--On Sunday, December 29, 2002 02:18:42 -0800 Andrew Morton
[off-list ref] wrote:

We would like to not hold i_shared_lock across the zap_pte_range() call
anyway, for scheduling latency reasons.

But I suspect that i_shared_lock is the only thing which prevents the
vma from disappearing while truncate is playing with it.

umm...  I think we can just turn i_shared_lock into a semaphore.  Nests
inside mmap_sem.

I've stared at this for a couple of days and don't see any better solution.

Dave McCracken

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/