I wish ext2 (and others) would mark the filesystem clean and flush
pages to disk when filesystem activity stops. Then supermount would
work better and a crash would leave most filesystems marked clean.
>> This should also extend to hard disks that have gone south,
>> which was the problem the original poster was experiencing.
>
> If a *device* vanishes - by being unplugged for instance - there's
> really nothing the kernel can do - it most likely has cached pages
> for that device which it cannot dispose of. This is what causes
> the deadlocks.
Cached pages may be stored in the bitbucket to save RAM.
I really mean that, because 40% of an executable is not
much good! Data is somewhat more useful, but it will be
lost anyway. It is better to recover the resources.