From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040312 Galeon/1.3.14 Description of problem: After entering suspend to memory with: echo "mem" > /sys/power/state which works OK, the notebook won't come back properly. The power comes on but the screen remains blank. CAPS LOCK and NUM LOCK change states fo LEDs but the keyboard does nothing else. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Enter suspend to memory. 2. Try to power back up. Additional info: I'll attach files as per bug 121020.
Created attachment 100027 [details] dmesg
Created attachment 100028 [details] lsmod
Created attachment 100029 [details] lspci -n
Created attachment 100030 [details] lspci
Please retest with kernel-2.6.5-1.358 or newer. There have been several major fixes that may have improved this situation for you. If it fails, please retest with the same version but i586 kernel.
Nope. Same deal with 358 i686. Let me try the i586...
With i586 keyboard doesn't even respond to CAPS LOCK, NUM LOCK etc. The lights come on (i.e. various LEDs like power button LED etc.), I can even hear the CPU fan going after a while, but the machine is otherwise dead. Hope this helps. Do you need any other info about the machine's hardware? BTW, even when I do the suspend to memory from run level 1, it does the same thing (i.e. running X doesn't seem to be the cause of this issue). Suspend to disk does nothing at all.
If it's of any help, the vanilla 2.6.6 does more or less the same thing with suspend to memory. The machine goes into suspend, but when it comes back, it is dead although the LEDs come on. No keys on the keyboard do anything. The main difference that I can see is that "echo disk > /sys/power/state" in fact _does_ write something out to the swap partition. I'm still unclear how to get that back (the resume=/dev/xxx doesn't seem to work), but hey, it's better than nothing :-)
Suspend-to-disk is not compiled in FC kernels for some reason. Neither the new (/sys/power) nor the old method (/proc/acpi). I was wondering if this issue qualifies for bugzilla entry but I guess the maintainers had something in mind.
suspend to disk is quite really badly broken unfortinalty. Ok new kernel uploaded to http://people.redhat.com/arjanv/2.6 it works better for my machine, but I do have to unload the ehci_hcd module before i can suspend for some reason (still investigating that)
Just FYI, kernel-2.6.6-1.435 is still a no go on this notebook.
Kernel 2.6.8-1.521 still doesn't do this right.
Still the case with FC3T2, kernel 2.6.8-1.603.
Please do not expect us to investigate your issue. You must push upstream kernel.org and use bugzilla.kernel.org. If they manage to fix this, then the FC kernel will have it when we upgrade to the latest upstream kernel.
I was just reporting this so that we have a full history of it. But, if you don't wan't me to do that, I can stop.
Reporting is fine, so others see this report. But if you want a solution, bugzilla.kernel.org and the upstream ACPI project is where you must seek help.
No worries. Shall open one today and post a link to it here.
Kernel bug: http://bugzilla.kernel.org/show_bug.cgi?id=3556
suspend2 works fine, so I'll close this. It seems video related anyway - radeon driver needs to soft boot the cards or something like that.