Comment 4 for bug 1195855

Unfortunately the kernel log isn't telling us much about what's going on. To help figure out blocked suspend problems on manta, rather than grabbing /proc/wakelocks you'll need to grab the output of two different files.

/sys/power/wake_lock will print out any active usersapce wake locks. /sys/kernel/debug/wakeup_sources provides a lot of information, including which in-kernel wakeup sources are "active" (an active wake-up source blocks suspend, and this is the mainline analog of in-kernel wake locks). If you capture both of these files when you're in that state I should be able to deduce who is keeping the device from suspending.