From: SMTP%"davidg@balder.com" 26-OCT-1997 20:23:48.93 To: "'ahaslam@ihug.co.nz'" , "'Joe N. Turner'" CC: "'NTFSD'" Subj: [ntfsd] point of FATsample (was IFS not in .....) Return-Path: owner-ntfsd@atria.com Received: by arisia.gce.com (UCX V4.1-12C, OpenVMS V7.1 VAX); Sun, 26 Oct 1997 20:21:29 -0500 Received: from gw.atria.com (gw.atria.com [192.88.237.2]) by mercury.mv.net (8.8.8/mem-971025) with SMTP id AAA11772 for ; Sun, 26 Oct 1997 00:01:48 -0400 (EDT) Received: by gw.atria.com id Sat, 25 Oct 1997 21:52:00 -0400 Received: from ivy-balder.Goebel-Enterprises by gw.atria.com id Sat, 25 Oct 1997 21:51:55 -0400 Received: by ivy-balder.Goebel-Enterprises with SMTP (Microsoft Exchange Server Internet Mail Connector Version 4.0.993.5) id <01BCE177.1405E620@ivy-balder.Goebel-Enterprises>; Sat, 25 Oct 1997 18:52:01 -0700 Message-ID: From: "David Goebel (Balder Technology Group, Inc.)" To: "'ahaslam@ihug.co.nz'" , "'Joe N. Turner'" Cc: "'NTFSD'" Subject: [ntfsd] point of FATsample (was IFS not in .....) Date: Sat, 25 Oct 1997 18:51:59 -0700 X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.993.5 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-ntfsd@atria.com Precedence: bulk I hate to continue this thread, but I was offline for a few days and felt a need to help people understand the value of the kit. -----Original Message----- From: Joe N. Turner [SMTP:jturner@pac.sc.ti.com] Sent: Friday, October 17, 1997 10:02 AM To: Tony Mason Cc: brad.baker@pegasus-ofs.com; ntfsd@atria.com Subject: Re: Re[2]: Why IFS kit not available in Europe .... Secondly, FAT source is availble from a veriety of sources; ..... The point of the fastfat source in the IFS kit is not to document FAT's on disk structure. That can be done on an 8.5 X 11 sheets of paper (maybe two sheets for FAT32 with LFNs). The point is that the fastfat sources shows you how to write a good citizen cached file system. It is true that sometimes NTFS does present slightly different semantics to filter drivers than FAT. This is mostly because NTFS can do more and has additional complexities it has to deal with (compression being an excellent example). When there are cases of FAT and NTFS semantics diverging for no good reason, these are considered bugs and fixed if possible. The bottom line is that FAT and NTFS have much more in common than different in how they interface with the system. This is the value of the IFS kit and fastfat sources. Take a look at how FatMoveFile() guarantees that there are no threads current in the non-cached path for the file while a segment is being moved, and further more does this without holding the PagingIo resource exclusive across I/Os which could lead to deadlocks. Or how an extending cached write-though understands that the recursive non-cached paging I/O write is not responsible for updating the ValidDataLength but rather knows the top-level request will handle it and acquires resources appropriately. Or for that matter, even how fastfat makes good use of ValidDataLength at all since there is no ValidDataLength stored on disk. This is the really hard stuff that was learned the hard way over many years. LINUX source for FAT would be zero help in implementing your own cached file system on NT while the fastfat source is tremendously helpful. In fact, almost all you have to do is replace the on-disk layout code (not very much) with your own and you're done. -----Original Message----- From: brad.baker@pegasus-ofs.com [SMTP:brad.baker@pegasus-ofs.com] Sent: Friday, October 17, 1997 7:43 AM To: mason@osr.com Cc: ntfsd@atria.com Subject: Re[2]: Re[2]: Why IFS kit not available in Europe ... Given that all of the file systems under NT use the IFS Microsoft must require internal documentation of this interface for their own engineers. I can not believe that the IFS continues to be carried around only in the heads of certain engineers and the source code itself. Believe it. David - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [ To unsubscribe, send email to ntfsd-request@atria.com with body UNSUBSCRIBE (the subject is ignored). ]