Maciej W. Rozycki wrote:
> > No, "-o sync" is too slow. You still want write-back cacheing to
> > coalesce blocks, and the elevator to optimise seeks. In addition, you
> > want to always keep light/motor on while there are pending writes, and
> > make the write flush timeout quite small when the floppy is idle.
>
> Well, doesn't the elevator work for "-o sync"? Floppy is slow enough it
> wouldn't make any inconsistency if writes get reordered for the duration
> of the motor being active (i.e. when the LED is lit).
Writes being reordered and coalesced is fine. That's disabled by "-o
sync" when you do a sequence of operations such as
open/write/write/write/close, open/write/close. "-o sync" forces every
operation to wait until it's finished writing to the disk before you can
continue with the next one.
> > Also, auto-unmounting after a few seconds fails if the directory is
> > anybody's current directory. But the floppy light has gone off. In
> > this case it's not safe to remove the floppy. That rather defeats the
> > point of auto-unmounting: now the user has to type a command to see if
> > it's safe to remove the floppy.
>
> Is that a problem? You still have the "dirty" flag set for filesystems
> that support it but data is otherwise consistent. Fat filesystems are
> clean anyway.
Well it's just as much a problem as now. You can already remove a disk
after writing it, without unmounting, provided all the data reached the
disk. The problem is when you insert the next disk :-)
-- Jamie
-
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 : Fri Jun 23 2000 - 21:00:21 EST