From mboxrd@z Thu Jan 1 00:00:00 1970 From: linux@arm.linux.org.uk (Russell King - ARM Linux) Date: Mon, 28 Apr 2014 11:28:06 +0100 Subject: [PATCH RFC 6/8] component: remove old add_components method In-Reply-To: <20140428070727.GC19455@ulmo> References: <20140426230025.GZ26756@n2100.arm.linux.org.uk> <20140428070727.GC19455@ulmo> Message-ID: <20140428102806.GH26756@n2100.arm.linux.org.uk> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, Apr 28, 2014 at 09:07:28AM +0200, Thierry Reding wrote: > I'm wondering if there may be an advantage to keeping both interfaces. > Even if currently what all implementations do is essentially creating > the match table at probe time there may be use-cases where that doesn't > work so well. Keeping both interfaces makes this whole change is pointless, because then there's no way to avoid having to rebuild the tracking of which components belong to which master - and we might as well stick with what we have. Moreover, one of the other points Laurent raised is that we need to be able to do partial binds for some subsystems, and for that to work we need more information held within the component helpers and the teardown/rebuild of the master/component relationships to be eliminated. The last point is that the repeated teardown/rebuild is already being used as a justification to go off and write a completely different infrastructure instead... it seems some people deem this to be far too wasteful of our billion cycles per second CPUs. -- FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly improving, and getting towards what was expected from it.