Hi group,
I've taken this problem to gentoo-user who can usually suggest a fix for a problem--but not this time.
In a nutshell, modprobe looks in the wrong dir when the PC boots. I won't bore you with all the detail(well, maybe some of it). You could look in the archive for this past week under "modprobe looks in wrong dir" for even more :)
It all started when I did a portage --sync then emerge -Du gentoo-sources. I noticed that /etc/modules.autoload.d/kernel-2.6 had been overwritten by an empty page. So I reloaded it. And re-booted.
Keep in mind that I had not compiled the new kernel yet. I was still using 2.6.12-r6
When it booted some of the modules failed to load. In dmesg these were flagged with Unknown symbol< >...disagrees about version of symbol < >... errors. These were almost all net modules. ppp_async, ppp_generic, unix, forcedeth, 8250 etc. Sound modules, USB modules, agp modules all loaded successfully.
Someone said save .config do a make mrproper and recompile. Did not work. #modules-update did not work. #emerge -Du glibc/gcc did not work.
So I compiled the new, 2.6.16-r3 kernel did a make && make modules_install. And updated the symlink to point to the new sources. And copied over bzImage to /boot/vmlinuz. Like always.
This time, even more modules failed to load: nvram and ie3374(?) IIRC.
But the PC *does* boot and allow me to log in.
Then I ran uname and modprobe and noticed it was still using the old kernel.
But that's very strange because it actually boots from the new kernel which I configured and compiled while in the dir with the new sources. And $ls /lib/modules displays both module dirs, the *.12* and the *.16*
One clue? When I ran #gcc-config, according to my notes it gave me an error:
could not locate i686-pc.linux-3.44 in '/etc/env.d/gcc'
But that was because it should have been looking for *3.45. I don't recall exactly what I did next but after I did what I did #gcc-config didn't complain.
Can someone help me?
-Maxim
