From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from bombadil.infradead.org ([198.137.202.133]:51366 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726218AbfEOHy2 (ORCPT ); Wed, 15 May 2019 03:54:28 -0400 Date: Wed, 15 May 2019 00:54:27 -0700 From: Christoph Hellwig Subject: Re: [PATCH 3/6] xfs: use locality optimized cntbt lookups for near mode allocations Message-ID: <20190515075427.GI29211@infradead.org> References: <20190509165839.44329-1-bfoster@redhat.com> <20190509165839.44329-4-bfoster@redhat.com> <20190510173110.GC18992@infradead.org> <20190513154544.GE61135@bfoster> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20190513154544.GE61135@bfoster> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: Brian Foster Cc: Christoph Hellwig , linux-xfs@vger.kernel.org On Mon, May 13, 2019 at 11:45:44AM -0400, Brian Foster wrote: > Hmm, I had attempted some tighter integration with the xfs_btree_cur on > a previous variant and ran into some roadblocks, the details of which I > don't recall. That said, I may have tried to include more than just this > active state, ran into header issues, and then never really stepped back > from that to explore more incremental changes. > > I think extending the priv union with something for the allocation code > to use makes sense. Your suggestion has me wondering if down the road we > could genericize this further to a bc_valid state or some such that > quickly indicates whether a cursor points at a valid record or off into > space. That's a matter for another series however.. Yep. > You mean to create tree specific cursor structures of which > xfs_btree_cur is a member, then the tree specific logic uses > container_of() to pull out whatever it needs from cur..? I'd need to > think more about that one, but indeed that is beyond the scope of this > work. Yes.