Created attachment 5760 [details]
testing h/w

CC: (none) => westel

Mageia-5-beta2-i586-DVD, Sat Dec 20 22:13:39 CET 2014

same on my hardware , see comment #1 for hardware

(In reply to ben mcmonagle from comment #2)
> Mageia-5-beta2-i586-DVD, Sat Dec 20 22:13:39 CET 2014
> 
> same on my hardware , see comment #1 for hardware

@ Ben

That's an Intel 810 and later card, too, but you don't seem to have a ThinkPad, correct? (I was wondering whether it was ThinkPad-related)

Btw, I've been having hard crashes (not even alt+sysrq keycombinations work any more, there is nothing in the logs) can they be related?

They did at least happen several times when moving the mouse and clicking (or about to click) on something. It maybe even be that I was always doing that when the system froze. It did at least happen in FF and TB, but also outside those applications.
 
This never happened with 3.17* kernels.

I have this problem on 2 desktop systems, both are i5 quad core units, 
both with asus M/b, 8G, using onboard video.
the older unit is my home one, the undrerun text only occurs when booting on the Mga5 32+64  systems, 
Mga4 32+64 does not display the underrun, there appear to be no operation issues during normal use, except that dbus startup problem.

my other newer unit is Mga4-64 only, no operation issues with it at all.

(In reply to Marja van Waes from comment #4)
> Btw, I've been having hard crashes (not even alt+sysrq keycombinations work
> any more, there is nothing in the logs) can they be related?
> 

Happened again after I had started "mgasync" in a konsole (that does nothing but sync my local mirror on a USB disk with a public Mageia mirror), and walked away to do some other things.

This time, I see a screen full of text, ending with:
Kernel panic - not syncing: Fatal exception in interrupt
Kernel Offset: (can't read all that follows)
drm_kms_helper: panic occurred, switching back to text console

Well, drm does sound related to drm, to me ;-)

Again not even Alt+SysRq keycombinations didn't work and there is nothing in the logs at all. 

Sorry for the bad screenshot, I won't make it worse by compressing it to be able to attach it:

http://waesvanm.home.xs4all.nl/Mageia/screenshots/Bug_14869_20141228_184854.jpg

same error messages still there, don't know yet about the crashes.

Source RPM: kernel-desktop-3.18.1-2.mga5 => kernel-desktop-3.18.1-3.mga5

Same issue here, it's been there for a while. It doesn't seem to have much impact on the boot, but it breaks plymouth for some seconds and that's not pretty to have such a huge "ERROR" on a black screen when booting :-)

I get:
[    2.361951] [drm:cpt_set_fifo_underrun_reporting] *ERROR* uncleared pch fifo underrun on pch transcoder A
[    2.361953] [drm:cpt_serr_int_handler] *ERROR* PCH transcoder A FIFO underrun

(see attached dmesg | grep drm for context)


Seems to be related to this kernel.org report: https://bugzilla.kernel.org/show_bug.cgi?id=79261

Also: https://www.libreoffice.org/bugzilla/show_bug.cgi?id=86478

CC: (none) => remi

Created attachment 5793 [details]
dmesg | grep drm

CC'ing Thierry since it seems to be a bug with the intel driver.

CC: (none) => thierry.vignaud

It is a kernel drm issue where it cant properly filter underruns, somtheing that got more exposed in 3.18 series kernels.

There is a patchset upstream 3.19... to fix this but it changes a lot of code, so I did a stripped down version that fixed the issue for Marja.

I need to either backport the filtering code or extend the stripped down version to cover more underruns or maybe switch the error to debug level...

The error should otherwise be harmless, it's more of an "informal nature"

Thanks for the update Thomas.

So maybe the "flickering" I see with plymouth is not related to this issue at all, even if it's the one displayed when it flickers? I explain:
- At boot, it starts with this FIFO underrun error (two lines on black screen), then the graphical plymouth boot starts. After two bubbles of the animation, it flickers back to the black screen with the two error lines for a couple of seconds, and then the graphical boot goes on.

If you can confirm that it's just a coincidence that we see the error message from this bug during the flickering, I'll open another bug report against plymouth or mageia-theme.

(In reply to Thomas Backlund from comment #11)

> 
> There is a patchset upstream 3.19... to fix this but it changes a lot of
> code, so I did a stripped down version that fixed the issue for Marja.
> 

It fixed a similar issue - bug 14867 - on a different laptop, this error on this laptop did not get fixed ;-)

@ Rémi

On this laptop, I've been having hard freezes (± daily until a few days ago) of which I was wondering whether they were related. Nothing in the logs.
Did you get any hard crashes?

@ Marja

Nope, I think what you describe in comment 4 is a symptom of another bug (that I don't experience here).

Sorry, Thomas, got different but similar underruns in return :-( 


[root@DenkBlok3Cauldron5a1 marja]# uname -r
3.19.0-desktop-0.rc7.2.mga5
[root@DenkBlok3Cauldron5a1 marja]# journalctl -b | grep underrun
Feb 03 17:00:04 DenkBlok3Cauldron5a1 kernel: [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A
Feb 03 17:00:04 DenkBlok3Cauldron5a1 kernel: [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun
[root@DenkBlok3Cauldron5a1 marja]#

Changing the summary because I think it changed from "cpt..." to "intel..." for all, please complain if I'm wrong!!


(In reply to Rémi Verschelde from comment #15)
> @ Marja
> 
> Nope, I think what you describe in comment 4 is a symptom of another bug
> (that I don't experience here).

Yeah, you're right, it doesn't happen anymore since several weeks.

Source RPM: kernel-desktop-3.18.2-1.mga5 => kernel-desktop-3.19.0-0.rc7.2.mga5
Summary: kernel/intel: [drm:cpt_* FIFO underrun => kernel/intel: [drm:intel_* FIFO underrun

I have a Compaq Presario V2030US

Intel 855GME Integrated controller
In the Mageia 5beta2 only show :

feb 09 19:05:25 sarasofia.sanpedrovalle kernel: [drm:i8xx_irq_handler] *ERROR* pipe B underrun

But i upgrade to Mageia 5beta3 this day feb 14 using the urpmi.removemedia and addmedia and after reboot i look in the journal and I get two: 

feb 14 23:11:33 sarasofia.sanpedrovalle kernel: [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
feb 14 23:11:33 sarasofia.sanpedrovalle kernel: [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun

I don't know if I updated the intel driver in the febrary 9 or 10, only tonight i gets again these errors....

I think the maintainer again switch this to the debug mode

CC: (none) => neoser10

fixed since kernel-3.19.2-2.mga5

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

Not for me:

# journalctl -b | grep underrun
avril 14 18:05:51 localhost kernel: [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A
avril 14 18:05:51 localhost kernel: [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun

# uname -r
3.19.3-desktop-1.mga5

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

or for me

this is weird, I cannot imagine it is related, but after I had switched to 16bpp for bug 15493, the underruns returned... I hadn't had them since March 25


......
mrt 25 17:05:55 Mga5RC_EFI kernel: [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A
mrt 25 17:05:55 Mga5RC_EFI kernel: [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun
apr 16 17:57:38 Mga5RC_EFI kernel: [drm:intel_set_cpu_fifo_underrun_reporting [i915]] *ERROR* uncleared fifo underrun on pipe A
apr 16 17:57:38 Mga5RC_EFI kernel: [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
apr 16 17:57:38 Mga5RC_EFI kernel: [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A
apr 16 17:57:38 Mga5RC_EFI kernel: [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun
<more of the same as above, until I swtiched back to 24bpp>

tty2-6 had gone crazy, but there was nothing odd in tty1 with KDE 

I'll switch back to 16bpp again, to see whether I can reproduce it

Created attachment 6297 [details]
journalctl -a | grep underrun

with 16bpp, tty2-6 are crazy and completely unusable again

and the underruns are back

In the MGA4 release kernel, this fifo underrun does not shown, but in the updated kernel (installed in the MGA5RC public available) this message is again, I will try install the MGA4 release kernel (3.12.21) in a clean install, test for this message and post results

(In reply to Mauricio Andrés Bustamante Viveros from comment #23)
> In the MGA4 release kernel, this fifo underrun does not shown, but in the
> updated kernel (installed in the MGA5RC public available) this message is
> again, I will try install the MGA4 release kernel (3.12.21) in a clean
> install, test for this message and post results

Thanks for the proposal but it's not really needed AFAICT, as per comment 11 it is established that this error shows since kernel 3.18.

Hello QA

This comment is informational only
I found that kernel 3.16.1 downloaded from kernel.org, configured with the .config file from 3.14 MGA4 release kernel, show the fifo underrun issue, diffing the 3.15.10 kernel files (the last kernel without problems) and the 3.16.1, see a little error in an static void function that refers to GEN2 intel graphics (attachment) because in the same file i found the same section but with boolean value false, not true as stated in the refered line, edited that boolean to false too, and only I have the set underrun reporting.

Created attachment 6468 [details]
File sections that have errors in the same functions

tmb, what about patch from comment #25?

Whiteboard: 5beta2 => 5beta2 5rc

It's not a fix, it only hides the issue

Really this is not a fix, the fifo underruns exist time ago, but, the kernel module maintainer, decided enable the reporting in the info level not in debug level, and this is breaking bootsplash (Plymouth), I think he want finding hardware glitches and freezes, but leave as is in a production kernel, mageia kernel maintainer must hide this fifo underrun and how??? or we must wait to upstream repair this ???

I will switch the messages to debug level so they can be accessed when needed without rebuilding kernel

Should be "fixed" / hidden as of: kernel-3.19.8-2.mga5

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

Those are back in kernel-4.1.7-1.mga5 in Core Updates Testing.

Status: RESOLVED => REOPENED
Resolution: FIXED => (none)
Source RPM: kernel-desktop-3.19.0-1.mga5 => kernel-desktop-4.1.7-1.mga5

Reported by Fred James on QA Discuss ML. Adding him to CC.

These messages flashes by in terminal window during boot, just before login dialog screen.

[    1.012968] [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A
[    1.012982] [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun

Hello again

I have the Compaq Presario V2030US PM064UA#ABA (that was used to test MGA5 Gnome) and a Latitude Dell 6430 (Used in my work, propietary of my work) but i can use that machine with other SATA disk to test MGA6, this Lati 6430 is Intel HD Graphics (i do not remember the model exactly) but i propose this core i5 4th gen to MGA6.

This lappy contains a intel HD Graphics, Broadcom WIFI, EFI capabilities (unknown specification) & TPM but is not used

Tomorrow i will install MGA6 from NET and report in the MGA6 QA team mailing list

Does anyone still see those underrun messages?

I haven't in a long time

[root@cldrn_64 marja]# journalctl -a | grep underrun
[root@cldrn_64 marja]# rpm -qa | grep kernel-desktop-latest
kernel-desktop-latest-4.6.2-2.mga6
[root@cldrn_64 marja]#

(In reply to Marja van Waes from comment #35)
> Does anyone still see those underrun messages?
> 
> I haven't in a long time
> 
> [root@cldrn_64 marja]# journalctl -a | grep underrun
> [root@cldrn_64 marja]# rpm -qa | grep kernel-desktop-latest
> kernel-desktop-latest-4.6.2-2.mga6
> [root@cldrn_64 marja]#

:( 

every boot up for Mga5. - UEFI, intel 810 or later, latest kernel

No errors here i assume it's fixed in newer Mageia 5 kernel builds.

journalctl -a | grep underrun
[root@localhost ozky]# rpm -qa | grep kernel-desktop-latest
kernel-desktop-latest-4.4.13-1.mga5

CC: (none) => ozkyster

Mass-reassigning all bugs with "kernel" in the Source RPM field that are assigned to tmb, to the kernel packagers group, because tmb is currently MIA.

Assignee: tmb => kernel

(In reply to Otto Leipälä from comment #37)
> No errors here i assume it's fixed in newer Mageia 5 kernel builds.
> 
> journalctl -a | grep underrun
> [root@localhost ozky]# rpm -qa | grep kernel-desktop-latest
> kernel-desktop-latest-4.4.13-1.mga5

fixed in cauldron, too:

[root@cldrn_64 marja]# journalctl -ab | grep underrun
[root@cldrn_64 marja]# 

Closing

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