From mboxrd@z Thu Jan 1 00:00:00 1970 Date: 18 Dec 2001 07:01:43 -0000 Message-ID: <20011218070143.15173.qmail@mailFA5.rediffmail.com> MIME-Version: 1.0 From: "amey d inamdar" Reply-To: "amey d inamdar" Subject: Stealing memory pages. Content-type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8BIT Content-Disposition: inline Sender: owner-linux-mm@kvack.org Return-Path: To: linux-mm@kvack.org List-ID: Hi, We are a group of four students, working on a project, "Implementation of Network RAM". We want to add the remote pages to the address space of a process ( by Modifying page fault handler). For the same reason we need a fixed allocation of a pool of page frames on each machine, which will serve as source for NRAM pages. This pool on each machine will be handled by a "Server" which will be implemented as a kernel module. So at the initialization only, we have to allocate the page frames to the module. We did it successfully as follows: 1) After setting up mem_map ( array of page *), while freeing individual page we didn't give last few pages to the buddy deallocator. 2) We individually marked all those pages non-reserved and stored virtual address of start of the first page frame. (__va(page)). The virtual address is part of kernel address space. 3) Now our server module will use this address and total no of pages, to manage allocation of pages to a remote process. My question is that, whether blocking such virtual address space inside the kernel can cause harm to its functionality? The machine is still working fine, but are there any ill-effects of such page frame stealing? thank you in anticipation. - Amey -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/