From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Clements Subject: Re: async writes in raid1 Date: Tue, 15 Apr 2003 14:14:07 -0400 Sender: linux-raid-owner@vger.kernel.org Message-ID: <3E9C4BEF.84E49DE5@SteelEye.com> References: <200304141924.h3EJO4x07957@oboe.it.uc3m.es> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Return-path: To: ptb@it.uc3m.es Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids "Peter T. Breuer" wrote: > > "A month of sundays ago Paul Clements wrote:" > > * allowing the bitmap to be rescaled -- i.e., changing the amount of > > data that each bit represents > > I really don't agree with that :-). But maybe it can be done! It's > just I personally see no way anyone can count in anything except 1K > blocks with the current raid code the way it is. > > Originally I wrote the patch with more generality, but realized that the > generality could never be made use of - there are too many assumptions > in the extant raid code that one block = 1KB. The translation from 1K blocks to whatever-sized bitmap blocks (or "chunks") will be done only when marking or unmarking the bitmap (and incrementing/decrementing the block counters). The block counters allow us to know when multiple 1K block writes are actually on the same "chunk" so that everything is handled correctly. -- Paul