[Hardy] [regression] ehci_hcd.ko breaks suspend-to-ram

Hi Petr,

Per the kernel team's bug policy, can you please attach the following information. Please be sure to attach each file as a separate attachment.

* cat /proc/version_signature > version.log
* sudo lspci -vvnn > lspci-vvnn.log

Also, can you attach your dmesg output (dmesg > dmesg.log) after an attempted suspend/resume cycle as outlined here: https://wiki.ubuntu.com/DebuggingKernelSuspend .

For more information regarding the kernel team bug policy, please refer to https://wiki.ubuntu.com/KernelTeamBugPolicies . Thanks again and we appreciate your help and feedback.

(but the problem was alredy present on kernel 2.6.24-13)

Of course, here it is.
(but the problem was alredy present on kernel 2.6.24-13)

Here is lspci, the dmesg will take me little more time.

Here is dmesg.

I tried to unload some modules from kernel before suspending, and I find out, that the problem is in ehci_hcd module (after unloading this and only this module, the resume works again).

I can confirm this bug on a Thinkpad T61 8889-2BG with Hardy 2.6.24-15. On resume, the backlinght turns on but nothing more happens. With 2.6.24-12 resume worked fine although, sometimes, instead of the password dialog a white screen was displayed.
After removing ehci_hcd resume worked again in 2.6.24-15 (many thanks for the hint). Unfortunately, blacklisting this module in acpi-support did not help as Hardy does not use this file. Does anybody know where I can blacklist this module in Hardy?

@Karl Martin

AFAIK Hardy replaces (to some extend) acpi-support with pm-utils. You might use pm-utils config files to blacklist ehci_hcd. See http://en.opensuse.org/Pm-utils#Configuration

I've got a T61 myself -- but cannot really confirm this bug. I can confirm the blank screen after resuming with kernel -12, though.

@Fabian Neumann

Thank you very much for the information. I just found out that resume is working fine when nothing is connected to the usb ports. However, if I connect my hub with mouse, keyboard and dvd burner to the usb port resume does not work anymore. Even with these devices connected resume was working fine in 2.6.24-12.

Unfortunately, kernel version 2.6.24-15 does not solve the other bug mentioned above displaying sometimes a white screen instead of the password dialog after resuming. I am wondering whether this bug is related to the nvidia driver or to X. Do you have a NVIDIA card too?

Yes, I think this is a bug in the binary nvidia driver. Just search for "ubuntu nvidia blank screen". Cannot name a bug, there're probably many of them...

But I mixed some things up. Resume works for me, but quick-user-switching gives me a blank screen. Haven't tested it on all kernel configurations, as I don't need it that often.

No, I have intel card. It seems to me, tha this bug is issue only of the ehci_hcd module.

I think that the bug reported initially in this thread is caused by the ehci_hcd module.

The other bug that I mentioned seams to be different from the one mentioned by Fabian Neumann. I do not see a black screen after resume but a white screen where the mouse pointer is correctly displayed and can be moved. However, the screen is otherwise completely white and no password dialog is displayed. I can then enter the password blindly and after that the desktop is shown correctly. As the mouse pointer is shown correctly on the white screen I am not sure that this bug is related to the nvidia driver. Perhaps there is a problem with the initialization of X.

hi i can confirm this bug on a clevo m72r. the workaround unloading the ehci_hcd module works too for me.

With 2.6.24-16 is it still broken.

confirm that this is still broken in 2.6.24-16

USB ports not working after resume on Dell Inspiron E1705/9400 with 2.6.24-16.

version.log:
Ubuntu 2.6.24-16.30-generic

Fix mentioned by Fabian Neumann seems to work. http://en.opensuse.org/Pm-utils#Configuration

sudo nano /etc/pm/config.d/config

SUSPEND_MODULES="button ehci_hcd"

Reboot, and it seems to work pretty well.

I can confirm this bug with Ubuntu 2.6.24-16.30-generic on a Samsung X20 laptop. The workaround mentioned, i.e. putting the line

SUSPEND_MODULES="ehci_hcd"

in some file in /etc/pm/config.d/ works, however only when suspending using the kde-guidance-powermanager applet. It doesn't work when pressing the sleep button. Probably the sleepbutton-event is still handled by the old scripts in /etc/acpi ?!?

Same kernel, same problem on my Lenovo 3000 V200 notebook.

c_t: "putting the line SUSPEND_MODULES="ehci_hcd" in some file in /etc/pm/config.d/ works"

Yep, works fine for me :)

I would suggest a file name like "/etc/pm/config.d/WORKAROUND" to remove this file later manually. Here my one-liner solution:

sudo echo SUSPEND_MODULES=\"ehci_hcd\" > /etc/pm/config.d/WORKAROUND

I don't know why but since today's upgrade suspend/resume works fine for me even *without* removing the ehci_hcd module... Although it doesn't seem like linux-image has been upgraded today :/... Can anybody confirm that suspend is working again?

It is still broken for me (2.6.16). Had you some usb devices plugged-in before suspending?

I even suspended my laptop using
$ echo mem > /sys/power/state
(so no pm-utils or other stuff should interfere) and all my usb devices (keyboard, mouse, external hd) are working perfectly (before and after suspend). I have no idea why this is... I have no files in /etc/pm/ at all and MODULES="" in /etc/default/acpi-support .

$ cat /proc/version_signature
Ubuntu 2.6.24-16.30-generic

When I have USB devices plugged into my IBM Advanced Mini Dock, it breaks suspend unless I have ehci_hcd in MODULES. However, if I do add ehci_hcd to MODULES, the USB ports on my dock become unavailable forevermore after a suspend.

Ok, it's broken again for me too. I can't explain, why it worked for some days, but since upgrading to 8.04 final and rebooting ehci_hcd must be blacklisted again in order to get usb devices working after resume.

$ cat /proc/version_signature
Ubuntu 2.6.24-16.30-generic

I can confirm this bug on a Clevo M720R, C2D 2.4 GHz, Intel 965GMA onboard graphics.

I may have just been getting lucky with the unload of ehci_hcd, because resume just failed again with USB devices plugged into my dock.

The 'SUSPEND_MODULES="ehci_hcd"' works well for me unless I'm running a virtual machine. So as part of my suspend I need to suspend all of them as well first.

The behavior have not changed with 2.6.24-17-generic :( The workaround is still necessary.

i confirm this problem

using solution from Alex Strabala (https://bugs.launchpad.net/ubuntu/+source/linux/+bug/211572/comments/19)
everything works perfect

Confirming the problem. System Dell XPS m1330 Nvidia 8400M GS. 2.6.24-16-generic.
Suspend worked OK, but there was no USB after resume.
Proposed fix (https://bugs.launchpad.net/ubuntu/+source/linux/+bug/211572/comments/19) worked fine.

Confirming the problem.

this is working for me:
http://ubuntuforums.org/showthread.php?t=782782

hp nx6125
cat /proc/version_signature
Ubuntu 2.6.24-16.30-generic

I updated to Intrepid, and it seems to be fixed in "kernel 2.6.26-1.5".

Still present in 2.6.24-19. If anything is plugged into my USB hub, the resume from suspend process hangs at a blank screen with a blinking cursor and does not respond. As others have mentioned unloading ehci_hcd before suspending fixes the problem.

Petr, thanks for testing Intrepid and the feedback. I'm going to mark this "Fix Released" against Intrepid. I'll open a nomination for a Hardy stable release update, but the likelyhood of a SRU ( https://wiki.ubuntu.com/StableReleaseUpdate ) at this point it not very high I don't think. Thanks.

This problem also affects my Toshiba Tecra M3. Suspend/resume worked under Gutsy but fails with all release Hardy kernels and the current Intrepid kernel. They would resume to a blank screen without accepting any input necessitating a hard reboot.

Adding this to /etc/pm/config.d/config:

SUSPEND_MODULES="ehci_hcd"

fixes the problem. LSPCI output is attached. /proc/version_signature is "Ubuntu 2.6.27-7.16-generic"

I found this out using the pm_trace function, the relevant portion of dmesg output is:

[ 1.851860] Magic number: 0:250:259
[ 1.851865] hash matches /build/buildd/linux-2.6.27/drivers/base/power/main.c:350
[ 1.851884] tty ttyx1: hash matches
[ 1.851958] pci 0000:00:1d.7: hash matches
[ 1.852017] rtc_cmos 00:08: setting system clock to 2068-04-03 03:15:29 UTC (3100648529)

I presume this bug is the correct place to file the issue, if anyone wants to report under Intrepid please do so :).

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.

The bug doesn't appear any more (in fact, for long time now). Feel free to open it again, if somebody is experiencing it.