Re: [airo.c bug] Couldn't allocate RX FID / Max tries exceeded when issueing command

On Thu, 2006-11-02 at 13:30 -0500, Ivan Matveich wrote:

On 11/2/06, Dan Williams [off-list ref] wrote:

quoted

It appears that the driver cannot talk to your card; see the "max tries
exceeded when issueing command".  Did this card work previously with a
kernel?  Can narrow down which kernels have problems and which don't?

It spontaneously stopped working about a week after I bought the

That does not bode well for the card itself.

laptop and installed Linux. I tried kernel 2.6.12 and it had the same
problem. (Let me know if you'd like me to try a specific version.)

2.6.12 should be old enough, but maybe if you could try 2.6.9 that would
definitely be old enough to rule out recent airo changes.

I'm hoping that the card has simply got itself into some kind of
invalid state, and not failed altogether.

quoted

It's a bit hard to figure out what firmware you have because the driver
can't talk to the card; can you boot under Windows and determine that
using the Cisco wireless utility?  You also need to flash the card under
Windows, not Linux, ideally to a version of firmware greater than
5.60.08.

I haven't run Windows in many years, so that's problematic. What's the
most straightforward way to boot into a Windows environment sufficient
to run the Cisco wireless utility?

You get to either install windows on your laptop, or pull the card
(pcmcia or minipci, either way) and put it into a windows computer.

quoted

reloading the driver (rmmod airo; modprobe airo) should reset the card.

Yeah, it unfortunately doesn't help. (Nor does rebooting or resetting
the bios.) I noticed a suspiciously relevant commit in the airo.c git
log:

    [wireless airo] reset card in init

    without this patch after an rmmod, modprobe the card won't work anymore
    until the next reboot.

    This patch seem safe to apply for all cards as the bsd driver already do
    that.

    I had to add a timeout because strange things happen (issuecommand will
    fail) if the card is already reseted (after a reboot).

    PS : it seems there are missing reset when leaving monitor mode...

    Signed-off-by: Matthieu CASTET [off-list ref]

Do you know which kernel version that patch first appeared in?

and that makes me wonder if there might be some kind of subtle bug in
the card initialization sequence that manifests itself with my
particular card/firmware.

Could be, it would be quite helpful to get the version of firmware you
are using.  Versions earlier than 5.30.17 are known not be flaky or even
not work with newer Linux kernel drivers since Cisco changed the
firmware interface in that version.  That said, I've seen cards in the
5.00.xx range work with the latest drivers, but not well.  These older
firmware versions sometimes lock up sponaneously and require a reboot,
updating to later firmware post-5.30.17 fixes that problem.

I think I'll burn a freebsd livecd today and see if their kernel works.

That would be a great idea, let us know what the results are, especially
if you cna figure out which firmware version you have, or if the card
itself is really just dead.

Thanks,
Dan