home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.ultrix
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!ames!haven.umd.edu!decuac!pa.dec.com!engage.pko.dec.com!nntpd.lkg.dec.com!nntpd2.cxo.dec.com!nabeth!alan
- From: alan@nabeth.enet.dec.com (Alan Rollow - Alan's Home for Wayward Tumbleweeds.)
- Subject: Re: Very large file systems ?
- Message-ID: <1992Dec14.214641.18792@nntpd2.cxo.dec.com>
- Lines: 56
- Sender: alan@nabeth (Alan Rollow - Alan's Home for Wayward Tumbleweeds.)
- Reply-To: alan@nabeth.enet.dec.com (Alan Rollow - Alan's Home for Wayward Tumbleweeds.)
- Organization: Digital Equipment Corporation
- References: <johnd.723948393@cortex.physiol.su.oz.au>
- Date: Mon, 14 Dec 1992 21:46:41 GMT
-
-
- In article <johnd.723948393@cortex.physiol.su.oz.au>, johnd@cortex.physiol.su.oz.au (John Dodson) writes:
- >Newsgroups: comp.unix.ultrix
- >From: johnd@cortex.physiol.su.oz.au (John Dodson)
- >Subject: Very large file systems ?
- >
- >I just got a couple of Hitachi DK517C-37 scsi drives to go on my DecStn 5000/200
- >running Ultrix 4.2
- >
- >The drive reports its size as 5611599 * 512 byte blocks.
- >
- >Further testing produced...
- >
- >$ mkfs /dev/rrz0c 4194303
- >
- > [ Example deleted, but it is the clue to what is going on... ]
- >
- >What is so magic about "4194304" blocks ? AH! it's 2^22 !
- >
- >In the 4.3bsd sources for the fs struct...
- > long fs_size; /* number of blocks in fs */
-
- While the number of blocks is limited to the size of a signed long
- (ULTRIX has the same declaration: /usr/include/ufs/fs.h) programs
- like newfs(8), fsck(8), dump(8), etc must treat devices as files.
- As you noted in a later follow-up of your own, the "where" argument
- of lseek(2) is a 31 bit integer and largest offset within a file
- is limited to (2^31)-1 bytes.
-
- >
- >so I should be able to have at least 2^31 (or even 2^32) blocks in a file
- >system, is Ultrix different ? or is Ultrix mkfs brain dead ?
- >or am I brain dead ? ;-)
-
- You'll discover BSD has the same limitation as ULTRIX. Some vendors
- have used the non-standard solution of inventing special seek functions
- that allow either larger offsets or do block addressing. Another
- solution on 32 bit systems might be to make long and int be different
- size and use either emualation libraries or smarter compilers to
- provide the different code. This would likely make long arithmetic
- slow. I believe it was what used on the PDP 11s to support 32 bit
- longs and 16 bit ints.
-
- It's worth noting that OSF/1 on Alpha AXP systems will use 64 integers
- for long and that the appropriate changes have been made to support
- large file and file system sizes.
- >
- >
- >John Dodson, Dept of Physiology,
- >johnd@physiol.su.oz.au University of Sydney,
- >Phone +61-2-692-3277 NSW 2006
- >Fax +61-2-692-2058 Australia.
- >
- --
- Alan Rollow alan@nabeth.cxo.dec.com
-
-