Re: [patch 08/22] reduce l_icloglock roundtrips

[Christoph Hellwig <hch@xxxxxxxxxxxxx>]

On Wed, Dec 03, 2008 at 01:56:14PM +1100, Dave Chinner wrote:
> On Tue, Dec 02, 2008 at 11:04:38AM -0500, Christoph Hellwig wrote:
> > All but one caller of xlog_state_want_sync drop and re-acquire
> > l_icloglock around the call to it, just so that xlog_state_want_sync can
> > acquire and drop it.
> 
> Not really a performance problem, though, because the one caller
> that doesn't already hold the lock is the common path (i.e.
> xlog_write()); the others are in the  unmount path, so lock traffic
> is not there is not performance critical....
> 
> Hence I'm not sure this is really necessary.

I don't think it's absolutely nessecary, but it's a lot cleaner this
way..