[Partial solution] Accessing a device through an offset

Martin Maney <[email protected]> Mon, 7 Jun 2004 22:29:41 -0500
Newsgroups gmane.linux.admin.managers
Message-ID <[email protected]>
Thanks to Mike Brodbelt for twisting my head frontwards; losetup was
the thing I had seen before and couldn't remember.  Unfortunately it
turns out not to be a very useful solution, since the standard loopback
device doesn't deal with partitions (that is, there's just loop0; no
loop0a, loop0b, etc.)

There is an enhanced loopback device from Nasa, but it's old and seems
to be stale (last released patch is for 2.4.22), requires a modified
losetup userland tool, and may have some other issues.  Apparently
this sort of thing will be easy, or at least not so challenging, in 2.6
thanks to the device mapper framework, but since I'm looking to support
a 2.4 environment at this time, it's not of immediate use to me.

enhanced loop: ftp://ftp.hq.nasa.gov/pub/ig/ccd/enhanced_loopback/patches

And I'll mention Jason Boxman's document about loopback mounting
devices, which seems to summarize all that you can do short of having
the device mapper:

http://www.trekweb.com/~jasonb/articles/linux_loopback.html

Thanks to Mike and Jason, I've taken this as far as I apparently can
within the bounds of what I'm willing to put into this now.  I have
confirmed that 3ware's mirror pairs simply have a 1024 sector header,
and I can mount at least some of the partitions using a regular loop
mount with the approriate offset.  This doesn't go too far, though,
since partitions past the first 2G and the LVM-managed partition that
is most of the drive's raw space remain unaccesable by this method.

Revised plan: balance the cost of a spare controller against the quite
low failure rate for non-mechanical parts like that.  Maybe keep an eye
out for bargins on ebay?

Thanks, folks.

-- 
the warfare on the cutting edge of any science draws attention
away from the huge uncontested background, the dull metal heft
of the axe that gives the cutting edge its power.  -- Dennett
_______________________________________________
LinuxManagers mailing list - http://www.linuxmanagers.org
submissions: LinuxManagers-35TzE1X9F6582KRnZfj+bdi2O/[email protected]
subscribe/unsubscribe: http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers