Description of problem: When the PCMCIA yentaq_socket module is loaded, APM suspend ceases to function. Version-Release number of selected component (if applicable): kernel-2.6.3-1.118 How reproducible: Always Steps to Reproduce: 1. modprobe pcmcia_core yenta_socket 2. apm -s 3. Actual results: LCD panel blanks, the system emits a dual beep (indicating APM failure, I suppose), and immediately returns from the suspend state. Expected results: System should enter APM suspend state. Additional info: - hardware : IBM ThinkPad A30p - ACPI suspend-to-ram/disk does not function, hence kernel parameter 'acpi=off' is invoked. - rmmod'ing yenta_socket allows APM suspend, even from within X.
Created attachment 98316 [details] relevant /var/log/messages extract Timestamps : 09:45:25 - successful APM suspend & wakeup 09:46:25 - modprobe yenta_socket 09:46:44 - failed APM suspend
Created attachment 98317 [details] lspci -v output
After upgrading to the "unified FC/EL" kernel-2.6.3-2.1.242, resume-after-suspend hangs the machine hard in both APM and ACPI mode. Mr. Nottingham suggests the 4G/4G mode to be the culprit. Does it suffice to set CONFIG_X86_4G, CONFIG_X86_4G_VM_LAYOUT and CONFIG_HIGHMEM4G to 'n' to be able to test a 4G-disabled kernel ?
Erroneous behaviour still present in kernel-2.6.4-1.300 with disabled 4G/4G (bug #117032, comment #18) in APM mode (ACPI mode does not exhibit this problem).
Tested with FC2t3, kernel 2.6.5-1.349 : the system still refuses to fully suspend when module 'yenta_socket' is loaded. Contrary to previous kernels, the system seems to partially suspend, that is : the screen blanks, drive & fans seem to go down, but the moon crest symbol on the laptop (visually indicating the suspend modus) does not light up. System suspends perfectly when yenta_socket is rmmod'ed.
is this still causing problems with current kernels ?
Having not used APM for a while (trying to get stable ACPI running), I'll have a look at this with the kernel as supplied in the FC3 final release.
Problem still present in kernel 2.6.9-1.681_FC3, with a regression, i.e. the behaviour is again as described in the initial description (no suspend, dual beep), regressing from comment #4 (partial suspend).
fixed in the -698 kernel in updates-testing ?
Tested with kernel-2.6.9-1.698_FC3 ; - Without yenta_socket loaded, 'apm -s' works fine ; - With yenta-socket loaded, 'apm -s' seems to suspend the machine (LCD backlight switches off, drives seem to spin down), but the "suspend" mooncrest symbol on the laptop does not light up ; apmd reports successfull suspend/resume cycle. /var/log/messages : Dec 8 11:56:47 dmbr042 network: Shutting down loopback interface: succeeded Dec 8 11:56:47 dmbr042 apmd[1878]: User Suspend Dec 8 11:57:48 dmbr042 kernel: PCI: Found IRQ 11 for device 0000:00:1d.0 Dec 8 11:57:48 dmbr042 kernel: PCI: Sharing IRQ 11 with 0000:01:00.0 Dec 8 11:57:48 dmbr042 kernel: PCI: Sharing IRQ 11 with 0000:02:00.0 [... other IRQ 11 sharing messages] Dec 8 11:57:55 dmbr042 kernel: PCI: Sharing IRQ 11 with 0000:02:02.0 Dec 8 11:57:57 dmbr042 netfs: Mounting other filesystems: succeeded Dec 8 11:57:57 dmbr042 netfs: Mounting other filesystems: succeeded Dec 8 11:57:57 dmbr042 apmd[1878]: Normal Resume after 00:01:10 (100% unknown) AC power
As I recently upgraded my IBM ThinkPad A30p platform to a more recent notebook (Dell Inspiron 9300), which does not seem to support APM, I am unable to verify whether this issue is fixed in the latest kernel packages.