Long gone and now totally screwed

if · n00b Joined: 15 Jun 2010 Posts: 23

I have been away from the forum for two years and three months. Also away from my Gentoo box for one and nine months.
During this either short or long absence, a lot seems to have changed to me.
I even had very little idea where to start for update.
I remember some commands for update like eix-sync and emerge -u package for sure.
First, I thought it would be better to update gcc to 5.0.2 and I did it.
Then moved to firefox and stuffs, and then gettext. (Up until this point, it was perfect.)
Then what? System world seemed to have lots of conflict, so I gave up after successfully (?) emerged udev by temporally removing sys-init scripts and then back on (this thread was helpful).
Next, openrc seemed nice to me (anyway, I used it for a while somehow). So I tried. It seemed okay though.
After rebooting, startx shows me X desktop. But no input device works (maybe this thread would work but not yet tried).
Confronted with the situation, I still had hope that networking would help.
But, I found no interface except lo on the box. It used to have two I/F, eth0 and wlan0.
I read this seemingly helpful thread but there is even no enpXsY I/F besides lo.
The kernel is still at the beginning of 2014, that is 3.12.x of gentoo-sources and 3.13.x of git-sources, not yet recompiled at the moment.

Sorry for not providing accurate info with detail, but I am writing this on the same box, from the other side (you know what I mean).

So, here are my issues.
1. No input device works on X.
2. No network connection even in CUI.

What do you suggest my next move to be?
Do I need issue old MAKEDEV style command for enpXsY to appear?
Or should I reconfigure the old kernel sources to fit latest init scripts?

Any thought would be much appreciated.
Thank you.

Keruskerfuerst · Posted: Fri Sep 04, 2015 4:41 pm Post subject:

After such a long time of non-updating, you should reinstall your system.

NeddySeagoon · Posted: Fri Sep 04, 2015 4:53 pm Post subject:

if,

Lets fix networking first.

You say you updated openrc. Did you do etc-update afterwards to get any changed init scrips and config files?
If you say no, thats a problem, if you say yes thats a problem too.
Some updates to openrc require the matching scripts and config files, so you need to run etc-update but in the process, you must keep your configuration information.
Keep that in mind as you read on.

Your interfaces probably still exist as the kernel has not been updated.

if · n00b Joined: 15 Jun 2010 Posts: 23

Thank you guys.

On my second thought, fresh install is the option. But so far I would like to stick to it.

ifconfig -a gives only lo.
lspci shows ethernet controller (Intel 82566MM).
I noticed network device module was not loaded. Manual insertion did not work.
insmod /lib64/modules/3.13.0_rc8/kernel/drivers/net/ethernet/intel/e1000e/e1000e.ko failed to be loaded with error message "unknown symbol in module".
ixgb.ko was loaded. Still, ifconfig -a does not show eth0, only lo.
Changing gcc profile back to 4.8.x won't work.

Like I said, kernel was not changed. I do not know why e1000e module failed to be loaded. The module was not touched this occasion.

NeddySeagoon · Posted: Sat Sep 05, 2015 10:23 am Post subject:

if,

insmod is brain dead - it does exactly what you ask and no more.
modprobe is a better choice. It resolves module dependencies too, so if e1000e deponds on another module, modprobe e1000e will load the entire dependency tree.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

if · n00b Joined: 15 Jun 2010 Posts: 23

With the help of modprobe (I totally forgot but remembered), now both eth0 and wlan0 exist and they started after issuing /etc/init.d/net.* restart command (restart... because they are both failed in initial boot).
The trouble now is... DHCP is not working. ping to outside world does not return any response.
Reading this document, my conf files have been set properly, it seems.
/etc/init.d/net is vacant. actual conf info is stored in /etc/wpa_supplicant/wpa_supplicant.conf and I think it worked fine a year ago.
One thing that comes to my mind is I edited /etc/sudoer when I did etc-update and commented out %wheel entry. I found ctrl_interface_group belonged to wheel in the conf file. Now I changed sudoer file back to the one before (wheel line is active), I think this part is okay.
Another clue is when running /etc/init.d/net.* commands, they both warn "net.* has started, but is inactive".

I think it is getting so close to finally connect to the world.
DHCPCD is working fine, at least via message.
What is wrong with the box now?

krinn · Watchman Joined: 02 May 2003 Posts: 7470

NeddySeagoon · Posted: Sat Sep 05, 2015 2:41 pm Post subject:

if,

krinn has asked for the IP addresses on your interfaces. That will help.
What is in /etc/resolv.conf ?

No nameservers will mean you can't get anywhere by name, so ping google.com will fail but ping 173.194.113.102 may work.
They are both google.com.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

if · n00b Joined: 15 Jun 2010 Posts: 23

Sorry for sharing so little info. As is right now, I am writing this from Windows (the box is dual boot with Gentoo).
I have been restarting both OS, going back and forth. All messages, responses are either memorized in my head or written down if not complicated. Detailed log kind info cannot be shared until at least network on Gentoo turns back normal.
So please bare with me.

/etc/resolv.conf has only one line "nameserver 192.168.0.1" (besides comments), which is correct in my setting.

Ping does give back some response if you say so.
That is, when using www.google.com, it is "ping: unknown host www.google.com" and when using 173.194.113.102, it is "connect: Network is unreachable".

Other things I noticed this time, iwconfig shows correct ESSID and Access Point in my home network for wlan0. wlan0 address is shown in inet6.
"dhcpcd -r 192.168.0.2" returns following messages:
version 6.2.0 starting
dev: loaded udev
no valid interfaces found
no interfaces have a career
forked to background, child pid 7102

NeddySeagoon · Posted: Sat Sep 05, 2015 5:23 pm Post subject:

if,

Run

if · n00b Joined: 15 Jun 2010 Posts: 23

dhcpcd eth0 returns:
dhcpcd[6988]: sending commands to master dhcpcd process.

ifconfig -a returns eth0 address in inet6 style. (wlan0 in inet6 too.)

route -n returns:
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
127.0.0.0 127.0.0.1 255.0.0.0 UG 0 0 0 lo

/etc/init.d/net.eth0 restart gives:
Bringing down interface eth0
Stopping ifplugd on eth0
Removing addresses
Bringing up interface eth0
Starting ifplugd on eth0
Backgrounding
Warning: net.eth0 has strted, but is inactive

ping 173.194.113.102 gives:
connect: Network is unreachable.

This time, during the whole process, only LAN cable to eth0 is connected, WLAN switch for wlan0 is always off.
Before that, after confirming wlan0 module was loaded, I tested mainly on wlan0, not eth0.

So the problem is eth0 is not recognized in routing table?

if · n00b Joined: 15 Jun 2010 Posts: 23

Reading this document, I finally managed to connect to outside world!

What I did was...
modprobe e1000e
/etc/init.d/net.eth0 restart
ifconfig eth0 192.168.0.2 broadcast 192.168.0.255 netmask 255.255.255.0 up
route add default gw 192.168.0.1

Then, ping www.google.com returned actual response from the server.
So, I think network problem is solved except that the procedure has to be done manually.
Thank you all for your help!

The next problem is no input on X.
After loading evdev (which is unloaded after boot), still the problem persists.
X.org.log will be coming soon...

if · n00b Joined: 15 Jun 2010 Posts: 23

Okay, so Xorg.0.log is this.
https://bpaste.net/show/d2f3577c67e5

NeddySeagoon · Posted: Sun Sep 06, 2015 9:01 am Post subject:

if,

Your network setup can be automated. The documentation is in several places.
To minimise your googling, add e1000e to /etc/conf.d/modules - read the comments in the file.

There is a static net setup in /usr/share/doc/netifrc-0.3.1/net.example.bz2 or whatever your version is.
All the entries in that file are commented out so you can uncompress it to /etc/conf.d/net then set whatever you need.
Do preserve your own net file first.

From your Xorg log

if · n00b Joined: 15 Jun 2010 Posts: 23

Adding module names in /etc/conf.d/modules did not load them automatically.
I still have to manually modprobe those in the file, e1000e, and wireless modules (rfkill, cfg80211, mac80211, iwlegacy, wal4965).
I have not tested netifrc yet.

As for Xorg-server, the config file /etc/X11/xorg.conf was in fact all commented out, but not vacant.
But to make sure, I renamed it to /etc/X11/xorg.conf.not_in_use.
And after that, I reinstalled xf86-input-evdev and the server acted the same, no input available at all (dock applets work).

Then, I noticed my layman setting became messy, so I once removed x11 overlay and newly installed.
Actually, I had got an error while emerging old xorg-server-9999 and decided to renew the overlay.
Some prerequisites were asked to be installed, xproto, xtrans, randrproto, fontsproto.
Eventually xorg-server was installed with the new overlay.
Then startx, but the result is the same. No input available.
I should have also report the Xorg.0.log but please give me some time.

During the update, I also upgraded or reemerged udev, cause I thought udev listing something warning message in the log told me the clue.
/etc/init.d/udev --nodeps restart did not change the situation.

Anyway, this is my emerge --info (repository and its first messy errors cut out. please see the link if you need.)
https://bpaste.net/show/133b076c7ce0

NeddySeagoon · Posted: Mon Sep 07, 2015 3:36 pm Post subject:

if,

Module loading works here,

Tony0945 · Posted: Mon Sep 07, 2015 4:31 pm Post subject:

Buffoon · Veteran Joined: 17 Jun 2015 Posts: 1369 Location: EU or US

Modules won't be loaded if modules are not run by openrc.

BTW, INPUT_DEVICES="mouse keyboard evdev" is OK, but overkill. One needs either "mouse keyboard" or "evdev", not all of them.