I just booted of the Gentoo 2005.0 x86 Universal Livecd. Everytime ipw2200 module doesnt find the firmware on boot. I have rmmod and modprobe it again for it to find the firmware.
Steps to Reproduce:
1.Boot of the 2005.0 x86 Universal LiveCD
ipw2200 module fails to locate firmware
Firmware should be located and device should be created
Portage seems to carry a different tgz than the one on the ipw2200 site, so the patch failed for me with the one portage downloaded. Downloading the one from the ipw2200 site and redigesting and applying the patch works for me.
I believe this is because /usr/lib/hotplug/firmware/firmware is a symlink to /lib/firmware. It should be /usr/lib/hotplug/firmware linking to /lib/firmware, not /usr/lib/hotplug/firmware/firmware.
Doubt that is the reason - looks like a duplicate of bug #90440 to me. Reporter, please see if the tip mentioned there fixes this issue for you.
Just by doing an rmmod and modprobe of ipw2200 fixes the problem, I didnt need to modify or add timeout value in /sys/class/firmware.
Perhaps hotplug is run before the firmware directory is mounted?
Hotplug is run after the mounts, which occur before runlevel default. I honestly think that Anthony was right RE: comment #2 here. I fixed the issue where the firmware was showing up in the wrong place in catalyst, so this should be resolved for the next release, however, we will need testers to find out if this is the case.
/usr/lib/hotplug/firmware is not used by >=sys-apps/hotplug-20040923. The new location is /lib/firmware. Why is the symlink even there?
Because there were module ebuilds that had not been updated at the time of the release...
Created attachment 59844 [details]
gkrellm window screenshot after gkrellm disappears
Comment on attachment 59844 [details]
gkrellm window screenshot after gkrellm disappears
sorry wrong bug report
It seems to me that the bug #90440 is indeed related to this bug, if not identical.
*** Bug 90440 has been marked as a duplicate of this bug. ***
(In reply to comment #4)
> Just by doing an rmmod and modprobe of ipw2200 fixes the problem, I didnt need
to modify or add timeout value in /sys/class/firmware.
I can confirm this. reinserting the module solves the problem.
Is it bug in ipw module or here? Do you need some testing?
Hopefully this is fixed in the newly added net-wireless/ipw2200-1.0.9. It contains a work-around for a firmware loading race.
I seemed to have the same problem after installing kernel 2.6.15 r1
The text underneath solved the problem for me:
quoting from http://www.openthought.org/blosxom.cgi/Blog/Computers/Organizations/
So, instead of compiling 'firmware_helper' as I had initially suggested above (which doesn't seem to work anymore anyways), just create the file '/etc/udev/rules.d/999-firmware.rules', and add the following line in it:
ACTION=="add", SUBSYSTEM=="firmware", RUN+="/sbin/udev_run_hotplugd"
Then rmmod ipw2200 and the ieee80211* modules, insmod ipw2200, and you should be good to go.
Mostly to help Google catch this, the new error I was received was:
ipw2200: Intel(R) PRO/Wireless 2200/2915 Network Driver, 1.0.8
ipw2200: Copyright(c) 2003-2005 Intel Corporation
ipw2200: Detected Intel PRO/Wireless 2200BG Network Connection
ipw2200: Unable to load ucode: -62
ipw2200: Unable to load firmware: -62
ipw2200: failed to register network device
ipw2200: probe of 0000:03:03.0 failed with error -5
This problem is still present with the 2006.0 release media. Still, the problem can be worked around by doing a "modprobe -r ipw2200; modprobe ipw2200" after boot up.
According to the ipw2200 driver log messages, the firmware files can not be found when it first attempts to load them (-2 = -ENOENT = "No such file or directory"). I still suspect that /lib/firmware/ipw*.fw is not present when the ipw2200 module is first loaded...
On a side note: the kernel used on the 2006.0 release media contains version 1.0.8 of the ipw2200 driver, which doesn't include the firmware loading race fix found in 1.0.9 and later versions - but seeing that the driver says -ENOENT and not -EIO, I don't think the race is the main issue here.
Perhaps should be its own bug, but I think the udev package ought to include the lines found at the link below, as well as in the comments here.
I've added a /etc/udev/rules.d/99-firmware.rules with the one-liner and now, finally, my ipw2200 works without a hitch. Huzzah! Could we get this into the udev package?
Please file a bug for the firmware thing separately, It really should be fixed in upstream udev, abd our udev maintainer is also one of the main udev developers.
Can anyone verify this with 2006.1?
it still fails with 2006.1.
dmesg says: 'unable to load firmware' code= -2
However if you 'rmmod ipw2200' and 'modprobe ipw2200' it can find firwmare and works fine after that.
This happens with udev 103, the recently stabilized one as well.
Ok, seems to work for udev-103 now when
First, I downgraded back to udev-094, and then
CONFIG_PROTECT="-*" emerge udev
It seems that when udev upgrades, some junk from the previous configuration gets left behind. Downgrading and then upgrading with configprotects removed fixed the issue.
(In reply to Comment #21, Comment #22)
Unmerge hotplug; if it fixes the problem, move to Bug 147006...
(In reply to comment #24)
> (In reply to Comment #21, Comment #22)
> Unmerge hotplug; if it fixes the problem, move to Bug 147006...
Unmerging hotplug did not help - only after steps in #23 things started to work.
We should probably get some of the recent udev versions stabilized quickly (104-r3?) and included on release media, it fixes quite a bunch of bugs...
Is this still an issue with 2007.0?
I tried with the i686 livedvd on my IBM Thinkpad T43, with an Intel Pro Wireless 2915 A/B/G.
The firmware loaded fine. However, the device did not show up as a network interface at all. ipconfig -a did not show it (it usually shows up as eth1), nor did iwconfig.
There were no errors, including in kernel logs. In fact it all looked quite normal.
I had to manually modprobe -rav ipw2200 and then modprobe ipw2200 to get this device to work. It then behaved as expected: Showing up as eth1 in both ifconfig and ipconfig. There were no firmware loading errors.
I did not test if it actually worked as I am having major problems with my access point.
So, it seems to work but most users are not going to know to remove the modules then re-add them.
Meh. s/ipconfig/ifconfig .
Also, before I unloaded and reloaded the modules, the card did show up in lspci.
The card would show up in lspci whether there are drivers loaded or not.
OK, we've switched to the in-kernel modules for 2008.0. Has anyone tried 2008.0 Beta 1 on their machine with this card?
Can someone test this with 2008.0 final?
I'm closing this bug as we've been waiting for feedback for over 2 years now.
If you can reproduce this bug with the latest available weekly ISO, feel free to reopen the bug.