On Thu, Feb 24, 2000 at 04:13:13PM +0100, Jamie Lokier wrote:
> Stephen C. Tweedie wrote:
> > > fsck isn't the only program which wants this info. I'm writing a
> > > faster version of find, and one thing I'd like to do is scan different
> > > disks in parallel if the search ends up spanning different disks. Now
> > > that we have multiple request queues :-)
> >
> > e2fsck isn't the only program which can get this info --- anyone can
> > parse fstab to obtain the same data.
>
> I cannot rely on fstab for this, because e2fsck combines fstab info with
> its own internal knowledge -- so fstab rarely serialises different
> partitions. No fstab I have ever seen serialises partitions. Just the
> difference between / and the rest.
The passno is a rather suboptimal way of telling somebody what partitions
can be processed in parallel. Just consider the example from the other mail.
(If I was not clear, please ask me to elaborate ...)
The reasone for this is: You try to tell fsck (or whom ever) what it can do
in parallel: All devs with the same passno. That does not work.
If you want to make it work? You have to specifiy which devs may NOT be
processed in parallel!
We could change the semantics in order to create a clean approach, finally.
Or let the fsck use it's own idea, but than allow for improvements of it.
Regards,
--
Kurt Garloff <[EMAIL PROTECTED]> Eindhoven, NL
GPG key: See mail header, key servers Linux kernel development
SuSE GmbH, Nuernberg, FRG SCSI, Security
PGP signature
