA 927D EBB3 987B 3751