aboutsummaryrefslogtreecommitdiff
path: root/src/util/pty/ChangeLog
diff options
context:
space:
mode:
Diffstat (limited to 'src/util/pty/ChangeLog')
-rw-r--r--src/util/pty/ChangeLog166
1 files changed, 166 insertions, 0 deletions
diff --git a/src/util/pty/ChangeLog b/src/util/pty/ChangeLog
index ff815b7..3dac1a3 100644
--- a/src/util/pty/ChangeLog
+++ b/src/util/pty/ChangeLog
@@ -1,3 +1,169 @@
+2001-11-28 Tom Yu <tlyu@mit.edu>
+
+ * update_utmp.c (PTY_GETUTXENT): Fix typo. Thanks to Shawn
+ Stepper. [fixes krb5-build/1020]
+
+2001-11-19 Tom Yu <tlyu@mit.edu>
+
+ * update_utmp.c (pty_update_utmp): Patch from Garry Zacheiss to
+ kludge around cases where we need to use more than 2 characters of
+ LINE in order to avoid conflicts in UT_ID.
+
+2001-11-05 Tom Yu <tlyu@mit.edu>
+
+ * Makefile.in (LIBMINOR): Bump due to changes to internals.
+
+2001-09-25 Tom Yu <tlyu@mit.edu>
+
+ * pty-int.h: Fix up botched merge: cause prototypes for
+ ptyint_update_wtmp{,x} to be correct.
+
+2001-09-07 Tom Yu <tlyu@mit.edu>
+
+ * update_utmp.c (pty_update_utmp): Remember to chop off leading
+ "/dev/" for the non-sysV case. Handle lseek() returning non-zero
+ yet non-negative values (it usually does... :-), so that we can
+ actually write somewhere not at the beginning of the utmp file if
+ necessary.
+
+ * update_utmp.c (pty_update_utmp): Don't copy host if it's a null
+ pointer.
+
+ * dump-utmp.c (print_ut): Use size of ut_name field, not ut_user,
+ which may not exist, for width when printing ut_name field value.
+ Specify width when printing hostname, it may be unterminated.
+ (main): Move utp and utxp declarations closer to their usages, and
+ make both conditionalized so they're not declared if they're not
+ used.
+
+ * getpty.c: Make pty_getpty() into ptyint_getpty_ext(), which has
+ an extra argument that determines whether to call grantpt() and
+ unlockpt() on systems that support it. The new pty_getpty() will
+ simply call the extended version. This is to support some
+ wackiness needed by pty_paranoia.c tests.
+
+ * pty-int.h: Add prototype for ptyint_getpty_ext().
+
+ * pty_paranoia.c: Add rant about ptys and quirks therein. Needs
+ to be updated somewhat. Add some more paranoia for the case where
+ we actually succeed in opening the slave of a closed master and
+ then succeed in opening the same master. This program will get
+ rewritten at some point to actually see what things result in EOFs
+ and under what conditions data will actually get passed between
+ master and slave.
+
+ * pty_paranoia.c: New file; do many paranoid checks about ctty
+ handling by the pty drivers.
+
+ * Makefile.in: Add rules for pty_paranoia and check-paranoia,
+ which runs pty_paranoia.
+
+ * configure.in: Define REVOKE_NEEDS_OPEN for Tru64. Add support
+ for program building and run flags for the sake of pty_paranoia.
+
+ * open_slave.c: Fix somewhat; AIX doesn't like opening the ctty
+ twice, so only do initial open if we special-case it in
+ configure.in, e.g. for Tru64.
+
+ * logwtmp.c: Delete code under "#if 0". Fix reversed test for
+ loggingin. Don't forget to set the ut_tv or ut_time for the
+ entry.
+
+ * update_utmp.c: Update rant about Tru64; remove fetching of
+ ut_user from old entry. The existence of the old ut_user in the
+ logout entry in wtmp was confusing last.
+
+ * cleanup.c: Call update_utmp() with the correct pid to assist in
+ finding the old utmp entry.
+
+ * open_ctty.c: Reformat somewhat and revise comment.
+
+ * open_slave.c: Rework significantly. Primarily, keep a fd open
+ to the slave if we need to reopen the slave device following
+ vhangup() or revoke(), to accommodate various OS quirks.
+
+ * update_utmp.c: Revise history section somewhat to document more
+ HP-UX brokenness. Search via ut_pid before searching via
+ ut_line. Copy stuff around because entuxent() will clobber some
+ things.
+
+ * void_assoc.c: Revise comment and reformat somewhat.
+
+ * open_slave.c (pty_open_slave): If revoke() present on system but
+ VHANG_FIRST is not defined, declare local variable.
+
+ * dump-utmp.c: Fix some off-by-one errors. Handle cases where we
+ have utmpname() but not utmpname().
+
+ * pty-int.h: Fix typo; VHANG_first -> VHANG_FIRST.
+
+ * open_slave.c (pty_open_slave): Add workaround for Tru64 v5.0,
+ since its revoke() will fail if the slave isn't open already.
+
+ * cleanup.c (pty_cleanup): Delcare local variable only if
+ VHANG_LAST defined.
+
+ * logwtmp.c (pty_logwtmp): Only declare local variables if
+ logwtmp() not available on system.
+
+ * update_utmp.c (pty_update_utmp): Fix typo (OWRONLY ->
+ O_WRONLY).
+
+ * update_wtmp.c (ptyint_update_wtmpx): Add missing semi-colon in
+ code path if PTY_UTMP_E_EXIT and PTY_UTMPX_E_EXIT exist.
+
+ * configure.in: Fix some quoting of shell variables when passing
+ to "test". Reorder some logic in consistency checks to validate
+ cache variables against "yes" to account for possible empty or
+ nonexistent values.
+
+ * pty-int.h: Fix conditional prototype of update_wtmp().
+
+ * update_wtmp.c: Fix conditional compilation of update_wtmp() to
+ cover the case where we have setutxent() but don't have updwtmpx()
+ and WTMPX_FILE, as is the case on some Linux installations.
+
+ * configure.in(K5_CHECK_UT_MEMBER): Fix typo in previous; make
+ sure to include the correct header when checking structure
+ members.
+
+ * configure.in: Many changes to support the rewriting of the utmp
+ pieces of libpty. Do a large amount of checking for consistency
+ of various utmp and utmpx APIs as currently understood. See rant
+ in update_utmp.c.
+
+ * dump-utmp.c: Rewrite; now has capability to use utmp{,x}name()
+ to extract entries from utmp and utmpx files. Adjusts field
+ widths when printing as appropriate.
+
+ * libpty.h: Update call signature for update_utmp() and logwtmp();
+ make prototypes unconditional.
+
+ * logwtmp.c: Rewrite. Use pututline() or pututxline() API
+ whenever possible.
+
+ * pty-int.h: Update call signatures for update_wtmp{,x}(); make
+ prototypes unconditional.
+
+ * sane_hostname.c: Use the autoconf-correct macro names.
+
+ * update_utmp.c: Rewrite. Basically, use functions from the
+ pututline() or pututxline() API whenever possible, to avoid
+ lossage. Inserted large rant about the conjectured history of BSD
+ utmp, sysV utmp, and utmpx, as well as documentation about some
+ known quirks.
+
+ * update_wtmp.c: Rewrite. Add new function ptyint_logwtmpx() that
+ takes a utmpx rather than a utmp, so it can fail to lose data
+ converting to and from utmp.
+
+ [many pullups from trunk]
+
+2000-03-24 Ken Raeburn <raeburn@mit.edu>
+
+ * configure.in: Check for alpha*-dec-osf* instead of
+ alpha-dec-osf*.
+
1999-10-26 Tom Yu <tlyu@mit.edu>
* Makefile.in: Clean up usage of CFLAGS, CPPFLAGS, DEFS, DEFINES,