[i945GM] X freezes typing quickly in Kmail when desktop effects are enabled

[Francis J. Lacoste]

Binary package hint: xserver-xorg-video-intel

Since I ugpraded to Jaunty Beta, I experience common "lock-up".

When the display is "locked-up", the mouse pointer moves on the screen, but the pointer doesn't change, the keyboard doesn't do anything, and the display doesn't get any update.

I can SSH in the box and kill -9 the Xorg server. After that either nothing happens, or the screen becomes black. In all cases, the display is still non-functional. A shutdown -r usually doesn't complete either and I have to power-cycle the laptop.

This happen in various circumstances, but the most "reliable" way to trigger it is by creating a new email in Kmail and typing in the subject field very fast. It also happened in various other situations: like opening a new tab in firefox, starting up gvim, or just on my first log-in after the upgrade.

ProblemType: Bug
Architecture: i386
DistroRelease: Ubuntu 9.04
Package: xserver-xorg-video-intel 2:2.6.3-0ubuntu4
ProcEnviron:
 LANGUAGE=fr_CA.UTF-8:fr_FR:fr:en_GB:en
 PATH=(custom, user)
 LANG=fr_CA.UTF-8
 LC_MESSAGES=en_CA.UTF-8
 SHELL=/bin/bash
ProcVersion: Linux version 2.6.28-11-generic (buildd@palmer) (gcc version 4.3.3 (Ubuntu 4.3.3-5ubuntu4) ) #38-Ubuntu SMP Fri Mar 27 09:00:52 UTC 2009
SourcePackage: xserver-xorg-video-intel
Uname: Linux 2.6.28-11-generic i686

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile 945GM/PM/GMS, 943/940GML and 945GT Express Memory Controller Hub [8086:27a0] (rev 03)
 Subsystem: Dell Device [1028:01d8]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)
 Subsystem: Dell Device [1028:01d8]

[Bryce Harrington (bryce)]

Hmm, nothing out of the ordinary in Xorg.0.log, although I notice some differences between the log and log.old in input device setup messages. Perhaps you just had different input devices attached in each case?

What would help in analyzing this bug is to have an error message of some sort. Not all lock-up bugs will cause an error message to be printed, but if one is, it'll help identify the problem more quickly.

Can you look in the following places for an error message, after reproducing the bug:
  * /var/log/gdm/*
  * /var/log/syslog
  * /var/log/kern.log
  * /var/log/messages
  * /var/log/Xorg.0.log or Xorg.0.log.old
  * dmesg

Look for some unusual message appearing at the same time that you first noticed the lock-up.

[Bryce Harrington (bryce)]

With lockup bugs, usually what's happening is the GPU has hung due to some bad register setting.

There is a tool for taking a snapshot of the GPU's register settings, `intel_reg_dumper`, which is available as part of the xserver-xorg-video-intel-dbg package. So install that, and then run this tool in two cases:

1. With the display working properly

 sudo intel_reg_dumper > reg_good.txt

2. After reproducing the lockup (you'll need to ssh into the box from a 2nd machine):

 sudo intel_reg_dumper > reg_bad.txt

Then attach both files to this bug report. We can send the report upstream for there to see if a fix can be identified.

[Francis J. Lacoste (flacoste)]

So I didn't find any interesting error messages apart maybe the following which appeared at the end of my Xorg.0.log:
exaCopyDirty: Pending damage region empty!

But the register trick does show a difference:

--- /home/francis/reg_bad.txt 2009-04-02 21:15:04.000000000 -0400
+++ /home/francis/reg_good.txt 2009-04-02 21:13:36.000000000 -0400
@@ -137,7 +137,7 @@
 (II): FBC_CONTROL2: 0x00000000
 (II): FBC_FENCE_OFF: 0x00000000
 (II): FBC_MOD_NUM: 0x00000000
-(II): MI_MODE: 0x00000000
+(II): MI_MODE: 0x00000200
 (II): MI_ARB_STATE: 0x00000040
 (II): MI_RDRET_STATE: 0x00000000
 (II): ECOSKPD: 0x00000307

[Francis J. Lacoste (flacoste)]

And here is the "bad" registers dump.

Btw, it seems that the KMail trick is a very reliable way to reproduce the lock-up. I've also found a "work-around" for this very common source of lock ups, i paste the subject line from another place, that way there is a single update to the window title which seems to be related to the problem.

[Bryce Harrington (bryce)]

Created an attachment (id=24517)
XorgLog.txt

[Problem]
X locks up in various circumstances, most easily reproduced when typing quickly into KMail's subject field.

[Original Report]
Since I ugpraded to Jaunty Beta, I experience common "lock-up".

When the display is "locked-up", the mouse pointer moves on the screen, but the pointer doesn't change, the keyboard doesn't do anything, and the display doesn't get any update.

I can SSH in the box and kill -9 the Xorg server. After that either nothing happens, or the screen becomes black. In all cases, the display is still non-functional. A shutdown -r usually doesn't complete either and I have to power-cycle the laptop.

This happen in various circumstances, but the most "reliable" way to trigger it is by creating a new email in Kmail and typing in the subject field very fast. It also happened in various other situations: like opening a new tab in firefox, starting up gvim, or just on my first log-in after the upgrade.

Btw, it seems that the KMail trick is a very reliable way to reproduce the lock-up. I've also found a "work-around" for this very common source of lock ups, i paste the subject line from another place, that way there is a single update to the window title which seems to be related to the problem.

ProblemType: Bug
Architecture: i386
DistroRelease: Ubuntu 9.04
Package: xserver-xorg-video-intel 2:2.6.3-0ubuntu4
ProcEnviron:
 LANGUAGE=fr_CA.UTF-8:fr_FR:fr:en_GB:en
 PATH=(custom, user)
 LANG=fr_CA.UTF-8
 LC_MESSAGES=en_CA.UTF-8
 SHELL=/bin/bash
ProcVersion: Linux version 2.6.28-11-generic (buildd@palmer) (gcc version 4.3.3 (Ubuntu 4.3.3-5ubuntu4) ) #38-Ubuntu SMP Fri Mar 27 09:00:52 UTC 2009
SourcePackage: xserver-xorg-video-intel
Uname: Linux 2.6.28-11-generic i686

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile 945GM/PM/GMS, 943/940GML and 945GT Express Memory Controller Hub [8086:27a0] (rev 03)
 Subsystem: Dell Device [1028:01d8]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)
 Subsystem: Dell Device [1028:01d8]

[Bryce Harrington (bryce)]

Created an attachment (id=24518)
reg_good.txt

[Bryce Harrington (bryce)]

Created an attachment (id=24519)
reg_bad.txt

--- /home/francis/reg_bad.txt 2009-04-02 21:15:04.000000000 -0400
+++ /home/francis/reg_good.txt 2009-04-02 21:13:36.000000000 -0400
@@ -137,7 +137,7 @@
 (II): FBC_CONTROL2: 0x00000000
 (II): FBC_FENCE_OFF: 0x00000000
 (II): FBC_MOD_NUM: 0x00000000
-(II): MI_MODE: 0x00000000
+(II): MI_MODE: 0x00000200
 (II): MI_ARB_STATE: 0x00000040
 (II): MI_RDRET_STATE: 0x00000000
 (II): ECOSKPD: 0x00000307

[Bryce Harrington (bryce)]

Hi Francis,

I've forwarded your bug upstream to https://bugs.freedesktop.org/show_bug.cgi?id=21041 - please subscribe to this bug report in case upstream needs further information or wishes you to test something. Thanks ahead of time!

[Steve Beattie (sbeattie)]

This may affect more hardware than just i945s, I had a very similar lockup (X frozen, mouse still moved, able to ssh in) this weekend with the following hardware:

00:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 0c)
        Subsystem: Lenovo Device 20b5
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+
        Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0
        Interrupt: pin A routed to IRQ 2295
        Region 0: Memory at f8100000 (64-bit, non-prefetchable) [size=1M]
        Region 2: Memory at e0000000 (64-bit, prefetchable) [size=256M]
        Region 4: I/O ports at 1800 [size=8]
        Capabilities: [90] Message Signalled Interrupts: Mask- 64bit- Queue=0/0 Enable+
                Address: fee0300c Data: 41c9
        Capabilities: [d0] Power Management version 3
                Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
                Status: D0 PME-Enable- DSel=0 DScale=0 PME-
                Bridge: PM- B3+
        Kernel modules: intelfb

[Bryce Harrington (bryce)]

Mmm, looks to me like this is a dupe of 20520... same chip, same symptoms, same difference in the registers...

[Bryce Harrington (bryce)]

Steps to reproduce are different though; what do you think?

[Botond Szász (boteeka)]

I also experienced the same lock-up using a Dell Inspiron 1525, with Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller.

I can reproduce it with Alt+Tab -ing multiple times. Using compiz and the Shift Switcher plugin.

From the posts above and my experiences I would say that this is related to the keyboard input handling code somewhere in the X driver.

[Eric Anholt (eric-anholt)]

FWIW, that MI_MODE bit just means "the ring is busy" (meaning hung in this case)

[Bryce Harrington (bryce)]

Botond, how are you determining that you have the same lockup? You have quite different hardware than the original reporter.

[Gunni (fgunni)]

For me the bug occured after an update around april 1st +- some days.
Its not really reproducable, When i watch a movie on mythtv after some time display gets black. Power button seems to trigger a good shutdown.
No reaction on ctrl+alt+backspace or ctrl+alt+F1 etc.
Its on an GMA 3100 (i965?)
May this be the same bug, or should i open a new?
I will try to get some logs later.

[Bryce Harrington (bryce)]

> May this be the same bug, or should i open a new?

If you would like your bug examined, you should open a new bug. Since you have different hardware, we will not be considering your issue when troubleshooting this one, so unless you're absolutely certain you have the same bug, by not filing a new report, your bug will probably not be looked into.

[Martin Konecny (martin-konecny)]

I can confirm the exact same freeze symptoms (mouse that moves but doesnt change when you hover it over different objects) when opening a new tab in firefox (happens only sometimes).

[Martin Konecny (martin-konecny)]

I should probably add that I have the same graphics hardware (as well as a dell laptop)

00:00.0 Host bridge: Intel Corporation Mobile 945GM/PM/GMS, 943/940GML and 945GT Express Memory Controller Hub (rev 03)
00:02.0 VGA compatible controller: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03)
00:02.1 Display controller: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)

[Bryce Harrington (bryce)]

Okay, so this is a dupe of 339091

-(II): MI_MODE: 0x00000000
+(II): MI_MODE: 0x00000200

[Bryce Harrington (bryce)]

Please take a look at bug 339091 and see if you agree this is a dupe. The steps to reproduce are somewhat different. But the MI_MODE value seems like the smoking gun here.

[Bryce Harrington (bryce)]

MI_MODE bit just means "the ring is busy", so not a dupe

[Bryce Harrington (bryce)]

Is there additional information/debugging that can be done for this class of bug?

[Carl Worth (cworth)]

(In reply to comment #6)
> Is there additional information/debugging that can be done for this class of
> bug?

Absolutely!

Ben Gamari wrote some excellent patches to our i915 kernel driver that provide through debugfs a dump of the current batchbuffer of commands submitted to the GPU. I did some recent brushing-up of those patches, and they should be going out to the mainline Linux kernel soon.

If you get yourself a driver with those debugfs hooks, then whenever you encounter a GPU hang, you can send us a dump of the batchbuffer causing the hang. From there, Eric's written some batch-buffer parsing code, which I will be packaging up as a new intel_gpu_dump tool that will make the hex-dump from debugfs human-readable. But for now, just sending us the hex dump should be extremely valuable. We've already verified this week that the dump can make it very easy to diagnose an otherwise mystic hang.

This is all going to be easier once all of the debug stuff is nicely upstream and available on the machines of all of our users. Until then, I wrote up some notes on where to get the code and how to use it here:

https://bugs.freedesktop.org/show_bug.cgi?id=17638#c28

I hope that helps, and I'll look forward to any dumps you can provide.

Thanks,

-Carl

[Bryce Harrington (bryce)]

Excellent news, and hopefully that should make this whole class of bug tractable. I've added your notes to our "debugging Freezes" page:

 https://wiki.ubuntu.com/X/Troubleshooting/Freeze

I won't have time to do the packaging work for this for Jaunty since we're nearing release, but hope we can get all that in for Karmic.

Meanwhile, the user reports that disabling compiz makes the freezes stop, so for the time being he has that as a workaround.

[Bryce Harrington (bryce)]

Before people 'me-too' onto this bug, be aware that X can freeze in numerous different circumstances, and lacking an error message or exact reproduction steps, it can be hard distinguishing your freeze from someone elses. Until we have this bug pinpointed more specifically, don't 'me-too' to this bug; file a new one.

[Francis J. Lacoste (flacoste)]

Some more interesting information:

- I'm not able to reproduce the lock-up if I disable 'Desktop effects' (so this is a nice work-around which also gives a performance bonus)

- I'm able to SSH in the box when this happens. Killing the X process doesn't give back the display, it will either do nothing or leave the screen black with bars on it.

- If I shutdown (over SSH) sometimes it works, but most of the time it just hangs at some point (unable to say where since SSH is now down and I have no display).

[Francis J. Lacoste (flacoste)]

I confirm it is still hapenning in the latest package: 2:2.6.3-0ubuntu8.

I also confirm that it has the same behaviour under EXA or UXA.

The only stable work-around seems to be to disable desktop effects.

[znmeb (zznmeb)]

I'm getting similar phenomena with Jaunty Beta x86_64 on an HP Pavilion dv4 laptop (Turion64 X2 dual-core with 4 GB). It runs for a while, then the mouse works, but apps refuse to start and windows don't repaint. The keyboard appears inactive; ctl-alt-F1 doesn't give me a console. Eventually I have to power cycle.

I started with ext4 ... then reloaded with ext3 after the first lockup. That didn't help. So I reloaded again. At that point, I had the both the ATI driver and Compiz running. After the next lockup, I shut down Compiz and power cycled. Another lockup. So I reloaded again without the ATI driver. The only thing "non-standard" now is the Broadcom wireless driver.

I started up apt-get upgrade over the wireless and it locked up again after about half an hour. I tried a shutdown and it said it was going to do it, but it didn't, and I ended up having to power cycle.

I'm going to bring it back up and see if I can complete the apt-get upgrade. If the lockup goes away after that, I'll post here. Otherwise I'm going to have to switch to another machine.

[znmeb (zznmeb)]

It looks like the apt-get upgrade fixed things ... I've gone over half an hour with no lockups, doing essentially the same thing I was doing before, which is installing lots of packages. So I guess the install process should be:

1. Install
2. Boot up and kill X
3. Run "apt-get update; apt-get upgrade" as root from a console
4. Reboot to get the new kernel and X.

[Bryce Harrington (bryce)]

znmeb, you probably had some unrelated bug. Since you say you'd had ext4 installed, I suspect your reversion procedure was incomplete. In any case, many different freezes can look alike, and there's not strong evidence you had exactly the same issue as Francis.

[Eric Anholt (eric-anholt)]

Can't reproduce it here with banging on keys in kmail subject field with KMS on master of everything.

[znmeb (zznmeb)]

This is an ext4 bug? I thought that was a different one. Ah well ... I can go back to ext4 and see if it comes back :)

[Bryce Harrington (bryce)]

znmeb, well there was a known ext4 freeze with symptoms I gather are indistinguishable from this, so it just seems awfully coincidental that you had it installed at one time, and are seeing freezes. Like, maybe you still had it loaded in the kernel or something.

Anyway, we're getting off topic from Francis' problem.

@Francis, I believe the kernel team put out a kernel last night with fixes for -intel freeze problems. Would you mind re-testing to see if that happened to catch your issue as well? (Keeping my fingers crossed...)

[Francis J. Lacoste (flacoste)]

I'm very sorry to report that the lock-up is still happening in the same circumstances. The kernel package I have installed (from today) is:

linux-image-2.6.28-11-generic: 2.6.28-11.41

Let me know if it's another one I should try. I'll disable Desktop Effects again in the mean time.

[Meeko (bjorn-aarset)]

I can reproduce this with 2.6.28-11-generic #41-Ubuntu SMP Wed Apr 8 04:39:23 UTC 2009.
Often happen when having compiz with Ubuntu Tweak (I have an effect so when I move my mouse down to the left corner it shows the same as "Shift-Alt-Up"), and often freezes when loading webpages or opeing programs while activating the "show windows" compiz effect. More frequent in later releases.

[Bryce Harrington (bryce)]

@Meeko (and Francis if you care to test too), I've packaged a modified mesa that I think may reduce the frequency of freezes:

https://edge.launchpad.net/~bryceharrington/+archive/purple

I won't guarantee it eliminates freezes entirely, but hopefully should change it from "once every few hours" to "once every couple days".

[Andres Mujica (andres.mujica)]

The triggering action for this bug could be misleading, but as taking a closer look on the descriptions, comments and logs from the original bug reporter this is a dupe from bug #359392

Specially the comment https://bugs.edge.launchpad.net/ubuntu/+bug/353898/comments/18

@Francis, can you take a look onto that bug, and run the repro.sh and metacity_stress.sh scripts there? also please follow the instructions from https://edge.launchpad.net/~ubuntu-x-swat/+archive/x-freeze-test/

If the system hangs with repro.sh using compiz and doesn't with metacity_stress.sh let the dupe state as it is. If is different, please unmark it as dupe.

Thanks in advance.

[Andres Mujica (andres.mujica)]

> If the system hangs with repro.sh using compiz and doesn't with metacity_stress.sh let the dupe state as it is. If is > different, please unmark it as dupe.

he, i mean the opposite, but just let me now please how it goes.

[fixture (universald)]

k, this is bugging the sh*t out of me! Jaunty has gotten progressively less stable since Feb!

I am on a Thinkpad X61 (generic intel hardware, integrated graphics) running 64bit, and have been grappling with freezes for the last couple weeks.

1) this is not a kernel freeze, you can move the mouse, but no keyboard responses, i.e., caps lock has no response. When I ssh into the frozen machine, you can login, htop shows you that nothing suspicious is going on at all. No bad cpu usage, no bad mem usage, etc.

2) but the onset of these freezes on powertop does seem to be associated with random CPU frequency and p-state hikes, i.e., out of the god damn blue, with no application running, CPU runs like hell in powertop and burns lap

3) this is not kernel specific, I have using bleeding edge kernels from ubuntu and compiled 2.6.29, problem persist

4) I theorize problem is in the Xorg drivers, but the bleeding edge intel xorg drivers freezes as well

5) the sata controller does seem to have suspicious activities sometimes, though this is not necessarily connected to freezes. The only time I noticed sata problems, the sata problem preceded the freeze by about 20 mins.

kernel log below,

Apr 22 00:35:54 xerox-L kernel: [ 185.318021] ata1.00: configured for UDMA/133
Apr 22 00:35:54 xerox-L kernel: [ 185.318031] ata1: EH complete
Apr 22 00:35:54 xerox-L kernel: [ 185.325636] sd 0:0:0:0: [sda] 625142448 512-byte hardware sectors: (320 GB/298 GiB)
Apr 22 00:35:54 xerox-L kernel: [ 185.331453] sd 0:0:0:0: [sda] Write Protect is off
Apr 22 00:35:54 xerox-L kernel: [ 185.331459] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
Apr 22 00:35:54 xerox-L kernel: [ 185.358675] sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
Apr 22 00:35:59 xerox-L kernel: [ 189.975183] CE: hpet increasing min_delta_ns to 15000 nsec
Apr 22 00:36:00 xerox-L kernel: [ 191.417034] ata1.00: configured for UDMA/133
Apr 22 00:36:00 xerox-L kernel: [ 191.417047] ata1: EH complete
Apr 22 00:36:00 xerox-L kernel: [ 191.422879] sd 0:0:0:0: [sda] 625142448 512-byte hardware sectors: (320 GB/298 GiB)
Apr 22 00:36:00 xerox-L kernel: [ 191.422928] sd 0:0:0:0: [sda] Write Protect is off
Apr 22 00:36:00 xerox-L kernel: [ 191.422933] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
Apr 22 00:36:00 xerox-L kernel: [ 191.422995] sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
Apr 22 00:37:48 xerox-L kernel: [ 298.789689] ath5k phy0: noise floor calibration timeout (2412MHz)
Apr 22 00:38:25 xerox-L kernel: [ 335.494013] ata1.00: configured for UDMA/133
Apr 22 00:38:25 xerox-L kernel: [ 335.494023] ata1: EH complete
Apr 22 00:38:25 xerox-L kernel: [ 335.511046] sd 0:0:0:0: [sda] 625142448 512-byte hardware sectors: (320 GB/298 GiB)
Apr 22 00:38:25 xerox-L kernel: [ 335.516190] sd 0:0:0:0: [sda] Write Protect is off
Apr 22 00:38:25 xerox-L kernel: [ 335.516197] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
Apr 22 00:38:25 xerox-L kernel: [ 335.529125] sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
Apr 22 00:38:28 xerox-L kernel: [ 339.187032] ata1.00: configured for UDMA/133
Apr 22 00:38:28 xerox-...

[fixture (universald)]

sorry dudes for the post, I did not read the previous posts. UXA is now enabled with compiz, fingers crossed.

[Maurizio (mauriziosilverii)]

My notebook is an Acer Aspire 5315 with Intel i960 video card. I have the same problem. Screen block/freeze. For me the problem is random.

[Carl Worth (cworth)]

I'm marking this bug as NEEDINFO since we haven't been able to reproduce the bug and we are aware that several hangs have been fixed in the latest code.

Bryce, if you can get a report from our latest driver and let us know if the issue persists, then that would be very helpful. Otherwise, I'll just assume the bug is fixed and close this bug report within a couple of weeks.

-Carl

[Bryce Harrington (bryce)]

Please test the package in this PPA, which has a fix for X freeze issues like this one.

  https://edge.launchpad.net/~bryceharrington/+archive/green/

[Carl Worth (cworth)]

Closing and assuming the issue is fixed as described above.

Please reopen if it's found that the latest bug still has this issue.

-Carl

[Bryce Harrington (bryce)]

Upstream has closed the bug because they can't reproduce it and believe it to be fixed. For that reason, and since there has been no comment or testing feedback on this in a couple months, I'm closing this bug as well.

[Alex Valavanis (valavanisalex)]

Jaunty has reached end-of-life, so I'll close this report. The bug is marked as being fixed in later versions of Ubuntu