Menu
▾
▴
#1 Pb when insmod (init_module not found)
When I'm trying to install your module : insmod
epiafb.o I've have an error in init_module not found.
I've looked at the code, there is no init_module or
cleanup_module in the code, so we can't do insmod
whith this module.
Is it possible to load in directly in the kernel (mv
epiafb.o trident.o :-) and try to install it by force
?
Or what can I do ?
Is the append=trident:mode directive still valid for
your driver in the /etc/lilo.conf ?
Logged In: NO
When I boot my mandrake 8.2 system with a vga= kernel
parameter, which puts it into framebuffer mode I think, i can
do nothing with thiis epiafb driver. However, when I boot the
system without any kernel parameters, and then do insmod
fbgen and then insmod epiafb.o .. it inserts ok. Doesnt work,
but it does insert ok.
Logged In: YES
user_id=103183
Pb?
epifb_init() is the module init function. What version kernel
are you using?
I haven't tried building it as part of the kernel, altho in theory,
with the right makefile, it ought to be plain sailing.
Logged In: YES
user_id=638222
Pb = problem :-)
I'm using the 2.4.19 kernel.
How do i modifie the makefile to install the module directly in
the kernel ?
Logged In: YES
user_id=103183
Ah ;)
Well, you've got a recent kernel, I can't understand why you
can't insmod. Did you build epiafb as root? Did you insmod
fbgen before you tried to insmod epiafb? If you could mail me
a log of the errors you get, that might help.
Logged In: YES
user_id=638222
Yes, I've build epia as root, I've gor the epiafb.o.
Then I've insmod fbgen, it works
Then insmod epiafb.o (as root)
I don't remember exactly the message, but it's complaining
about the init_module fonction which doesn't exists.
I'll post the real message tomorrow.
I've read the documentation about module in linux and it's
says that a init_module and cleanup_module fonction is
mandatory to insmod module ?
Logged In: YES
user_id=103183
Yes, it has the init_module and cleanup_module, they're just
not called "init_module" and "cleanup_module".
Look for documentation on "module_init()" and "module_exit
()" macros. Thus in epiafb_driver.c:
module_init(epiafb_init);
module_exit(epiafb_exit);
In other words, epiafb_init() is equivalent to init_module() and
epiafb_exit() is equivalent to cleanup_module().
Logged In: YES
user_id=638222
The error message is :
tridentfb: EPIA framebuffer 0.1 initializing
epiafb.o : init_module: No such device
Logged In: YES
user_id=103183
This indicates that the epiafb driver can't find your hardware.
You are running this on an EPIA 800 or EPIA 5000 board?
Logged In: YES
user_id=638222
EPIA 800
the tridentfb work with my system, I don't understand
Logged In: YES
user_id=103183
Possibly your board has different vendor/device IDs. I'd
assumed that the VIA boards would all be the same.
Can you do "cat /proc/pci" and
also "cat /proc/bus/pci/devices" and mail me the output from
them?
Logged In: YES
user_id=638222
EPIA 800
the tridentfb work with my system, I don't understand
Logged In: YES
user_id=103183
tridentfb supports a wide range of vendors and chipsets.
epiafb on the other hand only supports a single vendor
(Trident) and a single chipset (CyberBladei1).
I'm guessing that your board has something different for
these values, so if you post me those /proc results I can
check, and if necessary, add your hardware info to the driver.
cheers
ant
cat /proc/pci/devices
cat /proc/pci
Logged In: YES
user_id=638222
Here is my value
cheers
Logged In: YES
user_id=103183
Thanks. Your values are exactly the same as mine :-(
I've studied the kernel PCI code, and it appears that if the IDs
match then the only logical way that ENODEV can be
returned is if the epiafb_probe() routine is failing.
Unfortunately, this routine has a number of failure points. Can
you try the following?
1- edit epiafb's Makefile so that it says "DEBUG=1" at the top
2- make clean
3- make
4- try to fire it up in the normal way
5- mail me the tail of you /var/log/messages
You should see msgs from the driver in /var/log/messages.
I've just noticed that they're prefixed with "tridentfb" - oops!
Hopefully this will shed some light on it.
cheers
Logged In: YES
user_id=638222
I've tried to compile in debug mode, but it was complaining for
the undeclaration of the crt variable.
I've had static int crt; juste behind the declaration of bpp.
It compile now and this is the result in the /var/log/messages
when I'm trying to insmo epiafb.o
Nov 2 03:55:12 DivxBox kernel: epiafb_init:mode=640x480,
bpp=8, crt=0
Nov 2 04:03:51 DivxBox mcedit: /dev/gpmctl: No such file or
directory
Nov 2 04:05:19 DivxBox mcedit: /dev/gpmctl: No such file or
directory
Nov 2 04:05:23 DivxBox mcedit: /dev/gpmctl: No such file or
directory
Nov 2 04:13:57 DivxBox kernel: tridentfb: EPIA framebuffer
0.1 initializing
Nov 2 04:13:57 DivxBox kernel: epiafb_init:mode=640x480,
bpp=16, crt=0
Logged In: YES
user_id=479772
I have an almost similar problem.
When I do "insmod epiafb.o" I get "epiafb.o: init_module: No
such device",
in /var/log/messages, with debugging on I get:
Nov 3 21:24:46 mao kernel: tridentfb: EPIA framebuffer 0.1
initializing
Nov 3 21:24:46 mao kernel: epiafb_init:mode=640x480,
bpp=8epia_pci_probe:request_region failed!
System is Redhat 8.0 with kernel 2.4.18-14.
Any ideas?
Just let me know if you need any more information...
Cheers
Robert
Logged In: YES
user_id=103183
rawburt: sounds to me like you already have a framebuffer driver installed, so the new one can't claim the memory resources as the old one already has them.
mcdermott: that's it? I'm concerned that there are no debug msgs from the pci_probe() routine. Are you sure that you have PCI support enabled in the kernel sources that you are using?
cheers
ant
Logged In: NO
I had the same problem no (init...) when i started mandrake in
fbdev mode, when i start mandrake without fbdev I can insert
the epiafb.o without problems
Logged In: YES
user_id=479772
blowback
You were absolutely right. I had a framebuffer driver
installed. Without that driver it works fine.
Thanks a lot!
Robert
Logged In: YES
user_id=850691
how can I unload the compiled-in tridentfb driver? or do I
have to recompile the whole kernel again?
/rnb