Timeline
12/06/97:
- 16:32 Changeset [10881] by
- Move responsibility for suppress-network-daemons, in.named, and config_afs to the init.d scripts, so that we can be called from do-update for a non-OS update. Take the version we're updating to as an argument, instead of picking it up from /etc/athena/version.
- 16:30 Changeset [10880] by
- Get version and newvers as arguments from update_ws. Responsibility for rebooting has been transferred here. If we're not doing an OS update, do the finish-update right away and only reboot to finish up an auto-update; don't reboot in the middle.
- 16:25 Changeset [10879] by
- Instead of exporting NEWVERS and VERSION, pass them as arguments to do-update. Transfer responsibility for rebooting to do-update.
- 16:21 Changeset [10878] by
- Move responsibility for suppress-network-daemons and config_afs here, so that /usr/athena/lib/update/finish-update can be invoked from do-update in the absence of an OS update.l Also fix some minor nits: invoke finish-update from /srvd, and don't put quotes around the literal string "Manual".
- 16:21 Changeset [10877] by
- Move responsibility for named and config_afs here, so that /usr/athena/lib/update/finish-update can be invoked from do-update in the absence of an OS update.l Also fix some minor nits: invoke finish-update from /srvd, and don't put quotes around the literal string "Manual".
- 16:18 Changeset [10876] by
- Blow away session files, logging if we find any non-zero-length ones. XXX There is a (preexisting) race conditin if someone logs in after reactivate starts. We should probably set up an /etc/nologin for a while.
- 16:13 Changeset [10875] by
- Back out the change to switch the Sun backspace and delete.
- 16:06 Changeset [10874] by
- Make /etc/default/su a configfile.
- 16:06 Changeset [10873] by
- Add /etc/default/sulogin, in case an administrator wants to set PASSREQ back to true.
- 16:04 Changeset [10872] by
- Make single-user mode more robust by setting PASSREQ=no in /etc/default/sulogin. With this setting, sulogin will fork /sbin/sh without asking for the root password, allowing you to recover if the passwd file is munged.
- 15:56 Changeset [10871] by
- Substitute variables without assuming what their default values are (mainly for consistency with /etc/init.d/athena and with SGI stuff). Substitute in MACHINE and SYSTEM variables.
- 15:53 Changeset [10870] by
- Preserve the NETDEV and SYSTEM variables in /etc/athena/rc.conf.
- 15:52 Changeset [10869] by
- Preserve the SYSTEM variable in rc.conf.
- 15:52 Changeset [10868] by
- Don't hardcode the MACHINE variable; substitute it at install time. Add a SYSTEM variable, also substituted at install time.
- 15:51 Changeset [10867] by
- Don't hardcode the MACHINE and SYSTEM variables; substitute them in at install time.
- 15:42 Changeset [10866] by
- Use free(), not cfree().
- 11:39 Changeset [10865] by
- Garbage collect.
- 11:27 Changeset [10864] by
- Garbage collect.
12/05/97:
- 16:20 Changeset [10863] by
- perl now supports "make check" as an alternative name for "make test". Use it. Set PERL_SKIP_TTY_TEST to punt tty tests, which don't work when we don't have a tty.
12/03/97:
- 17:05 Changeset [10862] by
- Use strerror.
- 17:04 Changeset [10861] by
- Nuke errno defines.
- 17:01 Changeset [10860] by
- Use signal(), not weirdo sigset().
- 17:01 Changeset [10859] by
- Use strerror; fix some socket warnings.
- 17:00 Changeset [10858] by
- NetBSD's stdio is a bit different.
- 17:00 Changeset [10857] by
- Nuke some horrible unnecessary setbuf code.
- 16:59 Changeset [10856] by
- Use strerror().
- 16:59 Changeset [10855] by
- Conditionalize on SYSV for ustat. The ffs header file moved on NetBSD. Take into account validuser() renaming.
- 16:56 Changeset [10854] by
- Assume POSIX and call setsid().
- 16:56 Changeset [10853] by
- Use strerror; define validuser() and checkhost() for all systems.
- 16:55 Changeset [10852] by
- Need sys/types.h on NetBSD.
- 16:55 Changeset [10851] by
- Assume POSIX. Simplify utmp vs. utmpx handling using a macro. Get utmp/wtmp/lastlog paths right for more systems. Fix a buffer overrun in findidle(). Fix an uninitialized structure field in doall().
- 16:52 Changeset [10850] by
- Use strerror.
- 16:51 Changeset [10849] by
- Need to use single quotes around $(foo).
- 16:50 Changeset [10848] by
- Include system header files before local ones.
- 16:50 Changeset [10847] by
- Use InstallNonExecFile to ensure that target directories exist.
- 16:49 Changeset [10846] by
- Simplify multi-platform support by only looking at the platform we're currently on. Remove compiled menu support.
- 16:48 Changeset [10845] by
- Remove a couple of trailing spaces after backlashes.
- 16:47 Changeset [10844] by
- Use strerror. Use HOSTTYPE environment variable to set bindir, or use compiled-in default. Assume POSIX.
- 16:44 Changeset [10843] by
- Use strerror.
- 16:44 Changeset [10842] by
- Use InstallNonExecFile to ensure that target directories exist. Nuke compiled menu support; we don't use it and it makes it harder to deal with the supported platforms field in menus.
- 16:43 Changeset [10841] by
- Assume POSIX signals.
- 16:42 Changeset [10840] by
- Use InstallNonExecFile to ensure that target directories exist.
- 16:41 Changeset [10839] by
- Nuke awful platform-dependent defines in favor of a HOSTTYPE define.
- 16:40 Changeset [10838] by
- Add HOSTTYPE to local imake defines.
12/02/97:
- 15:47 Changeset [10837] by
- Pull up changes from revs 1.12 and 1.13 to Athena 8.1 release branch.
- 15:45 Changeset [10836] by
- Create on Athena 8.1 release branch.
- 15:43 Changeset [10835] by
- Pull up change from rev 1.2 to Athena 8.1 release branch.
- 15:37 Changeset [10834] by
- Pull up change from rev 1.29 to Athena 8.1 release branch.
- 15:29 Changeset [10833]
- This commit was manufactured by cvs2svn to create tag 'xntp3-5_91'.
- 15:29 Changeset [10832] by
- This commit was generated by cvs2svn to compensate for changes in r10831, which included commits to RCS files with non-trunk default branches.
- 15:29 Changeset [10831] by
- ftp://ftp.udel.edu/pub/xntp/xntp3-5.91-export.tar.gz
- 15:24 Changeset [10830] by
- send a HUP to inetds after having possibly replaced their conf file
12/01/97:
- 22:16 Changeset [10829] by
- Don't tell root about the wonderful features of not having its own .xsession file.
- 20:13 Changeset [10828] by
- Garbage collect.
- 20:13 Changeset [10827] by
- athena/bin/mon only builds itself on AIX and BSD. Nuke it.
11/30/97:
- 10:19 Changeset [10826] by
- Update aclocal.m4 from master.
- 10:14 Changeset [10825] by
- Check for possible Kerberos dependencies.
11/28/97:
- 14:41 Changeset [10824] by
- Propagate from Xsession.dt some compatibility for internationalization for the display of xsession.msg. For Irix 6.3, message files moved down into a "C" subdirectory. Required to remain compatible with 5.3, we will fall back to the non-C subdir if necessary. Only call xconfirm on files that actually exist.
11/27/97:
- 17:48 Changeset [10823] by
- There's already a seperate rc scriptfor AFS that will start the AFS server if necessary.
- 17:46 Changeset [10822] by
- Make Athena modifications consider AFSSRV in addition to AFSCLIENT (and deal with a server that isn't a client too)
11/26/97:
- 16:23 Changeset [10821]
- This commit was manufactured by cvs2svn to create branch 'athena-8_1'.
- 16:23 Changeset [10820] by
- Config file to turn off some daemons.
- 16:23 Changeset [10819] by
- Install new files in /etc/config to prevent a number of daemons in Irix 6.3 from being executed. Also renamed CONFIG to SGICONFIG for compatibility with the platform script.
11/25/97:
- 01:06 Changeset [10818] by
- From mwhitson: document -l option.
11/24/97:
- 18:45 Changeset [10817] by
- Don't nuke /usr/local - SGI puts stuff there.
11/22/97:
- 14:27 Changeset [10816] by
- Add spaces after '.I'.
- 14:27 Changeset [10815] by
- Rename ?.nroff to qmark.nroff to placate some versions of make.
- 14:26 Changeset [10814] by
- Add Linux and NetBSD support.
- 14:25 Changeset [10813] by
- Use more portable (if less elegant) method for setting ONLCR for pager.
- 14:25 Changeset [10812] by
- Use strerror; don't echo keystroke in wait_for_key().
- 14:24 Changeset [10811] by
- Need termios.h for struct winsize on NetBSD.
- 14:23 Changeset [10810] by
- Simplify AFS libraries; use -lcompat on NetBSD.
- 14:23 Changeset [10809] by
- Use strerror.
- 14:23 Changeset [10808] by
- Remove terrible unneeded code for finding edsc.
11/21/97:
- 14:50 Changeset [10807] by
- fix a function name
11/20/97:
- 17:20 Changeset [10806] by
- Disallow account operations on invalid usernames. In cleanup functions, check if the session record exists before getting the session record. Don't bother checking for root account in al_acct_revert(), since the session record for root accounts won't exist.
- 17:18 Changeset [10805] by
- Disallow invalid usernames from logging in.
- 17:18 Changeset [10804] by
- Prototype new functions.
- 17:16 Changeset [10803] by
- Add an al__username_valid() to check for illegal characters in a username.
- 17:16 Changeset [10802] by
- Add an al__record_exists() to see if there is an existing record for a user.
- 14:51 Changeset [10801] by
- athena-login
- 14:50 Changeset [10800] by
- get krb5 and krb4 tickets with password, and destroy them afterwards
- 13:58 Changeset [10799] by
- Don't give an error if you can't find a ftp/hostname principal, only give error if you can't find host/hostname.
- 13:57 Changeset [10798] by
- Try to negotiate encryption before logging the user in (in case we need a password.)
- 12:40 Changeset [10797] by
- Don't do anything if we're root.
11/19/97:
- 18:18 Changeset [10796] by
- Make /var/athena/sessions on the packs. Mode 700 so that people can't see who's been logged in and when by looking at the mod times.
- 17:37 Changeset [10795] by
- krb5 may need -lgen for regexp stuff
- 15:52 Changeset [10794] by
- small security fix (originally from Matt Power)
- 15:52 Changeset [10793] by
- Fix make depend rule
- 15:44 Changeset [10792] by
- do chown later
- 13:37 Changeset [10791] by
- Add kermit.
- 11:55 Changeset [10790] by
- work in Athena build system
11/18/97:
- 22:05 Changeset [10789] by
- From epeisach: initialize p. Oops.
- 22:04 Changeset [10788] by
- From epeisach: need <afs/ktime.h> for compatibility macro.
- 17:09 Changeset [10787] by
- s/configure:/prepare:/
- 17:06 Changeset [10786] by
- Sync with reality of do.sh script.
- 14:40 Changeset [10785] by
- Revert last change; we're going to do it differently.
- 14:32 Changeset [10784] by
- fix Makefile syntax
- 14:10 Changeset [10783] by
- Whups, we need to check for networking functions for Solaris.
- 14:00 Changeset [10782] by
- Add note for kermit. Clarify that the $COMPILER is also the linker.
- 13:28 Changeset [10781]
- This commit was manufactured by cvs2svn to create tag 'kermit-6_0'.
- 13:28 Changeset [10780] by
- This commit was generated by cvs2svn to compensate for changes in r10779, which included commits to RCS files with non-trunk default branches.
- 13:28 Changeset [10779] by
- 6.0 from http://www.columbia.edu/kermit
- 13:26 Changeset [10778] by
- Add ssh.
11/17/97:
- 20:28 Changeset [10777] by
- Oops, hit another setuid case.
- 19:48 Changeset [10776] by
- setuid to real_uid, not owner_uid. Closes a security hole, and allows root to do attaches with ticket files owned by root.
- 17:40 Changeset [10775] by
- Add media lab and zone, fix Cygnus
- 17:10 Changeset [10774] by
- document "-a cred"
- 17:07 Changeset [10773] by
- implement "-a cred"
- 17:07 Changeset [10772] by
- implement "-a cred" use net_write instead of printf for al warnings
- 17:06 Changeset [10771] by
- close stdout and stderr before exec'ing
- 15:59 Changeset [10770] by
- Garbage collect gnu malloc.
- 15:58 Changeset [10769] by
- Add some sed magic to pretend that SGI has a modern rpcgen. Yech.
- 15:52 Changeset [10768] by
- From epeisach: don't display error message if we have no srvtab.
- 14:19 Changeset [10767] by
- Only set record->exists if the record really did exist.
- 12:24 Changeset [10766] by
- --enable-athena is no more.
- 11:26 Changeset [10765] by
- Use ktime_DateToLong for compatibility with 3.3 libraries. The argument is still an int32, which is potentially broken; the reason is that Transarc was boneheaded and defined ktime_DateToLong to ktime_DatetoInt32 to do backward compatibility in 3.4a, and I don't want to break things on new platforms.
- 11:25 Changeset [10764] by
- Use an autoconf build system.
- 11:23 Changeset [10763] by
- ANSIfy a bit.
11/15/97:
- 18:59 Changeset [10762] by
- Add an explanatory comment.
- 15:26 Changeset [10761] by
- Remove bogus input redirects from /dev/tty.
- 13:44 Changeset [10760] by
- Fix a waitpid return value check
- 02:31 Changeset [10759] by
- al__get_session() now guarantees an extra slot in the pids array. Use that slot to add sessionpid.
- 02:30 Changeset [10758] by
- Fix multiple problems: * We never free buf in al__get_session_record(). * There were a number of direct returns which did not clean things up properly. * On a warning return, we need to block signals. * We sometimes try to malloc 0 bytes, which can return NULL, yielding a false error. Add to the contract that we allocate an extra slot in the pid and gid arrays; useful (in the pid case) for acct.c.
11/14/97:
- 19:04 Changeset [10757] by
- Use atexit() functions to destroy tickets and call al_acct_revert. Work around Solaris lossage with libucb and grantpt.
- 17:33 Changeset [10756] by
- Garbage collect.
- 17:30 Changeset [10755] by
- Rename server calls to *_svc() to conform to ANSI C rpcgen stubs. We no longer need to put underscores before calls to client stubs, or cast the result, because client stubs have different names than server calls.
- 17:29 Changeset [10754] by
- We no longer need to put underscores before calls to client stubs, or cast the result, because client stubs have different names than server calls.
- 17:29 Changeset [10753] by
- Rename server calls to *_svc() to conform to ANSI C rpcgen stubs. Remove an abortive piece of Solaris compatibility; we deal by defining PORTMAP in the include file instead. We no longer need to put underscores before calls to client stubs, or cast the result, because client stubs have different names than server calls.
- 17:27 Changeset [10752] by
- Use strerror.
- 17:27 Changeset [10751] by
- Hack for NetBSD sunrpc. Not sure if this ought to be necessary.
- 17:26 Changeset [10750] by
- Rely on native sunrpc support.
- 17:25 Changeset [10749] by
- Use rpcgen -C to generate stubs with different names for client and server calls (and prototypes, and stuff). We can punt local_prot_clnt.c then, and more importantly, we're compatible with some modern versions of rpcgen (e.g. the ones in Linux and NetBSD).
- 17:23 Changeset [10748] by
- Punt link_with_fxrpc check; we want to use native sunrpc support.
- 17:22 Changeset [10747] by
- Hack to get backward compatibility declarations for Solaris. I'm not sure if there's a better way of doing this, since tirpc didn't really catch on outside of Sun. Maybe Solaris 2.6 will render this change unnecessary?
- 17:21 Changeset [10746] by
- Rely on native sunrpc.
- 17:21 Changeset [10745] by
- Build the server everywhere, but don't build rpc3.9 any more; we're going to rely on native sunrpc. Also punt fxtest.
- 17:18 Changeset [10744] by
- Update aclocal.m4 from master.
- 17:10 Changeset [10743] by
- Fix -I/-L botch in ATHENA_UTIL_COM_ERR and ATHENA_UTIL_SS. For sigvec test, put "-lc -lucb" at end of LIBS to avoid libucb lossage. Give afs prefix in AFS_DIR for things like libdes and util.a.
- 14:24 Changeset [10742] by
- What a bogus "sanity check"; why shouldn't you be able to configure perl without a tty? Nuke it.
11/13/97:
- 18:40 Changeset [10741] by
- Clean up. Nuke old platform support, indent code consistently, switch on HOSTTYPE (`/bin/athena/machtype`), do a better job of cleaning temporary areas, clean /var/athena/tmphomedirs, use /var/athena instead of /var/tmp to store the ps_data file on Solaris.
- 18:26 Changeset [10740] by
- If the user has no primary group entry, don't generate a warning, but don't add the user to any groups. If the user doesn't have a grplist entry, don't generate a warning; return a one-element group list with the primary group.
- 17:11 Changeset [10739] by
- Multiple fixes: * gids weren't being checked for whether they are numeric. (I made a function to parse up to the gid field, which unfortunately required me to un-const-poison several of my pointers.) * strncmp is safer than memcmp where it's used. * some code was printing a gid_t as a %d. * updating the group file wasn't checking ferror() and fclose() status * al__read_line() return status wasn't checked * retrieve_hesgroups didn't clean up grplistvec if *grplistvec is false * There was a confusing comment about writing out an edited line. * The code should check for zero-length group names and non-numeric gids.
- 16:59 Changeset [10738] by
- s/WNOTMPDIR/WNOHOMEDIR/, which is more clear.
- 16:59 Changeset [10737] by
- Multiple fixes: * local_pwd wasn't claned up in two places. * AL_WNOTMPDIR is more clearly read as AL_WNOHOMEDIR. * We shouldn't free hes_pwd if it's NULL. * The status-26 check for attack is obsolete with the current library interface. * The malloc for the temporary directory should use strlen(PATH_TMPDIRS). * We don't want to nuke the temporary directory here. (We do want to try to rmdir it if we failed to copy the temp files in, though.) We'll do it from a cron job. * The temporary directory creation code would create "tmphomedir" in the user's home directory. It's difficult to come up with a simple solution to this problem, because you don't want to introduce security holes where the user could play around with symlinks while you're diddling with the filesystem as root. I changed the code to copy each file individually, doing the copy as the user. It didn't add too much code.
- 16:57 Changeset [10736] by
- Multiple fixes: * The hesiod passwd entry wasn't always freed. * The shadow file code was assuming that the shadow file was in sync with the passwd file. * lock_passwd() didn't call ulckpwdf() on System V systems when fopen() returned NULL. * It also didn't remove the temporary file when fdopen() returned NULL. * It was using the wrong mode for the lock file on master.passwd systems. * update_passwd() used a gratuitous vfork(). * update_passwd() wasn't checking ferror(), and neither were the pieces of code to update the shadow file. * update_passwd() wasn't calling ulckpwdf() in all cases. * It also didn't remove the temporary file when rename() failed. * The waitpid() call should check for EINTR. * buf wasn't freed in change_passwd_homedir(). * There was an extra hesiod cleanup in the handler of al__add_to_passwd(). * The shadow temp file should be mode 600. * Non-readable files were created with fopen() and fchmod'd, which is a security hole (the attacker has a window to open the file and read the contents written out later). * The "skip to homedir" comment preceeds a confusing block of code, and needed expansion.
- 16:56 Changeset [10735] by
- Multiple fixes: * al_strerror() made assumptions about the order of error codes. * There was no explanation of why we had the al__getpw* functions. * Comments after #else statements should be consistent between util.c and passwd.c. * passwd lookup routines should check for zero-length usernames and for non-numeric uids and gids in the passwd file. * The BSD 4.4 lookup function should be in sync with the libc version as far as how to decide which database to open. * There was a misformatted line. * The al__read_line() comment needed clarification. * In one case, strcpy() makes more sense than memcpy().
- 13:12 Changeset [10734] by
- Merge with perl 5.004_04. Previous local modifications are no longer necessary; will revert RCS file to vendor branch after this commit.
- 10:56 Changeset [10733] by
- For perl 5, redo the prepare rule. Generate config.sh using ./Configure -ds -E, use sed to add $$SRVD to all of the install paths, and then run ./Configure -S. Also, h2ph is built in utils/h2ph now, for the install rule.
- 10:53 Changeset [10732] by
- These files replaced by versions in utils directory in perl 5.
- 10:52 Changeset [10731] by
- No longer needed for perl build.
- 10:36 Changeset [10730] by
- Merge with perl 5.004_04. Previous local modifications are no longer necessary; will revert RCS file to vendor branch after this commit.
- 10:16 Changeset [10729] by
- Explain how to delete removed files after an import, and revert files to the vendor branch.
11/12/97:
- 22:15 Changeset [10728] by
- The change is to replace the two listed medialab servers with the actual current system.
- 21:41 Changeset [10727]
- This commit was manufactured by cvs2svn to create tag 'perl-5_004_04'.
- 21:41 Changeset [10726] by
- This commit was generated by cvs2svn to compensate for changes in r10725, which included commits to RCS files with non-trunk default branches.
- 21:41 Changeset [10725] by
- This file went away in perl 5.004_04.
- 20:50 Changeset [10724] by
- This commit was generated by cvs2svn to compensate for changes in r10723, which included commits to RCS files with non-trunk default branches.
- 20:50 Changeset [10723] by
- Import of perl 5.004_04
- 16:16 Changeset [10722] by
- Athena-login changes (including some krb4 stuff)
11/11/97:
- 14:33 Changeset [10721] by
- Fix syntax.
- 14:33 Changeset [10720] by
- Nuke bogus i386-conditionalized code.
- 14:32 Changeset [10719] by
- Include lex code after we're sure yacc declarations have been output. We're doing direct shell globbing; no need to convert patterns to regexps.
- 14:31 Changeset [10718] by
- Punt regexps for a real shell glob match.
- 14:31 Changeset [10717] by
- Ditch alloca for malloc.
- 14:30 Changeset [10716] by
- Get declarations from stdlib.h. Include lex code after we're sure the yacc declarations have been output.
- 14:26 Changeset [10715] by
- Use strerror().
- 10:43 Changeset [10714] by
- Automatically generated; nuke.
- 10:38 Changeset [10713] by
- Stop building athena/lib/gdb. With moira in its own locker, we shouldn't need it any more.
- 10:35 Changeset [10712] by
- Update aclocal.m4 from master.
- 10:31 Changeset [10711] by
- Always put include/kerberosIV in CPPFLAGS, in case a direct Kerberos 4 distribution needs it. NetBSD is the main known example. Only do so if the directory exists, though. Look for krb5 compatibility before diret krb4 support. This is not the best plan; at some point the code should try to decide which version exists in the specified prefix. But that might not be simple enough.
11/10/97:
- 21:26 Changeset [10710] by
- Fix two errors with reading off the end of an array, and make the deletion of creatorpid in al_acct_revert() more robust.
- 20:11 Changeset [10709] by
- If we are invoking login, do not chown the forwarded ticket file, since login will take care of it and it creates a security hole to try to take care of it twice. Ergo, do the chown in krshd.c and not in forward.c.
11/07/97:
- 13:16 Changeset [10708] by
- Pull up change from rev 1.6 to Athena 8.1 release branch.
- 13:15 Changeset [10707] by
- sis isn't an attach and run script any more.
Note: See TracTimeline
for information about the timeline view.