Unable to build acx111 driver on new kernel 2.6.20.2925.fc6-i586

[LaChild]

Hey everyone wondered if you guys could give me a hand. I am attempting to re-build the acx111 driver for the new kernel except the build is generating the following errors.

/var/lib/dkms/tiacx/0.4.7-3/build/common.c:6620:14: error: macro "INIT_WORK" passed 3 arguments, but takes just 2
/var/lib/dkms/tiacx/0.4.7-3/build/common.c: In function ‘acx_init_task_scheduler’:
/var/lib/dkms/tiacx/0.4.7-3/build/common.c:6619: error: ‘INIT_WORK’ undeclared (first use in this function)
/var/lib/dkms/tiacx/0.4.7-3/build/common.c:6619: error: (Each undeclared identifier is reported only once
/var/lib/dkms/tiacx/0.4.7-3/build/common.c:6619: error: for each function it appears in.)
make[1]: *** [/var/lib/dkms/tiacx/0.4.7-3/build/common.o] Error 1
make: *** [_module_/var/lib/dkms/tiacx/0.4.7-3/build] Error 2
make: Leaving directory `/usr/src/kernels/2.6.20-1.2925.fc6-i586'

The build worked fine on 2.6.18 and 2.6.19. Not sure what changed in the new kernel or what to do with this error, to get my wireless card working again.

Thanks in advance for any help you can provide.
Lachild

[Figment]

I'm not positive that this is the problem, but is your processor really an i586? For i686-capable processors like the Pentium, you should be using a i686 kernel, but this is not always installed properly with FC6. See here for instructions on how to fix:
http://fedoraproject.org/wiki/Bugs/FC6Common

[LaChild]

Thanks for the reply. I reinstalled FC6 with I686 and that did not fix it. Still gives the same error only stating the I686 message instead of I586.

This solution did however fix a problem I was having on a seperate computer with it's network card though.

Thanks again, and if anyone can think of anything else please let me know.

[rbwerst]

I have encountered exactly the same problem with my ACX111...

I can still compile the modules under the 2.6.19 Kernel, but the samer error messages are returned t me when I use the same commands under the 2.6.20, and it happened right after doing the update.

I can still boot into the old kernel from the Grub menu, and the wireless card will work if I recompile and reinsert it so all is not lost as yet, I would appreciate a heads up if a solution is found for this bug.

BTW is there a way to change the default settings for Grub so that it automatically boots into the 2.6.19 kernel until the problem is fixed?

Thanks

[Seve]

Originally Posted by rbwerst

BTW is there a way to change the default settings for Grub so that it automatically boots into the 2.6.19 kernel until the problem is fixed?

Thanks

Hello:
You can edit your /etc/grub.conf file and change the default from X to Y.
It starts counting at 0
You also may want to edit your /etc/yum/pluginconf.d/installonlyn.conf
and set your
tokeep=2 [which is the default] to
tokeep=4
This will protect you if another kernel comes along and gets installed and removes your working 2.6.19

Seve

[rbwerst]

I truly appreciate your help Seve...changing "default=0" to "default=1" enabled me to boot into the previous kernel automatically -- and the wireless card still works with it as well.

It was a real pain to get the ACX111 working in the first place so it is really nice to have it up and running again. There is still a problem with the boot process hanging near the end when the computer is started with the card inserted -- I have to allow the boot to finish then insert the card -- but it's a small price to pay to actually have the ability to use my wireless network.

I suppose I shouldn't complain in that all of the problems have helped me a lot in re-learning how to use Linux.

RIchard

[LaChild]

Well it looks like all we can do is simply wait for the next kernel or wait for a new DKMS package...

Has anyone tired the Diers kmod-acx module? If so is it working in 2.6.20?

[rbwerst]

Originally Posted by LaChild

Has anyone tired the Diers kmod-acx module? If so is it working in 2.6.20?

Had not even heard of the Diers module...might be worth checking out though for other Distros or when Fedora 7 is released.

[LaChild]

Sorry... Let me rephrase... It's a precompiled Kernel Modual of the same driver we are compiling through DKMS. It's just through the Dries respository instead of freshrpms. It's located at: http://dries.ulyssis.org/rpm/ They have two files kmod-acx-common and acx-kmod.

Not sure if it works but worth looking into.

[jwp_jfo]

I've replaced the lines

INIT_WORK(&adev->after_interrupt_task, acx_e_after_interrupt_task,
adev->ndev);

in the following file

/var/lib/dkms/tiacx/0.4.7-3/source/common.c

with

INIT_WORK(&adev->after_interrupt_task, acx_e_after_interrupt_task);

and builded again with

dkms build -m tiacx -v 0.4.7-3

It worked for me

[rubio]

I followed your hint on removing the 3rd argument to INIT_WORK and was able to get the tiacx driver to compile and install, but immediately on attempting to load it the machine dies with a spinlock error. The trace shows it's having issues at or near the acx_e_after_interrupt_task function; this seems unlikely to be mere coincidence.

jwp_jfo, were you able to use the driver successfully without issues following your tweak? I can't imagine that much changed between .2925 and .2933 but trying to narrow this down.

Thanks

[wrz]

Looks like the INIT_WORK macro is different in 2.6.20, and neither the dries nor freshrpms packages will work. I guess it's best to notify the authors of the driver and/or rpm and wait for an update.

HOWEVER. There is this little patch which might do the trick - I guess one should apply the four lines by hand and pray for the best ;-) Haven't tried it yet myself.

http://www.kernel.org/pub/linux/kern...breakage.patch

[Orion Blastar]

I get the spinlock in Fedora 7. Do I have to disconnect the wireless card to prevent it and modify the driver?

It would be a lot easier if Red Hat supported the ACX driver in Fedora 7. UBuntu 7.04 already supports it and I might just make the switch to avoid the headaches.

[Orion Blastar]

Just apply the patch, make sure there are no typos. Fedora 7 works with it. The spinlock happened because I tried the first patch and not the second one.I advise against the first patch with one line change and use the second patch with four lines changed. The lesson here is to read all posts before trying something.

I see to have problems connecting to my wireless network, but the driver appears to work.
I used a tool at http://wifi-radar.systemimager.org/ to scan and it detected my network. But I cannot get an IP address from it. I chekced the WEP key and it is in hex 0x format, the name of my router etc are all set correctly as well. Does this driver have issues with Fedora 7?

[jqwire]

Hi,

I've been following this thread and tried both patches as indicated by Orion on a Fedora 6 with 2.6.20-1.2962 kernel, but the system hangs as soon as I plug the PCI board.

Then I went through all the code in common.c and replaced all references to:
acx_device_t *adev = (acx_device_t*)data;
by
acx_device_t *adev =
container_of (work, acx_device_t, after_interrupt_task);

since the work_struct data field doesn´t hold any possible reference to the acx_device structure. Expectedly, the kernel still hangs.

I suspect that there is really a compatibility issue here. Nothing guarantees that the code that calls
acx_e_after_interrupt_task(struct work_struct *work) is really passing a pointer to a work_struct.

Does anyone know if this problem has been solved somewhere else?

Thank's a lot for helping.