Committed! Plus I did even more work on the code. I made it work
with -s, added a usleep() in the loop, revamped the indexing code
even more, and made -a work better.
When I wrote the code originally I tried to mask the start and end indices
too early and that created a lot of confusion. The kernel's write index
is a full 32 bit integer... the variable is only masked when the kernel
needs to look up an entry. I made the ktrdump code do the same thing
pretty much and that greatly improved its readability.
(Now happily ktrdump -a -l'ing on my test box!)
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
