I'm trying to get the LVM driver working over a raid5
under 2.4.0-test1-ac2. There seems to be a problem with the
LVM and raid5 make_request functions.
"What we have here is a failure to comunnicate."
LVM remaps the blocks by changing bh->b_rsector.
Raid5 then ignores that field and looks at bh->b_blocknr.
As a result, LVM LVs on a raid5 are broken. Seems this
could be fixed one of three ways:
1) Have LVM change bh->b_blocknr.
2) Have Raid5 use bh->b_rsector instead of computing
the sector from bh->b_blocknr.
3) Something else.
Which of these is the Right Way (TM)? I went with
option two. Near the beginning of raid5_make_request, I
changed:
new_sector = raid5_compute_sector(bh->b_blocknr*(bh->b_size>>9),
To:
new_sector = raid5_compute_sector(bh->b_rsector,
This seems to work. Is this the right way to fix
this? Or should I do something else? If this is the right
way, can someone make this change?
Brian Kress
kressb@icp.siemens.com
-
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 : Sat May 27 2000 - 12:43:40 EDT