Hello, an update this morning and after that I register several problems:
1) delay in login session (lightdm) 2) delay/fail of logout/shutdown from inside gui 3) unable to open wifi connections from gui
verified both with mate and cinnamon
For 1) and/or 2) I see May 15 16:29:12 ope46 journal[4284]: Failed to open CK session: Timeout was reached May 15 16:29:12 ope46 lightdm[4160]: Error writing to daemon: Bad file descriptor May 15 16:29:12 ope46 lightdm[4160]: Failed to write utmpx: Permission denied
For 3) I see May 15 16:57:15 ope46 NetworkManager[819]: <info> [1526396235.9934] audit: op="connection-activate" uuid="2f2235c1-a162-48c8-ac14-5c 50d6b1f7a7" name="VodafoneMobileWiFi-04CE26" pid=4679 uid=1000 result="fail" reason="Not authorized to control networking."
Previously I got window where I had to insert root password and then able to start the wifi connection
Packages involved this morning update
pam-kwallet.x86_64 5.12.5-3.fc27 systemd-pam.x86_64 234-11.git5f8984e.fc27
systemd-udev.x86_64 234-11.git5f8984e.fc27
Anyone else?
Thanks, Gianluca
On Tue, 15 May 2018 17:05:36 +0200 Gianluca Cecchi wrote:
Anyone else?
I haven't noticed those specific problems, but pam being messed up was certainly the reason NIS logins weren't working:
https://bugzilla.redhat.com/show_bug.cgi?id=1575297
Possibly the new authselect stuff has messed up more than just NIS?
On Tue, May 15, 2018 at 5:05 PM, Gianluca Cecchi gianluca.cecchi@gmail.com wrote:
Hello, an update this morning and after that I register several problems:
- delay in login session (lightdm)
- delay/fail of logout/shutdown from inside gui
- unable to open wifi connections from gui
verified both with mate and cinnamon
I decided to update this system from f27 to f28 and all went well from the update itself point of view. But still the same problems like the lats updates.
From mate I select a wifi network and then select connect I get the error window with:
" Connection activation failed (1) Not authorized to control networking "
Going into /etc/pam.d and comparing with a f27 system not already updated (it is up to date at 06 May), I see this I don't know if can influence my problems
system-auth-ac old: password requisite pam_pwquality.so try_first_pass local_users_only retry=3 authtok_type=
new: password requisite pam_pwquality.so try_first_pass retry=3 authtok_type=
password-auth-ac old: password requisite pam_pwquality.so try_first_pass local_users_only retry=3 authtok_type=
new: password requisite pam_pwquality.so try_first_pass retry=3 authtok_type=
BTW: with latest updates there is also a problem with audio card, not detected any more. From hwinfo point of view I see this:
snd_hda_codec_hdmi: /devices/pci0000:00/0000:00:1b.0/hdaudioC0D3 snd_hda_codec_hdmi: module = snd_hda_codec_hdmi snd_hda_codec_generic: module = snd_hda_codec_generic snd_hda_codec_realtek: module = snd_hda_codec_realtek snd_hda_codec_realtek: /devices/pci0000:00/0000:00:1b.0/hdaudioC0D0
The pc is an Asus U36SD
Running alsa-info.sh in the ouptut I see
pcilib: sysfs_read_vpd: read failed: Input/output error
and in the generated file I see
!!Soundcards recognised by ALSA !!-----------------------------
0 [PCH ]: HDA-Intel - HDA Intel PCH HDA Intel PCH at 0xdf000000 irq 34
!!PCI Soundcards installed in the system !!--------------------------------------
00:1b.0 Audio device: Intel Corporation 6 Series/C200 Series Chipset Family High Definition Audio Controller (rev 05)
!!Advanced information - PCI Vendor/Device/Subsystem ID's !!-------------------------------------------------------
00:1b.0 0403: 8086:1c20 (rev 05) Subsystem: 1043:1ba3
but in mate and cinnamon nothing as hardware device Up to three days ago I could choose between analog and hdmi
Full output file here: https://drive.google.com/file/d/1ZXEwzHM9XgOsqiUv-OF2zJ-UqbZrbGHG/view?usp=s...
Any help on these problems? Thanks in advance, GIanluca
On Wed, 16 May 2018 10:09:04 +0200 Gianluca Cecchi gianluca.cecchi@gmail.com wrote:
BTW: with latest updates there is also a problem with audio card, not detected any more. From hwinfo point of view I see this:
snd_hda_codec_hdmi: /devices/pci0000:00/0000:00:1b.0/hdaudioC0D3 snd_hda_codec_hdmi: module = snd_hda_codec_hdmi snd_hda_codec_generic: module = snd_hda_codec_generic snd_hda_codec_realtek: module = snd_hda_codec_realtek snd_hda_codec_realtek: /devices/pci0000:00/0000:00:1b.0/hdaudioC0D0
The pc is an Asus U36SD
Running alsa-info.sh in the ouptut I see
pcilib: sysfs_read_vpd: read failed: Input/output error
and in the generated file I see
!!Soundcards recognised by ALSA !!-----------------------------
0 [PCH ]: HDA-Intel - HDA Intel PCH HDA Intel PCH at 0xdf000000 irq 34
!!PCI Soundcards installed in the system !!--------------------------------------
00:1b.0 Audio device: Intel Corporation 6 Series/C200 Series Chipset Family High Definition Audio Controller (rev 05)
!!Advanced information - PCI Vendor/Device/Subsystem ID's !!-------------------------------------------------------
00:1b.0 0403: 8086:1c20 (rev 05) Subsystem: 1043:1ba3
but in mate and cinnamon nothing as hardware device Up to three days ago I could choose between analog and hdmi
Full output file here: https://drive.google.com/file/d/1ZXEwzHM9XgOsqiUv-OF2zJ-UqbZrbGHG/view?usp=s...
Any help on these problems?
Can't help with the network permissions problem, but for the sound problem, you can install pavucontrol, and check which device is set as default. Since non-deterministic instantiation of drivers is the norm, it is possible that the order of your sound devices has changed, and the old default now points to the hdmi device. An easy try, at least. alsa-info shows that the pci device is there. If you have a wav file, you can try aplay --D plughw:1,0 [wav file] to see if the device is working.
Thanks Stan for your feedback. Actually it seems I've narrowed the origin of the problem. I have been using Mate as DE for many months without any problem. Only recently (when still in F27 and some days before the start of the problems described in this thread) I switched from gdm to lghtdm because I had some lagging events and I tried to eliminate at all gnome/gdm/gnome-shell dependency
Now (I'm in F28) it is simply sufficient to re-enable gdm and connectusing Gnome as DE to see that all is ok: - logout/shutdown buttons without problems (previously a logout originated an automatic relogin...) - switch wifi on/off and other components (such as external usb disks) without getting authorization errors - audio adapter is seen ok with its analog and hdmi settings (and it works too... ;-)
So it seems that the problem doesn't show with default Fedora environment but only with the combination I outlined above. I also tried Cinnamon (with lightdm) and same problems.
Possibly gdm in its start enables some permission for normal users that lightdm (at least in latest updates) doesn't provide? I'm going also to test other combinations, to narrow down better (eg lightdm with gnome DE session, ecc...)
Gianluca
On Thu, 17 May 2018 10:54:57 +0200 Gianluca Cecchi gianluca.cecchi@gmail.com wrote:
Thanks Stan for your feedback. Actually it seems I've narrowed the origin of the problem. I have been using Mate as DE for many months without any problem. Only recently (when still in F27 and some days before the start of the problems described in this thread) I switched from gdm to lghtdm because I had some lagging events and I tried to eliminate at all gnome/gdm/gnome-shell dependency
Now (I'm in F28) it is simply sufficient to re-enable gdm and connectusing Gnome as DE to see that all is ok:
- logout/shutdown buttons without problems (previously a logout
originated an automatic relogin...)
- switch wifi on/off and other components (such as external usb disks)
without getting authorization errors
- audio adapter is seen ok with its analog and hdmi settings (and it
works too... ;-) So it seems that the problem doesn't show with default Fedora environment but only with the combination I outlined above. I also tried Cinnamon (with lightdm) and same problems.
Possibly gdm in its start enables some permission for normal users that lightdm (at least in latest updates) doesn't provide? I'm going also to test other combinations, to narrow down better (eg lightdm with gnome DE session, ecc...)
Excellent! And you know you have a working fallback, which is always nice when testing. It's strange that a desktop manager would affect this, since it is the desktop that performs the actions that are having problems. Maybe you eliminated some gnome dependency that cinnamon and mate were depending on? Or maybe you are right, and it is some permission or configuration that lightdm is not completing as gdm does.
Hi.
On Thu, 17 May 2018 10:19:52 -0700 stan wrote:
On Thu, 17 May 2018 10:54:57 +0200 Gianluca Cecchi gianluca.cecchi@gmail.com wrote:
Now (I'm in F28) it is simply sufficient to re-enable gdm and connectusing Gnome as DE to see that all is ok:
- logout/shutdown buttons without problems (previously a logout
originated an automatic relogin...)
- switch wifi on/off and other components (such as external usb disks)
without getting authorization errors
- audio adapter is seen ok with its analog and hdmi settings (and it
works too... ;-)
Possibly gdm in its start enables some permission for normal users that lightdm (at least in latest updates) doesn't provide?
I don't think it's that. See below.
We are using lightdm and noticed also this problem.
It's strange that a desktop manager would affect this, since it is the desktop that performs the actions that are having problems.
This is normal since those actions rely on having a proper session/seat defined to be authorized by polkit and that is the desktop manager that setup that.
I found that the culprit is the RPM pam-kwallet-5.12.5-3.fc27: if you erase it this problem disappear. You indicated this RPM in your initial post.
With pam-kwallet installed, the journal shows:
lightdm[10869]: pam_kwallet5(lightdm:session): pam_kwallet5: final socket path: /tmp/kwallet5_fm.socket lightdm[10869]: pam_kwallet5(lightdm:session): pam_kwallet5-kwalletd: Couldn't listen in socket lightdm[10869]: pam_kwallet5(lightdm:session): pam_kwallet5: Impossible to write walletKey to walletPipe lightdm[10869]: pam_kwallet(lightdm:session): pam_kwallet: pam_sm_open_session lightdm[10870]: pam_kwallet(lightdm:session): pam_kwallet: final socket path: /tmp/kwallet_fm.socket lightdm[10870]: pam_kwallet(lightdm:session): pam_kwallet-kwalletd: Couldn't listen in socket lightdm[10870]: pam_kwallet(lightdm:session): pam_kwallet: Impossible to write walletKey to walletPipe
### Fail to create session:
lightdm[10870]: pam_systemd(lightdm:session): Failed to create session: Access denied lightdm[10870]: pam_unix(lightdm:session): session opened for user fm by (uid=1005)
### Fail to register in lastlog and btmp
lightdm[10870]: pam_lastlog(lightdm:session): unable to open /var/log/lastlog: Permission denied lightdm[10870]: pam_lastlog(lightdm:session): unable to open /var/log/btmp: Permission denied
It works with gdm since it doesn't include pam_kwallet in its pam configuration files (of course :-) ), unlike lightdm:
/etc/pam.d/lightdm:
auth substack system-auth -auth optional pam_gnome_keyring.so -auth optional pam_kwallet5.so -auth optional pam_kwallet.so .. -session optional pam_gnome_keyring.so auto_start -session optional pam_kwallet5.so -session optional pam_kwallet.so session include system-auth
I tried to put system-auth before the pam_kwallet* in the session part: nmcli works, but logout no. In this case the socket is put under /run/user/$UID that is created before, but still "Couldn't listen insocket"
I haven't tried to put the pam_kwallet* last in the session part.
sddm is perhaps subject to this bug since it includes also the pam_kwallet* modules. I haven't tested it.
It is safe to suppress the pam_kwallet RPM since it fails anyway to spawn kwalletd and no other RPM depends ont it.
Nice side effect for an optional module :-(
On Mon, May 21, 2018 at 3:36 PM, Francis.Montagnac@inria.fr wrote:
Hi.
On Thu, 17 May 2018 10:19:52 -0700 stan wrote:
On Thu, 17 May 2018 10:54:57 +0200 Gianluca Cecchi gianluca.cecchi@gmail.com wrote:
Now (I'm in F28) it is simply sufficient to re-enable gdm and connectusing Gnome as DE to see that all is ok:
- logout/shutdown buttons without problems (previously a logout
originated an automatic relogin...)
- switch wifi on/off and other components (such as external usb disks)
without getting authorization errors
- audio adapter is seen ok with its analog and hdmi settings (and it
works too... ;-)
Possibly gdm in its start enables some permission for normal users that lightdm (at least in latest updates) doesn't provide?
I don't think it's that. See below.
We are using lightdm and noticed also this problem.
So I'm not alone...
Perhaps I didn't express clear enough. With lightdm I had the problem (as you confirmed you had too) With gdm I don't have the problem at all
I found that the culprit is the RPM pam-kwallet-5.12.5-3.fc27: if you
erase it this problem disappear. You indicated this RPM in your
initial post.
I can try and verify if it is the same for my environment
In the mean time I took the occasion to switch back again to gdm as DM and gnome-session as DE I have to say that it seems more usable and less cpu consuming as in the time when I abandoned it (especially gnome-shell process that often became mad...) Also, after disabling tracker related things (see the other thread), it seems to me that all in all this configuration in F28 on my laptop (Asus U36SD) consumes less (battery lasts more) than the previous F27+lightdm+Mate DE I will continue to test and eventually switch again if not satisfied. The important is to preserve the power of choice and ability to change if needed ;-)
It is safe to suppress the pam_kwallet RPM since it fails anyway to spawn kwalletd and no other RPM depends ont it.
Nice side effect for an optional module :-(
Indeed!
Thanks for your comments and help
Gianluca
-
Francis.Montagnac＠inria.fr -
Gianluca Cecchi -
stan -
Tom Horsley