Path: utzoo!attcan!uunet!lll-winken!lll-lcc!ames!mailrus!husc6!ogccse!blake!uw-beaver!microsoft!w-colinp
From: w-colinp@microsoft.UUCP (Colin Plumb)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Faster File Systems
Summary: Auto-recognising disk formats
Keywords: BOOH
Message-ID: <117@microsoft.UUCP>
Date: 21 Dec 88 06:01:13 GMT
References: <10716@s.ms.uky.edu>
Reply-To: w-colinp@microsoft.UUCP (Colin Plumb)
Organization: Microsoft Corp., Redmond WA
Lines: 31
Confusion: Microsoft Corp., Redmond WA

In article <10716@s.ms.uky.edu> sean@ms.uky.edu (Sean Casey) writes:
>Heck, I'd be happy with something that wouldn't thrash when two processes
>use disk at the same time.

I'll see what I can do.  This basically boils down to adequate cacheing.

Technical questions:

So far, I think the way to auto-recognise my floppies is to patch the
RootNode RestartSeg pointer to first go to my validator, which checks to
see if it's my disk format, then passes control to the existing validator,
which will handle the existing file system (and FFS, in 1.4).

Questions:
- Is this the right thing to do?  Is there a better way to stick
my file system in, or will there be one in 1.4?

- How to lock the RootNode structure?  Since this field gets read rarely and
modified even more rarely, I could go for years and never lose a race condition,
but I would like to do it right.  Forbid()/Permit()?

- What info does the disk validator get called with?

- What does the disk validator do, besides fsck and adding an entry to
the device list?

- Is there a way to get the slightly magic df0: and df1: aliases to work?
I don't really know how they work.

Thanks for the help!
-- 
	-Colin (uunet!microsof!w-colinp)