xref: /linux/Documentation/filesystems/hpfs.rst (revision d99ff463ecf651437e9e4abe68f331dfb6b5bd9d)
1.. SPDX-License-Identifier: GPL-2.0
2
3====================
4Read/Write HPFS 2.09
5====================
6
71998-2004, Mikulas Patocka
8
9:email: mikulas@artax.karlin.mff.cuni.cz
10:homepage: https://artax.karlin.mff.cuni.cz/~mikulas/vyplody/hpfs/index-e.cgi
11
12Credits
13=======
14Chris Smith, 1993, original read-only HPFS, some code and hpfs structures file
15	is taken from it
16
17Jacques Gelinas, MSDos mmap, Inspired by fs/nfs/mmap.c (Jon Tombs 15 Aug 1993)
18
19Werner Almesberger, 1992, 1993, MSDos option parser & CR/LF conversion
20
21Mount options
22
23uid=xxx,gid=xxx,umask=xxx (default uid=gid=0 umask=default_system_umask)
24	Set owner/group/mode for files that do not have it specified in extended
25	attributes. Mode is inverted umask - for example umask 027 gives owner
26	all permission, group read permission and anybody else no access. Note
27	that for files mode is anded with 0666. If you want files to have 'x'
28	rights, you must use extended attributes.
29case=lower,asis (default asis)
30	File name lowercasing in readdir.
31conv=binary,text,auto (default binary)
32	CR/LF -> LF conversion, if auto, decision is made according to extension
33	- there is a list of text extensions (I thing it's better to not convert
34	text file than to damage binary file). If you want to change that list,
35	change it in the source. Original readonly HPFS contained some strange
36	heuristic algorithm that I removed. I thing it's danger to let the
37	computer decide whether file is text or binary. For example, DJGPP
38	binaries contain small text message at the beginning and they could be
39	misidentified and damaged under some circumstances.
40check=none,normal,strict (default normal)
41	Check level. Selecting none will cause only little speedup and big
42	danger. I tried to write it so that it won't crash if check=normal on
43	corrupted filesystems. check=strict means many superfluous checks -
44	used for debugging (for example it checks if file is allocated in
45	bitmaps when accessing it).
46errors=continue,remount-ro,panic (default remount-ro)
47	Behaviour when filesystem errors found.
48chkdsk=no,errors,always (default errors)
49	When to mark filesystem dirty so that OS/2 checks it.
50eas=no,ro,rw (default rw)
51	What to do with extended attributes. 'no' - ignore them and use always
52	values specified in uid/gid/mode options. 'ro' - read extended
53	attributes but do not create them. 'rw' - create extended attributes
54	when you use chmod/chown/chgrp/mknod/ln -s on the filesystem.
55timeshift=(-)nnn (default 0)
56	Shifts the time by nnn seconds. For example, if you see under linux
57	one hour more, than under os/2, use timeshift=-3600.
58
59
60File names
61==========
62
63As in OS/2, filenames are case insensitive. However, shell thinks that names
64are case sensitive, so for example when you create a file FOO, you can use
65'cat FOO', 'cat Foo', 'cat foo' or 'cat F*' but not 'cat f*'. Note, that you
66also won't be able to compile linux kernel (and maybe other things) on HPFS
67because kernel creates different files with names like bootsect.S and
68bootsect.s. When searching for file thats name has characters >= 128, codepages
69are used - see below.
70OS/2 ignores dots and spaces at the end of file name, so this driver does as
71well. If you create 'a. ...', the file 'a' will be created, but you can still
72access it under names 'a.', 'a..', 'a .  . . ' etc.
73
74
75Extended attributes
76===================
77
78On HPFS partitions, OS/2 can associate to each file a special information called
79extended attributes. Extended attributes are pairs of (key,value) where key is
80an ascii string identifying that attribute and value is any string of bytes of
81variable length. OS/2 stores window and icon positions and file types there. So
82why not use it for unix-specific info like file owner or access rights? This
83driver can do it. If you chown/chgrp/chmod on a hpfs partition, extended
84attributes with keys "UID", "GID" or "MODE" and 2-byte values are created. Only
85that extended attributes those value differs from defaults specified in mount
86options are created. Once created, the extended attributes are never deleted,
87they're just changed. It means that when your default uid=0 and you type
88something like 'chown luser file; chown root file' the file will contain
89extended attribute UID=0. And when you umount the fs and mount it again with
90uid=luser_uid, the file will be still owned by root! If you chmod file to 444,
91extended attribute "MODE" will not be set, this special case is done by setting
92read-only flag. When you mknod a block or char device, besides "MODE", the
93special 4-byte extended attribute "DEV" will be created containing the device
94number. Currently this driver cannot resize extended attributes - it means
95that if somebody (I don't know who?) has set "UID", "GID", "MODE" or "DEV"
96attributes with different sizes, they won't be rewritten and changing these
97values doesn't work.
98
99
100Symlinks
101========
102
103You can do symlinks on HPFS partition, symlinks are achieved by setting extended
104attribute named "SYMLINK" with symlink value. Like on ext2, you can chown and
105chgrp symlinks but I don't know what is it good for. chmoding symlink results
106in chmoding file where symlink points. These symlinks are just for Linux use and
107incompatible with OS/2. OS/2 PmShell symlinks are not supported because they are
108stored in very crazy way. They tried to do it so that link changes when file is
109moved ... sometimes it works. But the link is partly stored in directory
110extended attributes and partly in OS2SYS.INI. I don't want (and don't know how)
111to analyze or change OS2SYS.INI.
112
113
114Codepages
115=========
116
117HPFS can contain several uppercasing tables for several codepages and each
118file has a pointer to codepage its name is in. However OS/2 was created in
119America where people don't care much about codepages and so multiple codepages
120support is quite buggy. I have Czech OS/2 working in codepage 852 on my disk.
121Once I booted English OS/2 working in cp 850 and I created a file on my 852
122partition. It marked file name codepage as 850 - good. But when I again booted
123Czech OS/2, the file was completely inaccessible under any name. It seems that
124OS/2 uppercases the search pattern with its system code page (852) and file
125name it's comparing to with its code page (850). These could never match. Is it
126really what IBM developers wanted? But problems continued. When I created in
127Czech OS/2 another file in that directory, that file was inaccessible too. OS/2
128probably uses different uppercasing method when searching where to place a file
129(note, that files in HPFS directory must be sorted) and when searching for
130a file. Finally when I opened this directory in PmShell, PmShell crashed (the
131funny thing was that, when rebooted, PmShell tried to reopen this directory
132again :-). chkdsk happily ignores these errors and only low-level disk
133modification saved me.  Never mix different language versions of OS/2 on one
134system although HPFS was designed to allow that.
135OK, I could implement complex codepage support to this driver but I think it
136would cause more problems than benefit with such buggy implementation in OS/2.
137So this driver simply uses first codepage it finds for uppercasing and
138lowercasing no matter what's file codepage index. Usually all file names are in
139this codepage - if you don't try to do what I described above :-)
140
141
142Known bugs
143==========
144
145HPFS386 on OS/2 server is not supported. HPFS386 installed on normal OS/2 client
146should work. If you have OS/2 server, use only read-only mode. I don't know how
147to handle some HPFS386 structures like access control list or extended perm
148list, I don't know how to delete them when file is deleted and how to not
149overwrite them with extended attributes. Send me some info on these structures
150and I'll make it. However, this driver should detect presence of HPFS386
151structures, remount read-only and not destroy them (I hope).
152
153When there's not enough space for extended attributes, they will be truncated
154and no error is returned.
155
156OS/2 can't access files if the path is longer than about 256 chars but this
157driver allows you to do it. chkdsk ignores such errors.
158
159Sometimes you won't be able to delete some files on a very full filesystem
160(returning error ENOSPC). That's because file in non-leaf node in directory tree
161(one directory, if it's large, has dirents in tree on HPFS) must be replaced
162with another node when deleted. And that new file might have larger name than
163the old one so the new name doesn't fit in directory node (dnode). And that
164would result in directory tree splitting, that takes disk space. Workaround is
165to delete other files that are leaf (probability that the file is non-leaf is
166about 1/50) or to truncate file first to make some space.
167You encounter this problem only if you have many directories so that
168preallocated directory band is full i.e.::
169
170	number_of_directories / size_of_filesystem_in_mb > 4.
171
172You can't delete open directories.
173
174You can't rename over directories (what is it good for?).
175
176Renaming files so that only case changes doesn't work. This driver supports it
177but vfs doesn't. Something like 'mv file FILE' won't work.
178
179All atimes and directory mtimes are not updated. That's because of performance
180reasons. If you extremely wish to update them, let me know, I'll write it (but
181it will be slow).
182
183When the system is out of memory and swap, it may slightly corrupt filesystem
184(lost files, unbalanced directories). (I guess all filesystem may do it).
185
186When compiled, you get warning: function declaration isn't a prototype. Does
187anybody know what does it mean?
188
189
190What does "unbalanced tree" message mean?
191=========================================
192
193Old versions of this driver created sometimes unbalanced dnode trees. OS/2
194chkdsk doesn't scream if the tree is unbalanced (and sometimes creates
195unbalanced trees too :-) but both HPFS and HPFS386 contain bug that it rarely
196crashes when the tree is not balanced. This driver handles unbalanced trees
197correctly and writes warning if it finds them. If you see this message, this is
198probably because of directories created with old version of this driver.
199Workaround is to move all files from that directory to another and then back
200again. Do it in Linux, not OS/2! If you see this message in directory that is
201whole created by this driver, it is BUG - let me know about it.
202
203
204Bugs in OS/2
205============
206
207When you have two (or more) lost directories pointing each to other, chkdsk
208locks up when repairing filesystem.
209
210Sometimes (I think it's random) when you create a file with one-char name under
211OS/2, OS/2 marks it as 'long'. chkdsk then removes this flag saying "Minor fs
212error corrected".
213
214File names like "a .b" are marked as 'long' by OS/2 but chkdsk "corrects" it and
215marks them as short (and writes "minor fs error corrected"). This bug is not in
216HPFS386.
217
218Codepage bugs described above
219=============================
220
221If you don't install fixpacks, there are many, many more...
222
223
224History
225=======
226
227====== =========================================================================
2280.90   First public release
2290.91   Fixed bug that caused shooting to memory when write_inode was called on
230       open inode (rarely happened)
2310.92   Fixed a little memory leak in freeing directory inodes
2320.93   Fixed bug that locked up the machine when there were too many filenames
233       with first 15 characters same
234       Fixed write_file to zero file when writing behind file end
2350.94   Fixed a little memory leak when trying to delete busy file or directory
2360.95   Fixed a bug that i_hpfs_parent_dir was not updated when moving files
2371.90   First version for 2.1.1xx kernels
2381.91   Fixed a bug that chk_sectors failed when sectors were at the end of disk
239       Fixed a race-condition when write_inode is called while deleting file
240       Fixed a bug that could possibly happen (with very low probability) when
241       using 0xff in filenames.
242
243       Rewritten locking to avoid race-conditions
244
245       Mount option 'eas' now works
246
247       Fsync no longer returns error
248
249       Files beginning with '.' are marked hidden
250
251       Remount support added
252
253       Alloc is not so slow when filesystem becomes full
254
255       Atimes are no more updated because it slows down operation
256
257       Code cleanup (removed all commented debug prints)
2581.92   Corrected a bug when sync was called just before closing file
2591.93   Modified, so that it works with kernels >= 2.1.131, I don't know if it
260       works with previous versions
261
262       Fixed a possible problem with disks > 64G (but I don't have one, so I can't
263       test it)
264
265       Fixed a file overflow at 2G
266
267       Added new option 'timeshift'
268
269       Changed behaviour on HPFS386: It is now possible to operate on HPFS386 in
270       read-only mode
271
272       Fixed a bug that slowed down alloc and prevented allocating 100% space
273       (this bug was not destructive)
2741.94   Added workaround for one bug in Linux
275
276       Fixed one buffer leak
277
278       Fixed some incompatibilities with large extended attributes (but it's still
279       not 100% ok, I have no info on it and OS/2 doesn't want to create them)
280
281       Rewritten allocation
282
283       Fixed a bug with i_blocks (du sometimes didn't display correct values)
284
285       Directories have no longer archive attribute set (some programs don't like
286       it)
287
288       Fixed a bug that it set badly one flag in large anode tree (it was not
289       destructive)
2901.95   Fixed one buffer leak, that could happen on corrupted filesystem
291
292       Fixed one bug in allocation in 1.94
2931.96   Added workaround for one bug in OS/2 (HPFS locked up, HPFS386 reported
294       error sometimes when opening directories in PMSHELL)
295
296       Fixed a possible bitmap race
297
298       Fixed possible problem on large disks
299
300       You can now delete open files
301
302       Fixed a nondestructive race in rename
3031.97   Support for HPFS v3 (on large partitions)
304
305       ZFixed a bug that it didn't allow creation of files > 128M
306       (it should be 2G)
3071.97.1 Changed names of global symbols
308
309       Fixed a bug when chmoding or chowning root directory
3101.98   Fixed a deadlock when using old_readdir
311       Better directory handling; workaround for "unbalanced tree" bug in OS/2
3121.99   Corrected a possible problem when there's not enough space while deleting
313       file
314
315       Now it tries to truncate the file if there's not enough space when
316       deleting
317
318       Removed a lot of redundant code
3192.00   Fixed a bug in rename (it was there since 1.96)
320       Better anti-fragmentation strategy
3212.01   Fixed problem with directory listing over NFS
322
323       Directory lseek now checks for proper parameters
324
325       Fixed race-condition in buffer code - it is in all filesystems in Linux;
326       when reading device (cat /dev/hda) while creating files on it, files
327       could be damaged
3282.02   Workaround for bug in breada in Linux. breada could cause accesses beyond
329       end of partition
3302.03   Char, block devices and pipes are correctly created
331
332       Fixed non-crashing race in unlink (Alexander Viro)
333
334       Now it works with Japanese version of OS/2
3352.04   Fixed error when ftruncate used to extend file
3362.05   Fixed crash when got mount parameters without =
337
338       Fixed crash when allocation of anode failed due to full disk
339
340       Fixed some crashes when block io or inode allocation failed
3412.06   Fixed some crash on corrupted disk structures
342
343       Better allocation strategy
344
345       Reschedule points added so that it doesn't lock CPU long time
346
347       It should work in read-only mode on Warp Server
3482.07   More fixes for Warp Server. Now it really works
3492.08   Creating new files is not so slow on large disks
350
351       An attempt to sync deleted file does not generate filesystem error
3522.09   Fixed error on extremely fragmented files
353====== =========================================================================
354