15 Apr
2005
15 Apr
'05
10:59
Dozens of different kinds of hardware are probed - which might or might not be available, thus requiring some (frequently heuristic) hardware probing - and waiting for answers for some time. Some hardware subsystems have looong possible timeouts - SCSI (an early 80s era technology) is a notorious example, USB (connectivity for all kinds of possibly very dumb and possibly very slow devices) is another.
Just curious: Do you have any idea how much of the boot time is spent waiting for hardware timeouts? Could this be 60% ? Is there any work on trying to run tasks concurrently on boot? (Subject to a task interdependencies, of course.) Regards, Håkon Hallingstad