Talk:Unix File System

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Universal File System vs. Unix File System[edit]

I'm studying up for my CompTIA A+ exam, and they had "UFS" defined as "Universal File System", not "Unix File System", likely due to the fact that it covers much more than just Unix? Or is that just errata in my study materials? —This unsigned comment was added by (talkcontribs) 2006-04-03 21:19:18 UTC.

It's a question of context. Apparently there's a "Universal File System" in NetWare: [1].
There's also a site that claims that "UFS" in the UNIX sense stands for Universal File system - [2] - but they're wrong. Guy Harris 21:16, 3 April 2006 (UTC)

Soft Updates in UFS1[edit]

AFAIK softupdates have been part of freebsd since ver 3, thus long before UFS2! The preceding unsigned comment was added by (talk • contribs) 11:15, 26 January 2004.

Rewrite, merge with FFS[edit]

I have largely rewritten this. Someone needs to run a spellchecker over me. --ssd 06:57, 12 Jul 2004 (UTC)

Additions this article could now use:

  • Additions to UFS after FFS (I think they just increased field widths and added reserved fields)
  • More complete list of vendors who extended UFS
  • More complete list of proprietary features added to UFS for each vendor (as with above comment)

I'd add these, but I only have data for System V, Solaris, and Linux. SGI uses XFS instead. The preceding unsigned comment was added by Ssd (talk • contribs) 06:57, 12 July 2004.

Should this and FFS just be combined? As I remember, the first "UFS" was just the SunOS 2.0 version of the 4.2BSD FFS; 2.0 was the first SunOS version with a VFS layer into which file systems could be plugged (as it was the first one with more than one file system to plug into it, becaue 2.0 introduced NFS), and Sun called it UFS at that point. (It was definitely "UFS" in 4.0; I think that went back to 2.0.) Other commercial UN*X vendors probably picked up the Sun NFS code, which included a VFS layer, and probably also picked up the UFS name if they had FFS. The split of the code into FFS and UFS layers in BSD happened later. -- Guy Harris 01:17, 22 September 2005 (UTC)
I agree, merge. —Claunia 03:54, 2 January 2006 (UTC)
I agree also, merge! The preceding unsigned comment was added by (talk • contribs) 13:17, 25 January 2006.
One more for merge. The same merge just happened in the german wikipedia. -- Kvedulv 20:04, 16 March 2006 (UTC)
Me, too: merge. (Aside: Am I right in thinking that UFS is the "native" file system of FreeBSD, OpenBSD and NetBSD? If so, the article should probably mention them.) —Chris Chittleborough 09:47, 23 March 2006 (UTC)

UFS still in OSX?[edit]

AFAIK support for UFS has been removed from OSX. Can someone confirm this? The preceding unsigned comment was added by Oneiros (talk • contribs) 01:52, 13 February 2006.

It is still as of 10.4.4.
Please sign messages.
Claunia 04:49, 13 February 2006 (UTC)

Mac OSX currently imposes a 4GB file limit when it formats a new disk under UFS. —Preceding unsigned comment added by (talk) 01:13, 27 July 2009 (UTC)

Copied from Comparison of BSD operating systems discussion[edit]

The information about the BSDs and UFS/UFS2/ffs is confusing. I know that Open doesn't yet have UFS2, but does Free really have no UFS? And is ffs really unique to Open? --Oneiros 00:12, 14 April 2006 (UTC)

...and, as per my comment above, it's not clear it belongs here anyway, as it's also in Comparison of operating systems.
At least some of the confusion is confusion about what FFS and UFS are. They're all descendants of Kirk McKusick's Berkeley Fast File System. Back in the early '80's, Sun developed one of the early virtual file system (VFS) implementations, so that UN*X system calls that manipulate files could, instead of going directly to the code for the one and only file system implementation, be directed to different file system implementations. This let those system calls work with both the Berkeley FFS (which was the local on-disk file system used by SunOS at the time) and NFS. Sun called the Berkeley file system "UFS", for "UNIX File System".
Later, the Berkeley FFS code was split into two parts - one that implemented lower-level inode or "container" functions, and one that implemented UN*X semantics (directories, permissions, ownership, etc.) atop that lower level, and the higher-level code was made to work atop the Berkeley LFS as well. The bottom part was called "FFS", presumably because the bulk of the changes between the old UNIX file system and the Berkeley Fast File System were at the container level (the higher-level changes were largely to implement symbolic links and large file names, although symbolic links could be trivially added to the old UNIX file system as well), and the top part was called "UFS", perhaps out of a combination of Sun's influence and the fact that most of the "UN*X" semantics were implemented at that layer.
In answer to your two questions:
  1. FFS is not unique to OpenBSD;
  2. FreeBSD does have UFS. Guy Harris 00:28, 14 April 2006 (UTC)
However, to my knowledge, OpenBSD is the only one on that list that still uses the old ffs as a default. Armedblowfish 01:25, 14 April 2006 (UTC)
The {FFS,LFS}/UFS code split occurred, as far as I know, in 4.4BSD, and, as far as I know, OpenBSD split from NetBSD after NetBSD updated to a 4.4-Lite base. OpenBSD has a sys/ufs source directory with ffs, fs, mfs (for memory-based containers), and ufs, so it has the same code structure as FreeBSD, NetBSD, and DragonFlyBSD in that regard, meaning it doesn't have the old pre-split FFS. Perhaps they call it FFS, but it's still UFS.
Are you thinking of UFS1 vs. UFS2 here? Guy Harris 01:50, 14 April 2006 (UTC)
OpenBSD definitely still calls it ffs. See a relevant OpenBSD FAQ. Also, they specifically state that they do not support UFS2. I have OpenBSD, and when I installed Debian, I remember having to do something special in order to mount OpenBSD's ffs partitions. Will try to look it up. Armedblowfish 02:52, 14 April 2006 (UTC)
What they call FFS in that FAQ is the OpenBSD version of UFS1 (as opposed to UFS2). Was it any different trying to mount NetBSD or FreeBSD UFS1 partitions on Debian? (If not - or if you didn't try to mount them - the fact that Debian required special action in order to mount OpenBSD UFS1 partitions doesn't necessarily mean OpenBSD is any more different from the other BSDs than they are from each other.)
Note also that according to the NetBSD fstab man page, the fstab entries for local "McKusick File System" (at this point, I'm tempted to use that term, to avoid FFS/UFS/etc. confusion) partitions have "ffs" as the file system type, just as they do in OpenBSD. Perhaps that's "ffs" as distinct from "lfs", with both of them offering "ufs" semantics - i.e., "ffs" means "UFS atop FFS", and "lfs" means "UFS atop LFS". (Amusingly enough, NetBSD and OpenBSD also implement ext2 as a container layer for UFS!) Guy Harris 03:08, 14 April 2006 (UTC)
A Linux man page specifies different "ufs types". (Search for "Mount options for ufs"). Ufstype 44bsd (4.4BSD) is apparently the one Free/Open/NetBSD use. This brings up the question of whether to call it UFS, UFS1, or FFS? Since ffs is the one that OpenBSDers and NetBSDers (and maybe FreeBSDers?) use, it would make sense to call it by what the people who use it call it, with an appropriate footnote. Armedblowfish 03:42, 14 April 2006 (UTC)
FreeBSD uses "UFS1" exclusively, OpenBSD uses "FFS" mostly but also "UFS" in some places (and in the code). Call it "UFS1" or "FFS" and add a footnote explaining the alternative names. All the BSDs versions of UFS1 are broadly similar and originate in the same codebase, but I believe disklabel and superblock differences now mean that FreeBSD's is not completely compatible with NetBSD/OpenBSD. I believe NetBSD and OpenBSD are compatible, and Solaris' is slightly different again. NicM 07:46, 14 April 2006 (UTC).
Thanks for the updates and clarification. Please don't forget to also update Unix File System.--Oneiros 11:39, 14 April 2006 (UTC)

Linux and UFS[edit]

I think Linux 2.6.20 has read/write UFS2 support. Source: . --Ysangkok 19:40, 27 April 2007 (UTC)

Also see here. --Ysangkok 22:06, 27 April 2007 (UTC)

Block = Cluster[edit]

Is the here mentioned block to the Unix File System what a cluster is to FAT? --Abdull (talk) 16:46, 1 February 2008 (UTC)

Empty infobox[edit]

Why is the infobox empty? TMC1221 (talk) 22:35, 25 February 2008 (UTC)

Thank you for your suggestion. When you feel an article needs improvement, please feel free to make those changes. Wikipedia is a wiki, so anyone can edit almost any article by simply following the Edit this page link at the top. The Wikipedia community encourages you to be bold in updating pages. Don't worry too much about making honest mistakes — they're likely to be found and corrected quickly. If you're not sure how editing works, check out how to edit a page, or use the sandbox to try out your editing skills. New contributors are always welcome. You don't even need to log in (although there are many reasons why you might want to). Guy Harris (talk) 23:22, 25 February 2008 (UTC)

Author of UFS[edit]

When McKusick started to work on UFS (in late 1981), UFS has been created and under development by Bill Joy since nearly 2 years. Joy started UFS development in early 1980. Why do people call McKusick the author and hide Bill Joy? Schily (talk) 12:29, 20 July 2010 (UTC)

You have reliable third-party sources documenting this? FWIW, the original FFS paper is M. K. McKusick, W. N. Joy, S. J. Leffler, and R. S. Fabry (August 1984). "A Fast File System for UNIX". ACM Transactions on Computer Systems. 2 (3): 181–197.  It does not describe the division of labor, nor does it cite prior work published by any of the paper's four authors. McKusick is the only one of the paper's authors who is still involved in development of this code, and the UFS1 superblock magic number is McKusick's birthday, not Joy's. By the normal principles of attribution in the field, if Joy gets credit it seems clear that Leffler should as well, with the assumption that Bob Fabry is listed as an author because he was their supervisor. (One could ask him, I suppose, but that would not meet WP:RS.) 121a0012 (talk) 02:27, 21 July 2010 (UTC)
I checked the output of sccs -R log CSRG_Archive_4/sys/ufs/
Fabry did not edit anything in the UFS source, If you only look at the time between 1980 and August 1984, Joy, Leffler and McKusick contributed a similar amount of work and as mentioned already, the first nearly 2 years only show edits from Bill Joy. --Schily (talk) 16:52, 22 July 2010 (UTC)
Is the SCCS archive considered a reliable secondary source? I'd certainly say it's reliable, but it's a primary source. Are there any secondary sources? 121a0012 (talk) 01:54, 23 July 2010 (UTC)
It is a reliable primary source as there is no motive to hack these SCCS history files. I prefer primary sources as many secondary sources are incorrect. --Schily (talk) 15:59, 23 July 2010 (UTC)
Wikipedia prefers secondary sources. 121a0012 (talk) 03:04, 24 July 2010 (UTC)
This is obviously a wrong decision. Secondary sources are usually biased and this can only be circimvented if you have plenty of independend secondary sources that allow to create an "average". Primary sources created by automated systems like SCCS are more reliable. --Schily (talk) 18:46, 25 July 2010 (UTC)

UFS and ZFS[edit]

@Intgr: Hello, yesterday I read some paragraphs of the ZFS article and saw the following sentence:

"While it (ZFS) is also faster than UFS, it can be seen as the successor to UFS."

because UFS and ZFS have good function in the field of the data integrity and citing to the above sentence, I added "ZFS" to "See also" section. ویرایشگر-1 (talk) 11:20, 22 October 2015 (UTC)

@ویرایشگر-1: Ok, thanks for the explanation. That would be a reasonable rationale for adding that link.
Except that I went to the ZFS page and looked at the two provided references for this claim: [3] [4]. Neither of them contain the word "successor" (on any of the pages), nor can I find anything else in there to imply that ZFS is the successor to UFS.
AFAICT the only relationship they have is that SunOS, which had inherited UFS from earlier Unices (SysV I believe?), sought to replace the aging file system with something more recent.
But I'm far from an expert on this subject. If you can find some more substantial source to support this, then I'm fine with re-adding the link. -- intgr [talk] 13:05, 22 October 2015 (UTC)
It is the other way round: Svr4 inherited UFS from SunOS-4 as the Svr4 kernel is a SunOS-4.0 descendent. I would not see ZFS as a successor of UFS in general. If ZFS is a real successor of another filesystem it is the successor of WOFS, the first Copy on Write filesystem, because the WOFS authors read my Diploma thesis before starting to implement ZFS. Schily (talk) 13:16, 22 October 2015 (UTC)