[mec]
> Ultimately I would like to see the link order dependencies specified
> explicitly in some way. (foo.o needs to come before bar.o et
> cetera), and then have Rules.make guarantee to honor that through
> list manipulation. But not in 2.4.
When Linus did the link order thing, I thought at the time that it
would be neat to have every init function be able to return -EAGAIN and
it would then be put back on the end of the queue. The master init
function would keep cycling through the list until one entire run with
no success returns.
Like you do with bubble sort. (:
In the SCSI devices case, every driver that wants to run before the
1542 (for example) decrements an "extern int pre_1542", and 1542 itself
returns -EAGAIN until it sees the same value of pre_1542 twice in a
row.
Thinking about it again, though, I'm not sure that this would be
powerful enough to make it useful in the SCSI case, where you *really*
want "do this one after other SCSI devices but before post-SCSI
things". Just a few thoughts.
Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Feb 07 2000 - 21:00:13 EST