KDE4 latest unstable

[Seeker]

Now that does it!
This KDE4 is so buggy and unstable that it is barely usable! :P

It is flawless in freezing whole system and bringing my server to it's knees and to complete shutdown, then reboot process begins! :f

Dolphin is good at it!
Also sometimes I can't open text files using Kate or Kwrite

It's apps., are all buggy and if I am lucky, crash only itself without bringing complete server down.

Last crash: I've watched south park, with SMPlayer and after 10 minutes of just watching(not touching PC) - freeze and crash - reboot.


And my home dir is full of .core files:

drkonqi.core
gdb.core
kdeinit4.core


And sometimes I get offer to create a bug report which then gets classified as unusable and offers me to install trace software(which would possibly make KDE4 even more crashier).

This is a total nightmare! :P

[adamk]

Seriously, if your entire system is crashing while using KDE, there is a bug in your video card drivers or X server, and that's where you should start looking. An X application should *never* crash your entire machine.

Are you using KDE's built in desktop effects? What video card and drivers are you using?

Adam

[Seeker]

Well..., today things seems pretty stable as KDE crashed server ONLY once!
Weeeho! ;)

Crash NEVER occurs when I use X (Xterm)
So serious stuff, like root, I do using Xorg.

KDE4 is used only by regular users.

Card is radeon 9600XT

[robbak]

Well, in the words of a red-haired mythbuster, "Well thare's yar prablaem."

Which driver are you using for that ati card?
Are you using desktop effects?

Oh, and you never run X as root, ever.

And KDE, of course, uses xorg.

[phoenix]

Did you install from ports or from packages?

If from ports, did you mess around with CPUTYPE or CFLAGS in /etc/make.conf?

If you did, comment out those two options, and rebuild all your ports (and kernel and world if needed). Especially if you changed CFLAGS. You really, really, really should not do that!! :)

If you are using Desktop Effects in KDE, try disabling them. I had lots of problems with DE and KDE 4.1 and 4.2 (works great in 4.3) on Linux using nVidia 9400, and with KDE 4.1 through 4.3 on FreeBSD using onboard Intel. The Radeon 9600 is pretty old, and may not have enough oomph for all the DE stuff (I've never even bothered testing them with my Radeon 7000 system. Things run smooth with DE disabled.)

Which version of X.org are you running?

Are you running a custom kernel, or GENERIC? If custom, have you tried using GENERIC for a while to see how things work? If you mucked around with CFLAGS, be sure to recompile your kernel and world as well with CFLAGS commented out.

[LateNiteTV]

if X is locking up, you may want to edit your xorg.conf and add the following
Option "XkbOptions" "terminate:ctrl_alt_bksp"
to InputDevice section so you can kill the xserver without rebooting.

[Seeker]

There is no problem with X
%X -version

X.Org X Server 1.6.1
Release Date: 2009-4-14
X Protocol Version 11, Revision 0
, just KDE.

KDE4 is from ports. Kernel is GENERIC.
All configs were left at theirs default values.
While installing other ports recursively, I did saw that some ports did have checked optimized CFLAGS by default.

Finally, regarding CFLAGS...
Are you saying, that I should always avoid using them in ports(or anywhere else) even if checked by deafult?

[adamk]

There is no problem with X
%X -version

X.Org X Server 1.6.1
Release Date: 2009-4-14
X Protocol Version 11, Revision 0
, just KDE.


If your whole system is crashing to the point that it reboots, then there is a significant bug in Xorg or your kernel (or, quite possibly, your 3D driver, which has a kernel module). KDE4 might be doing something to cause the crash, but the kernel should never let the system go down like that.

Adam

[mickey]

If your whole system is crashing to the point that it reboots, then there is a significant bug in Xorg or your kernel ...

... or faulty hardware ...

[phoenix]

Finally, regarding CFLAGS...
Are you saying, that I should always avoid using them in ports(or anywhere else) even if checked by deafult?

CFLAGS being set by the port is fine, as the port maintainer has checked that the software runs correctly with those settings, and it only affects that one port.

Editing CFLAGS in /etc/make.conf is rarely a good idea, and can lead to all kinds of hard-to-diagnose errors, with the world, the kernel, and ports.

Gentoo's "optimise everything, -O100 is the best" mentality leads to issues in FreeBSD. :)

[adamk]

... or faulty hardware ...

Ideally faulty hardware wouldn't crash the operating system, but I have yet to see an operating system handle that properly :-) So I agree, that's another possibility.

Adam

[Seeker]

Anyway, there is no crashes for a 3 days now. ;)

I did NOT edited CFLAGS in /etc/make.conf nor do I use custom kernel.

I guess tracing and debuging .core files and that one of a kerenel itself would give a result, which I don't feel like doing as it is not critical issue.
+ On WinXP there were issues with graphic card as drivers would every here and there reset card, so there MIGHT be a faulty AGP hardware.

[mickey]

+ On WinXP there were issues with graphic card as drivers would every here and there reset card, so there MIGHT be a faulty AGP hardware.

AGP is also famous for causing real headaches with some mainboard's chipsets, when the driver would enable things like 'AGP FastWrite'.

[Seeker]

I've just been editing PHP code in Kate.
Sys got frozzen.
None of shortcuts did a trick ctrl+alt+(del,esc,backspace)
After 15 minutes passed, I could only reset whole PC.

To my shock, after reboot and launching kde and opening Kate and then that file with PHP code...
It was totally empty
All 1000 un-backed up lines of code went to hell!!
And I save frequently, while coding
I can't believe this!!!
Now, this kind of stuff haven't hit me even on XP.

I stll HOPE, it is SOMWHERE, or I'll never use KDE again.
Has it been transfered somwhere, as sos procedure, before/during/after crash.

I mean.. seriously - COMPLETELY EMPTY FILE!!! :\

[DutchDaemon]

I don't know or use Kate, but if it supports auto-saving, it will probably not save to the document itself, but to a backup file, which may or may not be in the directory of your original document, and may or may not be named similarly to your original document.

If you remember the exact time of the crash, do the following:

Suppose date and time of crash are as follows: 10 November 2009, 08:30:00. To be safe, subtract an hour from that, so 07:30:00.


cd /tmp
touch timestamp
(convert time of crash -/- 1 hour to be used by touch:)
touch -t 200911100730.00 /tmp/timestamp
(find files later than that date/time)
find / -Bnewer /tmp/timestamp > /tmp/files.txt


Now investigate /tmp/files.txt and see if you find anything resembling the document you lost.

[Seeker]

I don't know or use Kate, but if it supports auto-saving, it will probably not save to the document itself, but to a backup file,
True!
As soon as you hit save, it immediately saves file AND creates backup file in a same directory with added "~" at the end of file like:
File: coding.php
Backup: coding.php~


...
If you remember the exact time of the crash, do the following:

Suppose date and time of crash are as follows: 10 November 2009, 08:30:00. To be safe, subtract an hour from that, so 07:30:00.


cd /tmp
touch timestamp
(convert time of crash -/- 1 hour to be used by touch:)
touch -t 200911100730.00 /tmp/timestamp
(find files later than that date/time)
find / -Bnewer /tmp/timestamp > /tmp/files.txt


Now investigate /tmp/files.txt and see if you find anything resembling the document you lost.
Thanx!
Done!
Nothing found!

I can't believe this!!!

Completely [erased backup file and original NULLED (empty)!
And keep in mind that kate is advanced text editor!

I feel like being hacked!

This is total disaster!
I am NOT using some beta or RC, so this was NOT supposed to be happening! :\

Is this really it??

Does this mean I can go, an start hitting my head at the wall???

[DutchDaemon]

You did use the correct date/time, I presume? I suddenly noticed that my example date/time was in the future ;)

If so, I can only assume that fsck unlinked these files that were probably in use at the time of the crash. One final path to take may be looking in any lost+found directories.

find / -type d -name "lost+found"

The files in those directories won't have recognisable names, so use more/less and quotes on these files and see what turns up, e.g.
more "#07019222"

[Seeker]

You did use the correct date/time, I presume? I suddenly noticed that my example date/time was in the future ;)

Yes, your example was 1 day in a future.
That would yield 0 results ;)

If so, I can only assume that fsck unlinked these files that were probably in use at the time of the crash. One final path to take may be looking in any lost+found directories.

find / -type d -name "lost+found"

The files in those directories won't have recognisable names, so use more/less and quotes on these files and see what turns up, e.g.
more "#07019222"
That yielded no result.
I did executed that command as a root.

Nada!

Is there anything else I can do?
You've mentioned fsck unlinked those files...
Maybe to use fsck somehow?

Or is this it? :r

An during 2 last post I had 2 sys crashes!
Now that does it!

I've found this 2 files:
/var/crash/info.12
/var/crash/vmcore.12
How to make them useful?
If procedure too complex then give me a link to procedure

[DutchDaemon]

Search (http://forums.freebsd.org/search.php) the forums for 'vmcore', and choose 'Show results as posts'. A few ways to extract information from vmcores will show up.

As to your lost file: other than low-level forensics, I don't think there is another tool I can think of right now, and seeing you're working on that system, the possible remnants of it have probably been overwritten by now.

[Seeker]

... As to your lost file: other than low-level forensics, I don't think there is another tool I can think of right now, and seeing you're working on that system, the possible remnants of it have probably been overwritten by now.

This is very, VERY BAD!
And I am very, VERY ANGRY!

I understand loss of precious file that was open in Kate, but NOT and it's BACKUP!!
Kate is one that is linking to both of them!

Am I right, if I target my anger on KDE's Kate?
So I'll get rid of them forever and take something super tested an safe.

That is what UNIX is about - SAFE! (and much more)
And KDE's Kate doesn't represent that!

WinXP NEVER did this to me!

Man, I am so pissed off right now! :P

[graudeejs]

pardon me, I haven't seen this being mentioned:
Have you looked at /var/tmp ?
I don't use KDE and friends either, so I don't know, but perhaps your file is there :)
It's worth checking

[true]

Am I right, if I target my anger on KDE's Kate?
No, Kate and KDE have absolutely nothing to do with this.

The reason why you lost your files is that the actual data was still not written to disk at the time the system crashed. Find out more about Soft Updates at http://en.wikipedia.org/wiki/Soft_updates

[Seeker]

pardon me, I haven't seen this being mentioned:
Have you looked at /var/tmp ?
I don't use KDE and friends either, so I don't know, but perhaps your file is there :)
It's worth checking

It has already been covered by find command.

No, Kate and KDE have absolutely nothing to do with this.
I disagree.
Your description would apply, if it would be only about edited file, that was not saved.

Only modification was supposed to be lost!
Whole file was ZEROED - EMPTY! :\

The reason why you lost your files is that the actual data was still not written to disk at the time the system crashed.
And here is a nail in Kate's coffin.
There WAS a backup in same dir with added("~"), at it's end, that was NOT opened in editor!
And it was completely LOST!

Only Kate had a contact/interaction with backup file!!
Where is it!
Retarded Kate!

This has never occurred on winXP! :r

I still can't believe this has happened!
I am totally not coding anything anymore, until I get my laptop back!
I simply don't have trust in file system anymore.
Find out more about Soft Updates at http://en.wikipedia.org/wiki/Soft_updates
Are you claiming, that if I would shift from default ufs to that of soft updates, this will not happen anymore??
The tunefs(8) program can be used to fine-tune a file system.
This program has many different options, but for now we are only concerned with toggling Soft Updates on and off, which is done by:
# tunefs -n enable /filesystem
# tunefs -n disable /filesystem

As to your lost file: other than low-level forensics, I don't think there is another tool I can think of right now, and seeing you're working on that system, the possible remnants of it have probably been overwritten by now.

I see, then let me reformulate question.
What kind of command would you use to retrieve thise files, just after reboot, in same situation/context?

Anyway, this thread is about crashing so here it is:
www.starforce.biz 5 core (http://www.starforce.biz/CORES/5.txt)
www.starforce.biz 6 core (http://www.starforce.biz/CORES/6.txt)
www.starforce.biz 7 core (http://www.starforce.biz/CORES/7.txt)
www.starforce.biz 8 core (http://www.starforce.biz/CORES/8.txt)
www.starforce.biz 9 core (http://www.starforce.biz/CORES/9.txt)
www.starforce.biz 10 core (http://www.starforce.biz/CORES/10.txt)
www.starforce.biz 11 core (http://www.starforce.biz/CORES/11.txt)
www.starforce.biz 12 core (http://www.starforce.biz/CORES/12.txt)

Yes, I Tested RAM --> memtest86+
It passed! 100%

[mickey]

An during 2 last post I had 2 sys crashes!

And still to me this appears to be likewise hardware failure.

[Seeker]

And still to me this appears to be likewise hardware failure.
And how would I isolate target hardware?
Hint: Crashes only started occur while in KDE

[adamk]

It's quite possible for a hardware problem or driver problem to only manifest itself under KDE4. Do the crashes happen if you disable direct rendering? How about if you disable the sound drivers? In the past, I've found that OSSv4 can be very tempermental, causing completely system crashes (though, to be fair, those backtraces don't look like they are coming from OSS, but it's worth a shot).

Adam

[DutchDaemon]

Note that two of your vmcore dumps show that csup caused the crash .. And the applications in all of your crashes were different, and not all were KDE or even X related. Seems to point to a more common issue, which is usually hardware. So RAM, disk, overheating, PCI cards, anything ..

number = vmcore
5. current process = 52286 (csup)
6. panic: ufs_dirbad: /usr: bad dir ino 1655620 at offset 1512: mangled entry
7. current process = 1613 (dolphin)
8. current process = 1550 (drkonqi)
9. current process = 1775 (csup)
10. current process = 1451 (opera)
11. current process = 3296 (smplayer)
12. current process = 15540 (kmail)


All cores seem to have this in common, but this may be typical for any vmcore image, I don't know ;) ->
Previous frame inner to this frame (corrupt stack?)

[Seeker]

While there was not KDE installed, there were only 3 crashes and all were caused by ral0 (ral driver) pci WiFi card.
And device timeout was common, so after googling, I learned that cause is a driver, which makes ral0 suck too much power so it end up starwed of power.

But, right now, I think I should benchmark my graphic card, as there were problems with it, even under XP
Port recommendation?

[adamk]

Perhaps try a moderately GPU intensive game like openarena: games/openarea ?

EDIT: There's even a time demo you can download and try:

http://dri.freedesktop.org/wiki/Benchmarking

[Seeker]

Perhaps try a moderately GPU intensive game like openarena: games/openarea ?

EDIT: There's even a time demo you can download and try:

http://dri.freedesktop.org/wiki/Benchmarking

No, no...
I would really prefer something nicer, like memtest for RAM, so I need specialised tool for GPU

And I completely avoid software, that requires lynux emulation.

Also, is ther a tool for seeing sys power consumption, logging perhaps? :stud

[adamk]

I am not aware of any memtest like application for GPUs.

But what makes you think that openarena requires linux emulation?

Adam

[Seeker]

I am not aware of any memtest like application for GPUs.

But what makes you think that openarena requires linux emulation?

Adam
Absolutely nothing. ;)
I just said that I avoid all ports that require linux emulation, in general.

I also don't feel like playing games.

[mickey]

How long has it been, since you assembled that hardware setup? In what environment is it running? Dusty? Do you smoke? Did you use some older components, that have been laying around for some time?

Often all it takes to get a stable system again are some Q-tips and some pure alcohol...

[Seeker]

How long has it been, since you assembled that hardware setup?
4 years ago.

In what environment is it running? Dusty? Do you smoke?

Nope, I don't smoke.
But sometimes, I do smoke a little bit of crack. ;)
I simply code better while on it! :)

Did you use some older components, that have been laying around for some time?

No, all are 4 years old.

Often all it takes to get a stable system again are some Q-tips and some pure alcohol...
I've cleaned dust in midi case, with aspirator, yesterday.

Anyway, I've unplugged cord to cooling fan on graphic card, 3 months ago, as it started to drive me nuts, with it's noise(which wasn't case before).
Would GPU overheating cause this?

Now, I have open midi case and if I decide to run KDE4, I plug in cord of cooling fan on graphic card.

How to fix fan?
Noise is terrible! It goes like this:
zzzzzzzzzzz-zz-zz-zzzzzzzzzzzzzzzzzzzzz-zz zz zz zzzzzzzzzzzzzzzzzzzz
intermittent...
I wana constant silence, as it was first 3 years.
:P

[graudeejs]

a) buy new fan, replace old
b) carefully try oiling it, with a little bit of oil....
c) put Very, very big and effective heat silk
d) put voter cooling system on it

I recommend a

[mickey]

4 years ago.

That is quite some time to accumulate dust, for contacts to oxidize, ... you name it ...

I've cleaned dust in midi case, with aspirator, yesterday.
I was not talking about cleaning the case, that is seldom the source of spurious system instabilities :P

I was talking about cleaning the electrical contacts, especially those on your add-on cards and memory modules.

Cleaning the case, should be part of your regular servicing procedures anyways :e

Anyway, I've unplugged cord to cooling fan on graphic card, 3 months ago, ...

Bad idea.

Would GPU overheating cause this?
Well, as KDE4 has a compositing window manager, which uses hardware 3D features of your graphics card, I would assume, that it puts more stress on your graphics card, than say using twm or 80x25 text mode console. More stress means more heat. No fan means even more heat, so yes, i would think this was absolutely possible.

How to fix fan?
Noise is terrible! It goes like this:
zzzzzzzzzzz-zz-zz-zzzzzzzzzzzzzzzzzzzzz-zz zz zz zzzzzzzzzzzzzzzzzzzz
intermittent...
I wana constant silence, as it was first 3 years.

I know... my old graphics card currently has it's third fan.

Get a replacement fan.
If that's not possible, get a replacement fan assembly.
If that's also not possible, get a replacement graphics card :f

[Seeker]

Well,after case is cleaned and open and fan is on, no more reboots. ;)

Now, the only annoyance is loudly fan.
It is a little hard to find a new fan for it, as it is custom for each graphic card.(CPU is different story)
I think I should disassemble and oil it, to shut it up. ;)

[graudeejs]

Just don't use too much oil

[mickey]

Well,after case is cleaned and open and fan is on, no more reboots. ;)
Surprise, surprise :e

Now, the only annoyance is loudly fan.
It is a little hard to find a new fan for it, as it is custom for each graphic card.

What make/model is your graphics card?

I think I should disassemble and oil it, to shut it up. ;)

Trust me, that wont last for long :P Once the fan's bearings are in a state, where they start making such a noise, oil wont help anymore.

[LateNiteTV]

try using vaseline on the bearings. its thicker than oil so it wont run all over the place and it can better lubricate damaged moving parts.

[mickey]

try using vaseline on the bearings. its thicker than oil so it wont run all over the place and it can better lubricate damaged moving parts.

Taking a closer look at an average fan as used in computer systems, you will notice that you wont even get close to the bearings without completely disassembling the fan. And if you get that far, without damaging the axle, it would probably make more sense, to replace the bearings. But that will usually cost more than a new fan. And given the size of an average GPU fan, I surely hope, you have a good magnifying glass at hand :e

For most standard graphics cards, like nvidia and ATI, a replacement cooler assembly should not be that hard to get.

Arctic Cooling for example has a line of products called ATI Silencer or NV Silencer, that use large 70mm fans, and exhaust the hot air directly out of the case. Only drawback is, that you cannot use the slot directly adjacent to the graphics card.

[Seeker]

Ok guys, this thread is a total mess!

What I mean, is that it could be split in 4,5 regular threads each with it's own theme.

It started with KDE crashing...
Then how to undo rm(ed) file and where to search for it on HD and how to do it.
...Passed to file system working and Soft Updates
Core dumping...
Touched installed ports and gaming..., linux emulation
..skipped to hardware and graphic card
Ending in detailing of hardware cleaning methods
and GPU fan alternatives...
And finally to debate moisturizing of loudly GPU fan.
Here by the way, some infidel deleted my post, related to Vaseline ambiguity issue, which, I am absolutely sure, would put a nail in the coffin, to this issue.

But even with that, this is considered as a solved matter.

I would like, to thank everyone, that helped me! ;)
No more KDE crashes at all. ;)

[DutchDaemon]

Right. Closed then.