Bruce Perens writes: > My hypothesis is that "apmd" is swapped out when the user-suspend signal > comes in (because there is so little RAM). It takes too long to run, and > thus the laptop doesn't suspend properly. I also notice that it is logging > resume, but not suspend.
> Is anyone else experiencing this? No. Another guess at what may be happening is that the computer shuts down during a write to the disk. AFAIK, the APM BIOS will signal to apmd that it has received a suspend request, and then wait for a confirmation that it can perform the actual suspend operation. If no confirmation arrives within 20 seconds, it will suspend regardless. apmd, upon receipt of a suspend request, calls sync, sleeps for 2 seconds, and then signals to the APM BIOS that the machine is ready for the suspend. Perhaps these 2 seconds aren't long enough and the suspend happens during the write to disk. (Not using apmd means that the disk isn't sync'ed at all.) A possible solution might then be to get the sources of apmd and make it sleep longer before suspending. The relevant sleep call is found in the apm_suspend function in apmlib.c. -- Olaf Weber -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .