KDE/sddm update - all new user accounts crash at login
Mark Knecht <[email protected]> Fri, 8 Apr 2016 10:18:58 -0700
| Newsgroups | gmane.linux.gentoo.amd64 |
|---|---|
| Message-ID | <CAK2H+eejGErbn6H08_k9X5bkGvf_1r-S5cv+0RaDbkjPbrj2jA@mail.gmail.com> |
I'm traveling on vacation and had an opportunity to update my mom's laptop. Being that the news about KDE5 plasma becoming standard and needing to watch out about app updates in the future not working in KDE4 I decided to bite the bullet and get the machine completely updated while I'm here. Overall the update process went moderately well - emerge wouldn't complete until I removed two packages for blocking issues, but other than that KDE5 came up OK ad actually the new plasma stuff seems to make things look quite nice on this machine. One thing I was interested in doing was moving my dad's use to this same machine so I did the normal useradd command from the install guide. First, I don't know how it was for kdm but with sddm, with it having pictures representing each user, sddm doesn't recognize new user accounts without being restarted. Once restarted however, my dad's name was there, I try to log in, and the whole process goes south with a message that plasma has died. If I hit the restart button at that point it just does it again and doesn't offer any more options. For context this is an Asus G73JW laptop, completely stable except for one package that emerge made me add use flags and needing to downgrade to nss-3.20.1 to run chrome. The kernel is 4.1.15-gentoo-r1 and it's running nvidia-drivers-361.28. I do not see any problems in sddm.log. Looking in /var/log/messages I'm finding two segfaults that don't occur when using accounts that previously existed: Apr 8 09:50:01 Rosie cron[6140]: (root) CMD (test -x /usr/sbin/run-crons && /usr/sbin/run-crons) Apr 8 09:50:03 Rosie acpid[2307]: client 5798[0:0] has disconnected Apr 8 09:50:15 Rosie acpid[2307]: client connected from 5798[0:0] Apr 8 09:50:15 Rosie acpid[2307]: 1 client rule loaded Apr 8 09:50:23 Rosie acpid[2307]: client 5798[0:0] has disconnected Apr 8 09:50:54 Rosie kernel: kactivitymanage[5919]: segfault at 7f093deb5c50 ip 00007f0928376071 sp 00007ffe78b30568 error 4 in libQt5Sql.so.5.5.1[7f0928362000+3f000] Apr 8 09:50:56 Rosie polkitd[2423]: Unregistered Authentication Agent for unix-session:/org/freedesktop/ConsoleKit/Session8 (system bus name :1.84, object path /org/kde/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8) (disconnected from bus) Apr 8 09:50:56 Rosie kernel: nvidia-modeset: Freed GPU:0 (GPU-ae014ee7-c17d-72a3-aa35-2dcd7c3050cf) @ PCI:0000:01:00.0 Apr 8 09:50:56 Rosie kernel: traps: ck-remove-direc[6248] trap int3 ip:7fdc33ff7313 sp:7ffea9594a20 error:0 Apr 8 09:50:56 Rosie kernel: ACPI Warning: \x5c_SB_.PCI0.PEG3.GFX0._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20150410/nsarguments-95) Apr 8 09:50:56 Rosie kernel: ACPI Warning: \x5c_SB_.PCI0.PEG3.GFX0._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20150410/nsarguments-95) The kernel trap appears to be related to consolekit. The libQt5Sql reference is from qtsql-5.5.1 Rosie ~ # equery belongs /usr/lib64/libQt5Sql.so.5.5.1 * Searching for /usr/lib64/libQt5Sql.so.5.5.1 ... dev-qt/qtsql-5.5.1 (/usr/lib64/libQt5Sql.so.5.5.1) Rosie ~ # Rosie ~ # locate ck-remove-direc /usr/lib64/ConsoleKit/ck-remove-directory Rosie ~ I can Alt-Ctrl-F1 to a console and restart xdm just fine. Any attempt to log into any new account results in the same crash. After restarting xdm I can log into any account that existed before the update but no new accounts work so far. Also, not a show stopper but if I boot the machine, switch to the console and come back to sddm the screen is corrupted but works. The date & time are gone, the pulldown in the upper left no longer says plasma, and some other text is missing or corrupted. sddm on this machine is not really ready for prime time. Any ideas where to look or what to try to get new accounts working? If I need to post more info please let me know. I wanted to capture this much and then go back to looking. It is currently Friday morning. I hope to solve this before Tuesday evening when I lave. Thanks, Mark