"Fossies" - the Fresh Open Source Software Archive

Member "ident2-v1.07_FINAL/CHANGES" (22 Jul 2005, 4932 Bytes) of package /linux/privat/old/ident2-v1.07_FINAL.tar.gz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file.

    1 --------- [07/22/2005
    2 + incorporated lsof patch which adds support for
    3   SunOS, NetBSD, OpenBSD, OSF/1, HP-UX
    4   (thanks inittab@netizenweb.com (Greg Schenzel))
    5 
    6 --------- [05/18/2004
    7 + buffer overflow in _getl fixed :(
    8 
    9 --------- [11/02/2003
   10 + fixed using data after it has been freed in sys/m_fbsd.c:m_get_uid()
   11   thanks to Ilya Novoselov <nullguid@t72.ru>
   12 
   13 --------- [08/18/2000
   14 + forgot to specifiy option 'n' in getopt_long
   15   thanks to Johan Ekenberg <johan@ekenberg.se> for pointing this out
   16 
   17 --------- [04/11/2000
   18 + correctly handles errno == EINTR when accept() returns -1
   19 + date formats in CHANGES file now Y2K compliant!
   20 
   21 --------- [11/2/99
   22 + port to FreeBSD
   23 + PID registration, under FreeBSD, is also /var/run
   24 
   25 --------- [9/18/99
   26 + changed all sprintf's to snprintf's
   27 + removed all potential security hazards (strcpy, memcpy, sprint, etc)
   28 + moved m_register_pid AFTER the initial fork()-to-become-daemon,
   29   otherwise wrong pid is registered.
   30 + d'oh. move m_reduce_rights AFTER m_register_pid, otherwise can't
   31   write pid to /var/run
   32 - NOTE, snprintf is not portable. must be removed in future.
   33 
   34 --------- [9/3/99]
   35 + performing security sweep. reevaluating choices
   36 + merged getopt_long support
   37 + added getopt_long detection to ./configure
   38 + wrote two help outputs (short and short+long versions)
   39 + implemented the --license parameter
   40 + implemented the --usage paramater
   41 
   42 --------- [8/30/99]
   43 + moved srand(time(NULL)) out of INIT_GLOBALS and into the random reply generator
   44   function. rand is seeded every time, which is flawed, but still harmless.
   45   this was moved because the parent never actually calls rand(), which sucks
   46   because the children will keep generating the same reply.
   47 + implemented new machine call, m_register_pid(), for use in daemon mode,
   48   on systems that wish to keep track of running system services.
   49 + PID registration, under Linux, in /var/run
   50 + switched to a ./configure based setup
   51 
   52 --------- [8/25/99]
   53 + changed m_linux.c to only setuid(). Under Linux, setuid() sets the effective
   54   userid as well as superuser. If this would be the case, seteuid() should fail.
   55 
   56 --------- [8/9/99]
   57 + added random replies
   58 + fixed standard unspecial replies (INIT_GLOBALS wasn't initting!)
   59 - ripped out config file parsers
   60 - ripped out all semblances of IP masquerading
   61 + rather than trying to drop userid to nobody and failing if it
   62   doesn't work, it only tries to change uid if uid is 0, elsewise,
   63   keep current uid.
   64 + reworded README
   65   
   66 --------- [6/11/99]
   67 - removed throttled daemon. I felt this included some very unnatural
   68   code and it was a pain to debug. it was a research project at most.
   69 + rewrote config file parser
   70 + rewrote command-line processor
   71 + included GNU information on top of all source files
   72 + code cleaning spree. Optimized, minimalized, etc.
   73 + fixed buffer handling in some areas which may have been dangerous
   74 + reorganized some general code into common.c
   75 + stablized threaded daemon
   76 + rewrote README
   77 + reorganized throttled daemon
   78 + moved accept() out into the child process so we don't have
   79   to do any nasty stuff like return file descriptors to the
   80   parent through exit().
   81 + adopted capitalize global variables convention
   82 + decided to start DATING these changes
   83 
   84 ---------- 0.999c
   85 + no idea why I thought the user of 'nobody' couldn't carry on identity
   86   functions. The dropping to nobody under Linux has been reenacted.
   87 + fixed a diverge from RFC1413. Thanks to Piotr Kucharski <chopin@sgh.waw.pl>
   88   for pointing out that END OF LINE is clearly defined by the RFC as CRLF
   89   and not LF. Apologies to anyone affected by this.
   90 + fixed an 'off by one' error in the throttled implementation causing a
   91   garbage char to appear at the end of an ident query.
   92 + command-line option to display compile-time options.
   93 + shamelessly copied a pidentd convention. If you have the file .noident
   94   in your home directory, it will not send a response to anyone requesting
   95   an ident that resolves to your username.
   96 + dropping userid to 'nobody' has been changed to instead drop the access
   97   level to a user defined variable (which SHOULD be a username).
   98 + moved quite a few variables out into seperate #defines. 
   99 + dropping userid to 'nobody' as a security measure is back, again.
  100 + IP masquerading works! Much love to everyone who helped the
  101   debugging/fixing process. Especially Hans.
  102   -Actually, it only sort of works. Linux 2.0 and 2.1 have
  103   differing levels of success. Kernel 2.2 is supposedly going to
  104   settle on a format once and for all.
  105 + rand() wasn't being seeded. 
  106 + for some reason I wasn't making use of #define USER_REPLIES.
  107 + dropping the userid to 'nobody' makes servicing ident requests
  108   impossible. I'm dumb. The function 'go_nobody' and all references
  109   to it were removed.
  110 + major MAJOR rewrite. This introduced some more natural (ie, not kludgy)
  111   code as well as more bugs! It should also be easier to extend and
  112   expand. Hopefully noone will get hurt this time. :)
  113