regression: suspend does not work on hp nw8240 with feisty

Another point: when machine hangs, fan is working like mad - it looks like the CPU load is high.

Please retest against 2.6.20-2 when it is available in the feisty archive.

Problem is still there on 2.16.20-2.

I can still reproduce this with 2.6.20-4-generic.

Since suspend was first fixed in dapper for this laptop (at least for console mode and fglrx), I have looked back for old bug. In bug 33827, the change that made the difference was introduced around 2006-04-30, as you can see from Matthew's comment (and my confirmation).

I have looked back to old change log from just before Apr 30: http://changelogs.ubuntu.com/changelogs/pool/main/l/linux-source-2.6.15/linux-source-2.6.15_2.6.15-21.32/changelog

And it looks to me that this might be related to this change by mjg59:

  * Fixes for swsusp with highmem, mjg59.

I now recall discussion with mjg59 on IRC when this was seen as issue with HP nw/nx laptops with 1GB RAM. Is it possible that this patch was dropped in Feisty?

The point is, that during version 5.04 I think, an IDE patch was included in the Ubuntu kernels (it's not in the official kernel, or at least was not for a long time), that calls _STM, _GTM and the like (I don't know what these commands exactly do?) to restore the IDE disk state after resuming from suspend to RAM, since many HP notebook BIOSes do not do this automatically.
I just tested the Feisty kernel and experienced exactly the same problem with my NX8220 (=NC8230) notebook. This is not a bug in the graphics driver resume code or some highmem things!
It looks to me as this patch was not yet incorporated into the new libata driver. Would be nice if you could port it if its not already available as a patch, since this would really make Feisty a no-option for me.

Sorry, I forgot: I tested using linux-image-2.6.20-8-generic.

I can reproduce this on an HP Pavilion dv8000. Exact same symptoms. I do have a broadcom 4318 using driverloader.

Thanks for reporting this bug. To make any progress with it, some more information will be needed. Please add (by commenting on the bug) to the bug report the output of the following commands: 'uname -a' 'sudo lspci -vvn' 'sudo dmidecode'. It would also be helpful if you could try to hibernate/suspend and after that fails, restart your system and attach /var/log/kern.log.0 as well. Thanks again for your contribution.

This bug only effects suspend to RAM, not hibernation, since on hibernation the BIOS reinitalizes the IDE controller and hard disks. By the way I found the patch, or at least a patch that does the same thing: https://sourceforge.net/mailarchive/forum.php?thread_id=7329939&forum_id=6102

I don't have the new kernel installed at the moment, so no uname -a from me...

and the dmidecode output...

Attached is the result of `sudo dmidecode`

`uname -a` yields: Linux larry 2.6.20-8-generic #2 SMP Tue Feb 13 01:14:41 UTC 2007 x86_64 GNU/Linux

Here's my lspci -vvn

Please read and try the suggestions contained in :

https://wiki.ubuntu.com/KernelSuspendDebugging

From dmesg:

[ 14.294210] ACPI: (supports S0 S3 S4 S5)
[ 14.294248] Magic number: 0:150:167
[ 14.294250] hash matches drivers/base/power/resume.c:28
[ 14.294283] hash matches device ttyq7

I just found this patch: http://lkml.org/lkml/2007/2/17/211
Will compile my own kernel and try it now...

Okay, the patch already was in the kernel source (release 2.6.20-9) and it is also enabled in linux-image-2.6.20-9-generic. However it does not seem to help for this problem.
I also disabled the libata_piix driver completely, so the PATA only piix was used again, but resuming still did not work.

The following procedure fixed the problem for me:
1. Disable the libata_piix kernel module (CONFIG_ATA_PIIX)
2. Apply the patch from http://marc.theaimsgroup.com/?l=linux-ide&m=116916643921306 , one line of the patch fails, it can easily be patched manually (just a line in the IDE makefile)
3. Activate the new CONFIG_BLK_DEV_IDEACPI kernel option
4. Compile and install the kernel and create a matching Initrd (update-initramfs)

Standby now works like before and the new kernel finally fixes bug #21574 (speakers do not work after suspending).

Unfortunately I have no idea how this could make its way into Feisty. Disabling libata_piix is probably not an option, since it also covers Intel SATA controllers, which are not covered by the plain old piix driver. Still, the IDE ACPI patch should be added! This (or a similiar) patch has been in Ubuntu kernels for some time now and I cannot understand why it was not applied to the new Feisty kernels anymore.
The right way probably is to fix this libata ACPI code. I think that there is some function that is needed on this HP notebooks during resume, that the new libata ACPI functions do not call yet. However I don't think that this could make its way into Feisty, could it? Unfortunately I don't have the time to get into kernel development in the moment :-(

I have a similar problem with feisty on an HP TC4200 Tablet PC (see bug #89779). I followed the procedure above (disabling the ata_piix module, applying the ide_acpi and rebuilding the kernel). I can now suspend and resume successfully with the new kernel.

As an aside, I had to apply the patch manually (too many hunks failed). I downloaded the kernel sources this morning, so they should be the latest. I was running 2.6.20-9-generic, but it looks like the kernel sources are 2.6.20-10 (or something; the new package built is called linux-image-2.6.20-ubuntu1.20070306_2.6.20-ubuntu1.20070306-10.00.Custom_i386.deb). This is my first time rebuilding an Ubuntu/Debian kernel, so I'm not sure of the naming convention (although I used --add-revision 20070306 when I built it). In any event, I've attached the patch that applies cleanly to the current sources.

I've got the same issues on an HP/Compaq nc6220 so this issue is looking like it affects a lot of HP laptop models.

2.6.20-9 with the above patch (which installed perfectly btw, thanks!!) and with CONFIG_ATA_PIIX disabled has resulted in perfectly functioning suspend. Also much like one of the above posters I now have working sound after suspend, neither Dapper nor Edgy could boast that!

Just to confirm that I'm seeing the same issue here on another NX8220.

Simple patch instructions for this bug and Bug #89779 can be found in Bug #75398.

This fixed NC4200 and NC6230 laptops for me.

tarek : )

I have removed duplicate link to bug 75398, as the 2.6.20-12 does not fix the bug for me.

There is visible improvement in the sense that HDD led is not on all the time and CPU is not under high load after resume, but I end up with just blank screen - no way to switch consoles, Ctrl-Alt-Del does not work... Caps lock and Num Lock keys turn on the led, therefore I suspect kernel works, just video is not properly restored.

I have tried Suspend from X (KDE) and console (no kdm/X running), but the same result. I have logged the execution of "sudo pmi action suspend" but nothing useful there. See attachment.

Similar problem also reported in bug 89779 (only on different HP laptop).

It looks like that we got most of needed patches in Ubuntu kernel from bug http://bugzilla.kernel.org/show_bug.cgi?id=7122, however there is one patch mentioned in this bugs in comments #37 #38 #41 #54 #63 #64, which does not seem to be applied yet to Linus's kernel, not Ubuntu kernel. Patch is attached in comment #38 and I will try it on my system.

With latest 2.6.20-14 kernel, suspend and resume works for me on HP nw8240.

All others that have reported this as "me too": can you install 2.6.20-14 kernel and try to reproduce. If this is still problem for your laptop, please open new bug with full info about the laptop (dmesg, lspci -vv) and reference this bug just for link.

Also confirmed that 2.6.20-14 fixes suspend / resume on the NC4200.

tarek : )

2.6.20-14 kernel appears to fix the hang on resume on my nx8220. I say appears as I still haven't figured out how to get a working display, but I can successfully run a "sudo reboot" blind so I assume that suspend/resume is fundamentally working.

2.6.20-14 also confirmed working on an HP NC6230

I hate to rain on the parade, but suspend STILL doesn't work on an
emachines M6810 with kernel 2.6.20-14-generic x86_64

This is feisty daily build 20070405, with the -14 kernel installed by
adept updater.

Richard: please open new bug. It seems that bug is fixed for (most/all) HP laptops and since you have HW from different vendor it makes sense to have separate bug.

See https://wiki.ubuntu.com/DebuggingKernelSuspend

My nc6220 is now working perfectly with 2.6.20-14. Awesome!! Thanks everyone who helped make it happen!

Oops, spoke too soon... With 2.6.20-14 the audio problem a few posters mentioned has come back - after resume there is no built-in sound. This isn't quite the same issue as the above but I doubt it's entirely coincidental that they're related.

For the audio problem see https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.15/+bug/21574 & https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/151111 .