From mboxrd@z Thu Jan 1 00:00:00 1970 From: Koji Sato Subject: Re: Physical block addresses? Date: Thu, 31 Jan 2008 15:25:05 +0900 Message-ID: <47A169C1.3070001@osrg.net> References: <196f72b10801280948uc9e1275o6c4bf1bf56cd26c2@mail.gmail.com> Reply-To: NILFS Users mailing list Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <196f72b10801280948uc9e1275o6c4bf1bf56cd26c2-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: users-bounces-JrjvKiOkagjYtjvyW6yDsg@public.gmane.org Errors-To: users-bounces-JrjvKiOkagjYtjvyW6yDsg@public.gmane.org Content-Type: text/plain; charset="us-ascii"; format="flowed" To: NILFS Users mailing list Hi, Sivan. > I am trying to figure out whether NILFS uses physical block addresses > as pointers in inodes and indirect blocks, but I could not figure this > out from the papers. > > Does it? If so, I assume that when cleaning a segment and copying live > data to a new segment, each relocated block causes the propagation of > dirty blocks (because some other blocks point to the old segment, so > they need to be modified, and so on. Is that correct? NILFS version 2 uses virtual block numbers to point to blocks. The virtual block numbers are translated into physical disk addresses when they are referenced. This mechanism prevents cascading updates of the file system when cleaning segments because only the virtual-to-physical mappings are updated. Thanks. -- Koji Sato Open Source Software Computing Project NTT Cyber Space Laboratories Nippon Telegraph and Telephone Corporation