Re: [PATCH] udev enhancements to use kernel event queue

[Kevin P. Fleming]

Greg KH wrote:

> But you also have to "hold" events for a bit of time in order to
> determine that things are out of order, or we have a gap.  So a bit of
> "complex" logic is in the works, but it's much less complex than if we
> didn't have that sequence number.
> 

OK, so the important point here is that while you probably delay events 
for a small amount of time (1-3 seconds maybe) to ensure that you aren't 
processing steps out of order, it's not likely that userspace is going 
hold event #1321 for an indefinite period of time just because it has 
never seen event #1320.

I can't wait to see this implementation; it's going to be interesting, 
to say the least. However, without the sequence numbers, it would very 
likely be impossible.