Proposed fix for PM deadlock on dpm_sem
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 8 lines · Paul Mackerras
I have been having trouble with deadlocks on
dpm_semwhen suspending and resuming, particularly if I remove a usb device while my powerbook is suspended. Thedpm_semdeadlocks also mean that USB can't deal with devices whose drivers don't have suspend/resume callbacks by simply disconnecting the device, as it did in the past. This meant that while I can get my dual G4 powermac desktop machine to suspend to ram and resume (using the powerbook sleep code), its USB keyboard wouldn't work after resume.
A while ago I posted some thoughts on locking in the driver model which are somewhat relevant here. Please take a look at
http://marc.theaimsgroup.com/?l=linux-usb-devel&m=109596905805858&w=2
It doesn't address exactly the same issues you're concerned with, but it does concern the way suspend/resume interacts with device addition/removal.
Alan Stern