His explanation:

 The socket on which MySQL listens for new connections on a blocking
  socket most of the time but is set to non-blocking during the
  accept() of the new connection.  Due to a bug in the kernel, the new
  socket returned by accept() is a blocking socket but returns the
  O_NONBLOCK flag when queried via fcntl(F_GETFL).  That is, the file
  descriptor and the underlying socket don't agree on the blocking
  mode.

  Since MySQL determines via fcntl(F_GETFL) that the socket is
  non-blocking, it expects the first read() in my_real_read to not
  block, so it doesn't enable the timeout alarm.  However, the read
  does block, and thus there's no timeout alarm.  The thread kill
  (which relies on rescheduling the timeout alarm) also does not work
  as a consequence.

The bug shows itself if you build MySQL with LinuxThreads support
(needed for SMP on FreeBSD).  Issuing a KILL command in MySQL won't be
"noticed" by the "killed" thread until it runs another query--that
makes KILL pretty useless.  And the wait_timeout doesn't work either.


vio/vio.c:
  FreeBSD patch by Jeremy Zawodny

into sinisa.nasamreza.org:/mnt/work/mysql-4.0

  Flush stderr if we run out of memory, so that the error message more probably finds its way to the error log


innobase/ut/ut0mem.c:
  Flush stderr if we run out of memory, so that the error message more probably finds its way to the error log

into sanja.is.com.ua:/home/bell/mysql/mysql-4.0

excluded double call of 'invalidate()'


mysql-test/r/innodb_cache.result:
  test of invalidation
mysql-test/t/innodb_cache.test:
  test of invalidation
sql/handler.cc:
  excluded double call of 'invalidate()'
sql/sql_delete.cc:
  fixed invalidation of query cache
sql/sql_insert.cc:
  fixed invalidation of query cache
sql/sql_update.cc:
  fixed invalidation of query cache

 - actually follow the hints in the comment above and "Remember that 
   regexps needs to quote [ and ] since this is run through m4"...


configure.in:
   - actually follow the hints in the comment above and "Remember that 
     regexps needs to quote [ and ] since this is run through m4"...

 - fix MYSQL_NO_DASH_VERSION if version number ends on one digit only
   ("comment" test failed when MySQL version was changed from 4.0.5
   to 4.0.5a)


configure.in:
   - fix MYSQL_NO_DASH_VERSION if version number ends on one digit only
     ("comment" test failed when MySQL version was changed from 4.0.5
     to 4.0.5a)

delete from table where column<=>NULL 
on indexed columns

Fix for shutdown on Mac OS X


include/my_tree.h:
  Try to optimize the cache buffer size needed for bulk_insert
myisam/mi_write.c:
  Try to optimize the cache buffer size needed for bulk_insert
mysql-test/r/bdb.result:
  Make test repeatable
mysql-test/t/bdb.test:
  Make test repeatable
mysys/tree.c:
  Try to optimize the cache buffer size needed for bulk_insert
sql/mysql_priv.h:
  Small optimization
sql/mysqld.cc:
  Fix for shutdown on Mac OS X
sql/sql_insert.cc:
  Try to optimize the cache buffer size needed for bulk_insert
sql/sql_yacc.yy:
  Call thd->strmake() instead of sql_strmake()
sql/table.cc:
  Try to optimize the cache buffer size needed for bulk_insert
sql/table.h:
  Try to optimize the cache buffer size needed for bulk_insert

into serg.mysql.com:/usr/home/serg/Abk/mysql-4.0

  Fix potential bus error in 64-bit computers in SHOW VARIABLES, if int is only 32-bit in them


sql/set_var.cc:
  Fix potential bus error in 64-bit computers in SHOW VARIABLES, if int is only 32-bit in them

  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp


sql/handler.h:
  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp
sql/sql_table.cc:
  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp
sql/ha_innodb.cc:
  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp
innobase/include/row0mysql.h:
  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp
innobase/row/row0mysql.c:
  Fix crash when InnoDB temp table is truncated + fix bug: do not X-lock rows under LOCK TABLES except if the table is temp

BitKeeper/etc/ignore:
  auto-union
BitKeeper/deleted/.del-depcomp:
  Delete: depcomp
config.guess:
  Auto merged
config.sub:
  Auto merged
configure.in:
  Auto merged
dbug/Makefile.am:
  Auto merged
heap/Makefile.am:
  Auto merged
install-sh:
  Auto merged
isam/Makefile.am:
  Auto merged
libmysql_r/Makefile.am:
  Auto merged
ltconfig:
  Auto merged
ltmain.sh:
  Auto merged
missing:
  Auto merged
myisam/Makefile.am:
  Auto merged
mysys/Makefile.am:
  Auto merged
mysys/mf_keycache.c:
  Auto merged
regex/Makefile.am:
  Auto merged
sql/Makefile.am:
  Auto merged
strings/Makefile.am:
  Auto merged

into genie.(none):/home/vva/work/mysqldump_xml/4.0

client/mysqldump.c:
  fix xml-output

configure.in:
  Auto merged

into geopard.(none):/storage/bk/my-new-4.0

Redo of changeset I commited, but failed to push


mysql-test/r/bool.result:
  Added tests for boolean operators in select part
mysql-test/t/bool.test:
  Added tests for boolean operators in select part

into mysql.com:/space/my/mysql-4.0-build


BitKeeper/etc/ignore:
  auto-union
sql-bench/test-create.sh:
  Auto merged

 - bumped up version number in configure.in
 - fixed a typo in test-create.sh


BitKeeper/etc/ignore:
  Added autom4te.cache/output.0 autom4te.cache/requests autom4te.cache/traces.0 bdb/dist/autom4te.cache/output.0 bdb/dist/autom4te.cache/requests bdb/dist/autom4te.cache/traces.0 innobase/autom4te.cache/output.0 innobase/autom4te.cache/requests innobase/autom4te.cache/traces.0 to the ignore list
configure.in:
   - Bumped up version number to 4.0.6 now that 4.0.5 is tagged (4.0.6 is
     now labelled "gamma", according to conversation with Monty)
sql-bench/test-create.sh:
   - typo fixed

into sanja.is.com.ua:/home/bell/mysql/mysql-4.0

mysql-test/r/query_cache.result:
  test of new status variable
mysql-test/t/query_cache.test:
  test of new status variable

sql/sql_select.cc:
  optimizer bug fixed. (what's the difference between s->records ans s->found_records ?)

 - "make distcheck" needs to be fixed


BitKeeper/etc/ignore:
  Added autom4te.cache/output.0 autom4te.cache/requests autom4te.cache/traces.0 bdb/dist/autom4te.cache/output.0 bdb/dist/autom4te.cache/requests bdb/dist/autom4te.cache/traces.0 innobase/autom4te.cache/output.0 innobase/autom4te.cache/requests innobase/autom4te.cache/traces.0 to the ignore list
acinclude.m4:
   - removed libtool.m4 (is part of libtool 1.4)
config.guess:
   - applied diffs from config.guess of automake 1.5
config.sub:
   - applied diffs from config.sub of automake 1.5
configure.in:
   - renamed "AM_PROG_LIBTOOL" -> "AC_PROG_LIBTOOL"
   - added "AM_PROG_AS" for automake 1.5
dbug/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
extra/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
heap/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
install-sh:
   - applied diffs from install.sh of automake 1.5
isam/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
libmysql_r/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
ltconfig:
   - small correction for new libtool
ltmain.sh:
   - applied diffs from ltmain.sh of libtool 1.4.2
merge/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
missing:
   - applied diffs from missing of automake 1.5
myisam/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
myisammrg/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
mysys/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
   - added $(EXEEXT) to test_charset
regex/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
sql/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
strings/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy
   - removed @CHARSET_SRCS@
vio/Makefile.am:
   - removed OMIT_DEPENDENCIES to make automake 1.5 happy

into serg.mysql.com:/usr/home/serg/Abk/mysql-4.0


scripts/mysqlhotcopy.sh:
  Auto merged
sql-bench/test-create.sh:
  Auto merged
support-files/mysql.server.sh:
  Auto merged

sql/ha_innodb.cc:
  Auto merged
sql/ha_myisam.cc:
  Auto merged
sql/sql_delete.cc:
  Auto merged
sql/sql_update.cc:
  Auto merged

Add support for LIMIT # OFFSET #
Changed lock handling:  Now all locks should be stored in TABLE_LIST instead of passed to functions.
Don't call query_cache_invalidate() twice in some cases
mysql_change_user() now clears states to be equal to close + connect.
Fixed a bug with multi-table-update and multi-table-delete when used with LOCK TABLES
Fixed a bug with replicate-do and UPDATE


BitKeeper/etc/ignore:
  added autom4te.cache/* bdb/dist/autom4te.cache/* innobase/autom4te.cache/*
include/my_alloc.h:
  Small improvement to alloc_root
libmysql/libmysql.c:
  Removed compiler warning
myisam/mi_page.c:
  Better DBUG message
mysql-test/r/multi_update.result:
  Added test with lock tables
mysql-test/r/rpl_replicate_do.result:
  Update results
mysql-test/r/rpl_rotate_logs.result:
  Make test independent of if t1 exists
mysql-test/t/multi_update.test:
  Added test with lock tables
mysql-test/t/rpl_rotate_logs.test:
  Make test independent of if t1 exists
mysys/my_alloc.c:
  Small imprevement to alloc_root
  (Don't free blocks less than ALLOC_MAX_BLOCK_ROOT (4K)
sql/ha_innodb.cc:
  More debug messages
sql/ha_myisam.cc:
  Safety change
sql/lex.h:
  Add support for LIMIT # OFFSET #
sql/lock.cc:
  Added assertion
sql/mysql_priv.h:
  Change of lock handling
sql/mysqld.cc:
  Added function clear_error_messages()
sql/sql_base.cc:
  Change lock handling by open_ltable() and open_and_lock_tables()
sql/sql_class.cc:
  Split THD::THD to two functions
  Move some code from cleanup() to ~THD:THD
  Add THD::change_user()
sql/sql_class.h:
  Prototype changes in class THD
sql/sql_delete.cc:
  Remove locking argument from mysql_delete()
  Locking type is now stored in TABLE_LIST
  Small code change to not call query_cache_invalidate() twice for transactional tables.
sql/sql_insert.cc:
  Remove locking argument from mysql_insert()
  Locking type is now stored in TABLE_LIST
  Small code change to not call query_cache_invalidate() twice for transactional tables.
  Don't use bulk insert if bulk_insert_buff_size is 0
sql/sql_parse.cc:
  Changes to make mysql_change_user() work as close+connect
  Changed command statistics to use statstics_increment to get more speed
  Update code to handle that locks is now stored in TABLE_LIST
sql/sql_update.cc:
  Remove locking argument from mysql_update()
  Locking type is now stored in TABLE_LIST
  Small code change to not call query_cache_invalidate() twice for transactional tables.
sql/sql_yacc.yy:
  Locking type is now stored in TABLE_LIST
  Added support for LIMIT # OFFSET # syntax
  Removed some wrong (never true) checks for SQLCOM_MULTI_UPDATE
mysql-test/t/rpl_replicate_do-slave.opt:
  Changed tables to use t1,t2,...
mysql-test/t/rpl_replicate_do.test:
  Changed tables to use t1,t2,...

  Print an error message if someone tries to call InnoDB startip more than once during process lifetime


innobase/srv/srv0start.c:
  Print an error message if someone tries to call InnoDB startip more than once during process lifetime