Re: speedup device flushing with devfs

[Andrew Clausen]

On Sat, Dec 14, 2002 at 12:07:59PM +0100, David Schweikert wrote:
> I noticed that a simple
> 
> # parted -s /dev/ide/host0/bus0/target0/lun0/disc print
> 
> takes on my machine 1.5 seconds to complete.
> I made a strace and I did see that parted does try to open
> part1, part2, ... part15, in that directory, which in turn triggers
> devfs module probing.

Hmmm.  What kind of module probing?  Rescanning the partition table?
(I don't have devfs here...)

> With the following simple patch, the time required is down to 0.05
> seconds. Is this a good thing or am I going to destroy my disks?

Why not structure the loop around scanning the directory, rather than
around iterating through partition numbers?

Cheers,
Andrew