Bug 19642 - 3 rectangle shapes with question mark instead of plymouth
Summary: 3 rectangle shapes with question mark instead of plymouth
Status: REOPENED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: i586 Linux
Priority: High normal
Target Milestone: ---
Assignee: All Packagers
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-10-23 21:10 CEST by Renato Dali
Modified: 2017-08-05 21:28 CEST (History)
12 users (show)

See Also:
Source RPM: plymouth
CVE:
Status comment:


Attachments
plymouth-debug.log of 5->6 upgraded VM in vbox (103.11 KB, text/plain)
2017-05-06 17:42 CEST, Rémi Verschelde
Details
journalctl -ab output when booting Mageia-6-rc-LiveDVD-Xfce-x86_64-DVD.iso in virtualbox. (174.57 KB, application/octet-stream)
2017-06-03 21:47 CEST, PC LX
Details
copy of /var/log/plymouth-debug.log when booting Mageia-6-rc-LiveDVD-Xfce-x86_64-DVD.iso in virtualbox. (100.37 KB, text/plain)
2017-06-03 21:48 CEST, PC LX
Details

Description Renato Dali 2016-10-23 21:10:50 CEST
Description of problem:

After the resolution of bug 19592, I noticed the proceeds to a screen where just three character boxes on a dark screen would be lit in progress to signal the progress of the boot process.

This appears to happen in stead of the graphical Mageia cauldron with bubbles appearing to mark the boot progress (still to be confirmed).

After all available updates were applied, such characters appear as white blocks with question marks inside.

It looks like the replacement character:

https://en.wikipedia.org/wiki/Specials_%28Unicode_block%29#Replacement_character

... but inside a rectangle.


Version-Release number of selected component (if applicable):


How reproducible:

This was tried in another PC without success. Possibly because the graphical boot screen was activated (Plymouth?) -- and possibly not in the PC where the current bug has been observed. I will try to investigate further.

Steps to Reproduce:
1.Install Mga6sta1.
2.Install package "grub2-mageia-theme-dejavu".
3. Install updates.
4. Set up boot to grub2 with text mode.

PS: Formerly, in order to save time, I was testing things on the original DVD image (Mga sta1).

The rationale was that I was not testing the distro but one particularl release to see whether it was ready to release. After pondering over the issue, I reversed my thinking and decided to apply all updates, so as to try to make Mageia 6 work more closer to what a new sta2 release.
Comment 1 Renato Dali 2016-10-23 21:33:18 CEST
Confirmed that on the affected machine the Splash screen with the Mageia logo where bubbles are added won't appear being replaced by the dark screen with three question marks used as progress indicator.

Priority and severity adjusted to reflect the minor importance of this bug (aesthetic only).

Severity: normal => minor
Priority: Normal => Low

Comment 2 Marja van Waes 2016-10-24 12:42:51 CEST
I think barjac mentioned at some time what causes those question marks, CC'ing him.

CC: (none) => marja11, zen25000

Comment 3 Barry Jackson 2016-10-25 13:11:54 CEST
Nope wasn't me.
I see them on a real h/w i586 (old pIII intel machine) that was booting fine until the more recent kernels appeared.

It was OK with 4.6.2 IIRC, after which boot fails without noacpi but I now see the 3 X "?" in place of plymouth.
Comment 4 Renato Dali 2016-10-26 03:33:30 CEST
This mostly affects the perception of the Mageia brand, otherwise it's a microscopic bug.

I only reported because the ? undefined chars give the impression of something internally wrong -- but I don't think that is the case -- and most old guys (the ones who have old computers!) will see it means nothing.
Comment 5 Marja van Waes 2016-10-26 13:43:28 CEST
Assigning to all packagers collectively, since there is no registered maintainer for plymouth.

Source RPM: (none) => plymouth
Summary: Weird characters during boot. => 3 diamond shapes with question mark instead of plymouth
Assignee: bugsquad => pkg-bugs
CC: (none) => anssi.hannula, fundawang, luigiwalser, mageia

Renato Dali 2016-10-26 23:19:22 CEST

Summary: 3 diamond shapes with question mark instead of plymouth => 3 rectangle shapes with question mark instead of plymouth

Comment 6 Renato Dali 2016-10-26 23:21:05 CEST
Marja, the link I provided was just to show the content of the shape, which is a rectangle/square instead of a diamond.
Comment 7 Marja van Waes 2016-10-27 00:03:45 CEST
(In reply to Renato Dali from comment #6)
> Marja, the link I provided was just to show the content of the shape, which
> is a rectangle/square instead of a diamond.

It is a square or rectangle, standing on one of its corners, too, when all angles are 90 degrees.

However, read the link you gave:
"often a black diamond with a white question mark"

Compare with the comment at the bottom of
https://en.wikipedia.org/wiki/Suit_%28cards%29#/media/File:French_suits.svg "

The four French playing cards suits used primarily in the English-speaking world: spades (â ), hearts (â¥), diamonds (â¦) and clubs (â£)

But I can understand your confusion, the expensive diamonds in rings and such have a very different shape ;-)

Summary: 3 rectangle shapes with question mark instead of plymouth => 3 rectangle shapes with question mark � � � instead of plymouth

Comment 8 Renato Dali 2016-10-27 01:20:36 CEST
Nice link, I wish more themes were available when I play Yukon/Kpatience...

Well, it seems I'm unable to convey what I saw, but here's a link which comes closer:

https://www.utf8icons.com/character/65533/replacement-character

Please take a look at the square titled GNU Unifont... that's what I saw (though the depicted one has a lot more jaggies).

BUT!

A few instants ago I applied today's update (new kernel etc.) and Plymouth works again. The three chars are no longer to be seen.

So I changed the current bug to RESOLVED, WORKSFORME (in part, because I won't be able to reproduce the bug again).

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

Comment 9 Marja van Waes 2016-10-27 10:53:42 CEST
@ Renato

I agree those are squares, not diamonds.


@ Barjac

Assuming it got fixed for you, too, so closing.

Please reopen if needed.

Resolution: WORKSFORME => FIXED
Summary: 3 rectangle shapes with question mark � � � instead of plymouth => 3 rectangle shapes with question mark instead of plymouth

Comment 10 Marja van Waes 2016-10-27 10:54:07 CEST
s/squares/rectangles/
Comment 11 Renato Dali 2016-10-27 12:27:21 CEST
Sorry, Marja and Barry. I did a big confusion, because I'm testing two machines at once.

I misremembered the error as occuring in my older (circa 2004) PC, The other, probably from 2009/2010 is the one which uses the sisimedia driver (SiS 671/771 integrated video card) -- this is the one showing the three question chars.

And they're still there... so we're back to where we were and I'll reopen the bug.

This is not very significant from a technical point-of-view, but could be from a Marketing angle, because people tend to associate such question marks with something gone wrong.

Sorry for any inconvenience caused and thank you for your attention and support -- I now and then open lesser bugs with the sole intent that Mageia looks good in the release version.

Status: RESOLVED => REOPENED
Resolution: FIXED => (none)

Yann Ciret 2016-11-06 11:28:54 CET

CC: (none) => mageia

Comment 12 Barry Jackson 2017-01-18 22:08:03 CET
@marja

This was not fixed for me either and seems to be similar to 
19539
18155

An upgrade of my Acer laptop (with NVIDIA graphics) from Mageia 5 to Cauldron today has produced the same problem.

I ended up searching here hoping to find a solution. :\

Maybe these bugs should be consolidated and given higher priority as this, while cosmetic, gives a terrible impression of the distro.
Comment 13 Renato Dali 2017-01-19 03:25:47 CET
IMHO, Barry is right about the bad impression -- I just hope it happens only in a few old computers.

Giuseppe did a much better description of the bug, I'm sorry I wasn't able to find his bug, entered 20 days before the present one. Since there is a conversation here I'm not quite sure about what to do.

Regarding the present situation, without knowing anything, I hypothesize the following:

1. The interrogation chars IMHO are actually some special ASCII char which is not defined in the font Mageia uses. I booted another (can't remember now which) showing three centered white squares (I believe Ubuntu did that in previous incarnations) -- so the solution for this might be find which chars are being used (probably by perusing Plymouth's source and looking for the hide-splash code (as Giuseppe has shown).

2. Regarding Plymouth not working, I venture (again just a conjecture) it didn't like the hardware and somehow concluded it was best not to mess with things it cannot understand... falling back to text mode boxes display. In my case the hardware is weird and uses a not very compatible driver (sisimedia) -- that's what led me to such a(n) hypothesis.
Comment 14 John Choate 2017-03-15 00:39:18 CET
Well, this has been happening on my machine ever since STA1 was released last summer. My machine is a few years old, but not a dinosaur:
MSI Z97 Gaming3 motherboard
Pentium G3258 dual-core processor
16GB DDR3 1600MHz
EVGS Geforce GTX 780
2x WDC 1TB 3.5" drives (blue & black), + 1TB Seagate 3.5"

I have only seen the proper boot splash, randomly, about 5 times in total since last July.

CC: (none) => jdchoate

Comment 15 Marja van Waes 2017-05-04 17:20:30 CEST
(In reply to Renato Dali from comment #13)
> IMHO, Barry is right about the bad impression -- I just hope it happens only
> in a few old computers.

Martin Whitaker has done things to workaround and/or fix this, he probably knows best who are affected, CC'ing him.

> 
> Giuseppe did a much better description of the bug, I'm sorry I wasn't able
> to find his bug, entered 20 days before the present one. Since there is a
> conversation here I'm not quite sure about what to do.
> 
That's bug #19539, and it's assigned to the Mageia tools maintainers. It's got normal priority and severity.

I agree that severity and priority should be increased for this report, doing so now.

Severity: minor => normal
CC: (none) => mageia
Priority: Low => Normal

Comment 16 Rémi Verschelde 2017-05-06 14:41:27 CEST
I got the same issue in VirtualBox after an upgrade from Mageia 5 to Mageia 6 RC using urpmi.

Priority: Normal => High

Comment 17 Martin Whitaker 2017-05-06 15:42:56 CEST
(In reply to Rémi Verschelde from comment #16)
> I got the same issue in VirtualBox after an upgrade from Mageia 5 to Mageia
> 6 RC using urpmi.

Does it persist after a reboot?

Please add 'plymouth.debug=stream:/dev/kmsg' to the boot command line in grub2, and attach the resulting /var/log/plymouth_debug.log
Comment 18 Rémi Verschelde 2017-05-06 17:42:25 CEST
Created attachment 9268 [details]
plymouth-debug.log of 5->6 upgraded VM in vbox

(In reply to Martin Whitaker from comment #17)
> Does it persist after a reboot?

Yes.

> Please add 'plymouth.debug=stream:/dev/kmsg' to the boot command line in
> grub2, and attach the resulting /var/log/plymouth_debug.log

Here it is. In case it matters, the bootloader is grub (legacy), as it's what was configured by default for Mageia 5 and this VM is an upgrade.
Comment 19 Martin Whitaker 2017-05-07 22:29:42 CEST
(In reply to Rémi Verschelde from comment #18)
> Created attachment 9268 [details]
> plymouth-debug.log of 5->6 upgraded VM in vbox

OK, you've hit the inherent race in the new plymouth device timeout scheme - see my initial comment in bug 19890. It should be fixable in a couple of ways:

1) Add the vboxvideo driver to the initrd. In /etc/dracut.conf.d/50-mageia.conf, change

  add_drivers+=" ahci "

to

  add_drivers+=" ahci vboxvideo "

then run 'dracut -f'

2) Increase or decrease the plymouth DeviceTimeout value in /usr/share/plymouth/plymouth.defaults and run 'dracut -f'

The first is the better fix, but I don't know if we can automate it. The second is what I've done on the Live DVDs.

I'm starting to think the best solution would be to revert the changes in plymouth. They are applied in a patch (sourced from upstream), so shouldn't be hard to remove.
Comment 20 Thomas Andrews 2017-05-09 03:30:41 CEST
Seeing the same thing after a 5->6 upgrade on real hardware, Intel Core2Duo, Intel 810 or later graphics, BCM4318 wifi.

CC: (none) => andrewsfarm

Comment 21 PC LX 2017-06-03 21:00:26 CEST
When booting the most recent live RC ISOs in VirtualBox, I'm seeing three "?" instead of the plymouth image.

CC: (none) => mageia

Comment 22 Martin Whitaker 2017-06-03 21:16:43 CEST
(In reply to PC LX from comment #21)
> When booting the most recent live RC ISOs in VirtualBox, I'm seeing three
> "?" instead of the plymouth image.

Please add 'plymouth.debug=stream:/dev/kmsg' to the boot command line, and attach the resulting /var/log/plymouth_debug.log, also the output from 'journalctl -ab'.

What host system are you using for VirtualBox?
Comment 23 PC LX 2017-06-03 21:47:55 CEST
Created attachment 9390 [details]
journalctl -ab output when booting Mageia-6-rc-LiveDVD-Xfce-x86_64-DVD.iso in virtualbox.
Comment 24 PC LX 2017-06-03 21:48:59 CEST
Created attachment 9391 [details]
copy of /var/log/plymouth-debug.log when booting Mageia-6-rc-LiveDVD-Xfce-x86_64-DVD.iso in virtualbox.
Comment 25 PC LX 2017-06-03 21:52:04 CEST
Host system information:

$ uname -a
Linux marte 4.4.70-desktop-1.mga5 #1 SMP Fri May 26 13:42:22 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
$ cat /etc/release 
Mageia release 5 (Official) for x86_64
$ rpm -qa | grep -E 'kernel|virtualbox' | sort
dkms-virtualbox-5.1.22-1.mga5
kernel-desktop-4.4.70-1.mga5-1-1.mga5
kernel-desktop-devel-4.4.70-1.mga5-1-1.mga5
kernel-desktop-devel-latest-4.4.70-1.mga5
kernel-desktop-latest-4.4.70-1.mga5
kernel-firmware-20160409-1.mga5
kernel-firmware-nonfree-20160914-1.mga5.nonfree
kernel-userspace-headers-4.4.70-1.mga5
nvidia340-kernel-desktop-latest-340.96-6.mga5.nonfree
virtualbox-5.1.22-1.mga5
virtualbox-doc-5.1.22-1.mga5
Comment 26 Martin Whitaker 2017-06-03 23:32:27 CEST
OK, you've hit the same problem as Rémi (see comment 19). But this shouldn't be happening with the Live DVD because the device timeout is set to 10 seconds. The plymouth debug log confirms that, but from the journal it is actually timing out after 5 seconds.

I should be able to fix this on the Live DVDs by adding the vboxvideo driver to the initrd.
Comment 27 Mageia Robot 2017-06-04 21:29:40 CEST
commit 2e82dec27838acd14667bd46005e6f2e10bf6d7b
Author: Martin Whitaker <mageia@...>
Date:   Sun Jun 4 20:16:12 2017 +0100

    live-dracut.conf: add vboxvideo to drivers in initrd
    
    Once more to stop plymouth falling back to text mode (mga#19642).
---
 Commit Link:
   http://gitweb.mageia.org/software/build-system/draklive-config/commit/?id=2e82dec27838acd14667bd46005e6f2e10bf6d7b
Comment 28 David Walser 2017-06-04 22:08:16 CEST
I thought that we were supposed to be using modesetting instead of vboxvideo in Cauldron now...
Comment 29 Martin Whitaker 2017-06-04 22:22:39 CEST
(In reply to David Walser from comment #28)
> I thought that we were supposed to be using modesetting instead of vboxvideo
> in Cauldron now...

We use the 'modesetting' X11 driver, but that sits on top of the 'vboxvideo' drm device driver. plymouth uses the drm devices.
Comment 30 David Walser 2017-06-04 22:49:48 CEST
Thanks for the clarification.
Comment 31 Samuel Verschelde 2017-07-10 16:18:31 CEST
Can it still happen and does it deserve an entry in the Errata?
Comment 32 Rémi Verschelde 2017-07-10 16:38:09 CEST
It's cosmetic (no actual "bug" that users should be aware of), so I don't think it needs to go in the errata.
Comment 33 Rémi Verschelde 2017-07-10 16:39:18 CEST
(In reply to Rémi Verschelde from comment #32)
> It's cosmetic (no actual "bug" that users should be aware of), so I don't
> think it needs to go in the errata.

Unless we want to document a workaround for users that might be bothered by the cosmetic issue. But it might very well be fixed by comment 27, so let's see what testers say.
Comment 34 David Walser 2017-07-10 18:21:13 CEST
I just saw someone say on IRC on the weekend that they saw the issue, so it might not be fixed, but they might have been using RC ISOs, so them again, maybe it is.
Comment 35 Thomas Andrews 2017-07-11 02:46:44 CEST
I saw it after a 5>6 upgrade install using the RC, but a clean install on the same hardware didn't do it. That makes me think it can still happen, but it's much rarer now.
Comment 36 Martin Whitaker 2017-07-15 13:49:45 CEST
If you are using the nvidia340 driver (and quite likely, any other driver that doesn't support kernel mode setting), one of the things that may cause this bug is not having a video mode (e.g. vga=791) specified on the kernel command line. Bug 21248 means this will happen if you install from a Live DVD.

Note that setting the video mode for grub2 is currently broken in drakboot (bug 21246), but you can work round that by manually adding a vga=... option in the "Append" text box.
Comment 37 Florian Hubold 2017-08-04 13:40:50 CEST
(In reply to Martin Whitaker from comment #36)
> If you are using the nvidia340 driver (and quite likely, any other driver
> that doesn't support kernel mode setting), one of the things that may cause
> this bug is not having a video mode (e.g. vga=791) specified on the kernel
> command line.

Does it have to be vga=xxx ? I've upgraded from 5 > 6 using grub2 and I have the video mode in /etc/default/grub as grub2 doesn't use vga= anymore from what I understood, also video mode was fine for mga5.

Since the upgrade to mga6 grub2 only shows a text menu instead of the graphical one and also any screen shown by plymouth has those rectangles with the question marks inside ...

Will see about the plymouth debugging from comment 22 and also the video mode selection.

CC: (none) => doktor5000

Comment 38 Martin Whitaker 2017-08-04 21:42:57 CEST
(In reply to Florian Hubold from comment #37)
> (In reply to Martin Whitaker from comment #36)
> > If you are using the nvidia340 driver (and quite likely, any other driver
> > that doesn't support kernel mode setting), one of the things that may cause
> > this bug is not having a video mode (e.g. vga=791) specified on the kernel
> > command line.
> 
> Does it have to be vga=xxx ? I've upgraded from 5 > 6 using grub2 and I have
> the video mode in /etc/default/grub as grub2 doesn't use vga= anymore from
> what I understood, also video mode was fine for mga5.

Do you mean you have a video= option instead of a vga= option? I've not tried that.

My comment was based on the behaviour of my machine that uses the nvidia340 driver. Without the vga= option, the driver doesn't create a legacy framebuffer device (and of course doesn't create a dri device either), so plymouth has nothing to display a graphical splash screen on. Adding the vga= option fixed this.

I've not seen this problem with any other drivers, but all the ones I use create dri devices and support kernel mode setting, and ignore the vga= option anyway.
Comment 39 Florian Hubold 2017-08-05 19:16:07 CEST
(In reply to Martin Whitaker from comment #38)

> Do you mean you have a video= option instead of a vga= option? I've not
> tried that.

No, I don't have that at all. I've only got the following for grub2, as vga= seems to be deprecated with grub2 from what I read, e.g. https://www.gnu.org/software/grub/manual/grub.html#gfxpayload

GRUB_CMDLINE_LINUX_DEFAULT="splash quiet nomodeset"
GRUB_GFXMODE=1280x1024x32
GRUB_GFXPAYLOAD_LINUX=keep
Comment 40 Martin Whitaker 2017-08-05 21:28:23 CEST
(In reply to Florian Hubold from comment #39)
> No, I don't have that at all. I've only got the following for grub2, as vga=
> seems to be deprecated with grub2 from what I read, e.g.
> https://www.gnu.org/software/grub/manual/grub.html#gfxpayload
> 
> GRUB_CMDLINE_LINUX_DEFAULT="splash quiet nomodeset"
> GRUB_GFXMODE=1280x1024x32
> GRUB_GFXPAYLOAD_LINUX=keep

This will set the initial video resolution, but when the video driver gets loaded, it gets to choose a new resolution. At least on my hardware, all the free drivers choose the native LCD panel resolution. The nvidia340 driver appears to depend on the vga= option, and if that isn't present, selects text mode. I don't have the necessary hardware to test if the other NVIDIA proprietary drivers do the same.

Note You need to log in before you can comment on or make changes to this bug.