Not sure this is a kernel issue, but the kernel maintainers will know how to find out :-)

@ Shelandy

Does it work when you boot with an older kernel?

CC: sysadmin-bugs => mageiatools, marja11
Component: Release (media or process) => RPM Packages
Assignee: bugsquad => kernel

I did not know that we can switch the network profile until recently that I can not get wpa2-enterprise connection work.  I had not tested it when I used the old kernels in the past and I am not able to test this feature now since I only keep the lasted version of kernel in my very tide SSD drive.

Please, after shutting your system down: 
* note the time
* boot your system two times
** where during the first boot you try to type the profile number when asked
** and the 2nd boot you don't.

After having successfully booted up the second time:
* please start a terminal/konsole
* become root.

Then type:

   journalctl --since="YYYY-MM-DD hh:mm" > journal.txt

Where YYYY-MM-DD and hh:mm match the date you're doing this + the noted time, e.g.: journalctl --since="2016-09-25 19:25" > journal.txt

Please attach journal.txt to this bug report

Created attachment 8468 [details]
hang after selecting a profile, continue with magic keys

I could reproduce this in cauldron

After creating a test profile with draknetprofile

[marja@cldrn_64 ~]$ ls -al /etc/netprofile/profiles/
totaal 16
drwxr-xr-x 4 root root 4096 sep 29 13:51 ./
drwxr-xr-x 4 root root 4096 feb  8  2016 ../
drwxr-xr-x 9 root root 4096 sep 29 13:51 default/
drwxrwxr-x 9 root root 4096 sep 29 13:51 test/
[marja@cldrn_64 ~]$ cat /etc/netprofile/current
test
[marja@cldrn_64 ~]$

I tried to switch from the activated test profile to the default one during boot up.

It prevented my system from finishing the boot process, it just hung.
However, I did not wait 10 minutes (intend to do that later).

I failed to see anything useful in the logs, but I'll attach logs anyway of the seeming freeze (when switching between plymouth and text mode, in both the request to make a choice kept being there, despite having chosen)

     Select Network Profile:
     (1)default (2)test*

In the attached logs I decided to try to continue with alt+sysrq+e around 19:00h, which succeeded
Despite having typed "1", the test profile was used.

I have strong doubts about this being a kernel issue, changing the assignee

Version: 5 => Cauldron
Assignee: kernel => mageiatools
Whiteboard: (none) => MGA5TOO

I following the suggest to make a test.  I boot up the 4.4.16 kernel at 3:54.  I have three profile, the first one is the default, So press "3"  when the network profile selection shown up.  I have waited for 6min.  No progress at all (I have benchmarked my system before: this ssd can be booed in 16 sec.) Since the boot is hang by pressing a number key, there is no way for me to get the  journal.txt. I can not even use crontol-alt-del to shut down.  I have to use power button to force it reboot at 4:03

Created attachment 8470 [details]
the log of the 2nd boot, (can not get the log of  1st one becuase it hung)

the log that I reboot the 4.4.16 kernel at 4:03pm
I can not get the previous on at 3:54 where I press "3"  when the network profile selection shown up.  It hung there over 6 mins. I have to use power button to force it reboot

(In reply to shelandy ting from comment #7)

> I can not get the previous on at 3:54 where I press "3"  when the network
> profile selection shown up.  It hung there over 6 mins. I have to use power
> button to force it reboot

Those logs should exist, and be either available using "--since=", as explained in comment 3, or using e.g.:

    journalctl -ab -1 > output.txt


-1 is the journal from 1 boot ago
-2 is the journal from 2 boots ago
-3 is the one from 3 boots ago
etc.

I hope there's more in your log than in mine.


Btw, typing the password for an encrypted partition during boot-up (in another cauldron on the same machine) does not give any problems at all here.

Typing the network selection gives problems both when seeing plymouth and when only seeing log messages with "splash=verbose".
The hang happens before an attempt is made to start a display manager (and, does, of course, also happen when logging into RL3).

When selecting the last activated network profile, by pressing the corresponding number the moment you're prompted to choose, _immediately_ the message appears that that network profile will be used (so much faster than when not choosing).

It is only typing the number of a _different_ network profile than the last activated one that the systems hangs for ever.

I've tried this both with network profile 1 as last used one, and with network profile 2 as last used one. (I have not tried typing a network profile name.)

Summary: network profile can not be selected by keyboard during the boot time => trying to switch to another network profile during boot makes the system hang forever (typing the number of the last used profile works ok)

Created attachment 8507 [details]
the journel file containing failing and sucessful boot info

line 1115 and line 1116 seems to be the turning point for failing boot

Oct 07 17:28:45 pipa kernel: input: SynPS/2 Synaptics TouchPad as /devices/platform/i8042/serio4/input/input17
Oct 07 17:30:07 pipa alsactl[814]: alsactl daemon stopped

while in line 2236, it goes on to load OSS Proxy

Oct 07 17:30:59 pipa kernel: input: SynPS/2 Synaptics TouchPad as /devices/platform/i8042/serio4/input/input17
Oct 07 17:31:00 pipa osspd[880]: OSS Proxy v1.3.2 (C) 2008-2010 by Tejun Heo <teheo@suse.de>

It looks like something happening after line 1115 kernel: input: SynPS/2 Synaptics TouchPad
where I can not make the boot go through, eventually I have to hit the power button to shut down the machine

CC: (none) => shelandy

@ shelandy ting

can you confirm that typing the number of the last used profile, works ok during boot?

(if not, then I'll revert the summary of this report and open a separate report for my findings)

Yes, I did try typing the number of last used profile, it will go through.  

So,I have confirmed that what Marja van Waes mentioned above regarding using the last working profile can be repeated on my machine.

This may be a red herring, but there is something very funny: since I created a second profile and switched to it:

sep 29 13:51:09 cldrn_64 draknetprofile[2772]: ### Program is starting ###
sep 29 13:51:18 cldrn_64 draknetprofile[2772]: launched command: /sbin/netprofile switch test
sep 29 13:51:20 cldrn_64 draknetprofile[2772]: Switching to "test" profile
sep 29 13:51:38 cldrn_64 draknetprofile[2772]: ### Program is exiting ###

Since then I've often seen many such messages:
sep 29 18:46:04 cldrn_64 pam_timestamp_check[5586]: PAM `/' permissions are lax


[marja@cldrn_64 ~]$ ll -d /
drwxrwxr-x 25 root root 4096 sep 29 13:51 //
[marja@cldrn_64 ~]$ stat /
  File: '/'
  Size: 4096         Blocks: 8           IO Block: 4096   directory
Device: 806h/2054d   Inode: 2            Links: 25
Access: (0775/drwxrwxr-x)   Uid: (    0/    root)   Gid: (    0/    root)
Access: 2016-01-18 23:49:24.000000000 +0100
Modify: 2016-09-29 13:51:19.000000000 +0200
Change: 2016-10-12 09:59:40.883815568 +0200
Birth:  -
[marja@cldrn_64 ~]$ 

I've now changed the / permissions to 0755, which immediately stopped those "PAM `/' permissions are lax" messages, but doubt that'll survive a reboot. If it does, then I'm curious whether that'll somehow magically help to allow switching to a different profile during boot.

(In reply to Marja van Waes from comment #13)

> 
> I've now changed the / permissions to 0755, which immediately stopped those
> "PAM `/' permissions are lax" messages, but doubt that'll survive a reboot.
> If it does, then I'm curious whether that'll somehow magically help to allow
> switching to a different profile during boot.

It does not survive a reboot, it's 0775 again. 

Sorry if this is totally unrelated. The time stamp may come from a very different package, but I don't know how to query which package has set the permissions on / :-(

The lax permissions I saw on / are not related to this bug.
I could reproduce (in a different cauldron) the / directory getting a timestamp that matches the time of when a second profile was created. 
However, this time the permissions on the root directory did not change, but this bug was valid anyway.

Booting a cauldron (fully updated Mageia-7 Beta-1 system; Kernel 5.0.3-desktop-2-mga7) I hit what very much looks like this problem:

- I have a choice between 3 network profiles;
- if, when the network selection question appears, I let the system use the (default) previously used profile, everything goes well;
- if I type-in anything specific (digit, letter ... ) the the system goes into a hang; the only way to get out of the hang is a complete restart (power cycle);
- in case the first key I hit is ESC, I obtain a console which illustrates the hang situation.

Note: I nearly ever use netprofile during boot, normally - if I need to change profile - I do this with drakconf once the the system is running. Therefore I have no idea whether the problem is new in Mageia-7 or whether it is the old bug still being around.

Drakconf is a perfectly valid work-around, but the nasty aspect of this bug is that the boot goes into a hang if the keyboard is touched during the critical period. How about simply de-activating the possibility to call netprofile during boot as long as this problem is not solved?

CC: (none) => juergen.harms