View previous topic :: View next topic |
Author |
Message |
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Nov 28, 2002 8:57 pm Post subject: Installation "gotchas" and current troubles. |
|
|
I've had quite a lot of issues trying to install Gentoo, some of which probably aren't typical but some others might run into so I thought I'd document them here.
Building a kernel with the input core (CONFIG_INPUT=m) as a module and your devices in-kernel will result in a failed build. This is a bug in the build system that's apparantly only fixed in the 2.5.x development series. When I first saw the kernel compile fail I figured I may just have an untested configuration that the gentoo-sources kernel didn't account for, so I emerge'd the redhat-sources kernel instead.
That build succeeded, but when I booted it, it locked up. I later found out that I ran into a bug with the RH 8.0 kernel which won't boot without an "apm=off" passed to it. So much for the widely hailed Redhat testing reputation.
In the mean time I then tried the vanilla sources kernel and it bombed out due to the input core issue -- so I knew gentoo-sources probably wasn't at fault, figured out the problem and set CONFIG_INPUT=y and that worked, and it booted it up fine.
Next I began emerging software and stuff that I wanted, but when it got to emerging gcc-3.2.1, I ran into a kernel bug which dumped this info to my root console:
Code: |
kernel BUG at page_alloc.c:115!
invalid operand: 0000
CPU: 0
EIP: 0010:[<c013806e>] Not tainted
EFLAGS: 00010282
eax: 00000000 ebx: 00005000 ecx: c02a4c80 edx: c11711d0
esi: c1145f14 edi: d536cffc ebp: 00004000 esp: c5983bf4
ds: 0018 es: 0018 ss: 0018
Process xgcc (pid: 10398, stackpage=c5983000)
Stack: ee4be76b c014be69 c103400c c02a50c0 c1145f14 00000000 c013d70f 00000000
00005000 c0000000 d536cffc 00004000 c012ca63 c1145f14 c0000000 00000005
c0000000 c0000000 c0316960 c47eec00 c012af4f c0316960 c47eebfc bfffb000
Call Trace: [<c014be69>] [<c013d70f>] [<c012ca63>] [<c012af4f>] [<c012e2a0>]
[<c014944e>] [<c01494d7>] [<c015d551>] [<c01300d0>] [<c015d2fe>] [<c0149b91>]
[<c014a3be>] [<c014b71f>] [<c0107346>] [<c0108a7b>]
Code: 0f 0b 73 00 ba 7b 26 c0 c6 46 24 05 8b 46 18 89 f5 83 e0 eb
<6>note: xgcc[10398] exited with preempt_count 1
kernel BUG at page_alloc.c:115!
invalid operand: 0000
CPU: 1
EIP: 0010:[<c013806e>] Not tainted
EFLAGS: 00010282
eax: 00000000 ebx: 000000af ecx: c02a4c80 edx: c12a7db4
esi: c1145f14 edi: c0317160 ebp: c63cf40c esp: d05e1f0c
ds: 0018 es: 0018 ss: 0018
Process cc1plus (pid: 10248, stackpage=d05e1000)
Stack: c100000c c10fa2d4 c103400c c02a50c0 00000202 ffffffff 00009c37 00000000
000000af 00000100 c0317160 c63cf40c c012b017 c1145f14 00000100 00000100
00100000 40c00000 40afb000 c63cf40c e5060260 e870f120 00273000 408fb000
Call Trace: [<c012b017>] [<c012e2a0>] [<c011a07b>] [<c011f5e0>] [<c011f7f6>]
[<c0108a7b>]
Code: 0f 0b 73 00 ba 7b 26 c0 c6 46 24 05 8b 46 18 89 f5 83 e0 eb
<6>note: cc1plus[10248] exited with preempt_count 1
|
What luck, although at least it didn't lockup my system. Now the issues I have left are:
1) For some reason, my /home is not mounted when I boot up and I have to do it manually (mount /dev/hda3 /home.)
2) For some reason, my network isn't started when I boot up and I have to run /etc/init.d/net.eth0 start manually. Same goes for sshd.
3) After setting my timezone, rebuilding the kernel and rebooting, uname -a still tells me: Linux <myhost> 2.4.19-gentoo-r9 #7 SMP Thu Nov 28 00:01:39 Local time zone must be set--see zic i686 GenuineIntel.
Anyone got any advice to fix this stuff? It's rather.....annoying. :)
Happy Thanksgiving. |
|
Back to top |
|
|
chatgris Guru
Joined: 14 Oct 2002 Posts: 383 Location: Canada
|
Posted: Thu Nov 28, 2002 9:12 pm Post subject: |
|
|
WEll, I can try to help you with some..
1) Can you post your fstab please?
2) did you add net.eth0 to your default runlevel?
3) I don't really know hwo to help you with this one, all I did was link the correct timzone to /etc/localtime..
Josh. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Nov 28, 2002 9:21 pm Post subject: |
|
|
Ah, yeah, I should have included my fstab - but I just don't see anything wrong with it:
Code: |
/dev/hda4 swap swap defaults 0 0
/dev/hda3 /home ext3 noauto,noatime 1 1
/dev/hda2 / ext3 noauto,noatime 1 2
/dev/hda1 /boot ext2 noauto,noatime 1 2
/dev/hdc1 /archive reiserfs defaults 1 2
/dev/hdb /cdrom auto ro,noauto,user,exec 0 0
/dev/fd0 /floppy auto noauto,user 0 0
proc /proc proc defaults 0 0
|
As for 2, I don't know if I added net.eth0 to my default runlevel. How do I add stuff like that? (I was going to check if I'd just been braindead and missed it in the install instructions, but www.gentoo.org isn't responding, frig.)
For 3:
Code: |
# ls -als /etc/localtime
0 lrwxrwxrwx 1 root root 30 Nov 27 16:32 /etc/localtime -> /usr/share/zoneinfo/US/Pacific
|
Which I did with ln -sf /usr/share/zoneinfo/US/Pacific /etc/localtime, which should be fine so far as I can see. Strange, I've had any of these problems before. |
|
Back to top |
|
|
chatgris Guru
Joined: 14 Oct 2002 Posts: 383 Location: Canada
|
Posted: Thu Nov 28, 2002 9:26 pm Post subject: |
|
|
Hmm, well, it looks like you have everything working well.. My guess is that since your kernel screwed up like that it's your kernels fault.. did you try compiling a vanilla-source and seeing how that working?
I guess the only help I can really give you is for problem number 2.
rc-update add net.eth0 default
that should do it. And yes it was in the install doc, you did miss it =).
Josh. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Nov 28, 2002 10:19 pm Post subject: |
|
|
OK, rc-update add net.eth0 default did the trick. It would be more convenient if that sort of info was all stored in /etc/rc.conf like [F][O]BSD and SuSE does it IMO.
I'll try some other kernels, but I did get the same time issue with the vanilla kernel afaik - so I'd guess it's not a kernel problem.
As for the mount thing, another thing I thought was weird was that I appear to have two tmpfs':
Code: |
$ df -h
Filesystem Size Used Avail Use% Mounted on
/dev/hda2 2.5G 1.4G 958M 60% /
tmpfs 1.0M 120K 904K 12% /mnt/.init.d
/dev/hdc1 57G 50G 7.5G 87% /archive
tmpfs 378M 0 377M 0% /dev/shm
/dev/hda3 6.8G 1.4G 5.0G 22% /home
/dev/hda1 15M 3.2M 11M 22% /boot
|
I guess /etc/init.d is mounted into a tmpfs at boot time. I haven't seen this in any other distros...hm |
|
Back to top |
|
|
chatgris Guru
Joined: 14 Oct 2002 Posts: 383 Location: Canada
|
Posted: Thu Nov 28, 2002 10:41 pm Post subject: |
|
|
You know, that IS very weird.. my fstab only has one tmpfs, and it is mounted on /dev/shm NOT /mnt/.init.d.
I'd try commenting out the tmpfs on mnt/init.d... Just to see what happens..
Can't hurt (we hope )
Josh. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Nov 28, 2002 10:57 pm Post subject: |
|
|
Yeah, same here. I forgot to list:
Code: |
# grep tmpfs /etc/fstab
# glibc 2.2 and above expects tmpfs to be mounted at /dev/shm for
# (tmpfs is a dynamically expandable/shrinkable ramdisk, and will use almost no
tmpfs /dev/shm tmpfs defaults 0 0 |
Which is the only tmpfs listed in my fstab. So not only does my /dev/hda3 not get mounted to home, but an invisible magic tmpfs gets mounted. wtf :)
Edit: A reboot somehow fixed the time issue now too. Go figure, I didn't change anything.
Edit 2: In /mnt/.init.d, all the files are of size zero and look like:
Code: |
borgy .init.d # ls -als
total 8
0 drw-r--r-- 13 root root 0 Nov 28 15:41 .
4 drwxr-xr-x 3 root root 4096 Nov 28 00:08 ..
0 drwxr-xr-x 30 root root 0 Nov 28 15:41 after
0 drwxr-xr-x 30 root root 0 Nov 28 15:41 before
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 broken
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 cache
0 drwxr-xr-x 10 root root 0 Nov 28 15:41 need
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 options
0 drwxr-xr-x 4 root root 0 Nov 28 15:41 provide
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 snapshot
4 -rw-r--r-- 1 root root 8 Nov 28 15:41 softlevel
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 softscripts
0 drwxr-xr-x 2 root root 0 Nov 28 15:41 started
0 drwxr-xr-x 4 root root 0 Nov 28 15:41 use
|
|
|
Back to top |
|
|
chatgris Guru
Joined: 14 Oct 2002 Posts: 383 Location: Canada
|
Posted: Thu Nov 28, 2002 11:07 pm Post subject: |
|
|
LOL.
Actually, yes I just did a mount and I have tmpfs on /mnt/init.d too...
Sorry I got confused I thought that was an fstab file haha been working too long today...
Oh well, as long as it's working now that's what really matters =).
If you have any more problems I'm sure someone who actually knows something (not me, I'm a n00b!) could help ya more. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Nov 28, 2002 11:16 pm Post subject: |
|
|
Oh, this is the problem with the mounting issue that someone noticed on IRC, but I apparantly didn't notice (I'm blind :)):
Code: |
/dev/hda3 /home ext3 noauto,noatime 1 1
|
noauto == don't automatically mount :) |
|
Back to top |
|
|
dufeu l33t
Joined: 30 Aug 2002 Posts: 924 Location: US-FL-EST
|
Posted: Fri Nov 29, 2002 12:09 am Post subject: |
|
|
FWIW and I Know Nothin'!
I believe that you're getting two tmpfs directories for good reason.
1) If you look in kernel settings in the file system support area, you probably have tmpfs checked <b>AND</b> have told the kernel to mount it at boot time.
2) You have also included it in your fstab.
I don't know if these are the actual reasons as I am far from being any kind of linux geek wannabe, but it seemed to make sense to me. _________________ People whom think M$ is mediocre, don't know the half of it. |
|
Back to top |
|
|
bendis n00b
Joined: 06 Nov 2002 Posts: 15
|
Posted: Fri Nov 29, 2002 1:08 am Post subject: |
|
|
Hi,
that weird /mnt/.init.d is mounted by /etc/init.d/checkroot and it looks like there is stored some sort of info for gentoo runlevel system (dependency checking etc.)
I think there's really no need to get rid of it!
Bendis |
|
Back to top |
|
|
chatgris Guru
Joined: 14 Oct 2002 Posts: 383 Location: Canada
|
Posted: Fri Nov 29, 2002 1:16 am Post subject: |
|
|
No need to get rid of it.. I thought that he had that line in his fstab, but it's done authmatically by the kernel it seems.
Josh. |
|
Back to top |
|
|
zhenlin Veteran
Joined: 09 Nov 2002 Posts: 1361
|
Posted: Fri Nov 29, 2002 1:17 pm Post subject: |
|
|
/mnt/.init.d is used by gentoo's wierd runlevel system (or should I say wonderfully organised?). It is mounted by /sbin/rc boot, which is in turn run by /sbin/init, which is invoked by the kernel.
Do not unmount it unless you want a full fsck on your next reboot... |
|
Back to top |
|
|
jeremy n00b
Joined: 17 Sep 2002 Posts: 5
|
Posted: Fri Nov 29, 2002 5:39 pm Post subject: |
|
|
I think you need to take noauto out of the options for your /home mount. That tells the system not to automatically mount your /home parition. Or at least that is what it means to me. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Fri Nov 29, 2002 10:54 pm Post subject: |
|
|
jeremy wrote: | I think you need to take noauto out of the options for your /home mount. That tells the system not to automatically mount your /home parition. Or at least that is what it means to me. |
Yup, mentioned that earlier. I was just copy/pasting and forgot to take that out, and then when I looked at it my mind just somehow didn't bother to read the options part, I was just looking to make sure the partition etc was right. Figures.
As for the time issue, I finally worked that out. One thing that drove me insane was that every time I rebooted, /etc/init.d/clock would write to my system/hardware clock of what it thought the time was (which was wrong.) So I didn't know it kept screwing up my system clock and had to go check it (even though I knew I had already checked that it was correct.) |
|
Back to top |
|
|
mawst Tux's lil' helper
Joined: 23 Nov 2002 Posts: 101 Location: MN, USA
|
Posted: Thu Dec 05, 2002 6:36 pm Post subject: zic! |
|
|
Ok, you going to tell us how you fixed that zic error? I'm getting the same thing and I also setup my timezone according to the installation guide correctly. And yes, date works but uname does not. |
|
Back to top |
|
|
Zadeh Tux's lil' helper
Joined: 31 Oct 2002 Posts: 131
|
Posted: Thu Dec 05, 2002 6:39 pm Post subject: |
|
|
Once you get your timezone set correctly, rebuild your kernel and reboot (also check to make sure /etc/init.d/clock didn't screw up your time in BIOS) and then all should be dandy. |
|
Back to top |
|
|
|