| Summary: | qt5 stack must be checked for sse2 code in 32bit builds | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Thomas Backlund <tmb> |
| Component: | RPM Packages | Assignee: | KDE maintainers <kde> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | High | CC: | cae, mageia, mageia, marja11, mkare |
| Version: | Cauldron | ||
| Target Milestone: | Mageia 6 | ||
| Hardware: | i586 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Source RPM: | qt5 stack | CVE: | |
| Status comment: | |||
| Bug Depends on: | |||
| Bug Blocks: | 19258, 19942 | ||
| Attachments: | Plasma desktop png | ||
|
Description
Thomas Backlund
2016-11-16 16:06:38 CET
Thomas Backlund
2016-11-16 16:07:11 CET
Priority:
Normal =>
release_blocker
Samuel Verschelde
2016-12-03 22:04:42 CET
Status comment:
(none) =>
Will be done as soon as possible i think we need to add:
%ifarch %ix86
-no-sse2 \
%endif
in other qt5 packagesStatus comment:
Will be done as soon as possible =>
(none) Something in the qt5 stack IS still requiring sse2. On and up-to-date Mga6 i586 system with an Athlon XP 3200+ which Does Not support sse2, plasma wil not run. When I try to launch plasma from the DM all I get is a screen with the Mga background and a movable cursor nothing else. I can switch to another tty and restart the DM launch gnome, xfce, lxde, etc; and they all run without problem. The issue also occurred when testing the pre-test Mageia-6-sta2-LiveDVD-Plasma-i586-DVD. CC:
(none) =>
cae Using QEMU, I have confirmed Charles's deduction that sse2 is to blame. Using qemu-system-i386 -cdrom Mageia-6-sta2-LiveDVD-Plasma-i586-DVD/Mageia-6-sta2-LiveDVD-Plasma-i586-DVD.iso -m 4G --enable-kvm -vga std -cpu qemu32 and forcing xdriver=modesetting, I can boot to a working desktop. Changing the -cpu option to "qemu32,-sse2" reproduces the behaviour Charles reports (eventually the Mageia Welcome screen gets displayed, but no desktop). CC:
(none) =>
mageia Assigned to David Geiger as decided during the KDE Maintainer Group meeting. Assignee:
kde =>
geiger.david68210 Because it is built using the qt5 stack this bug applies also to sddm. It will not work on a system whose CPU does not support see2.
Charles Edwards
2016-12-14 07:21:48 CET
Blocks:
(none) =>
19942
Thomas Backlund
2016-12-17 22:16:01 CET
Blocks:
(none) =>
19258 David, Nicolas, any progress, any ETA? Sorry I don't know how to fix that properly, so if someone else can do this job it would be great.
Samuel Verschelde
2017-01-11 18:01:43 CET
Assignee:
geiger.david68210 =>
kde i don't think we have any sse2 enabled now ( please test new rpms ). The issue still very much exists. I just tested on a drive with a new Mga6 install and tested again on this established and fully up-to-date Mga6 installation. Both installations are on seperate drives in the same system. [charles@XP3200 ~]$ lscpu Architecture: i686 CPU op-mode(s): 32-bit Byte Order: Little Endian CPU(s): 1 On-line CPU(s) list: 0 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 1 Vendor ID: AuthenticAMD CPU family: 6 Model: 10 Model name: AMD Athlon(tm) XP 3200+ Stepping: 0 CPU MHz: 1837.000 CPU max MHz: 1837.0000 CPU min MHz: 1287.0000 BogoMIPS: 3674.02 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow eagerfpu 3dnowprefetch vmmcall 1st Big red flag....if I try to use sddm as DM all I get is is an empty black screen. When I use lxdm I can select plasma and login....but then all I get is the default Mga6 background screen with a movable mouse cursor and nothing else. If I continue to let it run a small black square eventually appears in the top left hand corner of the Mga6 background. I can run gnome, xfce4, lxde, at el on this system, though gnome is sluggish so I primarily use xfce4 or lxde. Btw, the 32bit and noarch packages provided by qtdeclarative5-5.6.2-2.mga6.srpm, that was pushed by neoclust to fix this issue, are:
libqt5qml-devel-5.6.2-2.mga6.i586
libqt5qml5-5.6.2-2.mga6.i586
libqt5quick-devel-5.6.2-2.mga6.i586
libqt5quick5-5.6.2-2.mga6.i586
libqt5quickparticles-devel-5.6.2-2.mga6.i586
libqt5quickparticles5-5.6.2-2.mga6.i586
libqt5quicktest-devel-5.6.2-2.mga6.i586
libqt5quicktest5-5.6.2-2.mga6.i586
libqt5quickwidgets-devel-5.6.2-2.mga6.i586
libqt5quickwidgets5-5.6.2-2.mga6.i586
qtdeclarative5-5.6.2-2.mga6.i586
qtdeclarative5-doc-5.6.2-2.mga6.noarch
@ Charles
You knew that and you made sure you have version 5.6.2-2.mga6 for each of them that's installed, didn't you?CC:
(none) =>
marja11 (In reply to Marja van Waes from comment #10) > Btw, the 32bit and noarch packages provided by > qtdeclarative5-5.6.2-2.mga6.srpm, that was pushed by neoclust to fix this > issue, are: > > > libqt5qml-devel-5.6.2-2.mga6.i586 > libqt5qml5-5.6.2-2.mga6.i586 > libqt5quick-devel-5.6.2-2.mga6.i586 > libqt5quick5-5.6.2-2.mga6.i586 > libqt5quickparticles-devel-5.6.2-2.mga6.i586 > libqt5quickparticles5-5.6.2-2.mga6.i586 > libqt5quicktest-devel-5.6.2-2.mga6.i586 > libqt5quicktest5-5.6.2-2.mga6.i586 > libqt5quickwidgets-devel-5.6.2-2.mga6.i586 > libqt5quickwidgets5-5.6.2-2.mga6.i586 > qtdeclarative5-5.6.2-2.mga6.i586 > qtdeclarative5-doc-5.6.2-2.mga6.noarch > > @ Charles > > You knew that and you made sure you have version 5.6.2-2.mga6 for each of > them that's installed, didn't you? I made sure both the installation were fully up-to-date [charles@XP3200 ~]$ rpm -qa |grep libqt5quick libqt5quickparticles5-5.6.2-2.mga6 libqt5quick5-5.6.2-2.mga6 libqt5quickwidgets5-5.6.2-2.mga6 libqt5quicktest5-5.6.2-2.mga6 [charles@XP3200 ~]$ rpm -qa |grep qtdeclarative5 qtdeclarative5-5.6.2-2.mga6 [charles@XP3200 ~]$ rpm -qa |grep libqt5qml libqt5qml5-5.6.2-2.mga6 how can i test this on VBox ? IIRC something like this to modufy the vm you are testing with: VBoxManage modifyvm <vm_name> --cpuidremove sse2 does not work, i obtain VBoxManage: error: asking on VBox irc channel, they told me this is not possible to do this kind of actions :/ Try using qemu, as described in comment 3. "-cpu qemu32,-sse2" emulates a CPU *without* SSE2 instructions. just tried and in lscpu i don't have see2. Plasma starts, i can open the menu, dolphin, etc. # LC_ALL=C lscpu Architecture: i686 CPU op-mode(s): 32-bit Byte Order: Little Endian CPU(s): 1 On-line CPU(s) list: 0 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 5 Model name: Pentium II (Deschutes) Stepping: 2 CPU MHz: 1995.379 BogoMIPS: 3990.75 Hypervisor vendor: KVM Virtualization type: full Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pse36 mmx fxsr eagerfpu x2apic hypervisor I don't know what to tell you. I did a new clean install on a brand new drive on my test system this afternoon and still have the same issues. Neither sddm or plasma will run|work on that system but other tested DMs and WMs will work. I'm not, but if you are convinced that there is no SSE2 issue you are welcome to close this bug as "Works for me" We can look at re-opening should other users report similar problems. I have not forgotten about this bug. I just booted the pre-testing Mageia-6-sta2-LiveDVD-GNOME-x86_64-DVD.iso on my Athlon XP3200+ to see if the outcome is the same. Am attaching a picture taken from my phone of my plasma desktop. How does it look to you? Created attachment 8942 [details]
Plasma desktop png
I've just tested this by building a 32-bit Live Plasma DVD and running in qemu with sse2 disabled. Everything works for me. Maybe what Charles is seeing is a video driver problem. @Charles, is this the system which had problems when I switched the GNOME default from Wayland to Xorg? (In reply to Martin Whitaker from comment #22) > I've just tested this by building a 32-bit Live Plasma DVD and running in > qemu with sse2 disabled. Everything works for me. Maybe what Charles is > seeing is a video driver problem. > > @Charles, is this the system which had problems when I switched the GNOME > default from Wayland to Xorg? Yes it is. But as it applies to plasma I've tried with both nvidia cards, nouveau and nvidia304 driver, and with ati cards, ati driver; and none could run plasma or sddm. If no one else sees or can reproduce this problem I would have no issue with the bug being closed as "Works for me". i don't pass to find any "non see2" machine here :( i installed on really old machines ( like P4 2.8Ghz ) and it has sse2.
Nicolas Lécureuil
2017-03-22 14:02:30 CET
Priority:
release_blocker =>
High Please try new qtdeclarative5 to be tested: src.rpm: qtdeclarative5-5.6.2-5.mga6 media: core/release some issues to build it, i come back when ready Please test new qtdeclarative5 src.rpm: qtdeclarative5-5.6.2-8.mga6 media: core/release Should Be OK, please reopen if any pb Status:
ASSIGNED =>
RESOLVED |