During boot, in the step "Setting system clock ..." (and before the [Ok] of that step), the error message "modprobe: Can't open dependencies file ..." is displayed four times on the console. This is on a non-module system (statically linked kernel with no module infrastructure at all), so of course this file does not exist. The error was introduced with baselayout 1.8.6.8-r1, the previous stable baselayout version displayed no such error.
You cannot maybe be a bit more specific as to what is causing it ?
If possible, of course. Thanks.
Unfortunately, no. I tried to locate the modprobe call causing this messages in the /etc/init.d scripts, but I failed: There is a modprobe in "localmount" and in "modules", but they are both properly surrounded by if's checking for "/proc/modules" (which does not exist on my system). There is another modprobe in "pcmcia". This one is *not* properly protected by an if (it reads "/proc/modules" without checking for its existance!), but it does not seem to be the modprobe causing the messages: The messages still appeared even after I commented that "modprobe" out or put an if around it. I've no idea where these modprobes come from. Is there any way to trace script execution during initialization?
allmost the same for me: but i HAVE modules and they work as they should. Between starting metalog and pcmcia i get two warning messages: "Hint: insmod errors can be caused..." The name of the module which caused this message is not shown. This all happened after emerging baselayout 1.8.6.8-r1 - no kernel changes where made. I also tried baselayout 1.8.6.9 but the bootmessage remains
Do any of you have hotplug or module-init-tools installed ?
I've hotplug installed, but not module-init-tools.
Sorry, been MIA for a bit. Could you try to uninstall hotplug ? Thanks.
The four messages are also displayed when booting with hotplug uninstalled. However, I noticed that hotplug is a prerequisite to some other package (a dependency check with emerge wanted to re-install hotplug), perhaps that package (I didn't check which one) is responsible for the messages?
Right, on 2.4 it was caused by modules-update running the 2.5 stuff as well. Should be fixed in 1.8.6.10, please try.
And ?
Sorry for the long delay, I was on holidays. Emerging baselayout 1.8.6.10 didn't change anything, the messages are still there.
Then I do not know. Try to track where it comes from please.
modprobe is called by devfsd, twice for /dev/rtc and twice for /dev/misc/rtc. hwclock (started by rc) is running in parallel and most likely causing this.
Tried 2.6 kernel, no problem there.
Is this still an issue in current kernels and baselayouts? There hasn't been any activity in this bug for a long time.
The modprobe messages don't appear any more on my system (with 2.6.x kernel and current baselayout, hotplug, ...). Thanks!
Sh*t! Today, I updated my system to current (new baselayout 1.9.4-r1 and new module-init-tools 3.0-r2, among others), and the problem reappeared!
Problem solved. The messages are indeed caused by devfsd trying to autoload modules when nonexistant /dev/xxx entries are accessed (/dev/rtc in my case, I do not have the rtc interface driver compiled into my kernel), which of course fails in a non-modular system. I commented out LOOKUP .* MODLOAD in /etc/devfsd.conf, and the messages are gone.
all set now