Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins
|
2023-02-15
| ||
| 11:26 | Fix build and release version 1.5.0 Nearly ten years after 1.5.0beta has been in use I think it's probably OK :) Leaf check-in: 24fd6d62e4 user: crees tags: trunk | |
| 11:02 | Don't match parent directory when we only wanted to match this one. check-in: c4550e1873 user: crees tags: trunk, 1.5.0 | |
| 10:55 | Types were wrong and initialisation was incorrect with SHORTLINKS cfs_fh.c:315:78: error: incompatible pointer to integer conversion initializing 'int' with an expression of type 'void *' [-Wint-conversion] cfs_fileid rootnode={1,0,"\0\0\0\0\0\0\0\0","/NOWHERE/null",NULL,NULL,0,NULL,NULL};/* fileid=1; should be unique */ check-in: e6e97126fe user: crees tags: trunk | |
|
2013-05-15
| ||
| 16:46 | Revert [0b7d118875], sloppily done and incorrect check-in: 004262b7e9 user: crees tags: trunk | |
| 16:22 | Create beta check-in: 651f4d07de user: crees tags: trunk | |
| 16:20 | Add Changelog check-in: 3c14f1086f user: crees tags: trunk | |
|
2013-05-14
| ||
| 19:59 | Simplify Makefile, using target variables check-in: 0b7d118875 user: crees tags: trunk | |
| 19:48 | Add a datastore property, which is added to each crypted dir. At the moment, the version is 0, which makes the directories compatible with all versions of cfsd (which might not be the case). Once the version in DATASTORE is bumped, previous versions are unreadable with new cfs, and new versions are unreadable with old cfs. The most likely use of this is when architecture independence is hopefully added to directories check-in: 644b82009e user: crees tags: trunk | |
| 19:41 | Fix array bounds error causing .. in root of attached dir to refer to self check-in: aac04db8a8 user: crees tags: trunk | |
|
2013-05-13
| ||
| 18:16 | Extend typed passphrases to 256 characters, just as allowed when provided to stdin. Remove some buffer overflows... Use NFS_MAXPATHLEN where appropriate. While here, use sig_t rather than manually defining the signal type. check-in: 61fe14c2e9 user: crees tags: trunk | |
|
2013-05-12
| ||
| 17:35 | Audit cfsd for buffer overflows. Some of the userland utilities may still overflow, but it's not such a big deal-- they can be checked later. check-in: 03ea0fc977 user: crees tags: trunk | |
| 16:52 | Raise the theoretical limit on file size to uint32_t; 4G. Fixes SEGV with files over 2G, file size limit should be hit much above that. Thanks to: Gerrit Pape, Carlo Contavalli <ccontavalli@commedia.it> Security: CVE-2006-3123 check-in: db812beaa1 user: crees tags: trunk | |
| 16:45 | Use NFS_MAXPATHLEN where appropriate Obtained from: Debian; Gerrit Pape check-in: abc6c70e11 user: crees tags: trunk | |
| 11:14 | Replace possibly unsafe strcmp function with safer and faster direct comparisons Use macro is_dot_or_dotdot for checking for string being "." or ".." While here, use snprintf instead of a strcpy check-in: 034d95b8ed user: crees tags: trunk | |
|
2013-05-11
| ||
| 17:29 | Sort and trim headers-- does anyone use NeXT any more??? While here remove redundant function prototypes. check-in: 9eda39bf3f user: crees tags: trunk | |
| 17:21 | Annotate and typesafe pfh() check-in: aa74da8e23 user: crees tags: trunk | |
| 15:44 | Turns out the ls issue was because of the way the encrypted name on the local filesystem was represented; the encrypted name is changed to hex using %02x, and used as the filename. %02x expects u_char, so feeding it char makes it give strange output such as fffffe4 for perfectly normal chars. Big d'oh moment :/ Correctly fixes [06d1571c8d] check-in: 0f6d1c5ae8 user: crees tags: trunk | |
| 13:18 | Faster to directly check for . and .. rather than unsafe strcmp check-in: b92e3d0e19 user: crees tags: trunk | |
| 13:02 | Direct use of fhdata * causes pain; compiler complains about type mismatches when passing nfs_fh * to these functions. Instead use some accessor macros which do the same job, and mean that we can stop doing casts in non-obvious places. check-in: 01e32c5872 user: crees tags: trunk | |
| 12:10 | Change argument to u_char; this function deals with casting anyway check-in: d40ad569ad user: crees tags: trunk | |
| 11:32 | Add missing stdlib header for malloc check-in: 10c58d55d4 user: crees tags: trunk | |
|
2013-05-10
| ||
| 18:05 | Sort headers and remove some old cruft check-in: abbd6d9cea user: crees tags: trunk | |
| 17:59 | Revert e65d55e8e0 and 2a1c134e08 until fhdata * and nfs_fh * are better understood. PR: 06d1571c8d check-in: 70d80f86d4 user: crees tags: trunk | |
|
2013-05-08
| ||
| 19:02 | Revert 5c864b31; this caused all passphrases to be rejected. Check instead for return values of less than 1, which was the original intention check-in: 26e0adfa03 user: crees tags: trunk | |
|
2013-05-07
| ||
| 19:58 | Add debug macro, and use it to annotate cipher types check-in: 91b291c68c user: crees tags: trunk | |
| 19:28 | Fix a really stupid segfault (this bug was mine) check-in: 3c422f1459 user: crees tags: trunk | |
| 11:08 | Oops, reversed lt and gt resulting in errors every time check-in: bd77e329b4 user: crees tags: trunk | |
|
2013-05-06
| ||
| 19:10 | Use snprintf instead of strncpy; better error handling (i.e. it works) check-in: 93a9faa6da user: crees tags: trunk | |
| 18:23 | Silence compiler warnings check-in: 125fe375c0 user: crees tags: trunk | |
| 18:16 | Final type fixes check-in: cba569a942 user: crees tags: trunk | |
| 18:15 | Add prototype for checkkey, sort & fix includes check-in: 983c9f021a user: crees tags: trunk | |
| 18:10 | Strange typo here, looks as though buf was supposed to be a string check-in: 35f9e0431e user: crees tags: trunk | |
| 18:09 | Fix headers, add ccat.h check-in: 9af2787525 user: crees tags: trunk | |
| 18:01 | Hm, shouldn't declare main in headers check-in: a14cddc0bb user: crees tags: trunk | |
| 18:00 | Fix cast, header style, create cname header check-in: a2bfd727ec user: crees tags: trunk | |
| 17:56 | Style, add header, cast some xdrproc_t again as per cd2519f2c7 check-in: 12a0c680f1 user: crees tags: trunk | |
| 17:51 | More casts to u_char check-in: 769c2890fd user: crees tags: trunk | |
| 17:45 | Add headers for cmkdir.c, and create truerand.h check-in: bebd747d5c user: crees tags: trunk | |
| 17:36 | Another reversal of args to mcg_block_encrypt. I suppose no-one ever actually used this. check-in: 510a5709be user: crees tags: trunk | |
| 17:34 | BF_LONG can be cast from u_char check-in: ac9abb9dae user: crees tags: trunk | |
| 17:31 | Cast a hash to unsigned char. Arguments to mcg_block_decrypt are reversed... this probably never worked correctly. check-in: 62d96599cc user: crees tags: trunk | |
| 17:14 | Add string.h and make register int explicit check-in: 55cdefd0dd user: crees tags: trunk | |
| 17:05 | Fix style of includes Add fcntl.h for open Remove ifdef for sgttyb check-in: 8083294227 user: crees tags: trunk | |
| 16:59 | Cast some function defs; expecting (arg, ...) but giving (arg, arg). While here pass pointer instead-- looks as though that's what it should be doing. Hopefully harmless. check-in: cd2519f2c7 user: crees tags: trunk | |
| 16:44 | Prototypes were compiled in ages ago check-in: ecb6668a83 user: crees tags: trunk | |
| 16:16 | Add prototypes for getpass.c check-in: 6e4cf454e0 user: crees tags: trunk | |
| 16:12 | Rename fhstat to avoid collision with mount.h check-in: 6158f34976 user: crees tags: trunk | |
| 16:09 | fread never returns negative numbers; presume that it's checking for early EOF check-in: 5c864b31a9 user: crees tags: trunk | |
| 16:06 | This one was my fault... char not chat! check-in: fe570a4230 user: crees tags: trunk | |
| 16:05 | bcopy needs strings.h check-in: 173829001e user: crees tags: trunk | |