bug in ffs_vget/ufs_reclaim

[Ted Unangst]

in ffs_vget, there is a VREF(i_devvp) call a ways into the function.
before this happens, there are several vput(vp) calls for various errors.
ufs_reclaim() will call vrele(i_devvp), but maybe it was never VREF'd, 
meaning there will be a mismatch, eventually leading to a panic when the 
ref goes below zero.  is this correct?



-- 
we owe so much money we're not broke we're broken
we're so poor we can't even pay attention