Description of problem: Since about kernel 1622 or so, 'echo mem>/sys/power/state' no longer suspends to memory. The log reports the following kernel: Stopping Tasks: ======================================================== kernel: stopping tasks failed (1 tasks remaining) kernel: Restarting tasks...<6> Strange, kauditd not stopped kernel: done Version-Release number of selected component (if applicable): 2.6.13-1.1624_FC5 How reproducible: every time Steps to Reproduce: 1.at root prompt issue 'echo mem > /sys/power/state' 2. 3. Actual results: Blanks X to console, reports stopping tasks, but then returns to X again. Expected results: Blank X, stop tasks and then go into sleep mode Additional info: Also the stopping tasks takes a lot longer that it used to. Printing of that message used to fly by, now it takes about a second for each '=' to be printed.
*** Bug 172003 has been marked as a duplicate of this bug. ***
*** Bug 172019 has been marked as a duplicate of this bug. ***
The bug is closed. What version of kernel is this fixed in?
current rawhide should have it fixed. (It was fixed a few days back whilst I was away, so I'm not sure the exact revision it got fixed in).
I have kernel 2.6.14-1.1639 and it still doesn't suspend, but for a different reason. Now the messages are kernel: Stopping tasks: =================================| kernel: hci_usb 4-1:1.1: no suspend? kernel: hci_usb 4-1:1.0: no suspend? kernel: Could not suspend device 4-1: error -16 kernel: Some devices failed to suspend. So it still doesn't work for me. Should this bug be reopened or should I file a new one?
It's a new one -- the USB Bluetooth driver lacks a suspend method. I work around this on my PowerBook by running 'hid2hci --tohid' in the pre-suspend script, and 'hid2hci' again upon resume. Obviously that requires that your bluetooth device supports HID mode.
Dave, would you attach patch to fix this bug? I'd like to try it with vanilla kernel.
Created attachment 120761 [details] patch
Mine suspended, but doesn't come out of it, with some thread exceptions and stack traces dumped out on resume.
For me, works fine. I applied it to vanilla 2.6.14.
You still shouldn't have kauditd running unless you've specifically configured your system that way, though. If you do, that's a bug -- and may be causing your system to run _much_ slower than it needs to. Please check whether you have the 'audit' package installed, and whether auditd is started by default.
I haven't specically configured the system except through frequent updates to rawhide. How should audit be configured? 'audit' package is installed and 'auditd' is running (its configured to start at levels 2-5)
I thought it was decided that auditd shouldn't enable itself by default? Steve?
David, auditd doesn't enable itself by default on RHEL4 only. FC4 shipped before we decided to do that and there hasn't been a distribution release since then. I want rawhide to keep it on by default so that problems are found and reported. Looks like we found one.