From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hiroyuki KAMEZAWA Date: Wed, 22 Dec 2004 01:16:37 +0000 Subject: Re: [PATCH] modifing vmemmap initialization [1/1] : modifies init Message-Id: <41C8CAF5.1060902@jp.fujitsu.com> List-Id: References: <41C7CC9E.7080602@jp.fujitsu.com> In-Reply-To: <41C7CC9E.7080602@jp.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-ia64@vger.kernel.org Matthew Wilcox wrote: >>>This patch adjust calling sequence of virtual_mem_map_alloc(), >>>which is added by previous patch. >> >>This doesn't look like an improvement at all, the whole virtual memmap >>should not need core ifdefs. In fact I still hope it'll go away. > > > It doesn't need core ifdefs at all -- simply define a dummy > virtual_mem_map_alloc() and you can remove the ifdefs. > I'm thinking about pass "memmap allocation function" as a free_area_init_node()'s argument. Such args can be used in i386(or some) NUMA case, which doesn't use alloc_bootmem(). Anyway, my patch is not good. > (I also hope it'll go away. Where's CONFIG_NONLINEAR got to?) > Robert Picco's one was intended to avoid using CONFIG_VIRTUAL_MEM_MAP. I wrote another one, which uses CONFIG_VIRTUAL_MEM_MAP, and it was too complicated :( -- Kame