Kernel 2.6.20-13 acpi bug

Bug #96480 reported by isazi
70
Affects Status Importance Assigned to Milestone
linux-source-2.6.20 (Ubuntu)
Fix Released
High
Ben Collins
Nominated for Feisty by resistor

Bug Description

I've installed feisty, all was good untile kernel version 2.6.20-13, after that there's an acpi module freeze when i try to launch /etc/init.d/acpid:

[ 260.056000] input: Power Button (FF) as /class/input/input4
[ 260.140000] ACPI: Power Button (FF) [PWRF]
[ 260.232000] input: Lid Switch as /class/input/input5
[ 260.232000] ACPI: Lid Switch [LID]
[ 260.268000] input: Sleep Button (CM) as /class/input/input6
[ 260.268000] ACPI: Sleep Button (CM) [SLPB]
[ 260.472000] Using specific hotkey driver
[ 260.620000] No dock devices found.
[ 260.720000] BUG: at lib/kref.c:32 kref_get()
[ 260.720000] [<c01ee21d>] kref_get+0x3d/0x40
[ 260.720000] [<c01ed50f>] kobject_get+0xf/0x20
[ 260.720000] [<c025515e>] get_device+0xe/0x20
[ 260.720000] [<c0255768>] device_add+0x78/0x530
[ 260.720000] [<c01ed50f>] kobject_get+0xf/0x20
[ 260.720000] [<c01ed92b>] kobject_init+0x2b/0x40
[ 260.720000] [<eedee60b>] i2c_add_adapter+0x12b/0x280 [i2c_core]
[ 260.720000] [<eede5573>] acpi_ec_hc_add+0x122/0x16b [i2c_ec]
[ 260.720000] [<c02ecd1a>] cond_resched+0x2a/0x40
[ 260.720000] [<c0228a3d>] acpi_bus_driver_init+0x28/0x4b
[ 260.720000] [<c0229ad8>] acpi_bus_register_driver+0x82/0xb7
[ 260.720000] [<eec7e00a>] acpi_ec_hc_init+0xa/0x11 [i2c_ec]
[ 260.720000] [<c014416d>] sys_init_module+0x15d/0x1ba0
[ 260.720000] [<c0107a5d>] sys_mmap2+0xcd/0xd0
[ 260.720000] [<c01031f0>] sysenter_past_esp+0x69/0xa9
[ 260.720000] =======================
[ 260.720000] BUG: unable to handle kernel NULL pointer dereference at virtual address 0000000c
[ 260.720000] printing eip:
[ 260.720000] c01b7f4d
[ 260.720000] *pde = 00000000
[ 260.720000] Oops: 0000 [#1]
[ 260.720000] SMP
[ 260.720000] Modules linked in: i2c_ec i2c_core dock button battery container ac af_packet binfmt_misc rfcomm l2cap bluetooth i915 drm re
iserfs dm_crypt dm_mod fuse cpufreq_ondemand p4_clockmod speedstep_lib freq_table usbhid hid gspca videodev v4l2_common v4l1_compat ohci_hc
d pcmcia snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm snd_seq_dummy snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event yenta_socket rs
rc_nonstatic pcmcia_core snd_seq snd_timer snd_seq_device iTCO_wdt snd soundcore snd_page_alloc iTCO_vendor_support intel_agp psmouse agpga
rt shpchp pci_hotplug tsdev evdev ext3 jbd mbcache sg sr_mod cdrom sd_mod usb_storage libusual ata_piix ata_generic libata scsi_mod b44 mii
 generic ehci_hcd uhci_hcd usbcore thermal processor fan fbcon tileblit font bitblit softcursor vesafb capability commoncap
[ 260.720000] CPU: 0
[ 260.720000] EIP: 0060:[<c01b7f4d>] Not tainted VLI
[ 260.720000] EFLAGS: 00210292 (2.6.20-13-generic #2)
[ 260.720000] EIP is at create_dir+0x1d/0x1e0
[ 260.720000] eax: dd2becc0 ebx: dd2becc4 ecx: dd2becc4 edx: 00000000
[ 260.720000] esi: dd2becc0 edi: dd2becc4 ebp: 00000000 esp: e007fdc8
[ 260.720000] ds: 007b es: 007b ss: 0068
[ 260.720000] Process modprobe (pid: 5292, ti=e007e000 task=eaa74a90 task.ti=e007e000)
[ 260.720000] Stack: c0126b6b 00000014 00000000 dd2becc0 dd2becc4 dd2bed06 00000000 c01b884c
[ 260.720000] e007fdec 00000000 dd2becc0 c01ed7ba dd2becc0 c01edac7 e007fe2c 00000020
[ 260.720000] ecfbd2c0 dd2bec48 dd2bed00 dd2bed06 ecfbd248 c02557a7 dd2becc0 c0362eee
[ 260.720000] Call Trace:
[ 260.720000] [<c0126b6b>] printk+0x1b/0x20
[ 260.720000] [<c01b884c>] sysfs_create_dir+0x2c/0x70
[ 260.720000] [<c01ed7ba>] kobject_add+0x8a/0x1a0
[ 260.720000] [<c01edac7>] kobject_set_name+0x37/0xb0
[ 260.720000] [<c02557a7>] device_add+0xb7/0x530
[ 260.720000] [<c01ed50f>] kobject_get+0xf/0x20
[ 260.720000] [<c01ed92b>] kobject_init+0x2b/0x40
[ 260.720000] [<eedee60b>] i2c_add_adapter+0x12b/0x280 [i2c_core]
[ 260.720000] [<eede5573>] acpi_ec_hc_add+0x122/0x16b [i2c_ec]
[ 260.720000] [<c02ecd1a>] cond_resched+0x2a/0x40
[ 260.720000] [<c0228a3d>] acpi_bus_driver_init+0x28/0x4b
[ 260.720000] [<c0229ad8>] acpi_bus_register_driver+0x82/0xb7
[ 260.720000] [<eec7e00a>] acpi_ec_hc_init+0xa/0x11 [i2c_ec]
[ 260.720000] [<c014416d>] sys_init_module+0x15d/0x1ba0
[ 260.720000] [<c0107a5d>] sys_mmap2+0xcd/0xd0
[ 260.720000] [<c01031f0>] sysenter_past_esp+0x69/0xa9
[ 260.720000] =======================
[ 260.720000] Code: f6 ff e8 67 d0 f4 ff eb b6 90 8d 74 26 00 83 ec 1c 89 5c 24 0c 89 cb 89 74 24 10 89 c6 89 7c 24 14 89 df 89 6c 24 18 8
9 54 24 08 <8b> 42 0c 83 c0 70 e8 c8 58 13 00 31 c0 b9 ff ff ff ff f2 ae f7
[ 260.720000] EIP: [<c01b7f4d>] create_dir+0x1d/0x1e0 SS:ESP 0068:e007fdc8
[ 260.720000] <3>ibm_acpi: ec object not found

Revision history for this message
Shaw Terwilliger (sterwill) wrote :

I get a very similar crash (loading ACPI modules dies at kref_get()) on my Mac Pro workstation. 2.6.20-12 did not have this problem. I upgraded all packages at once, so I have the latest ACPI support tools.

I'm back to running 2.6.20-12 with a downgraded nvidia-glx until this gets resolved.

Tim Gardner (timg-tpi)
Changed in linux-source-2.6.20:
assignee: nobody → ubuntu-kernel-acpi
importance: Undecided → High
status: Unconfirmed → Confirmed
Revision history for this message
ooops (ooops) wrote :

I have the same oops. Maybe the following helps to locate the problem. My laptop hangs during boot because 'modprobe sbs' hangs, i.e. the oops is caused by the sbs-module.

Revision history for this message
isazi (isazi) wrote :

Same for me, the boot hangs during "modprobe sbs".
If i press enter i can go after, but the whole system is not so reliable from that moment on.

Revision history for this message
Christian Kellner (gicmo) wrote :

I have the exact same problem, ie stacktrace is identical. My machine is also a Mac Pro. 2.6.20-12 worked. 2.6.20-13 renders the system unusable.

Revision history for this message
soundray (ubuntubug) wrote :

I have the same problem running feisty x86_64 on a Mac Pro. 2.6.20-13.21 fails with a similar dmesg log.

When I boot this kernel and run gnome, I get "Failed to initialize HAL". Networking fails, and Suspend and Hibernate are unavailable from the Quit dialog.

Blacklisting sbs fixes the HAL, network and Suspend/Hibernate problems, but the kernel oops is still in the dmesg log (attached).

When sbs is blacklisted, running 'strace modprobe sbs' freezes at "read(4, "8448 1 asus_acpi, Live 0xfffffff"..., 1024) = 1024 \n nanosleep({0, 100000000}'

None of these problems occur with 2.6.20-12.20.

Revision history for this message
Joacim Haglund (joaha) wrote :

I have a very similar behaviour (perhaps identical). Could it be related to the "**WARNING** I2C adapter driver [] forgot to specify physical device; fix it!", that was in my dmesg from 2.6.20-12? I attach a file with snippets of both outputs for comparison.

Revision history for this message
Harri Kaimio (harri-kaimio) wrote :

I have similar problem as well (in Acer Travelmate 4502):

Boot hangs while loading ACPI modules, but the actula hangup happens when loading module i2c_ec (not sbs as seems to be the case for most people in this thread). After blacklisting i2c_ec boot goes smoothly. And after system is up and running, lsmod shows that ic2_ec is loaded anyway.

Revision history for this message
llpamies (llpamies-deactivatedaccount) wrote :

I attach my error with Acer Travelmate 4000.

Revision history for this message
Piotr Kęplicz (keplicz) wrote :

Also valid here on an Asus M6R. After blacklisting sbs the system boots fine, but I get a similar oops (however the system runs stable).

Revision history for this message
ooops (ooops) wrote :

FWIW this is also present in 2.6.20-14 on an Acer TM3200. Probably the i2c-ec-driver pulled in by sbs (sbs is read via ec)? In the working 2.6.20-12 I get the following warining message. At that point it hangs in 2.6.20-14. If you need further info/tests let me know.

[ 27.576000] **WARNING** I2C adapter driver [] forgot to specify physical device; fix it!
[ 27.576000] ACPI: EC HC smbus [SMBC]

Revision history for this message
Christian Kellner (gicmo) wrote :

Just wanted to let you know that blacklisting the i2c-ec driver helps to make boot work smoothly again *but* as soon as X11 starts (with the nvidia driver) the box freezes. I guess the nvidia driver needs i2c.

Revision history for this message
eis (launchpad-eiselin) wrote :

I'm having the exact same problem on my Acer 1681. I can only boot into the 2.6.20-12 kernel. The other versions hang at "Loading acpi modules".

BTW: the slow booting can be fixed by following the tips in this thread:
http://ubuntuforums.org/showthread.php?t=403008

Changed in linux-source-2.6.20:
importance: High → Critical
Revision history for this message
Onestone (onestone) wrote :

I confirm this. The laptop is an Acer 4502WLMi. Using kernel 2.6.20-12 everything is ok, but with 2.6.20-13 and 2.6.20-14 it fails to boot, hanging at "Loading ACPI modules".

Revision history for this message
SerGieZ (sergio-rodher) wrote :

I have the same problem on a Apple Mac Pro 2Ghz in i386 and amd64 2.20.13-xx and 2.20.14-xx kernels.
All is running in well in 2.20.12 kernel series.

Revision history for this message
Tim Gardner (timg-tpi) wrote :

Would one of you bug subscribers be willing to try an instrumented version of i2c_ec.ko for 2.6.20-14? It has a check that I hope will prevent a crash, and will also output an error message 'acpi_ec_hc_add: Uninitialized device parent.'

cd /lib/modules/2.6.20-14-generic/kernel/drivers/acpi
sudo mv i2c_ec.ko i2c_ec.ko.old
sudo wget http://linux.tpi.com/~rtg/i2c_ec.ko

Reboot and attach the output of dmesg. Thanks.

Revision history for this message
resistor (resistor) wrote :

I tried this on my Mac Pro, and it seems to work well. I've attached my dmesg log.

Revision history for this message
Tim Gardner (timg-tpi) wrote :

resitor: Thanks. Now, lets try to see if the problem is a startup race condition, or if it is relatively constant.

sudo dmesg -c
modprobe i2c_ec
dmesg > dmesg.txt

Please attach the output of dmesg.txt.

Revision history for this message
resistor (resistor) wrote :

I can't do that, because i2c_ec is loaded on boot. modprobe'ing it again does nothing.

Revision history for this message
Martin Kuehn (mkuehn) wrote :

Your fix for module i2c_ec.ko also works on my Acer 4001 WLMi.
See file "dmesg after booting 2.6.20.14".
After that I removed modules sbs and i2c_ec and modprobed sbs again.
Here is the output of dmesg:
[ 2345.200000] acpi_ec_hc_add
[ 2345.200000] acpi_ec_hc_add: Uninitialized device parent.

Revision history for this message
Tim Gardner (timg-tpi) wrote :

The committed fix will keep the module from faulting. However, it doesn't solve the problem wherein the ACPI I2C bus device fails to initialize.

Changed in linux-source-2.6.20:
status: Confirmed → Fix Committed
Revision history for this message
franzb (franz-burgmann) wrote :

The fixed module by Tim Gardner doesn't work for me, nothing changes at all, boot process stopps with the message "Starting ACPI services...".

Feisty works fine with either the kernel 2.6.20-14-generic in recovery mode or with 2.6.20-12-generic in standard mode. The mainboard is an Asus P4C800-E, CPU Pentium-M 760.

franz

Revision history for this message
Tim Gardner (timg-tpi) wrote :

franzb: Can you boot using kernel 2.6.20-14-generic in recovery mode and attach the output of dmesg? Please use the module I've supplied since it ought to prevent a kernel fault.

Revision history for this message
franzb (franz-burgmann) wrote :

sure!

franz

Changed in linux-source-2.6.20:
assignee: ubuntu-kernel-acpi → timg-tpi
Revision history for this message
ooops (ooops) wrote :

On a Acer TM3200 booting in emergency mode will not trigger anything since i2c_ec is pulled in later. In any case, emergency-booting and doing a manual 'modprobe i2c_ec' will produce the uninitialized parent message. I guess the i2c-bus is not initialized since a following 'modprobe sbs' returns after a few seconds but fails to produce any results. No kernel-message, no /proc/acpi/battery entry. So the i2c-bus is not working. Here is what I can see of the embedded controller:

$ cat /proc/acpi/embedded_controller/EC0/info
gpe: 0x1d
ports: 0x66, 0x62
use global lock: no

Revision history for this message
Brazovsky (brazovsky) wrote :

In my case the committed fix works, now i can boot from kernel 2.6.20.14.
But the boot takes more 5 or 6 seconds than the kernel 2.6.20.12

i leave my dmesg here, maybe it will help you.

Revision history for this message
Tim Gardner (timg-tpi) wrote :

Shall we continue the fishing expedition? Please get this updated kernel in the hope that it improves your ACPI/I2C behavior:

cd /boot
sudo mv vmlinuz-2.6.20-14-generic vmlinuz-2.6.20-14-generic.old
sudo wget -O vmlinuz-2.6.20-14-generic http://linux.tpi.com/~rtg/bzImage

Reboot and attach _all_ of the output from dmesg. This may those of you that were getting the 'Uninitialized device parent'.

Revision history for this message
Jason Straight (jason-jeetkunedomaster) wrote :

I've been trying to figure out more about why hald hangs on start: https://bugs.launchpad.net/ubuntu/+source/hal/+bug/92647

It seems that my problems with the 64 second hald hang on bootup are battery and thermal related with the kernel, modprobing battery takes a good 10-15 sec, thermal takes less but more than it should.

If you look at my hald log from the above mentioned bug you can see that it hangs when handling battery and other ACPI events.

I am running a vanilla 2.6.20.4 kernel, and I have this issue with that as well.

Revision history for this message
franzb (franz-burgmann) wrote :

dmesg output with the updated kernel in recovery mode. Still the same behavior (booting only with 2.6.20-14-generic in recovery mode or with 2.6.20-12-generic in standard mode)

franz

Revision history for this message
Martin Kuehn (mkuehn) wrote :

dmesg output with the updated kernel in standard mode.
Nothing's changed, I had to log in with ctlr-alt-f1 to save the ouptput of dmesg.
Theres annother strange effect in all 2.6.20-xx kernels including 2.6.20-12:
my ide harddisk is now /dev/sda. It should be /dev/hda.

mmk@penguin:/$ cat /etc/mtab
/dev/sda7 / ext3 rw,errors=remount-ro 0 0
proc /proc proc rw,noexec,nosuid,nodev 0 0
/sys /sys sysfs rw,noexec,nosuid,nodev 0 0
varrun /var/run tmpfs rw,noexec,nosuid,nodev,mode=0755 0 0
varlock /var/lock tmpfs rw,noexec,nosuid,nodev,mode=1777 0 0
procbususb /proc/bus/usb usbfs rw 0 0
udev /dev tmpfs rw,mode=0755 0 0
devshm /dev/shm tmpfs rw 0 0
devpts /dev/pts devpts rw,gid=5,mode=620 0 0
/dev/sda1 /media/Acer vfat rw,utf8,umask=007,gid=46 0 0
/dev/sda5 /media/Daten vfat rw,utf8,umask=007,gid=46 0 0
/dev/sda2 /media/System vfat rw,utf8,umask=007,gid=46 0 0
binfmt_misc /proc/sys/fs/binfmt_misc binfmt_misc rw 0 0

Revision history for this message
Tim Gardner (timg-tpi) wrote :

Gentlemen - Thanks for your help. Given the information that I've acquired from your efforts, I've decided that the ACPI I2C problems that you are having will not get fixed for this release. The ACPI subsystem has undergone serious architectural changes upstream. Hopefully a future release will fix these problems. In the meantime, to run Feisty set 2.6.20-12 as your default kernel.

Tim Gardner (timg-tpi)
Changed in linux-source-2.6.20:
assignee: timg-tpi → ubuntu-kernel-acpi
importance: Critical → Medium
Tim Gardner (timg-tpi)
Changed in linux-source-2.6.20:
status: Fix Committed → Confirmed
Revision history for this message
franzb (franz-burgmann) wrote :

Thank you for the information. I don't know how may users will be affected by this problem, but wouldn't it be better to release feisty with kernel 2.6.20-12 since it is not very nice to have a not booting live-cd on one side or a fresh installed but not booting system in the case of the text installation on the other side in a final release. Besides, at the moment the kernel 2.6.20-12 is not available on the repos anymore.

franz

Revision history for this message
Shaw Terwilliger (sterwill) wrote :

I share the same concerns as franzb. This bug could affect a large number of computers out there, and Ubuntu's reputation will suffer when the latest release doesn't even boot on many of the machines where older releases worked correctly. This was a problem with Edgy compared to Dapper for many people (luckily not me).

The number of people who have commented on this bug is small, but how many others are waiting to get hit during the next upgrade cycle?

Also, short of making sure 2.6.20-12 ships as an alternate kernel option in Feisty, I don't know how I could even install it on affected machines without making my own boot media, installing Feisty, then manually replacing the installed kernel with the old packages. And even then, I'd be prevented from upgrading any proprietary module packages indefinitely (nvidia-glx, in my case) until Ubuntu fixes their kernels.

This is a pretty disappointing resolution.

Revision history for this message
Tim Gardner (timg-tpi) wrote :

The number of users affected by this bug appear to be few, but its annoying nonetheless. Kernel releases subsequent to 2.6.20-12 fix a host of other problems which do affect a great many people, which is why we cannot release 2.6.20-12. I also asked how to get the 2.6.20-12 kernel:

"if they can manage to install, then they can download it from the "builds of" thing at the top left of https://launchpad.net/ubuntu/+source/linux-source-2.6.20/2.6.20-12.20 - e.g. https://launchpad.net/+builds/+build/312897 -> Resulting binaries -> https://launchpad.net/+builds/+build/312897/linux-image-2.6.20-12-generic -> http://librarian.launchpad.net/6895136/linux-image-2.6.20-12-generic_2.6.20-12.20_i386.deb"

In summary, do this:

wget http://librarian.launchpad.net/6895136/linux-image-2.6.20-12-generic_2.6.20-12.20_i386.deb
sudo dpkg -i linux-image-2.6.20-12-generic_2.6.20-12.20_i386.deb

Revision history for this message
eljefe6 (jesse-anderson) wrote :

"if they can manage to install" should say "if they can manage to boot the installer". I of course haven't tried that, but I believe they will not even be able to boot the Release installer. The instructions you put require a bootable kernel from the Release installer.

In order to install, someone would have to download the beta Feisty release with 2.6.20-12 or earlier and update everything but the kernel.

At my office we are all using Mac Pros and plan on having all of them triple boot. Right now, I am the only one triple booting. The rest are waiting on the full release of Feisty.

As of a few months ago, Feisty was the only distribution that worked remotely well on a Mac Pro. This will help Ubuntu gain quite a bit of market share.

This issue will affect all Mac Pro and some Acer users. For this reason, I urge you to reconsider this decision.

Revision history for this message
Jason Straight (jason-jeetkunedomaster) wrote :

As far as the batt, thermal and ac module hangs it doesn't look good for a kernel fix too soon. I had the same issues with 2.6.21-rc5 with mm patches as well.

Revision history for this message
franzb (franz-burgmann) wrote :

As mentioned, with the alternate cd it is possible to get through the whole installation process without any problem, but afterwards the boot process of the installed system stopps with the message "Starting ACPI services...". The live-cd doesn't boot properly, so an installation can't be accomplished.

As Shaw Terwilliger emphasizes, a behavior like this, even if hopefully not many users will be affected (if eljefe6 is right, then there will be a lot out there), could be detrimental for Ubuntu's reputation and I consider therefore to find a solution with which also unexperienced users will be able to run Feisty.

Franz

Revision history for this message
Shaw Terwilliger (sterwill) wrote :

Tim,

Thanks for the info about the Launchpad builds. Because many people having these ACPI problems will also be using the proprietary nvidia modules (most or all Mac Pro users), or other packages tied to the installed kernel version, we'll be effectively tied to old versions of packages that cross from the kernel into user-space. This will keep us from upgrading any packages higher on the dependency chain, the most common and most commonly changed of which is X. For this reason, we're basically stuck using Feisty beta packages from March 2007 indefinitely for anything tied to the kernel.

Revision history for this message
ooops (ooops) wrote :

Find attached the ooops generated using the kernel you provided above (https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/96480/comments/26) . As already told before this is booting the emergency target and manually pulling in the respective acpi modules with 'modprobe sbs'. Can you point me to the patches (git?) between 12 and 13 that affect i2c-ec? I can burn some time and cpu-cycles here to debug the problem.

Finally, I agree to the previous posters. Most probably, the bug will affect all Acer TM that have an sbs connected via an ec-controller. Introducing such a regression _after_ beta-release is not a good PR for ubuntu.

Revision history for this message
Brazovsky (brazovsky) wrote :

I have to say that I'm really disappointed with the decision because in lot of forums people are talking about this problem, and this isn't just a bug, this is a critical bug, lot of people haver acer tm laptops, mac pros, and i cant imagine in reviews people talking that Ubuntu didn't boot in their pcs.

And putting this bug now has medium is hide the big problem, noting more.

But thanks anyway Tim for trying :)

Revision history for this message
Gonçalão (gonvelho) wrote :

Installed Feisty 2 hours ago (kernel 2.6.20.14).
Same problem with ACPI stop.
I have a Acer Travelmate 4002 WLMI
I just can't believe that such a bug will be

Revision history for this message
Joacim Haglund (joaha) wrote :

For reference, mine isn't an acer or mac, it's a Packard Bell EasyNote A8 (pentium-m 750). It has its closest relatives in the A8 series and among some Benq systems, I believe (from alsa efforts with the audio codec).

Revision history for this message
chris (chris-evers) wrote :

I've got exactly the same problem on my Toshiba Satellite L10
Intel Celeron M 360 / 1.4 GHz with Intel 852GM chipset 108

Revision history for this message
Christian Kellner (gicmo) wrote :

So you are basically telling me that "serious architectural changes" have been backported from 12 -> 13 about 3 weeks before release during stabilize phase? It's the first time that I have a non bootable kernel since my slackware times long ago. Also how can we be so sure it only does affect "a few" people since that change was only 3 weeks ago?! First no response from ubuntu kernel folks for 2 weeks and then lowering the prio to medium and tell us we should just stick with -12 (forcing me to old nvidia-glx as well)? Why do you vendor patch what should be a stable kernel if you *seem* to not have a clue whats going on upstream. I appreciate all you efforts and I *love* ubuntu but this all really leaves me totally disappointed.

Revision history for this message
eljefe6 (jesse-anderson) wrote :

I thought the same thing as Christian. Wasn't the beta "freeze" designated so that only bug fixes were committed?

Revision history for this message
Onestone (onestone) wrote :

It's very disappointing to see this bug demoted to Medium severity. Don't underestimate the number of affected users. Feisty shouldn't be released in this state. IMHO the right decision is to make 2.6.20.12 the default version, and provide 2.6.20.14 in the repositories for whoever needs it.

Revision history for this message
Ralph (ralph-puncher-deactivatedaccount) wrote :

I have exactly the same problem (Bug# 98821)on an Acer Aspire 1414WLCi. I'm not happy staying on -12. With i2c-ec blacklisted the only problem that I have noticed I am experiencing with -14 is the inability to monitor the battery (cannot see it at all) and the power monitor thinks AC power is connected all the time even when it is disconnected.

Revision history for this message
Jason Straight (jason-jeetkunedomaster) wrote :

I belive the problem I have with the long delay loading some acpi modules is related to this, it also affects the 64 second delay loading hald, and other applications which use ACPI, it screws up the ability to suspend and resume correctly, because the scripts don't seem to finish before the system suspends, then things don't come back in the right order, and the machine resuspends because it thinks batt ran out our lid was closed, etc.

I just compiled my own kernel 2.6.19 to get away from it, but I have to say I was on the boat that this problem was a disgrace to allow feisty to release like this, I imagine there are a lot of laptops out there that are going to be affected, if not by a totally unbootable or uninstallable bug, then by a bug which makes using a laptop with feist on it a real PITA.

1. Hard release dates are stupid - it should be final when it's ready
2. Bleeding edge components and packages come at a price where there just hasn't been enough time to test against a broad range of users.

Revision history for this message
Josué Alcalde González (josuealcalde) wrote :

This bug affects all Mac Pro and a lot of Acer Laptops. There are also packard bell and toshiba laptops affected. The affected computers won't be able to boot a instalation cd and only expert users will be able to use feisty in that laptops.
There will be a lot of people who won't be able to install feisty and that is the worst of the bugs.
Perhaps it is not posible correct this bug in time, but I can't understand why it is not a critical bug.

Revision history for this message
Matthew Garrett (mjg59) wrote :

db2f0f088a056c4ccf9054747169802db2f9ae9a introduced a hunk into i2c_ec.c that isn't valid with 2.6.20's ACPI. Just delete the single line it adds to i2c_ec.c and it'll be reverted to the 2.6.20-12 level of functionality.

Revision history for this message
Matthew Garrett (mjg59) wrote :

Bug severity hasn't changed

Changed in linux-source-2.6.20:
importance: Medium → Critical
Revision history for this message
Brazovsky (brazovsky) wrote :

@Matthew Garrett

Is that a fix? I don't know if you are from the kernel team our if you just trying to fix the bug, but we all appreciate you're effort

Revision history for this message
Tollef Fog Heen (tfheen) wrote :

This needs to be fixed for 7.04; milestoning.

Revision history for this message
Patrick (patrick-voegeli) wrote :

Toshiba Satellite L10 Celeron M 1,5Ghz, intel 852gm, same problems with both -13 and -14. Vanilla 2.6.20.6 kernel works fine. Can I do something? I've read some workarounds, but maybe you need some other info now..

Patrick

Revision history for this message
SerGieZ (sergio-rodher) wrote :

It looks that the problem is afecting to an undeterminated number of users/machines. I notice the problem in the Mac Pro but not in the Toshiba A40 Pentium 4 Mobile. I can install Feisty Herd 5 and update everything but the kernel, but problem with sound reported at bug 89980 will still remains. In fact we don´t know if -13 and -14 kernels solve the sound problem due to this acpi bug. Some people has no problems (no acpi no sound bugs) with 2.6.21 vanilla kernel.

I want to offer my support in all I can help to try to solve this problem before the final release.

Thank´s to Tim and all the ubuntu guys for their great efford.

Changed in linux-source-2.6.20:
assignee: ubuntu-kernel-acpi → ben-collins
importance: Critical → High
status: Confirmed → Fix Committed
Revision history for this message
Brazovsky (brazovsky) wrote :

the update from this morning didn't fix the acpi problem in my case.
I have a acer tm 4002wlmi.

Can i do anything to help fixing this? if the answer is yes, just tell what you want to do in my laptop.

Matt Zimmerman (mdz)
Changed in linux-source-2.6.20:
importance: High → Critical
Revision history for this message
Tim Gardner (timg-tpi) wrote :

OK, lets revert to 2.6.20-12 behavior with respect to ACPI and I2C. This version of the I2C module requires that you boot the original 2.6.20-14 kernel, _not_ the test kernel mentioned in https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/96480/comments/26. It is also a preview of the fix that will appear in the kernel release candidate.

cd /lib/modules/2.6.20-14-generic/kernel/drivers/acpi
sudo mv i2c_ec.ko i2c_ec.ko.old
sudo wget -O i2c_ec.ko http://people.ubuntu.com/~rtg/i2c_ec.ko.2.6.20-14.platform_device

Please attach the full output of dmesg if you continue to experience problems.

Revision history for this message
ooops (ooops) wrote :

I get the new 'unitialized parent' instead of the old 'forgot to specify physical address' message. Consequently the acpi-module using i2c, namely sbs, is not working. Everything else is okay as is it was with your last i2c_ec. Are you sure you reverted also your intermediate change?

This is on a Acer TM3200 with 2.6.20-14.22. Cannot use 23 because of a bug and 24 is not available yet.

Revision history for this message
franzb (franz-burgmann) wrote :

Hello,

Unfortunately the situation didn't change yet. Attached you'll find the dmesg-output.

Franz

Revision history for this message
Patrick (patrick-voegeli) wrote :

The kernel boots fine here, but I don't get battery monitor..

Revision history for this message
Tim Gardner (timg-tpi) wrote :

I hate hard links. I'll repost the i2c_ec.ko module in a few minutes. Please try it again when it appears.

Revision history for this message
Patrick (patrick-voegeli) wrote :

Thanks for your great work ubuntu devs! New kernel -15 works fine here: SBS battery monitor works and the kernel boots fine.

Revision history for this message
Shaw Terwilliger (sterwill) wrote :

Patrick, are you using Tim's i2c_ec.ko module or the one that's in 2.6.20-15.24 to boot successfully?

Revision history for this message
Ben Collins (ben-collins) wrote :

Available in 2.6.20-15.25

Changed in linux-source-2.6.20:
importance: Critical → High
status: Fix Committed → Fix Released
Revision history for this message
Brazovsky (brazovsky) wrote :

It works!!!
My acer tm 4002wlmi is working fine!
Thanks for all, you are the best!

Revision history for this message
Onestone (onestone) wrote :

Kernel 2.6.20-15 fixed the issue. Thanks.

Revision history for this message
Martin Kuehn (mkuehn) wrote :

Just install the meta package linux-generic from the repos. It pulls in the linux-imasge and restricted modules for kernel 2.6.20-15.24.
It works fine with my Acer TM 4001 WLMi (battery-status, CPU frequency etc)

Revision history for this message
eljefe6 (jesse-anderson) wrote :

2.6.20-15 is working for me on the Mac Pro. Thanks.

Revision history for this message
eis (launchpad-eiselin) wrote :

2.6.20-15 is working for me too, on my Acer 1680. Great work, thanks!

Revision history for this message
Josué Alcalde González (josuealcalde) wrote :

It has been fixed in Acer TM 4001 WLMI. Good work. Thanks!

Revision history for this message
SerGieZ (sergio-rodher) wrote :

I can re-confirm that 2.6.20-15 works on Mac Pro. Thank you very much!

Revision history for this message
Jason Straight (jason-jeetkunedomaster) wrote :

Crap, my apci issues (the delays) are still there with -15, I'm going to file a new bug report for the delays.

Revision history for this message
Gonçalão (gonvelho) wrote :

My 4002WLMI is working great. I also noticed a major improvement in the battery life. Great work guys.
And I even didn't had to do nothing. The update manager fixed everything including the latest ubuntu kernel.
I have managed to convince some friends to change to Ubuntu. Now I think there are even more reasons.
Congrats!!

Revision history for this message
franzb (franz-burgmann) wrote :

For me the new kernel isn't working, still the same problems. That might have to do with my rather unusual hardware: Asus P4C800-E with Asus ct-479, Pentium-M 760.

But I'm glad to hear, that for many others the problem seems to be solved.

Franz

Revision history for this message
Ralph (ralph-puncher-deactivatedaccount) wrote :

2.6.20-15 is working for me too, on my Acer Aspire 1414WLCi. Great work, thanks!

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.