Created attachment 13228 [details]
Screenshot of an affected MCC window

Thank you TJ for the report, and the screenshot which says it all.
This is curious as being something that has appeared relatively recently.
The fact that it happens on different systems clearly matters: it is not down to a specific hardware platform.

Assigning to the mageiatools people, CC'ing tmb.

Assignee: bugsquad => mageiatools
CC: (none) => tmb

Also valid with the 5.17.4 desktop and desktop586 kernels, on a separate install on the same Dell hardware.

The issue appears to be video-related. Switching from the radeon driver to vesa allows drakconf to come up and act normally on the Inspiron, but it also takes away some of the eye candy, including the photo I was using as wallpaper.

I described the video as being radeon RV200. It is more formally a Mobility Radeon 7500. We have had problems with this gpu before, because of its limited capabilities. 

As for the VirtualBox guests, pursuing the video-related idea, I first tried changing graphics controllers from VMSVGA, but that made no difference - except to tell me they were all invalid settings. But, when I enabled 3d acceleration, all of the guests started acting normally.

This is a change, in both cases. These systems, both real and virtual, worked just fine a few weeks ago. The settings weren't changed, but somewhere along the line, something basic to all systems is now using higher graphics capabilities than it did before.

Perhaps try downgrading webkit2 from bug 30320 to see if that restores the
functionality.

CC: (none) => davidwhodgins

That did it. It seems that the update of webkit2 from 2.34.x to 2.36.x (Bug 30262) is what "broke" drakconf on my systems. Using this command restores function in the Inspiron:

urpmi --downgrade libjavascriptcore-gir4.0-2.34.6-1.mga8 libjavascriptcoregtk4.0_18-2.34.6-1.mga8 libwebkit2gtk4.0_37-2.34.6-1.mga8 libwebkit2gtk-gir4.0-2.34.6-1.mga8 webkit2-2.34.6-1.mga8 webkit2-jsc-2.34.6-1.mga8

The same command with "lib64" instead of "lib" makes 64-bit vbox guests act normally again, even when 3d acceleration isn't enabled.

Summary: drakconf comes up with a blank window on i586 real hardware and in VirtualBox guests of both arches => After webkit2 update, drakconf comes up with a blank window on i586 real hardware and in VirtualBox guests of both arches if 3d acceleration is not enabled

Confirming in mga8-64 in VirtualBox guest.

I wrote in Bug 30302#c14 that i did not see it in my virtualbox guest Mageia, but i see now it was not fully updated. Now updating it with all released updates, including webkit, MCC is empty just like TJ's screenshot.

But just clicking in MCC makes it display the content and it works.
After closing MCC, launching it again, it displays correctly.

Guest is installer all per defaults i think, except i choose LXDE desktop.
I see in drakx11 it is using "WMware virtual video card"  (Not "VirtualBox video card", as one would assume...?)  Composite mode activated.


The host is my workstation, using nvidia-current for its GPU.
VirtualBox settings for this guest: Controller: VMSVGA, 3D acceleration not enabled.

And yes when I enable that 3D acceleration, content in MCC is drawn correctly when launched.

CC: (none) => fri

Confirming downgrading below 2.36 is the workaround.

Tests on my Dell Precision M6300:
mga8-64, Plasma, nvidia driver "Geforce 8100 to GeForce 415"

It had not been updated for a while, was using webkit2-2.36.0-1.mga8.x86_64
Same symptoms: "empty" MCC until clicked. Restarting MCC, it shows OK.

I found that just suspending and waking it up, with MCC closed, make it show the problem again when launched.

The downgrade command in comment 5, modified for 64 bit, solves the issue.


(In reply to Thomas Andrews from comment #0)
> Dell Inspiron 5100, 32-bit P4, Radeon RV200 graphics, mga8-32 Xfce system
...
> There is no response to mouse clicks within the window, except to close it.

So we have two levels of problems, that behaviour on Inspiron 5100 is severe.

I suspect the severity of the symptoms would vary according to the capabilities of the video driver/hardware. 

For example, nvidia users with hardware that is no longer supported by anything but nouveau might be affected more severely than your Dell. And it could be that real 32-bit hardware is all affected more severely.

Unfortunately, our pool of potential test hardware for that theory is limited, to say the least.

I have 4 PC at home all fully updated.
- main PC: AMD Ryzen 5 3600 6-Core, ‎NVIDIA ‎GP108 [GeForce GT 1030]
- PC for tests: Intel I5-7600,      ‎NVIDIA ‎GP108 [GeForce GT 1030]
- old laptop Intel pentium 2020M,   ‎NVIDIA ‎GK208M[GeForce GT 740M]
- wife's PC: Pentium Gold G54500,   Intel Coffee Lake-S GT1[UHD Graphics 610]

The problem occurs only with the kernel 5.15.35-desktop-2.mga8 on the first and second PC. MCC has an empty window (see attachment 13228 [details]) but harddrake2, rpmdrake and other work if started from a konsole.

When MCC is started from a root konsole, this message is issued:
«
(WebKitWebProcess:213104): Gdk-ERROR **: 10:43:46.042: The program 'WebKitWebProcess' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadValue (integer parameter out of range for operation)'.
  (Details: serial 201 error_code 2 request_code 151 (GLX) minor_code 34)
  (Note to programmers: normally, X errors are reported asynchronously;
   that is, you will receive the error a while after causing it.
   To debug your program, run it with the GDK_SYNCHRONIZE environment
   variable to change this behavior. You can then get a meaningful
   backtrace from your debugger if you break on the gdk_x_error() function.)
»
When started with the previous kernel (5.15.32-desktop-1.mga8), the nvidia module is automatically recompiled and MCC works fine.

I hope this can help to solve this bug which seems linked to the kernel.

CC: (none) => jarillon

(In reply to Pierre Jarillon from comment #9)
> I have 4 PC at home all fully updated.
> - main PC: AMD Ryzen 5 3600 6-Core, ‎NVIDIA ‎GP108 [GeForce GT 1030]
> - PC for tests: Intel I5-7600,      ‎NVIDIA ‎GP108 [GeForce GT 1030]
> - old laptop Intel pentium 2020M,   ‎NVIDIA ‎GK208M[GeForce GT 740M]
> - wife's PC: Pentium Gold G54500,   Intel Coffee Lake-S GT1[UHD Graphics 610]
> 
> The problem occurs only with the kernel 5.15.35-desktop-2.mga8 on the first
> and second PC. MCC has an empty window (see attachment 13228 [details]) but
> harddrake2, rpmdrake and other work if started from a konsole.
> 

I guess you got caught by updating both kernel and nVidia driver....

If you  boot into 5.15.35, recreate the initrd with "dracut -f", and reboot, does it work better then ?

On the PC #1, I have just do dracut -f and reboot. Now MCC works fine.

On the PC #2, I reboot on the last kernel 5.15.35 (no nvidia recompilation) and MCC works as usual.

Continuing from Comment 6 - that machine have not been touched since then.
In VB settings I disabled 3Dacceleration, booted, confirmed problem.
$ uname -a
Linux localhost 5.15.35-desktop-2.mga8 #1 SMP Fri Apr 22 20:40:06 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
$ sudo dracut -f
...OK
reboot
-> same problem still.

And, I saw the problem on all of my affected hardware/VMs before updating the kernel to 5.15.35 or 5.17.4. I had hoped that updating the kernel and/or VirtualBox would help, but there was no effect.

Seems that the same/similar problem was already reported 2017 with bug 20193

No change with the update to webkit2-2.36.2-1.

*** Bug 20193 has been marked as a duplicate of this bug. ***

CC: (none) => yves.brungard_mageia

Because downgrading webkit2 makes the issue go away, I have been thinking that a bug in webkit2 is probably at fault. But after reading the duplicate bug, I'm wondering...

What if the bug has been in drakconf all along, and a bug in webkit2 hid it for a while? And if that webkit2 bug has now been fixed, the old drakconf bug is exposed once more?

Other people have been seeing this issue. It's been brought up in the QA mailing list. Raising the severity level, as for many of those affected it makes drakconf unusable.

Severity: normal => major

Seeing this in Mageia Cauldron. 

Attempted to take screenshot of the MCC main window and the missing text appeared exactly as the screenshot was taken, so some of the text is not in the screenshot.

CC: (none) => epp

Created attachment 13329 [details]
Magreia Cauldron MCC main screen with (what was) missing text until the screenshot was taken.

The problem reappeared with the last update of webkit2-2.36.6

If I run drakconf (aka Mageia Control Center), whether from the panel icon or using the command line, the window comes up normally but the inside is blank.

Real hardware intel gpa

CC: (none) => philippedidier

(In reply to Philippe Didier from comment #20)
> The problem reappeared with the last update of webkit2-2.36.6
> 
> If I run drakconf (aka Mageia Control Center), whether from the panel icon
> or using the command line, the window comes up normally but the inside is
> blank.
> 
> Real hardware intel gpa

I saw the same yesterday, only at first launch. When I commute to other tabs, it's OK, even for the first one. The second launch is OK.

It never went away on my 32-bit hardware. 

I suspect more than ever that the root problem is in drakconf, not in webkit2, but not being a developer I have no way of confirming or eliminating that suspicion.

Hi Thomas and Papoteur

I use 2 laptops and a tower computer
Everything was ok on the three of them before updating webkit2 yesterday :
I used Mgaapplet to update one of the laptops (but didn't yet update the two others computers)

The only difference between those installations is webkit2-2.36.6 and what depends on it on the updated one and webkit2-2.36.4 on the two others
The blank window appears only on the updated

NB from a console I can still launch the components of the CCM on the uodated computer
but the CCM window remains blank for each of them :  no icon whatever tab I choose in the vertical tray
  
As noone use the CCM everyday this bug may be hidden for some time ...

As of today, this problem is still affecting my 32-bit hardware, through several updates of webkit2. 

However, the behavior has changed, somewhat. Before, the empty window was completely unresponsive. Now, there is still a blank screen in drakconf, but if you look carefully when moving the mouse cursor in the drakconf window, you can see the cursor change as you pass over the areas where you would click to bring up the various tools. If you click on one of those areas, the tool will come up.

I suppose that's progress.

Whatever the root cause of this, also as of today, Cauldron is not affected on this 32-bit hardware. MCC comes up normally there. 

That gives me hope...

Please try installing task-gnome-minimal (don't switch to running it). And see
if that makes a difference.

Used urpmi to install it, then immediately tried to run MCC. No change. Then I rebooted into Xfce, just because, and tried MCC again, again with no change.

Well, no change to MCC. There IS a change. I now have two net_applet icons in the panel, one for wifi that says it's working, and the other that just says "Networking disabled." The wifi IS working.

(In reply to Thomas Andrews from comment #0)
> Description of problem:
> Dell Inspiron 5100, 32-bit P4, Radeon RV200 graphics, mga8-32 Xfce system
> using the desktop kernel.

Same GPU, and mga8-32 XFCE here, a Thinkpad T40 laptop;

Confirming exactly;
Was working in fresh install, now after full update it is per this bug.

Details:
When MCC window comes up list at left is OK but the main space is blank.
When moving the mouse pointer over where the drakrpm icon should be, pointer becomes an arrow but cklicking do not work now.

But: after first having clicked in the left pane, i can click item in bind when mouse pointer is an arrow.

Sidenote: Here, net_applet is OK.

This bug have been in mga8 for most of its life now, and users who upgrade to or fresh install and update as they should may hit this, so it should be in errata.

Keywords: (none) => FOR_ERRATA8

Meanwhile, my affected 32-bit hardware continues to work OK in Cauldron, even after a drakconf/drakxtools update earlier today.

Great then :)
I may test mga9 Live XFCE on my affected laptops when beta2 ISO releases.

For mga8: https://wiki.mageia.org/en/Mageia_8_Errata#Mageia_Control_Center

Keywords: FOR_ERRATA8 => IN_ERRATA8

Another setup experiencing this empty MCC:
GPU nvidia GTX750 (GM107) with default nvidia proprietary driver.

Workaround here is using nouveau

(Another problem with the nvidia driver is that acceleration seem to not work:  it is 2,5 times slower running teapot and eats CPU - I may try more settings and another driver later)

This and comment before is on mga9.

(In reply to Morgan Leijström from comment #30)
> Another setup experiencing this empty MCC:
> GPU nvidia GTX750 (GM107) with default nvidia proprietary driver.

(nvidia-current)

> Workaround here is using nouveau

...which is also faster, but fail suspend-resume.

Better: The driver modesetting is 50 % faster, do not have this bug, and suspend-resume works. 

But what works best for GTX750 is nvidia470 driver, in drakx11 listed as "Geforce 635 to 920": fastest, do not show this bug, and resume works from both suspend and hibernation.

Whiteboard: (none) => MGA8TOO
Version: 8 => Cauldron
Keywords: (none) => FOR_ERRATA9

Comment for comment 31: I have not checked elder webkit on this setup.  It is however most plausible the cause is same as other systems showing empty MCC, and same workaround works: select a more suitable GPU driver which also works better in other aspects anyway...

Weird, but may give a clue:

My new workstation build with old stuff and running Mageia 9:
GPU: nvidia GTX750 GPU with 1GB VRAM, nvidia 470 driver, 4k screen.

That VRAM is a bit small for this screen size.
After starting a normal bunch of apps, VRAM is up to 100% used.
(as seen by the nvidia X server settings app)
Some window moving etc may get sluggish, but no application misbehaves

 - except MCC, showing up empty as described in this bug.

If I then i.e close Firefox, MCC launch OK.

I dont know if relevant, but this is with the manually added line
    Option "TripleBuffer" "true"
added in Section "Screen" in /etc/xorg.conf
(done in order to see if it helped occasional distortion when resuming from suspend, which it did not, but it improved speed for teapot ~20%)



I did not ever see this with same graphics card when used for same tasks and with same screen, also used nvidia470, in Mageia 8.

Do someone still see this?

I have not noticed this for long on my main workstation system "svarten", neither with nvidia proprietary nor modesetting.  GTX 750Ti.
Now i have no line with "TripleBuffer" in /etc/X11/xorg.conf

This seem related:
Bug 32185 - webkit2: drakconf (mcc) over ssh fails to display its menu options on the right hand side.

---

Another quirk i have seen (latest yesterday) recently is that on first launch of MCC, the text of categories in left pane is black instead of white, whenever i select something all text there go white OK. After that it works per design i think:  When mouse hoover over a text it is black, other wise white.

See Also: (none) => https://bugs.mageia.org/show_bug.cgi?id=32185

I'm still seeing it in Mageia 8 on Foolishness, the hardware from Comment 0. 

I don't see it in Mageia 9/Cauldron, and haven't seen it in Cauldron through many updates of webkit2 and of drakconf.

I have become resigned that the only "cure" for this bug on that hardware is to upgrade to Mageia 9. Once Mageia 8 is EOL, and that install is no longer needed for testing purposes, that is exactly what I will do.

Does this work OK on Foolishness with mga9 Live?

(In reply to Morgan Leijström from comment #36)
> Does this work OK on Foolishness with mga9 Live?

I tried starting it from Mageia Welcome, the panel icon, and the command line, all in Live mode. It worked every time.

*** Bug 32194 has been marked as a duplicate of this bug. ***

CC: (none) => pejamoba

A new bug 32185 about drakconf missing display elements after a webkit2 update is similar: >=2.41.6. This bug is about >=2.36. I do not think they are duplicates, just the same theme.

*** Bug 32212 has been marked as a duplicate of this bug. ***

*** Bug 32240 has been marked as a duplicate of this bug. ***

CC: (none) => jeff

Now on my workstation I saw this.
Same system as in Comment 34, updated Mageia 9, nvidia-current.

Nothing to note in journal.
Closed mcc, opened again: OK
Go figure...

I have also seen users hit this on mga9.
Rising priority.

Priority: Normal => High
Whiteboard: MGA8TOO => MGA8TOO, MGA9TOO

https://wiki.mageia.org/en/Mageia_9_Errata#Mageia_tools

Keywords: FOR_ERRATA9 => IN_ERRATA9

Oh, for mga9 it is Bug 32185

Keywords: IN_ERRATA9 => (none)
Whiteboard: MGA8TOO, MGA9TOO => MGA8TOO

I also have the problem with a HP S7310n SFF (32 bit version of mageia).
Not only are the actions blank, but I can't click on where they should be located.

This is with the onboard video, which I think comes from the intel i915gv northbridge or the ich6 southbridge.

@Morgan Leijström Yes, nVidia GTX750Ti system newly upgraded to Mageia 9.  MCC is blank as reported in the original problem report.

CC: (none) => kbulgrien

I upgraded to M9 from M8 a while back. Went very well i thought!

This bug Just happened here. MCC in Plasma failed. White catagories to the left, blank window to the right, but mouse changes over where the icons should be, and clicking opens the applet. 

I have Nvidia GeForce GTX970 Graphics (Last NV card with true VGA socket?)

Had to go single user as root : CTRL +F2, Log in as Root, "INIT 3" then run MCC from commandline, install the "LATEST LEGACY..." Nvidia driver instead of whatever else NV driver was installed and it fixed it.

Switching to Different Kernels made no difference, the various rollbacks of stuff with URPMI were not possible/did not work, but the switch of driver type for nVidea graphics did.

Have run benchmarks and tested the 3d and it's all screaming fast, so, fix is a fix.

CC: (none) => playthatbeat

On my Dell laptop with fully updated Cauldron I fixed it by changing theme from Breeze Dark to Breeze Twilight.

Intel Core i7-3720QM, 8 GB RAM, Intel® HD Graphics 4000 i915 driver

(In reply to Morgan Leijström from comment #27)

> Same GPU, and mga8-32 XFCE here, a Thinkpad T40 laptop;
> 
> Confirming exactly;
> Was working in fresh install, now after full update it is per this bug.

...and now after upgrading to mga9 it shows OK.

Please note that in mga9 (not Cauldron), on this date and when fully updated, when using a GTX 750 Ti, this problem occurs when the NVidia New Feature driver is used (driver 545.*) and goes away when the NVidia Production driver (x11-driver-video-nvidia-current-550.54.14-2.mga9.nonfree).  In this instance, there is no need to down-grade webkit.

Please test drakconf-13.29-1.1.mga9 from core/updates_testing when it reaches the mirrors. This contains some wqrk-arounds that fix the issue for some people.

CC: (none) => mageia

Confirm as fixed on my mga9 install after installing drakconf-13.29-1.1.mga9.

I reported this bug for Mageia 8 on Foolishness, my Dell Inspiron 5100, but installing Mageia 9 made it disappear on that hardware, including after testing an update of webkit2 just a few minutes ago.

Since I can no longer reproduce the issue, of course I can't say if Martin's drakconf fixes it. But, I can say that after updating Foolishness to Martin's drakconf, it still works.

Bug cleaning....

This bug is set for mga8, it have just been sliding along...

For mga9 we have Bug 32185.  (where i reported drakconf update works for me)

Probably same:

The problem drakconf show seem to be triggered not only by webkit, but also depend on hardware, drivers, many things...


As this bug is set for mga8 we could close as wontfix.

Or we set it to mga9 if it is another issue than Bug 32185.
Or make them duplicates?

We also need one bug set to drakconf to handle the update with advisory and file list.

I think we can close this bug. Most of the affected systems now have Nvidia graphics, and the original was old Radeon, so I believe they may have different "fixes." The original system has been fixed, or if not exactly fixed no longer shows the symptoms. 

BTW, we need to get the workaround out, ASAP. My newest system, using nvidia-current 550.76, was affected this morning after getting the latest webkit2 update, so more user systems are bound to be affected now. The workaround made the symptoms go away.

Will not be fixed in mga8.
-but please try the manual workaround if yo use mga8:
See Bug 32185 Comment 23


For mga9 follow on in Bug 32185

Resolution: (none) => OLD
Status: NEW => RESOLVED

commit 6e1902789812a5b8fb9d3a2e4d5f3dba1d99b2bd
Author: Martin Whitaker <mageia@...>
Date:   Sat Sep 21 20:30:40 2024 +0100

    Work around WebKit2 bugs that stop right hand pane icons being displayed.
    
    Since WebKit2 2.36, the icons in the right hand pane are not displayed on
    numerous (but not all) combinations of graphics hardware and drivers. This
    seems to be due to bugs in the WebKit2 accelerated compositing mode. The
    workaround is to disable WebKit2 accelerated compositing.
    
    This workaround was already released as a temporary fix for mga9, but still
    appears to be needed. See mga#30332, mga#32185, and mga#33573.
---
 Commit Link:
   https://gitweb.mageia.org/software/control-center/commit/?id=6e1902789812a5b8fb9d3a2e4d5f3dba1d99b2bd

 Bug links:
   Mageia
      https://bugs.mageia.org/30332
      https://bugs.mageia.org/32185
      https://bugs.mageia.org/33573