lkml.org 
[lkml]   [1998]   [Jan]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: devfs
Rik van Riel writes:
> On Fri, 9 Jan 1998, Richard Gooch wrote:
>
> > Michael O'Reilly writes:
> > > "Albert D. Cahalan" <acahalan@cs.uml.edu> writes:
> > > >
> > > > Fine, /dev will be 4 to 8 GB. The linear search will be fun!
> > > > Seriously, a larger dev_t will _not_ fix this problem.
> > >
> > > This is truely nonsense. You don't need to create every possible
> > > device. Even in the current scheme, most people don't have a /dev/sdf*
> > > in there.
> > >
> > > You create the devices you need. It isn't terrible hard!
>
> Can you say 'scsidev'. Even on my (3-disk) system I use it,
> because I can call my disks by name...
> Also, I moved my ide devices into /dev/ide. Although there
> only are a few, you don't need block devices at a fixed place,
> and I don't need to see them on an ls.
>
> > Even if you had a programme which automagically creates /dev nodes,
> > you still have to run it (unless you do it at boot in which case the
> > boot process slows down that little bit more).
>
> That is just a _little_ bit. A clean reboot (on large systems)
> is carefully planned and rare. The extra 30 seconds the device
> creator is working don't really matter here. And the unclean
> reboot is going to take eons anyway (on large systems you usually
> have to check several GB's of ext2) so the 30 seconds don't
> matter...
>
> > The devfs solution is faster, and has the distinct advantage of being
> > actively written, whereas I'm not aware of a robust, generally
> > available auto-makedev (feel free to take that as a challenge:-).
>
> But how are you going to request modules? You can't write to
> a not-loaded device, because it's node doesn't yet exist.
> Of course you could make up loads of kludges to get around this
> problem, but it won't make it any prettier than the current
> solution.

When kerneld see a request for "sd_*" it should start loading SCSI
drivers. Once a SCSI driver is loaded that pops up the exact entry
that was requested, kerneld knows that that module is required to
satisfy all drives on that host.
Better yet, kerneld should maintain a database of device
entries. Whenever a module is loaded, all the new devices should
be associated with that module, so on future device requests kerneld
knows which module to load. This database could even be saved to disc
so that it can be used for later boots.

Regards,

Richard....

\
 
 \ /
  Last update: 2005-03-22 13:41    [W:1.119 / U:0.460 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site