[SOLVED] no connection with Intel Ethernet I217-LM

[rych]

With the e1000e replaced to 3.3.5.3-NAPI, which works on the l219 machine fine, on the problem l217 I first find out which ethX device the eth0 got renamed to, if any:

Code: Select all

root@porteus:~# ifconfig eth0 down
eth0: ERROR while getting interface flags: No such device
root@porteus:~# dmesg | grep -e eth -e e1000
[    4.341578] e1000e: loading out-of-tree module taints kernel.
[    4.341730] e1000e: Intel(R) PRO/1000 Network Driver - 3.3.5.3-NAPI
[    4.341731] e1000e: Copyright(c) 1999 - 2016 Intel Corporation.
[    4.341870] e1000e 0000:00:19.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[    5.519033] e1000e 0000:00:19.0 0000:00:19.0 (uninitialized): registered PHC clock
[    5.602009] e1000e 0000:00:19.0 eth0: (PCI Express:2.5GT/s:Width x1) 74:e6:e2:e2:a2:78
[    5.602014] e1000e 0000:00:19.0 eth0: Intel(R) PRO/1000 Network Connection
[    5.602060] e1000e 0000:00:19.0 eth0: MAC: 11, PHY: 12, PBA No: FFFFFF-0FF
[    5.633367] e1000e 0000:00:19.0 eth126: renamed from eth0
[    5.639724] e1000e 0000:00:19.0 eth7: renamed from eth126
[    9.262823] e1000e: eth7 NIC Link is Up 100 Mbps Full Duplex, Flow Control: None
[    9.262835] e1000e 0000:00:19.0 eth7: 10/100 speed: disabling TSO

Then I run the sequence below with the output (sorry for the wrong order of dhcpcd -k and dhcpcd, will try again tomorrow):

Code: Select all

root@porteus:~# ifconfig eth7 down
root@porteus:~# ifconfig eth7 up
root@porteus:~# dhcpcd -k eth7
dhcpcd not running
root@porteus:~# dhcpcd eth7
DUID 00:01:00:01:20:2b:71:b2:74:e6:e2:e2:83:3a
eth7: IAID e2:e2:a2:78
eth7: soliciting a DHCP lease
eth7: soliciting an IPv6 router
eth7: probing for an IPv4LL address
eth7: using IPv4LL address 169.254.113.179
eth7: adding route to 169.254.0.0/16
forked to background, child pid 1486

root@porteus:~# dhclient -v eth7
Internet Systems Consortium DHCP Client 4.3.4
Copyright 2004-2016 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/eth7/74:e6:e2:e2:a2:78
Sending on   LPF/eth7/74:e6:e2:e2:a2:78
Sending on   Socket/fallback
DHCPDISCOVER on eth7 to 255.255.255.255 port 67 interval 6
DHCPDISCOVER on eth7 to 255.255.255.255 port 67 interval 15
DHCPDISCOVER on eth7 to 255.255.255.255 port 67 interval 16
DHCPDISCOVER on eth7 to 255.255.255.255 port 67 interval 21
DHCPDISCOVER on eth7 to 255.255.255.255 port 67 interval 3
No DHCPOFFERS received.
No working leases in persistent database - sleeping.
root@porteus:~# cat /etc/resolv.conf
# Generated by dhcpcd
# /etc/resolv.conf.head can replace this line
# /etc/resolv.conf.tail can replace this line

And separately, a content of /etc/dhcpcd.conf with all the comment lines removed:

Code: Select all

root@porteus:~# cat /etc/dhcpcd.conf
hostname
duid
persistent
option rapid_commit
option domain_name_servers, domain_name, domain_search, host_name
option classless_static_routes
option ntp_servers
require dhcp_server_identifier
nohook lookup-hostname
nohook wpa_supplicant

dhcpcd -t 60 didn't make any difference.

EDIT: run a few times "dhcpcd -k eth7" followed by "dhcpcd eth7" -- no difference.

Porteus 2.1 32bit doesn't connect on the l217 either. Will try the 3.1 porteus as you suggest.

[neko]

New "ISO update Example" v4.9.9 with e1000e-3.3.5.3 was offered.
Please refer to http://forum.porteus.org/viewtopic.php? ... 232#p52232

The kernel of Porteus ISO can be update to version 4.9.9 with e1000e-3.3.5.3 easily without compiling.


---------------------------------------------------------
[How to update the kernel of Porteus ISO]
Please refer to the example of "kernel 4.10-rc6".
http://forum.porteus.org/viewtopic.php? ... =15#p52642


Thanks.

[rych]

Thanks @neko, for building the e1000e-3.3.5.3. I've downloaded a fresh copy of the iso I use, Porteus-XFCE-v3.2.2-x86_64.iso and tried to patch it with your Porteus Kernel Builder. But I can't select the kernel version in the GUI and then it doesn't run, because it complains that the kernel version is missing. Pointing it to either v4.9.9 folder or its parent, doesn't help. Maybe because I was doing it without network connection. Anyway, could you send me just the patched 000-kernel.xzm of my distro, if that the only thing that is patched inside the iso, so I could just replace it and try again.

[neko]

@rych
UP.Porteus-XFCE-v3.2.2-x86_64.iso is one of kernel version 4.9.9 with e1000e-3.3.5.3 driver.
"vmlinuz & 000-kernel.xzm" of Porteus-XFCE-v3.2.2-x86_64.iso were replaced.

Note:
"Porteus kernel builder" needs an internet connection in order to get the recent kernel version information.


Thanks.

[neko]

Temporary upload, UP.Porteus-XFCE-v3.2.2-x86_64.iso, was deleted.

[rych]

@neko
Could you upload somewhere either the UP.Porteus-XFCE-v3.2.2-x86_64.iso or, if it's too big, only the vmlinuz & 000-kernel.xzm? I'm back in the office and can now try it.
Thanks

[neko]

@rych

you can get "vmlinuz & 000-kernel.xzm" from v4.9.10.tar.xz.

v4.9.10.tar.xz (39 M)
http://www.mediafire.com/file/4qm2tdnl8 ... .10.tar.xz
md5sum: 8fb31ba70c8cfdd81869dd392bc23869 v4.9.10.tar.xz

Code: Select all

% su
# ls -1
Porteus-XFCE-v3.2.2-x86_64.iso
v4.9.10.tar.xz
# mkdir mnt
# mount -t iso9660 -o loop Porteus-XFCE-v3.2.2-x86_64.iso mnt
# cp mnt/porteus/base/000-kernel.xzm old
# umount mnt
# mount -t squashfs -o loop old mnt
# mkdir new
# cp -a mnt/* new/
# umount mnt
# rm -fr mnt old
# xz -dc v4.9.10.tar.xz | tar -x
# rm -fr new/lib
# mv v4.9.10/64/lib new/
# mv v4.9.10/64/vmlinuz .
# rm -fr v4.9.10
# mksquashfs new 000-kernel.xzm -b 256K -comp xz -Xbcj x86 -noappend
# rm -fr new
# ls -1
000-kernel.xzm
Porteus-XFCE-v3.2.2-x86_64.iso
v4.9.10.tar.xz
vmlinuz

For 32 bit,
Porteus-XFCE-v3.2.2-x86_64.iso ---> Porteus-XFCE-v3.2.2-i586.iso
v4.9.10/64/ ---> v4.9.10/32/

For "Porteus 2.1 32-bit ISO"
Porteus-XFCE-v3.2.2-x86_64.iso ---> "Porteus 2.1 32-bit ISO"
v4.9.10/64/ ---> v4.9.10/32/

For "Porteus 2.1 32-bit ISO" & "kernel version 4.10-rc8"
Porteus-XFCE-v3.2.2-x86_64.iso ---> "Porteus 2.1 32-bit ISO"
v4.9.10.tar.xz ---> v4.10-rc8+e1000e-3.3.5.3.tar.xz
v4.9.10/64/ ---> v4.10-rc8/32/
about v4.10-rc8+e1000e-3.3.5.3.tar.xz, refer to http://forum.porteus.org/viewtopic.php? ... 030#p52232


Thanks.

[rych]

OK, thanks to @neko, I've now tested his new v4.10-rc8+e1000e-3.3.5.3 kernel, (booted always fresh, confirmed that the new kernel and e1000e-3.3.5.3 were in fact loaded, tried the same commands to restart-re-enforce the eth0 and dhcp) but unfortunately DHCP doesn't connect anyway -- no difference in the output and logs.

[neko]

@rych
I was disappointed with unfortunate results.
I don't have the equipment.
Execuse me that I can not test directly by myself.

v4.10-rc8+e1000e-3.3.5.3 was built with 2 patch.
One is gotten from intel. (v4.10-rc8/e1000e-3.3.5.3)
Another is modifid by me. (v4.10-rc8/e1000e-3.3.5.3-v4.10)
This my modification might be not fit to kernel v4.10.

First example,
v4.9.10.tar.xz was built with the patch that is gotten from intel. (v4.9.10/e1000e-3.3.5.3)
My modification was not included.

Thanks.

[rych]

Tomorrow I'm going to try your v4.9.10.tar.xz then. Also, I'm going to run the script at the end of this page https://bbs.archlinux.org/viewtopic.php?id=191981 just before bringing the eth0 up again.

[rych]

Good news: it works! The combination of the new kernel/e1000e, resetting eth0, dhcp with resetting PCI inbetween (I suspect this is the key) gave me an IP address and a DHCP connection. I'm now going to try to narrow this result down to the necessary ingredient(s) only and will post the final, minimal solution.

(Today must be a lucky day: another USB stick worked too. I've tried a 3-4 years old puppy linux USB stick, and it picked up the network automatically even though it has a 2.X.X. version of e1000e driver.)

[neko]

@rych
Congratulation on your good news.
I expect that your report will explain "how to set up".

Thank you.

[rych]

Solution:
1. Boot as usual with your stock Porteus, no need to change the kernel or e1000e.
2. The network will be trying and complaining that it can't connect
3. While it's still trying, simply execute the script resetting the PCI:

Code: Select all

#!/bin/bash
#Get the PCI-Address of network card (Caution: This works ONLY with ONE NIC)
PCI=`/sbin/lspci | /bin/egrep -i 'network|ethernet' | /usr/bin/cut -d' ' -f1`
PCIPATH=`/usr/bin/find /sys -name *\${PCI} | /bin/egrep -i *pci0000*`
/usr/bin/logger -t "ResetNIC" "Resetting PCI NIC ${PCIPATH}"
#Reset the PCI Device completely (like Power-ON/Off)
echo 1 >${PCIPATH}/reset

4. After a few seconds the Network will suddenly connect!
5. Think where to put that script to be run automatically maybe.

Thanks @donald, @neko, @francois for helping me, and sorry for leading you astray with this. Thanks to you I've learnt something about resetting network, replacing drivers and kernels. I hope though I won't be needing that knowledge again anytime soon . Please keep up the good job maintaining the wonderful porteus and the community.

[donald]

Excellent!..

...sorry for leading you astray with this

Don't worry...it is the [SOLVED] mark that counts.