in the never ending quest to debug the 2.4 md code under UML I ran into a md
bug. ubd not only closes down on release, it also zeroes the blk_size entry.
md however tried to read from blk_size before having called lock_dev again -
woops. The attached patch fixes this; I'll try to get it into 2.4.21, but
maybe it is helpful for someone in the meantime.
Now, I'd still like to suggest to _not_ zero the blk_size entry and the
nr_sects in ubd_release(); they also make it disappear from /proc/partitions.
(Even though it uncovered the md bug) Can those two lines simply be dropped
without bad effects or is there a particular reason for it?
Lars Marowsky-Brée <lmb@...>
SuSE Labs - Research & Development, SuSE Linux AG
"If anything can go wrong, it will." "Chance favors the prepared (mind)."
-- Capt. Edward A. Murphy -- Louis Pasteur