Description of problem:
Installed F23 i686 KDE live, updated to latest F23, dnf system-upgrade'd to F24 with updates-testing enabled, booted. sddm shows only background and environment selector: https://www.happyassassin.net/temp/sddm24.png . This crash is apparent in system logs.
Version-Release number of selected component:
cmdline: /usr/bin/sddm-greeter --socket /tmp/sddm-:0-okCmbm --theme /usr/share/sddm/themes/01-breeze-fedora
pkg_fingerprint: 73BD E983 81B4 6521
pkg_vendor: Fedora Project
runlevel: N 5
Thread no. 2 (1 frames)
Created attachment 1149840 [details]
Created attachment 1149841 [details]
Created attachment 1149842 [details]
Created attachment 1149843 [details]
Created attachment 1149844 [details]
Created attachment 1149845 [details]
Created attachment 1149846 [details]
Created attachment 1149847 [details]
Created attachment 1149848 [details]
Created attachment 1149849 [details]
Created attachment 1149850 [details]
Created attachment 1149851 [details]
Created attachment 1149852 [details]
Created attachment 1149853 [details]
*** Bug 1329709 has been marked as a duplicate of this bug. ***
Note this may not relate to upgrade at all, it could just be a case of 'this always happens on i686'. We haven't had working F24 i686 images for months due to https://bugzilla.redhat.com/show_bug.cgi?id=1302071 , so we have not reached this point in testing. Tomorrow we should have a working i686 kernel, so we should be able to see if this happens with a clean F24 i686 KDE.
I'm gonna propose this as a Beta freeze exception for now, on the basis it may well happen on fresh installs. Not a blocker so long as it's i686 only, i686 images no longer block the release by policy.
Thread 1 (Thread 0xb36a9800 (LWP 1110)):
#0 0x00000000 in ?? ()
No symbol table info available.
#1 0xb6ff867c in drainMarkStack (markBase=<optimized out>, engine=<optimized out>) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/jsruntime/qv4persistent.cpp:220
#2 QV4::PersistentValueStorage::mark (this=0x81a28f98, e=0x81b23fb0) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/jsruntime/qv4persistent.cpp:234
markBase = <optimized out>
p = 0xab405000
#3 0xb6ee1ea4 in QV4::MemoryManager::mark (this=this@entry=0x81b24070) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/memory/qv4mm.cpp:384
markBase = 0xae201410
#4 0xb6ee2dc0 in QV4::MemoryManager::runGC (this=this@entry=0x81b24070) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/memory/qv4mm.cpp:546
#5 0xb6ee3f4c in QV4::MemoryManager::allocData (this=this@entry=0x81b24070, size=size@entry=32, unmanagedSize=16) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/memory/qv4mm.cpp:267
didGCRun = false
pos = <optimized out>
m = <optimized out>
header = <optimized out>
#6 0xb6fe7a4d in QV4::MemoryManager::allocManaged<QV4::String> (unmanagedSize=<optimized out>, size=<optimized out>, this=0x81b24070) at /usr/src/debug/qtdeclarative-opensource-src-5.6.0/src/qml/memory/qv4mm_p.h:102
o = <optimized out>
Discussed at 2016-04-25 blocker review meeting: .
This bug was accepted as Freeze Exception: if a targeted and tested fix for this is available soon we will consider including it (as this would be an upgrade criterion violation if i686 still blocked releases), but will be careful not to break x86_64
Confirming that this happens on a clean install also, thanks openQA:
note particularly the login screen appearance - https://openqa.fedoraproject.org/tests/14980/modules/_graphical_wait_login/steps/7 - exactly the same as I saw on the upgraded system. So it's clearly the same bug.
Also note that 32-bit KDE lives hang at a black screen when the desktop should appear:
I'm not sure if it's the same bug.
The backtrace matches my findings from bug #1331593, but this one at least has a better backtrace (debuginfo)
OK, after a bit of discussion, kde-sig highly suspected this is a qt5-qtdeclarative issue, most likely triggered by gcc6.
So, installed f24/i686 workstation, and installed @kde-desktop-environment on top of that to do additional testing, and found that building qt5-qtdeclarative with addtional flag: -fno-lifetime-dse , seemed to fix the crashes for me.
Here's a build that includes that for testing purposes:
*** Bug 1331593 has been marked as a duplicate of this bug. ***
After installing qt5-qtdeclarative-5.6.0-10.fc25.i686.rpm sddm looks as it should and plasma-desktop starts.
qt5-qtdeclarative-5.6.0-10.fc24 has been submitted as an update to Fedora 24. https://bodhi.fedoraproject.org/updates/FEDORA-2016-452ca0119f
qt5-qtdeclarative-5.6.0-10.fc24 has been pushed to the Fedora 24 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2016-452ca0119f
qt5-qtdeclarative-5.6.0-10.fc24 has been pushed to the Fedora 24 stable repository. If problems still persist, please make note of it in this bug report.