Bugzilla – Bug 117927
suspend-from-disk does not work if PCMCIA network-adapter is removed before restoring
Last modified: 2005-09-21 08:44:25 UTC
When I have my PCMCIA network-adapter plugged in when suspending to disk and try to restore without the adpater plugged in, the restoring fails after stating something like: All taks have been restored. CapsLock and ScrlLock blink periodically. If I have the adapter plugged in, it works perfectly. Maybe removeable devices should not be "saved" when suspending.
I solved the issue by using acpi=off, not sure if this works on PCs that do support ACPI though. I cannot test the latter as I do not own such a notebook. However this is another reason to make sure whether the notebook actually supports ACPI before enabling it by default! See bug 117820. If this works on a notebook that supports acpi with acpi=on, then this bug can be closed.
This is clearly one of the things you should not do, regardless of if you are using ACPI or APM: modifying the hardware during suspend is asking for trouble. You could use SUSPEND*EJECT_PCMCIA=yes in /etc/sysconfig/powersave/sleep to eject the card before suspend to work around this problem.
I do not think so! Imagine you are working somewhere where you have to use battery. You get a call, do not watch the system, it stays idle for some time and goes into suspend-to-disk. You pack your stuff, which means that you remove removeable things like USB and PCMCIA and go home, you turn on you notebook and it hangs. If there is a removeable device, ACPI it should be handled as removeable and not as static. Obviously I agree that removing some non-removeable-devices is not a good idea, yet removing removeable devices is their actual sense, as the term already states. ;)
(In reply to comment #2) > You could use SUSPEND*EJECT_PCMCIA=yes in /etc/sysconfig/powersave/sleep to > eject the card before suspend to work around this problem.
I know I can. Yet Joe Smith does not, because s/he does not know about it and relies on the _removable_ in removable-device. Hence I think that SUSPEND*EJECT_PCMCIA=yes should be default. If there is such a switch for USB, it should also be switched to yes by default. What disadvantages would it have, to set both of these to yes?
USB is "unplugged" by removing the host controller modules already. If you eject the PCMCIA card that holds your $HOME on an external drive, you won't be lucky after resume. This is why it is not set to yes. Also, it usually does not break. If it does, you have hardware with buggy drivers that should be fixed.