>Hi,
>
>> > The access to last_ino is serialized by spin lock, so that
>> > it is allowed to be not-atomic.
>> >
>> > Real bug is inside grow_inodes(), which calls prune_dcache() in turn.
>> > And nobody took care of poor dcache in 2.3, so that it stands now
>> > as bone in throat and all similar operations require big kernel lock.
>
>what is grow_inodes()? I can't see anything called that (in 2.3.30-1).
There isn't. When I rewrote the inode allocation from scratch I got rid of
such stuff. Now the icache gets shrunk only when the memory goes low and
now the icache is populated only by _caching_ entities.
>I.e. get_empty_inode() looks SMP-safe to me. What am I missing?
You aren't missing nothing, it's SMP-safe and can be safely called without
the big kernel lock.
Andrea
-
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/