Re: [PATCH 2/4] Convert epoll to a bitlock

[wli@movementarian.org]

On Mon,  2 Feb 2009 11:20:09 -0700 Jonathan Corbet <corbet@lwn.net> wrote:
>> Matt Mackall suggested converting epoll's ep_lock to a bitlock as a way of
>> saving space in struct file.  This patch makes that change.

On Tue, Feb 03, 2009 at 01:39:42PM -0800, Andrew Morton wrote:
> hrm.  bit_spin_lock() makes people upset (large penguiny people).  iirc
> it doesn't have all the correct/well-understood memory/compiler
> ordering semantics which spinlocks have.  And lockdep doesn't know about
> it.

ISTR the memory/compiler ordering semantics bits coming up when it was
still pte_chain_lock(), but not the entirety of it. I think
smp_mb__after_clear_bit() and/or smp_mb__before_clear_bit() turned out
to be needed in the unlock function, and they're there now with
clear_bit_unlock() et al. lockdep I'm less sure about, but suspect
the objects with bit locks embedded in them are too numerous to feasibly
track, e.g. there may be several bh's per-page, and how numerous they
are tends to be why they're using bit locks in the first place.


-- wli