____ _ _ | _ \| |_| |__ | |_) | __| '_ \ ``Just because everything is different | __/| |_| | | | doesn't mean anything has changed.'' |_| \__|_| |_| -- Irene Peter GNU Pth - The GNU Portable Threads ChangeLog ========= This is the list of all(!) changes to this Pth source tree. For a list of just the user-visible and/or major changes please have a look at the NEWS file. ____ ___ |___ \ / _ \ __) || | | | / __/ | |_| | __|_____(_)___/_________________________________________________________ Changes between 2.0.6 and 2.0.7 (22-Nov-2005 to 08-Jun-2006) *) Upgraded build environment from GNU shtool 2.0.3 to 2.0.6 and from GNU libtool 1.5.20 to 1.5.22. [Ralf S. Engelschall] *) Let AC_CHECK_PTH use "pth-config --libs --all" instead of just "pth-config --libs" to allow it to work under platforms like Solaris where additional libraries are required. [Ralf S. Engelschall] *) Adjusted all copyright messages for new year 2006. [Ralf S. Engelschall] Changes between 2.0.5 and 2.0.6 (05-Oct-2005 to 22-Nov-2005) *) Add a missing "break" in pth_attr.c:pth_attr_ctrl(). [Stefan Schippers ] *) Return an error if pth_kill() is called before pth_init(). [Martin Kraemer ] Changes between 2.0.4 and 2.0.5 (18-Feb-2005 to 05-Oct-2005) *) Upgraded build environment to GNU libtool 1.5.20 and GNU shtool 2.0.3 [Ralf S. Engelschall] Changes between 2.0.3 and 2.0.4 (03-Dec-2004 to 18-Feb-2005) *) Fix internal determination of maximum number in a size_t in the string formatter engine. [Ralf S. Engelschall] *) Upgraded build environment to GNU libtool 1.5.14 [Ralf S. Engelschall] *) Add quotes around function names for AC_DEFUN in pth.m4 to avoid warnings about "underquoted definitions". [Stephane Loeuillet ] *) Adjusted all copyright messages for new year 2005. [Ralf S. Engelschall] *) Do not insist on writing to the source directory on "make install" in order to more closely follow GNU standards. [Bruno Haible ] *) Remove "*.bak" and "*~" files under "make distclean" instead of "make clean". [Bruno Haible ] Changes between 2.0.2 and 2.0.3 (12-Sep-2004 to 03-Dec-2004) *) Resurrect the old "make striptease" command by fixing the configure run-time in the stripped down source tree and fixing the list of source files. [Ralf S. Engelschall] *) The pth_uctx_save() and pth_uctx_restore() API functions unfortunately were broken by design because they are C _functions_. This leads to one more deadly nesting on the run-time stack which effectively caused the pth_mctx_restore() in pth_uctx_restore() to return to the end of pth_uctx_save() but then the control flow unfortunately returns to the pth_uctx_restore() caller instead of the pth_uctx_save() caller because the call to pth_uctx_restore() had already overwritten the run-time stack position where the original return address for the pth_uctx_save() call was stored. The only workaround would be to #define pth_uctx_save() and pth_uctx_restore() as C _macros_, but this then would require that lots of the GNU Pth internals from pth_mctx.c would have to be exported in the GNU Pth API (which in turn is not acceptable). So, the only consequence is to remove the two functions again from the GNU Pth API. [Ralf S. Engelschall, Stefan Brantschen ] *) Enhance internal pth_mctx_save() if getcontext(3) is used for the machine context saving by better emulating the setjmp(3) style return code semantics. [Ralf S. Engelschall] *) Do not break in pth_msgport_find() if a message port was created with a NULL name. [Ralf S. Engelschall] *) Better error checking in pth_key_xxx() functions. [Ralf S. Engelschall] *) Added PTH_CTRL_FAVOURNEW control which allows the user to disable the favouring of new threads on scheduling to get more strict priority based scheduling behavior. [Ralf S. Engelschall, Vinu V ] *) Upgraded build environment to GNU libtool 1.5.10 [Ralf S. Engelschall] Changes between 2.0.1 and 2.0.2 (13-Jul-2004 to 12-Sep-2004) *) Added hard syscall wrappers for send(2) and recv(2). [NetBSD pkgsrc patches] *) Internally handle errors returned from pth_event() in order to pass them upstream to the caller in pth_high.c functions. [Ralf S. Engelschall, NetBSD pkgsrc patches] *) Fix syscall wrapper for sendto(2). [NetBSD pkgsrc patches] *) Use "(char *)NULL" instead of plain "NULL" in last argument to execl(2) to avoid GCC 3.5's "warning: missing sentinel in function call". [Ralf S. Engelschall] *) Workaround Autoconf/m4 problems in pth.m4 by using "m4_define" instead of just "define" and no using m4_undefine at all. [Dr. Andreas Mueller ] *) Avoid side-effects by defining _XOPEN_SOURCE only temporarily for inclusion of instead of permanently. [Graham Nash ] *) Workaround GCC 2.x and 3.5 (3.4 worked fine) macro parsing behaviour by using a pre-declared function pointer type "typedef int (*pth_event_func_t)(void *);" instead of using an inlined "int (*)(void *)". [Ralf S. Engelschall, Robert Anderson ] *) Fixed prototype for pthread_attr_setschedparam(3) from to use a "const struct sched_param *" argument. [Rafael Ávila de Espíndola ] *) Upgraded build environment to GNU libtool 1.5.8 and GNU shtool 2.0.1. [Ralf S. Engelschall] Changes between 2.0.0 and 2.0.1 (17-Feb-2003 to 13-Jul-2004) *) Fix handling/casting of "int (*)(void *)" callback functions. [Ralf S. Engelschall] *) Resolve symbol conflict in pth_string.c between pow10/round and GCC builtins. [Ralf S. Engelschall] *) Use GCC 3.3 option "-fno-strict-aliasing" (if available) under Autoconf option "--enable-debug" because mainly pth_mctx.c contains important and correct pointer casting constructs which are not acceptable in "strict aliasing" for GCC. [Ralf S. Engelschall] *) Upgraded build environment to GNU autoconf 2.59, GNU libtool 1.5.6 and GNU shtool 2.0.0. [Ralf S. Engelschall] *) Fixed implementation so that --enable-pthread now correctly implies --enable-syscall-soft. [Steve Alstrin , Ralf S. Engelschall] *) Replace some 0 by more politically correct NULL in pth_string.c [Ralf S. Engelschall] *) Adjusted all copyright messages for new year 2004. [Ralf S. Engelschall] Changes between 2.0b2 and 2.0.0 (04-Dec-2002 to 17-Feb-2003) *) Final source tree polishing before release. [Ralf S. Engelschall] *) Wrote summary for Pth 2.0 for the ANNOUNCE document. [Ralf S. Engelschall] *) Adjusted all copyright messages for new year 2003. [Ralf S. Engelschall] Changes between 2.0b1 and 2.0b2 (08-Nov-2002 to 04-Dec-2002) *) Upgraded built environment to GNU Autoconf 2.57. [Ralf S. Engelschall] *) Fixed compiler warnings under Linux: require inclusion of [Joseph Wayne Norton ] *) Fixed error handling in pth_init(3): it now correctly returns an error instead of abort(3)'ing. [Ralf S. Engelschall] *) Added soft syscall mapping for nanosleep(3) and usleep(3) functions also to the Pthread API. [Ralf S. Engelschall] Changes between 2.0b0 and 2.0b1 (07-Nov-2002 to 08-Nov-2002) *) Make pth_poll(3) more compliant to POSIX.1-2001/SUSv3 poll(2). [Ralf S. Engelschall, Nick Hudson ] *) Make pth_select(3) more compliant to POSIX.1-2001/SUSv3 select(2). This especially fixes the polling-only situation (timeout = (0,0)). [Ralf S. Engelschall] *) Make sure that in the even manager a polling-only select(2) call uses a correctly initialized timeout parameter. [Ralf S. Engelschall] *) Added soft and hard syscall mapping for nanosleep(3) and usleep(3) functions. [Ralf S. Engelschall] *) Fix poll(2) semantics: remove POLLRDNORM from polling result if POLLHUP is detected. [Paolo Bonzini ] *) Updated PORTING file to include community feedback from the last months. [Ralf S. Engelschall] Changes between 1.4.1 and 2.0b0 (27-Jan-2002 to 07-Nov-2002) *) Fix dropping of scheduler thread pools and this way fix memory leak. [Ralf S. Engelschall] *) Check for PTH_FDMODE_ERROR error conditions internally, too. [Ralf S. Engelschall] *) Use the new PTH_STATUS_FAILED event status in the scheduler's event-manager for filedescriptor events if the internal select(2) call returned with an error. Additionally this PTH_STATUS_FAILED is recognized by the high-level API functions (pth_select, etc) and produce the necessary POSIX conforming return codes (usually -1 and errno == EBADF). [Ralf S. Engelschall, Thanh Luu ] *) The function "int pth_event_occurred(pth_event_t)" was replaced with "pth_status_t pth_event_status(pth_event_t)" where pth_status_t can have values of PTH_STATUS_PENDING (replacing the old FALSE return value of pth_event_occurred), PTH_STATUS_OCCURRED (replacing the old TRUE return value of pth_event_occurred), and PTH_STATUS_FAILED (a new return value indicating an error in processing the event). This was scheduler/event-manager errors can be indicated which happended while processing the event. For backward compatibility reasons, a macro pth_event_occurred() was added. This will be removed soon. [Ralf S. Engelschall] *) Upgraded to latest version of the OSSP Autoconf macro AC_CHECK_EXTLIB in order to fix problems with the current Autoconf 2.54 environment. [Ralf S. Engelschall] *) Add Autoconf option --with-fdsetsize=NUM which allows to force a particular (usually higher than the default) FD_SETSIZE value for building Pth. Additionally Pth makes sure that the application using Pth does not use an even larger size. [Nick Hudson , Ralf S. Engelschall] *) Added thread attribute PTH_ATTR_DISPATCHES which (in bounded attribute objects) is incremented every time the context is switched to the associated thread. This can be used for statistical information. [Ralf S. Engelschall] *) Added a stand-alone sub-API for manual user-space context switching. It is somewhat modeled after the POSIX ucontext(3) facility and consists of an opaque data type pth_uctx_t and the management functions pth_uctx_create(), pth_uctx_make(), pth_uctx_save(), pth_uctx_restore(), pth_uctx_switch() and pth_uctx_destroy(). These functions are based on the same underlying machine context switching facility (pth_mctx) the threads in GNU Pth are using. This facility can be used to implement co-routines without a full real multithreading environment or even to implement an own multithreading environment. [Ralf S. Engelschall] *) Add a Pth variant of the new POSIX pselect(2) function, including soft and hard syscall mapping support for it. [Ralf S. Engelschall] *) More POSIX compliance for pth_select() in case of invalid timeout values and invalid filedescriptors. [Ralf S. Engelschall] *) Internally switch from "errno_shield {...}" to "pth_shield {...}" and from "return_errno(..)" to "return pth_error(...)" in order to make the internal error handling a little bit more consistent. [Ralf S. Engelschall] *) Cleaned and speeded up the pth_exit() processing. [Ralf S. Engelschall] *) Added POSIX-compliant sanity checks for bad filedescriptors to mostly all filedescriptor-based I/O functions in pth_high.c [Ralf S. Engelschall] *) Added pth_nanosleep() function. [Nick Hudson , Ralf S. Engelschall] *) Allow a NULL name for pth_msgport_create() in order to support locally scoped message ports which are not searched via pth_msgport_find(). [Ralf S. Engelschall, Matthew Mondor ] *) Added "hard syscall mapping" support for using SYS___sigprocmask14 instead of SYS_sigprocmask on NetBSD. [Nick Hudson ] *) Changes three direct sigprocmask(2) calls with the corresponding "hard syscall mapping" macro calls. [Nick Hudson ] *) Add Autoconf support for GNU Hurd. [B. Douglas Hilton ] *) Completely rewrote the "hard syscall mapping". Previously the internal syscall exit points were based on syscall(2) only. This is problematic because it by-passes the C library glue code which sometimes performs necessary assembly fiddling in order to call the underlying system calls correctly. Additionally, syscall(2) does not exists everywhere. Now the internal exit points are based on a by-syscall dynamically selected combination of RTLD_NEXT+dlsym(2), dlopen(2)+dlsym(2) and the known syscall(2) (in this fallback order). This way the "hard syscall mapping" is a lot more portable and flexible. [Ralf S. Engelschall, Jonathan Schilling ] *) Added support to pth_poll(3) for the poll(2) XPG.4 flags POLLD{RD,WR}{NORM,BAND}. [Jason Evans , Ralf S. Engelschall] *) Fixed a long-standing termination bug in pth_exit(3): The event handler of pth_exit(3) didn't let pth_exit(3) finish if there were any threads on the "dead queue" (where non-detached terminated threads are put). Instead it re-entered the scheduler which in turn aborted with "**Pth** SCHEDULER INTERNAL ERROR: no more thread(s) available to schedule!?!?". This is now fixed by not counting the "dead queue" for the determination whether the process as a whole should terminate or not. [Jonathan Schilling ] *) The manual page stated that system(3) is supported in the "Soft System Call Mapping", but in fact it was forgotten in the implementation. [Jonathan Schilling ] *) Add #define _PTHREAD_T to pthread.h to guard under SCO UnixWare 7 and OpenUNIX 8. [Jonathan Schilling ] *) Upgraded to Autoconf 2.54 environment and cleaned up internal Autoconf macro usage. [Ralf S. Engelschall] *) Fixed lots of English errors in the manual page. [Felix Berger ] *) Add optional support for OSSP ex based exception handling. GNU Pth (still) does not throw exceptions by itself, but handles the per-thread exception context of OSSP ex to make exception handling local to a thread. [Ralf S. Engelschall] *) Removed all generated files from CVS. Use OSSP devtool stuff to re-generate files on demand. Switched to Autoconf 2.52 and Libtool 1.4.2 environment. [Ralf S. Engelschall] _ _ _ / || || | | || || |_ | ||__ _| __|_(_) |_|_____________________________________________________________ Changes between 1.4.0 and 1.4.1 (24-Mar-2001 to 27-Jan-2002) *) Internally make sure an invalid file-descriptor (integer not between 0 and (FD_SETSIZE-1) does not lead to any segfaults or other undefined behaviour. Instead an error is returned and errno is set to EBADF, similar to what the OS functions do. Especially pth_poll() now return with this error (instead of skipping the fd) if an fd in the "struct pollfd" is invalid. [Ralf S. Engelschall, Archie Cobbs ] *) Correctly support PTH_FDMODE_NONBLOCK in pth_connect and pth_accept. [Archie Cobbs ] *) Fixed typos in manual page. [Michael Schloh v. Bennewitz , Takashi Ishihara ] *) For portability reasons changed definition of PTH_EXT_SFIO to 0/1 instead of FALSE/TRUE because some external definitions use a casted value and hence make trouble on plain #if constructs. [Staehli Patrik ] *) Fixed return value (number of occurred events) of pth_wait(). [David Dureau ] *) Replaced thread-unsafe usage of a static struct iovec in pth_writev_ev() with a thread-safe stack/heap-based solution. [Ralf S. Engelschall, Mark Burton ] *) Replaced antiquated PTH_FLAG_NOJOIN references with the correct PTH_ATTR_JOINABLE references in the manual page. [Takashi Ishihara ] *) Fixed a (not very subtle) bug in pth_writev_ev() that screwed up output if a partial write happened. [Mark Burton ] *) Fixed static initializers PTH_BARRIER_INIT and PTH_COND_INIT. [Shawn Wagner ] *) Typo fixes in pth.pod [] *) Upgraded to GNU shtool, version 1.5.4. [Ralf S. Engelschall] *) Fixed "make striptease": pth_string.c was not included and some commands which were removed at all (and this way causes syntax errors) will be now correctly commented out with ":" commands. [Paolo Bonzini ] *) Fixed pth.pod: a closing angle bracket was missing, leading to incorrect POD to XXXX formatting. [Ralf S. Engelschall] Changes between 1.4a3 and 1.4.0 (29-Jul-2000 to 24-Mar-2001) *) Added PTHREAD_PRIO_XXXX definitions to pthread.h for conformance to the POSIX/SUSv2 Pthread API. [Ralf S. Engelschall, Bill Apt ] *) Implemented the pthread_{set,get}concurrency() API parts of POSIX/SUSv2, although internally we are (allowed to be) free to not do anything based on the requested level. [Ralf S. Engelschall, Bill Apt ] *) Adjusted all pthread_attr_getXXXX() functions to use a "const pthread_attr_t *" as the first argument instead of "pthread_attr_t *" to fully-conform to POSIX/SUSv2. [Ralf S. Engelschall, Bill Apt ] *) Added ENOSYS-stubs for pthread_attr_{set,get}guardsize() to the Pthread API to be more complete with POSIX/SUSv2 specs. [Ralf S. Engelschall, Bill Apt ] *) Added still missing soft system call mapping to Pth and Pthread APIs for functions recv(2), send(2), recvfrom(2) and sendto(2). [Ralf S. Engelschall, Bill Apt ] *) Upgraded to GNU shtool 1.5.2 [Ralf S. Engelschall] *) Fixed an even-manager bug which causes a thread that calls pth_nap() to never woke up if the only elapsed event was a timer. [Archie Cobbs ] *) Added `#define _BITS_SIGTHREAD_H' to pthread.h to avoid inclusion of bits/sigthread.h (from signal.h) on Linux running glibc6 2.2. [Tomas Pihl ] *) Added support to Makefile.in for DESTDIR variable. This allows easier rolling of installation tarballs (for instance from within RPM or similar facilities) by using "make install DESTDIR=/tmp/pth". [Brad Smith , Ralf S. Engelschall] *) Implemented a pth_system(3) function which is a thread-aware clone of the POSIX system(2) function. [Ralf S. Engelschall] *) Fixed typos in pth.pod: "fd" -> "s" for pth_connect/pth_accept. [Sebastian ] *) Make --disable-shared the default under Solaris-2.[78]/x86, because it is known to segfault sporadically if Pth is built as a DSO. As traces showed, it is not a Pth problem, but it looks like a problem with the dynamic linker on Solaris/x86. The same Solaris versions on SPARC don't have this problem. [Ralf S. Engelschall] *) Updated copyright messages to cover new year 2001. [Ralf S. Engelschall] *) Fixed quoting in configure.in [Ralf S. Engelschall] *) Let pth_sleep(3) and pth_usleep(3) immediately return if an argument of zero is given. [Ralf S. Engelschall] *) Fixed pthread.pod: the newer pod2man versions seems to dislike embedded comments, so I moved them to the top of the file. [Ralf S. Engelschall] *) Changed CVS URL in HACKING document. [Ralf S. Engelschall] *) Mention http://www.mail-archive.com/pth-users@gnu.org/ in pth.pod and SUPPORT document. [Ralf S. Engelschall] Changes between 1.4a2 and 1.4a3 (01-Jul-2000 to 29-Jul-2000) *) Upgraded to GNU shtool 1.5.1 [Ralf S. Engelschall] *) Fixed (unused) pth_time_mul() function: operator & replaced by % [Tim Harris ] *) Use --disable-lock for ltconfig. [Ralf S. Engelschall] *) Fixed a few typos in pth.pod. [Thomas Klausner ] Changes between 1.4a1 and 1.4a2 (16-Apr-2000 to 01-Jul-2000) *) Upgraded to GNU Shtool 1.5.0 [Ralf S. Engelschall] *) Added OS/390 support to config.sub. [Greg Ames ] *) Upgraded rse-pmt.ps paper to latest version as it was published on USENIX 2000. [Ralf S. Engelschall] *) Stack boundary fixes for Interactive Unix support (--with-mctx-dsp=sjljisc). This allows one also to use this variant for Interix on Window-NT (a POSIX.1 compliant subsystem). [Kim Jongsu ] *) Upgraded to GNU Libtool 1.3.5 [Ralf S. Engelschall] *) Fixed config.param parsing: IF is now also allowed on VARIABLE=VALUE lines. [Ralf S. Engelschall] *) Allow for convinience reasons pth_usleep() to accept also arguments greater than 1000000. [Harvinder Sawhney ] *) Updated HACKING document. [Ralf S. Engelschall] *) Fixed warnings in pth_string.c related to va_arg() usage and implicit type conversions. [Ralf S. Engelschall] *) Merge from Pth 1.3.5: Fixed the checks in Autoconf: the logic was reversed and this way wasn't included on platforms were it existed and included where it wasn't present. [M. Lavasani ] Changes between 1.3.3 and 1.4a1 (10-Mar-2000 to 16-Apr-2000) *) Added a new feature to config.param: parameters can be extended with ``IF''. Then the preceeding parameter is only added to the command line if the returns 0. can be anything which is possible in a Bourne-Shell `if' construct's expression. [Ralf S. Engelschall] *) Fixed usage of `volatile' qualifier in pointer context. [Ralf S. Engelschall] *) Now check also for -Wno-long-long compiler option under --enable-debug, because pth_string.c contains `long long' stuff. [Ralf S. Engelschall] *) Now pth.h and pthread.h include the non-standard header on brain-dead platforms (like AIX) to get the definition of fd_set (which is required for the pth_select prototype). [Ralf S. Engelschall, Stian Seeberg ] *) Fixed auto-configuration for ISC and Win32/Cygwin platforms. [Ralf S. Engelschall, Giwon On ] *) Removed -L. from $(LDFLAGS) in Makefile.in, because this is not required (libtool already takes care of this). [Ralf S. Engelschall] *) Added a RPM spec file pth.spec which allows one to build RPM packages directly from the Pth distribution tarball through a simple `rpm -tb pth-1.X.Y.tar.gz' call. [Daniel Richard G. ] *) Removed the too explicit `-m 644' from the libtool/shtool installation command for libpth.la and instead use a `umask 022'. The reason is because some platforms require the shared libraries to be executable, so we cannot use an explicit mode. But we use the umask to make sure we don't result in world or group writeable files. [Ralf S. Engelschall, Daniel Richard G. ] *) Use full-path /sbin/sysctl in config.guess on FreeBSD because not all users have /sbin in their $PATH. [Jeff Trawick ] *) Added eight new I/O functions pth_{recv,recvfrom,send,sendto}[_ev](3) which correspond to the counterparts in UNIX98 (SUSv2). [Ralf S. Engelschall] *) Replaced for security reasons sprintf() and vsprintf() calls with own pth_snprintf() and pth_vsnprintf() functions from the new pth_string.c source. [Ralf S. Engelschall] *) Speeded up pth_ring_t handling by inlining code and by maintaining number of contained nodes explicitly to avoid an O(n) operation when the number of elements are requested. [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.9 [Ralf S. Engelschall] _ _____ / | |___ / | | |_ \ | |_ ___) | __|_(_)____/____________________________________________________________ Changes between 1.3.6 and 1.3.7 (01-Jul-2000 to 29-Jul-2000) *) Backport from GNU Pth 1.4a3: Upgraded to GNU shtool 1.5.1. This fixes especially the compilation problems under Solaris which were caused by a too unportable `shtool version' command from 1.5.0. [Ralf S. Engelschall] *) Backport from GNU Pth 1.4a3: Fixed (unused) pth_time_mul() function: operator & replaced by % [Tim Harris ] Changes between 1.3.5 and 1.3.6 (17-Apr-2000 to 01-Jul-2000) *) Backport from GNU Pth 1.4a2: Upgraded to GNU Shtool 1.5.0 [Ralf S. Engelschall] *) Backport from GNU Pth 1.4a2: Added OS/390 support to config.sub. [Greg Ames ] *) Backport from GNU Pth 1.4a2: Upgraded rse-pmt.ps paper to latest version as it was published on USENIX 2000. [Ralf S. Engelschall] *) Backport from GNU Pth 1.4a2: Upgraded to GNU libtool 1.3.5 [Ralf S. Engelschall] *) Backport from GNU Pth 1.4a2: Allow for convinience reasons pth_usleep() to accept also arguments greater than 1000000. [Harvinder Sawhney ] Changes between 1.3.4 and 1.3.5 (16-Apr-2000 to 17-Apr-2000) *) Fixed the checks in Autoconf: the logic was reversed and this way wasn't included on platforms were it existed and included where it wasn't present. [M. Lavasani ] Changes between 1.3.3 and 1.3.4 (10-Mar-2000 to 16-Apr-2000) *) Merged from Pth 1.4a1: Fixed usage of `volatile' qualifier in pointer context. [Ralf S. Engelschall] *) Merged from Pth 1.4a1: Now pth.h and pthread.h include the non-standard header on brain-dead platforms (like AIX) to get the definition of fd_set (which is required for the pth_select prototype). [Ralf S. Engelschall, Stian Seeberg ] *) Merged from Pth 1.4a1: Fixed auto-configuration for ISC and Win32/Cygwin platforms. [Ralf S. Engelschall, Giwon On ] *) Merged from Pth 1.4a1: Removed -L. from $(LDFLAGS) in Makefile.in, because this is not required (libtool already takes care of this). [Ralf S. Engelschall] *) Merged from Pth 1.4a1: Removed the too explicit `-m 644' from the libtool/shtool installation command for libpth.la and instead use a `umask 022'. The reason is because some platforms require the shared libraries to be executable, so we cannot use an explicit mode. But we use the umask to make sure we don't result in world or group writeable files. [Ralf S. Engelschall, Daniel Richard G. ] *) Merged from Pth 1.4a1: Use full-path /sbin/sysctl in config.guess on FreeBSD because not all users have /sbin in their $PATH. [Jeff Trawick ] *) Merged from Pth 1.4a1: Upgraded to GNU shtool 1.4.9 [Ralf S. Engelschall] Changes between 1.3.2 and 1.3.3 (24-Feb-2000 to 10-Mar-2000) *) Under Solaris and --enable-optimize we now also check for the compiler option -fast which is supported by the Sun vendor compilers. [Ralf S. Engelschall] *) Fixed AC_COMPILER_OPTION macro in aclocal.m4: it wasn't aware of the fact that running gcc with not supported options just leads to a warning (but gcc still compiles the test program and exists with a 0 return value). [Ralf S. Engelschall] *) Fixed semantics of pth_cond_notify() to match POSIX: if one notifies a condition variable which has still no waiter, the notify operation is a no-op. Previously the condition signal was remembered until a waiter arrived. This is now no longer the case. [Ralf S. Engelschall, Chris Leishman ] *) Allow pth_yield() to yield also to if is a freshly spawned thread, i.e. in state PTH_STATE_NEW instead of PTH_STATE_READY, directly after a pth_spawn. This allows one better control under co-routine programming. [Ralf S. Engelschall] *) Fixed internal pth_pqueue_delete() and pth_pqueue_tail() functions to pth_pqueue.c for convinience reasons. [Ralf S. Engelschall] *) API CHANGE: Changed pth_time(int,int) to pth_time(long,long) and pth_timeout(int,int) to pth_timeout(long,long), because pth_time_t is a struct timeval and this structure is defined always via two long's and not just int's. [Ralf S. Engelschall] *) Cleaned up source code even more by making sure "signed", "unsigned" and "const" qualifiers are used correctly and consistently. [Ralf S. Engelschall] *) Changed default stack size from 32KB to 64KB, because on Solaris and other platforms where one can set FD_SETSIZE to values up to 65535 an fd_set is up to 8KB. Pth internally (in the scheduler in pth_select_ev) has up to three fd_sets (up to 24KB) on the stack, so a 32KB stack is too risky. Nevertheless one can still spawn threads with smaller stacks by using PTH_ATTR_STACK_SIZE, of course. Only the scheduler thread remains with a 64KB stack, because this thread cannot be configured by the application. [Ralf S. Engelschall] *) Fixed "make depend": a backslash was missing. [Edwin Brown ] *) Fixed "make install" for pth.m4 and the situation where one compiles from a different sub-directory. [Raphael Bossek ] Changes between 1.3.1 and 1.3.2 (20-Feb-2000 to 24-Feb-2000) *) Do no longer use -woff in CFLAGS for IRIX 6.5.2 and above. [Edwin Brown , Ralf S. Engelschall] *) Replaced ``while (1)'' constructs with ``for (;;)'' because some compilers like this more and do not warn about constant expressions. [Ralf S. Engelschall] *) Added hint to pth.pod that ``pth_join(, NULL)'' is allowed. [Ralf S. Engelschall] *) Cancel and join the ticker thread explicitly in test_select.c [Edwin Brown , Ralf S. Engelschall] *) Removed unnecessary code in pth_event.c [Edwin Brown , Ralf S. Engelschall] *) Add special namespace workarounds to pthread.h.in for HPUX platforms where the pthread_kill() prototypes conflict without this. [M. Lavasani , Ralf S. Engelschall] Changes between 1.3.0 and 1.3.1 (19-Feb-2000 to 20-Feb-2000) *) Added HISTORY document where we now write down the evolution and release dates of Pth to have a concise history reference. [Ralf S. Engelschall] *) Added USERS document where we now collect references to software packages utilizing GNU Pth. [Ralf S. Engelschall] *) Fixed a subtle typo in pth.pod about thread-safe functions and finished documentation of pth_attr_set/pth_attr_get functions. [James Robinson , Jeremie ] Changes between 1.3b3 and 1.3.0 (13-Feb-2000 to 19-Feb-2000) *) Polished the various document files. [Ralf S. Engelschall] *) Use -w in $CFLAGS also on UnixWare 2.1, because its cc complains about "not reached" situations because of our macros. [Ralf S. Engelschall] *) Replaced "NOT REACHED" with "NOTREACHED" in comments to be more correct for lint(1). [Ralf S. Engelschall] *) Updated config.param and fixed its parsing procedure. [Ralf S. Engelschall] *) Fixed a few typos in pth.pod and adjusted example program to make sure it doesn't segfault accidently for the users. [Tim Harris , Raphael Bossek , Ralf S. Engelschall] Changes between 1.3b2 and 1.3b3 (28-Jan-2000 to 13-Feb-2000) *) Fixed PTH_EVENT_FUNC handling (it was not polled really in regular intervals if there were no other FUNC events which helped) by extending it with an interval time argument. This way a PTH_EVENT_FUNC event is likewise a PTH_EVENT_TIME event which calls the check function after elapsing and resets itself again if it failed. [Ralf S. Engelschall] *) Removed PTH_EVENT_PID because the waiting on a process id was not working as expected (the polling was not done in intervals) and this type of event doesn't really fit very well into the multi-threading environment of Pth (where the events should be thread-related and not process-related). So instead of fixing the pid waiting event handling through some unclean workarounds or kludges (it cannot be done very different), it was decided to kick it out at all. [Ralf S. Engelschall] *) Make sure pth_connect[_ev]() doesn't block by internally switching to non-blocking mode temporarily if necessary. [Chris Leishman , Ralf S. Engelschall] *) Fix pthread_cond_timedwait by making it more POSIX compliant: return ETIMEDOUT instead of 0 if the timeout occurred. [Emanuele Fornara , Ralf S. Engelschall] Changes between 1.3b1 and 1.3b2 (26-Jan-2000 to 28-Jan-2000) *) Let "make striptease" be aware of $TMPDIR and remove temporary directory from this dir, too. [Ralf S. Engelschall] *) Again cleaned up and enhanced the manual page pth.pod. [Ralf S. Engelschall] *) Added a few more errno_shield { ... } sections to prevent the destruction of errno values on error returns. [Ralf S. Engelschall] *) Added more complete Linux support for --enable-syscall-hard by using SYS_socketcall+SOCKOP_{accept,connect} if SYS_{accept,connect} doesn't exist. [Ralf S. Engelschall] *) Added pth_suspend() and pth_resume together with an additional SUSPENDED queue. This can be used to temporarily park threads in order to move them out of the schedulers scope. [Ralf S. Engelschall] Changes between 1.3a5 and 1.3b1 (15-Jan-2000 to 26-Jan-2000) *) Completely cleaned up the manual page. [Eric Hanchrow ] *) Use SYS__newselect instead of SYS_select under Linux and --enable-syscall-hard because SYS_select is a dummy stub which always just returns -1 and errno = EFAULT. [Artem Gr , Ralf S. Engelschall] Changes between 1.3a4 and 1.3a5 (08-Jan-2000 to 15-Jan-2000) *) Updated HACKING document. [Ralf S. Engelschall] *) Removed '+DAportable' from HPUX flags to avoid problems. [M. Lavasani ] *) Added a workround in Makefile.in (replaced "rm -r" with "rm" + "rmdir") for brain-dead AmigaOS. [Kriton Kyrimis ] *) Fixed example in Pth manual page: peer_len wasn't initialized. [Sami Niemi , Ralf S. Engelschall] *) Fixed cleanup handling in test_sig.c. [Ralf S. Engelschall] *) Fixed memory leaks in some test programs. [Ralf S. Engelschall] Changes between 1.3a3 and 1.3a4 (08-Jan-2000 to 08-Jan-2000) *) Translated The Open Group's SUSv2 Threading HTML documentation into POD format to form an appendix for pthread.pod. This is for convinience reasons when programming with the Pthread API of GNU Pth. [Ralf S. Engelschall] *) Fixed introduced debugging with "==== THREAD CONTEXT SWITCH ====" lines. They were printed always. [Ralf S. Engelschall] Changes between 1.3a2 and 1.3a3 (30-Dec-1999 to 08-Jan-2000) *) Added a PTH_CTRL_DUMPSTATE to pth_ctrl() which helps in debugging. a pth_ctrl(PTH_CTRL_DUMPSTATE, stderr) for instance writes out a summary page of the internal Pth library state to stderr. This is intended for debugging purposes only, of course. [Ralf S. Engelschall, Lubos Lunak ] *) Fixed destructor for statically initialized events which failed under asynchronous cancellation if those events were merged together at the time of cancellation. [Igor A. Minyukoff, Ralf S. Engelschall] *) Updated the INSTALL document. [Ralf S. Engelschall] *) Fixed a memory leak in pth_kill(): the TCBs (plus one stack) of the main and scheduler thread were not freed. [Pete , Ralf S. Engelschall] *) Added developer support for compiling and linking against the Dmalloc library (for debugging memory allocation) through the --with-dmalloc[=DIR] Autoconf option. [Ralf S. Engelschall] *) Fixed pth_compat.c: the #define for strerror was buggy. [Ralf S. Engelschall] Changes between 1.3a1 and 1.3a2 (09-Nov-1999 to 30-Dec-1999) *) Added first cut of experimental Win32 support through Cygwin B20.1. [Ralf S. Engelschall, Schizoid ] *) Fixed autoheader step by using AC_CHECK_HEADERS() instead of AC_CHECK_HEADER() because Autoconf's autoheader.m4 isn't aware of AC_CHECK_HEADER(). [Ralf S. Engelschall] *) Adjusted all copyright messages to include the forthcoming new year 2000, too. [Ralf S. Engelschall] *) Fixed $DIFS in pth-config.in and pthread-config.in (the tab was lost) and cleaned up pth-config.in a little bit. [Ralf S. Engelschall] *) Upgraded from GNU libtool 1.3.3 to 1.3.4 and from GNU shtool 1.4.6 to 1.4.7. [Ralf S. Engelschall] *) Fixed cleanup handling for main thread. Now pth_kill() and pth_exit() in the main thread work as expected. [Ralf S. Engelschall, Anton Umnikov ] *) Fixed Autoconf --host option. [Ralf S. Engelschall, Kent Overstreet ] *) Fixed pth_mctx.c by making sure ss_base is mapped to ss_sp only if sigaltstack(2) is used for the stack trick. [Kriton Kyrimis ] *) Fixed return code semantics for error situation in both pth_write(3) and pth_writev(3). [Rob Quinn , Ralf S. Engelschall] *) Stripped trailing whitespaces from all source files. [Ralf S. Engelschall] Changes between 1.2.0 and 1.3a1 (31-Oct-1999 to 09-Nov-1999) *) Changed AC_CHECK_STRUCTATTR macro from a too weak AC_TRY_EGREP to a AC_TRY_LINK based approach to make sure the ss_sp/ss_base checks do not fail on platforms where sys/signal.h contains both sigstack and sigaltstack defines. [Ralf S. Engelschall] *) Added a large "BUILD ENVIRONMENT" section to pth.pod which describes three ways to establish build environments for Pth based packages. [Ralf S. Engelschall] *) Updated the pth-config.pod and pthread-config.pod for missing option entries. [Ralf S. Engelschall] *) Added --datadir and --acdir options to pth-config. Especially --acdir can be used in conjunction with Autoconf's aclocal program to retrieve the AC_CHECK_PTH macro, e.g. aclocal --output- --acdir=`pth-config --acdir` prints the macro definition to stdout. [Ralf S. Engelschall] *) Added a new, very flexible and robust pth.m4 Autoconf macro file which is installed under $datadir/aclocal/ and can be used by third-party Autoconf scripts to locate GNU Pth. Syntax is: AC_CHECK_PTH([MIN-VERSION [,DEFAULT-WITH-PTH [,DEFAULT-WITH-PTH-TEST [,EXTEND-VARS [,ACTION-IF-FOUND [,ACTION-IF-NOT-FOUND]]]]]]). [Ralf S. Engelschall, Markus Fischer ] *) Added PTH_VERSION_{STR,HEX} to pth.h [Ralf S. Engelschall] *) Fixed AC_CHECK_EXTLIB macro. [Ralf S. Engelschall] *) Added a new developer target `make striptease' which uses the new striptease.{pl,mk} files to strip down the Pth source tree to a minimum (approx. 1/3 of size). The generated striptease.dir/ area can be then imported into third-party source trees and there blows up the stuff only as minimal as possible. [Ralf S. Engelschall] *) Cleaned up copyright messages in source files. [Ralf S. Engelschall] *) Rewritten the Linux guessing part in config.guess to create more useful GNU platform triples for Linux flavors. [Ralf S. Engelschall] _ ____ / | |___ \ | | __) | | |_ / __/ __|_(_)_____|___________________________________________________________ Changes between 1.2.2 and 1.2.3 (08-Jan-2000 to 04-Feb-2000) *) Backport from Pth 1.3b3: Fix pthread_cond_timedwait by making it more POSIX compliant: return ETIMEDOUT instead of 0 if the timeout occurred. [Emanuele Fornara , Ralf S. Engelschall] *) Backport from Pth 1.3b2: Added more complete Linux support for --enable-syscall-hard by using SYS_socketcall+SOCKOP_{accept,connect} if SYS_{accept,connect} doesn't exist. [Ralf S. Engelschall] *) Backport from Pth 1.3b2: Added a few more errno_shield { ... } sections to prevent the destruction of errno values on error returns. [Ralf S. Engelschall] *) Backport from Pth 1.3b1: Use SYS__newselect instead of SYS_select under Linux and --enable-syscall-hard because SYS_select is a dummy stub which always just returns -1 and errno = EFAULT. [Artem Gr , Ralf S. Engelschall] *) Backport from Pth 1.3a5: Removed '+DAportable' from HPUX flags to avoid problems. [M. Lavasani ] *) Backport from Pth 1.3a5: Added a workround in Makefile.in (replaced "rm -r" with "rm" + "rmdir") for brain-dead AmigaOS. [Kriton Kyrimis ] *) Backport from Pth 1.3a5: Fixed example in Pth manual page: peer_len wasn't initialized. [Sami Niemi , Ralf S. Engelschall] *) Backport from Pth 1.3a5: Fixed memory leaks in some test programs. [Ralf S. Engelschall] Changes between 1.2.1 and 1.2.2 (14-Nov-1999 to 08-Jan-2000) *) Backport from Pth 1.3a3: Updated the INSTALL document. [Ralf S. Engelschall] *) Backport from Pth 1.3a3: Fixed a memory leak in pth_kill(): the TCBs (plus one stack) of the main and scheduler thread were not freed. [Pete , Ralf S. Engelschall] *) Backport from Pth 1.3a3: Fixed pth_compat.c: the #define for strerror was buggy. [Ralf S. Engelschall] *) Backport from Pth 1.3a3: Fixed destructor for statically initialized events which failed under asynchronous cancellation if those events were merged together at the time of cancellation. [Igor A. Minyukoff, Ralf S. Engelschall] *) Backport from Pth 1.3a2: Upgraded from GNU libtool 1.3.3 to 1.3.4 and from GNU shtool 1.4.6 to 1.4.7. [Ralf S. Engelschall] *) Backport from Pth 1.3a2: Fixed pth_mctx.c by making sure ss_base is mapped to ss_sp only if sigaltstack(2) is used for the stack trick. [Kriton Kyrimis ] Changes between 1.2.0 and 1.2.1 (31-Oct-1999 to 14-Nov-1999) *) Backport from Pth 1.3a2: Fixed return code semantics for error situation in both pth_write(3) and pth_writev(3). [Rob Quinn , Ralf S. Engelschall] *) Backport from Pth 1.3a1: Changed AC_CHECK_STRUCTATTR macro from a too weak AC_TRY_EGREP to a AC_TRY_LINK based approach to make sure the ss_sp/ss_base checks do not fail on platforms where sys/signal.h contains both sigstack and sigaltstack defines. [Ralf S. Engelschall] *) Backport from Pth 1.3a1: Fixed AC_CHECK_EXTLIB macro. [Ralf S. Engelschall] *) Backport from Pth 1.3a1: Cleaned up copyright messages in source files. [Ralf S. Engelschall] Changes between 1.2b8 and 1.2.0 (26-Oct-1999 to 31-Oct-1999) *) Finally tested the package on all major Unix platforms to which I've direct access (see PORTING document for details). [Ralf S. Engelschall] *) Fixed a few remaining typos in pth.pod; adjusted README file for release and included ANNOUNCE document in source tree. [Ralf S. Engelschall] *) Fixed special Autoconf detections for UnixWare 2.x platform. [Ralf S. Engelschall] *) Fixed special Autoconf detections for Linux platforms. [Ralf S. Engelschall] *) Updated INSTALL document. [Ralf S. Engelschall] *) Added -w option to CFLAGS under HP-UX with CC. [Ralf S. Engelschall] Changes between 1.2b7 and 1.2b8 (22-Oct-1999 to 26-Oct-1999) *) Fixed Pthread compilation under NetBSD 1.4 and Solaris. The pth.h header has to skip the fallbacks (and not the pthread.h header) when compiling pthread.c. This subtle difference is important because ordering is important for correct namespace hiding. [Ralf S. Engelschall, John A. Maier ] *) Greatly enhanced AC_CHECK_ARGTYPE macro to support more vendor headers and their syntax. [Ralf S. Engelschall] *) Added two new Autoconf options: --enable-maintainer for enabling maintainer targets (now the x.pod -> x.N translation is disabled per default and is enabled with this option only) and --enable-tests which is on per default and can be used (as --disable-tests) to disable the building of the test programs. [Ralf S. Engelschall] *) Fixed autoheader processing with a new acheader.m4 file. [Ralf S. Engelschall] *) Stripped down shtool script to 7 of 17 ingredients to reduce source tree by approx. 40KB. [Ralf S. Engelschall] *) Updated acconfig.h: HAVE_NFDS_T was missing. [Ralf S. Engelschall] *) Fixed `make dist': CVS temporary files (.#xx) were included into the distribution tarball. [Ralf S. Engelschall] Changes between 1.2b6 and 1.2b7 (28-Sep-1999 to 22-Oct-1999) *) Add the .gdbinit file also to the distribution tarball. [Ralf S. Engelschall] *) Re-acquire the mutex corresponding to a condition variable when pth_cond_await() is cancelled, in order to restore the condition variable semantics. [Lubos Lunak ] *) Fixed cancellation point handling. [Lubos Lunak ] *) Added support for poll(2)'s POSIX nfds_t argument type. [Ralf S. Engelschall] *) Fixed AC_CHECK_ARGTYPE: it now handles wrapped lines, too. [Ralf S. Engelschall] *) Fixed regex in aclocal.m4: tabs were expanded to spaced. [Ralf S. Engelschall] *) Fixed a casting problem related to `void *' in pth_high.c which g++ 2.95.1 hated. [Ralf S. Engelschall] *) Replaced Autoconf AC_CHECK_FUNCS macro with an own local copy named AC_CHECK_FUNCTIONS which _always_ uses ``extern "C"'' if __cplusplus is defined. This way one can configure Pth even if $CC is a C++ compiler. [Ralf S. Engelschall] *) Updated rse-pmt.ps document. [Ralf S. Engelschall] *) Fixed a few typos in the manual page. [Ralf S. Engelschall] *) Fixed aclocal.m4: CFXXLAGS -> CXXFLAGS ;) [Alexandre Oliva ] Changes between 1.2b5 and 1.2b6 (21-Sep-1999 to 28-Sep-1999) *) Added platform support for GNU/Linux kernel 2.2 / glibc 2.1 for both Alpha and Intel platforms via sjlj/ssjlj/sas. (tested with both RedHat 6.0 and SuSE 6.1) [Ralf S. Engelschall] *) Added platform support for Tru64/OSF1 4.x and 5.0. [Ralf S. Engelschall] *) Added --prefix, --bindir, --libdir, --includedir and --mandir options to pth-config and pthread-config. Mainly for convinience and consistency reasons. [Ralf S. Engelschall] *) Avoid compiler complains in pth_mctx.c for assigning a `volatile sigset_t'. Instead use a plain memcpy() with sizeof(sigset_t). This now makes also IRIX' 6.5 cc happy. [Ralf S. Engelschall] *) Fixed the Autoconf AC_CHECK_MCSC check which tests whether SVR4/SUSv2 makecontext(2), swapcontext(2) and friends can be used for user-space context switching. It was broken because the uc_link and uc_stack structure attributes were not initialized correctly. [Ralf S. Engelschall, Ben Harris ] *) Updated dependencies in Makefile.in. [Ralf S. Engelschall] *) Mention the new pth-users@gnu.org mailing list in SUPPORT and added hint to Makefile.in. [Ralf S. Engelschall] Changes between 1.2b4 and 1.2b5 (17-Sep-1999 to 21-Sep-1999) *) Added new Autoconf macro AC_COMPILER_OPTION which is now used to check whether -pipe, -ggdb3 and -W really work before they are implicitly used. [Ralf S. Engelschall, Alexandre Oliva ] *) Make sure pthread.c compiles even if the platforms lacks `struct timespec'. The pthread_cond_timedwait() function will then still be unuseable by the application, of course. But that's the intended behavour, because the Pthread API cannot be changed at this point. [Michele Satriani , Ralf S. Engelschall] *) Document it more clearly that PTH_ATTR_DEFAULT is _not_ the values an pth_attr_init() creates. It means that priority, joinability and cancelstate are _inherited_. [Eric Newton ] Changes between 1.2b3 and 1.2b4 (17-Sep-1999 to 17-Sep-1999) *) Fixed test_httpd.c: highest port is 65535 and not 65553 ;) [Alex Fiori ] *) Fix Makefile.in: $(S)shtool -> $(srcdir)shtool [Ben Harris , Ralf S. Engelschall] *) Make Rhapsody happy with pth_errno.c by statically initializing the contained variables. [Ben Harris , Ralf S. Engelschall] Changes between 1.2b2 and 1.2b3 (04-Sep-1999 to 17-Sep-1999) *) Added first cut of support for building Pth outside it's source tree. [Ben Harris , Ralf S. Engelschall] *) Updated pth.pod for new socklen_t usage and updated NEWS file. [Ralf S. Engelschall] *) Officially switched from ``GNU Library General License v2.0'' to successor license: ``GNU Lesser General License v2.1''. [Ralf S. Engelschall] *) Enhanced ss_sp to ss_base mapping: There is now Autoconf support for the mapping decision. [Ralf S. Engelschall, Ben Harris ] *) Add `PTHREAD' also to defines for namespace protection. [Hans v. Sommerfeld ] *) Made the temporary errno saving more elegant with a new `errno_shield { .... }' construct. [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.6 [Ralf S. Engelschall] Changes between 1.2b1 and 1.2b2 (02-Sep-1999 to 04-Sep-1999) *) Ported Pth to another esoteric, anchient and brain-dead platform which needs jmp_buf fiddling: Interactive Unix (ISC) 4.0. [Ralf S. Engelschall, Brandon Reynolds ] *) Make sys/resource.h an optional header. [Ralf S. Engelschall] *) Added CPPFLAGS to Makefile.in and configure.in [Ralf S. Engelschall] Changes between 1.1.4 and 1.2b1 (30-Aug-1999 to 02-Sep-1999) *) Added TESTS document which shows success stories with Apache, MySQL, Perl, Python, OpenLDAP, pident, etc. [Ralf S. Engelschall] *) Enhanced pth-config/pthread-config scripts: first they are now concatenating the output of the options into a single string and print this at the end once. This way one can use things like LIBS="`pthread-config --ldflags --libs`" without getting an intermediate newline which causes trouble in various packages. Second per default only pth-specific CFLAGS/LDFLAGS/LIBS are printed. If one wants also additional flags one has used while building Pth one now can add the --all option. This is to avoid conflicts between Pth flags and flags of the application. [Ralf S. Engelschall] *) Added -w to CFLAGS for Solaris if $CC is not GCC to avoid the warnings about "end of loop not reached" because of our ``do { ...return;... } while(0)'' macros. [Ralf S. Engelschall] *) Fixed two bugs in internal pth_pqueue_delete() and one in pth_pqueue_insert(): it accidently also adjusted the priority queue top element if one deleted the last element of the queue and it calculated the priority of the following element incorrectly. [Jens Andersen , Ralf S. Engelschall] *) Optimized the signal mask handling: The application is now required to use pth_sigmask(3) and no longer can use sigprocmask(2) directly (as it's already the case for Pthreads where POSIX states that the behaviour of sigprocmask(2) is not defined in a MT application). The side-effect of this consistencency restriction is that internally the machine context switching no longer has to remember the signal mask on _every_ context switch. It's now sufficient that pth_sigmask(3) sets the mask to allow the scheduler to calculate its own mask. The result is a faster and more clean thread dispatching. Additionally a soft and hard syscall mapping for sigprocmask(2) was added, of course. [Ralf S. Engelschall] *) Added support for socklen_t to both avoid warnings under some platforms and to avoid problems on 64 bit machines. [Ralf S. Engelschall] *) Added UnixWare7 support to config.guess [Ralf S. Engelschall] *) Added config.param facility. [Ralf S. Engelschall] *) Cleaned up bootstrap stuff from configure.in and moved it to aclocal.m4. [Ralf S. Engelschall] *) Fixed autoconf's `--silent' flag: it wasn't passed to ltconfig.sh [Ralf S. Engelschall] *) Rewritten the "make test" message and refreshed the PORTING file to start with a new format which now includes also the machine context and stack growth direction. [Ralf S. Engelschall] *) Enhanced the interactive (not --enable-batch) mode: The Makefile now displays real hints what to do as the next step. All messages can be disabled via --enable-batch, of course. [Ralf S. Engelschall] *) Fixed aclocal.m4: Tabs were lost by accident. [Ralf S. Engelschall] *) Fixed a type warning in pth_debug.c: pid_t vs. int [Ralf S. Engelschall] _ _ / | / | | | | | | |_| | __|_(_)_|_______________________________________________________________ Changes between 1.1.5 and 1.1.6 (02-Sep-1999 to 28-Sep-1999) *) Backport from Pth 1.2b6: Fixed the Autoconf AC_CHECK_MCSC check which tests whether SVR4/SUSv2 makecontext(2), swapcontext(2) and friends can be used for user-space context switching. It was broken because the uc_link and uc_stack structure attributes were not initialized correctly. [Ralf S. Engelschall, Ben Harris ] *) Backport from Pth 1.2b5: Document it more clearly that PTH_ATTR_DEFAULT is _not_ the values an pth_attr_init() creates. It means that priority, joinability and cancelstate are _inherited_. [Eric Newton ] *) Backport from Pth 1.2b4: Fixed test_httpd.c: highest port is 65535 and not 65553 ;) [Alex Fiori ] *) Backport from Pth 1.2b4: Make Rhapsody happy with pth_errno.c by statically initializing the contained variables. [Ben Harris , Ralf S. Engelschall] *) Backport from Pth 1.2b3: Add `PTHREAD' also to defines for namespace protection. [Hans v. Sommerfeld ] *) Upgraded to GNU shtool 1.4.6 [Ralf S. Engelschall] Changes between 1.1.4 and 1.1.5 (30-Aug-1999 to 02-Sep-1999) *) Backport from Pth 1.2b1: Enhanced pth-config/pthread-config scripts: first they are now concatenating the output of the options into a single string and print this at the end once. This way one can use things like LIBS="`pthread-config --ldflags --libs`" without getting an intermediate newline which causes trouble in various packages. Second per default only pth-specific CFLAGS/LDFLAGS/LIBS are printed. If one wants also additional flags one has used while building Pth one now can add the --all option. This is to avoid conflicts between Pth flags and flags of the application. [Ralf S. Engelschall] *) Backport from Pth 1.2b1: Fixed two bugs in internal pth_pqueue_delete() and one in pth_pqueue_insert(): it accidently also adjusted the priority queue top element if one deleted the last element of the queue and it calculated the priority of the following element incorrectly. [Jens Andersen , Ralf S. Engelschall] *) Backport from Pth 1.2b1: Added -w to CFLAGS for Solaris if $CC is not GCC to avoid the warnings about "end of loop not reached" because of our ``do { ...return;... } while(0)'' macros. [Ralf S. Engelschall] *) Backport from Pth 1.2b1: Fixed autoconf's `--silent' flag: it wasn't passed to ltconfig.sh [Ralf S. Engelschall] *) Backport from Pth 1.2b1: Fixed aclocal.m4: Tabs were lost by accident. [Ralf S. Engelschall] *) Backport from Pth 1.2b1: Fixed a type warning in pth_debug.c: pid_t vs. int [Ralf S. Engelschall] Changes between 1.1.3 and 1.1.4 (27-Aug-1999 to 30-Aug-1999) *) Changed internal handling of spawned threads: they are now really inserted to the top of the priority queue to make sure a thread is guarrantied to be dispatched next when one calls pth_yield(NULL) directly after a pth_spawn(). This is especially interesting for programming with the co-routine paradigm where it allows one to better control the startup of a co-routine. [Ralf S. Engelschall, Jens Andersen ] *) Fixed pthread_cancel(): a switch statement was incorrect (three breaks were missing to stop falling through). [Jens Andersen ] *) Change internal `_pthread' prefix to `__pthread'. [Ralf S. Engelschall] *) Enhanced backward compatibility stuff for Pthread draft 4 (DCE Threads) in pthread.h.in: When _POSIX_BACKCOMPAT is defined before the pthread.h header is included one now gets a lot more backward compatibility mappings. [Ralf S. Engelschall] *) Optimize internal processing by inlining various small functions from pth_ring.c, pth_pqueue.c and pth_sched.c which are called a lot. [Ralf S. Engelschall] *) Optimized readline() in test_common.c [Ralf S. Engelschall] *) Added specific support for Linux/glibc/mc68000 to pth_mctx.c [Roman Hodek ] *) Updated INSTALL document for --enable-optimize. [Ralf S. Engelschall] Changes between 1.1.2 and 1.1.3 (23-Aug-1999 to 27-Aug-1999) *) Enhanced test program of AC_CHECK_STACKGROWTH in aclocal.m4 to make sure the test variables are not optimized away by the compiler and this way lead to incorrect check results. [Martin Kraemer , Ralf S. Engelschall] *) Fixed default stack size: was 32*16384 (=512KB) instead of the intended 32*1024 (=32KB). [Ralf S. Engelschall] *) Added --enable-optimize flag to configure with selected standard optimization flags of GCC. [Ralf S. Engelschall] *) Added SUPPORT file to source tree. [Ralf S. Engelschall] *) Added test_philo, Dijkstra's Five Dining Philosophers ;) [Ralf S. Engelschall] *) Add "auto" qualifier to PTH_STACKGROWTH test program to make sure the variables are placed into the stack. [Ralf S. Engelschall] *) Use predetermined $PLATFORM of configure also in Makefile [Ralf S. Engelschall] Changes between 1.1.1 and 1.1.2 (21-Aug-1999 to 23-Aug-1999) *) Fixed Makefile.in: pth_acdef.h and pth_acmac.h were not removed on "distclean" and this way they were accidently distributed. [Ralf S. Engelschall] *) Changed return type of pth_yield() from `void' to `int' to be able to return an error FALSE/EINVAL in case the `tid' argument specified an invalid or still not ready thread. [Kurt D. Zeilenga , Ralf S. Engelschall] *) Fixed description of pth_yield() in pth.pod. [Kurt D. Zeilenga , Ralf S. Engelschall] Changes between 1.1.0 and 1.1.1 (19-Aug-1999 to 21-Aug-1999) *) Changed pthread_yield_np() from `void' return type to `int' to allow the `#define sched_yield pthread_yield_np' work as expected, because sched_yield has `int' return type per SUSv2. [Kurt D. Zeilenga , Ralf S. Engelschall] *) Fixed problems related to pthread_exit() and pth_exit(): The PTH_EVENT_FUNC internally was used incorrectly by pth_exit() and the implicit initialization was forgotten in pthread_exit(). [Kurt D. Zeilenga , Ralf S. Engelschall] *) Added consistency check for `timeout' attribute of pth_poll and various other EINVAL related checks to pth_event.c. [Ralf S. Engelschall] *) Added consistency check for `abstime' attribute of pthread_cond_timedwait(). [Ralf S. Engelschall] Changes between 1.1b7 and 1.1.0 (18-Aug-1999 to 19-Aug-1999) *) Disable vendor typedefs in pthread.h.in because it caused problems under Solaris. [Ralf S. Engelschall] *) Cleaned up the sources a little bit more. [Ralf S. Engelschall] *) Updated rse-pmt.ps to v0.9.15 [Ralf S. Engelschall] Changes between 1.1b6 and 1.1b7 (18-Aug-1999 to 18-Aug-1999) *) Added test for nested thread operation to test_std.c [Ralf S. Engelschall] *) Updated rse-pmt.ps to talk about the sigsetjmp(3) problems. [Ralf S. Engelschall] *) Enhanced the heart of Pth: pth_mctx_set(). Instead of the real dispatching functions a plain setjmp(3)/longjmp(3) pair is used for the trampoline trick in order to avoid problems with sigjmp_buf-attached signal stacks on platforms where sigsetjmp(3) does more than it is usuallly expected. This especially fixes `thread spawns thread' problems under HPUX. [Ralf S. Engelschall, Aaron Metzger ] Changes between 1.1b5 and 1.1b6 (17-Aug-1999 to 18-Aug-1999) *) Fixed return value of pth_write/pth_read. [Eric Newton , Ralf S. Engelschall] *) Fixed example in manual page (pth.pod): pth_load() is now pth_ctrl(PTH_CTRL_GETAVLOAD, ...) [Eric Newton ] *) Fixed warning in pth_high.c related to implicit casting. [Ralf S. Engelschall] *) Added BS2000 support to config.guess/config.sub. [Martin Kraemer] *) Fixed a sigaltstack() call in pth_mctx.c: the return code was checked incorrectly. [Ralf S. Engelschall] *) Updated rse-pmt.ps [Ralf S. Engelschall, Eric Newton ] *) Fixed a nasty bug in the thread stack allocation: A stack was also allocated for the main thread, but not used. The problem was that an allocated stack has a guardian and this way was checked each time and could lead to a bogus STACK OVERFLOW for the "main" thread. [Ralf S. Engelschall, Martin Kraemer] Changes between 1.1b4 and 1.1b5 (13-Aug-1999 to 17-Aug-1999) *) Added draft of forthcoming paper ``Portable Multithreading - The Signal Stack Trick Of User-Space Thread Creation'' to source tree under rse-pmt.ps. Hackers are encouraged to read this if they want to understand Pth's internals. [Ralf S. Engelschall] *) Source cleanups: added `extern' qualifier to function prototypes in pth.h.in and pthread.h.in; make compilation silent with both gcc 2.95.1 C and C++ frontends by fixing implicit type conversion situations, etc. [Ralf S. Engelschall] *) Return EBUSY instead of EAGAIN for pth_mutex_acquire() when the "try" argument is TRUE. This fixes especially the POSIX semantics of pthread_mutex_trylock(). [Ralf S. Engelschall] *) Added pth_compat.c with strerror(3) replacement for SunOS. [Ralf S. Engelschall] *) Fixed pthread_{mutex,rwlock,cond}_destroy() functions. [Ralf S. Engelschall] *) Added support for `configure '. [Ralf S. Engelschall, Martin Kraemer ] Changes between 1.1b3 and 1.1b4 (11-Aug-1999 to 13-Aug-1999) *) Fixed blocking semantics of pth_writev(). [Ralf S. Engelschall] *) Fixed a very nasty bug in pth_select(): The write fd_set was passed twice to the scheduler (once as the write fd_set and once as the execptional fd_set). This caused problems, because the event loop this way deleted bits in the write fd_set by accident and the return semantic was broken. [Ralf S. Engelschall] *) Added fallbacks for UIO_MAXIOV to pth.h and pthread.h [Ralf S. Engelschall] *) Remove locking check in pth_mutex_init() because this would only work with static vars where ANSI C guarranties an initialization with 0 bytes. [Martin Kraemer ] *) Fixed pth_write() semantics. [Ralf S. Engelschall] Changes between 1.1b2 and 1.1b3 (10-Aug-1999 to 11-Aug-1999) *) Overhauled pthread.h to make it better compile-able also under platforms where a vendor Pthread library exists. [Ralf S. Engelschall] *) Reordered ingredients of configure.in. [Ralf S. Engelschall] *) Fixed aclocal.m4: -g was incorrectly stripped. [Ralf S. Engelschall] *) Enhance pth_write(): In non-blocking (default) mode it now mimics the behaviour of write(2) more correctly by iterating until all data is written or an error occurs. [Ralf S. Engelschall] Changes between 1.1b1 and 1.1b2 (07-Aug-1999 to 10-Aug-1999) *) Fixed ring walking inside both public pth_msgport_find() function and the internal pth_mutex_releaseall() function. [Laurent Vaucher ] Changes between 1.0.4 and 1.1b1 (03-Aug-1999 to 07-Aug-1999) *) Fixed test_select.c: a char has to be used for read(2) and not an int. [Ralf S. Engelschall] *) Fixed AC_CHECK_NSIG test and this way PTH_NSIG define. [Ralf S. Engelschall] *) Overhauled pth_mctx.c's internals according to latest algorithm of my paper ``Portable Multithreading - The Signal Stack Trick For User-Space Thread Creation''. Especially: sigstack had not restored stack correctly, the signal mask was not inherited correctly; Linux signal mask handling was incorrect, etc. [Ralf S. Engelschall] *) Adjusted variables in pth_mctx.c to use `volatile' qualifier. [Ralf S. Engelschall] *) Simplified Makefile.in a little bit by removing $(OBJS) [Ralf S. Engelschall] *) Replaced the run-time stack growth decision for the guardian in pth_tcb.c with a compile-time decision. [Ralf S. Engelschall] *) Completly rewritten the configure.in part which deals with the machine context (mctx) implementation: First SVR4/SUSv2 makecontext(2) and friends are generally preferred. Second, the user now can override the decisions via --with-mctx-{mth,dsp,stk}. Third, the pth_mctx.c source was enhanced by using skaddr_xxx() and sksize_xxx() macros which know through Autoconf checks how a stack has to be setup correctly for makecontext(2), sigaltstack(2) and sigstack(2). [Ralf S. Engelschall] *) Add support to pth_mctx.c for SVR4/SUSv2 makecontext(2)/swapcontext(2) and friends. This way Pth now is best prepared to cover forthcoming Unix flavors, because these are didicated and standadized functions for user-space context switching. [Ralf S. Engelschall] *) Completely overhauled aclocal.m4 [Ralf S. Engelschall] *) Use GNU config.guess/config.sub instead of GNU shtool guessos for wider support inside GNU libtool. [Ralf S. Engelschall] *) Replaced switch() over long variables (which aren't covered by traditional C) with an if-else construct in pth_event.c to be more portable. [Ralf S. Engelschall] *) Fixed timeout handling in pth_select(). [Ralf S. Engelschall] *) Don't free(2) application-supplied stacks on cleanups. [Ralf S. Engelschall] *) Don't allow an application to change the stack size or address on an pth_attr_t object when its attached to a thread (see pth_attr_of). [Ralf S. Engelschall] *) Added pth_ext.c for extensional functionality. As a first cut Sfio support is provided by an additional function pth_sfiodisc. This functions is always available, but only reasonably useable when Pth was built with Sfio support (--with-sfio option) and PTH_EXT_SFIO is then defined by pth.h. It is useful for applications which want to use the comprehensive Sfio I/O library with the Pth threading library. Then this function can be used to get an Sfio discipline structure (Sfdisc_t) which can be pushed onto Sfio streams (Sfio_t) in order to let this stream use pth_read(3)/pth_write(2) instead of read(2)/write(2). The benefit is that this way I/O on the Sfio stream does only block the current thread instead of the whole process. The application has to free(3) the Sfdisc_t structure when it is no longer needed. [Ralf S. Engelschall] *) Added `barrier' synchronization objects. These provide a way to let a group of threads wait at a barrier until the last reached the barrier. It's implemented on top of mutex and condition variable. [Ralf S. Engelschall] *) Replaced a bogus stdio fgetc() in test_select with pth_read(). [Ralf S. Engelschall] *) Fixed internal pth_debug(3): it now preserved errno to make sure there is no run-time difference in semantics just because of debug messages. [Ralf S. Engelschall] *) Overhauled the filedescriptor handling: 1. There is no longer a requirement to manually switch a filedescriptor into non-blocking mode in order to use it. This is automatically done temporarily inside Pth now. Instead when you now switch a filedescriptor explicitly into non-blocking mode, pth_read(3) or pth_write(3) will never block the current thread as it is the expected behaviour for non-blocking I/O. 2. pth_nonblocking(3) was replaced by a more convinient pth_fdmode(3) function which can perform both get and set operations at once. [Ralf S. Engelschall] _ ___ / | / _ \ | || | | | | || |_| | __|_(_)___/_____________________________________________________________ Changes between 1.0.4 and 1.0.5 (03-Aug-1999 to 10-Aug-1999) *) Backport from Pth 1.1b2: Fixed ring walking inside both public pth_msgport_find() function and the internal pth_mutex_releaseall() function. [Laurent Vaucher ] *) Backport from Pth 1.1b1: In pth_mctx.c the signal stack was not restored correctly. [Ralf S. Engelschall] *) Backport from Pth 1.1b1: Fixed test_select.c: a char has to be used for read(2) and not an int. And replaced a bogus stdio fgetc() in test_select with pth_read(). [Ralf S. Engelschall] *) Backport from Pth 1.1b1: Fixed AC_CHECK_NSIG test and this way PTH_NSIG define. [Ralf S. Engelschall] *) Backport from Pth 1.1b1: Fixed timeout handling in pth_select(). [Ralf S. Engelschall] Changes between 1.0.3 and 1.0.4 (30-Jul-1999 to 03-Aug-1999) *) Fixed two read/write calls: they didn't use the pth_sc wrapper. [Ralf S. Engelschall] *) Make pth_{read,write}() functions more POSIX compliant by returning 0 when the nbytes argument is given as 0. And make pth_{readv,writev}() functions more POSIX compliant by returning -1/EINVAL when the iovcnt argument is less or equal to 0. [Ralf S. Engelschall] *) Added missing readv/writev wrappers to pthread.c [Ralf S. Engelschall] *) Typo fixes in pth.pod [Eric Newton ] *) Fixed special `--disable-static --enable-shared' situation. [Felix von Leitner , Ralf S. Engelschall] *) Fixed `debug-xx' targets. [Ralf S. Engelschall] Changes between 1.0.2 and 1.0.3 (28-Jul-1999 to 30-Jul-1999) *) Fixed inconsistent and wrong API value: PTH_ATTR_NULL existed and was documented as PTH_ATTR_NONE. But actually the intended and intuitive value was PTH_ATTR_DEFAULT. !BE CAREFUL: This is an API change and requires adjustments in Pth applications! [Ralf S. Engelschall, Eric Newton ] *) Make sure pthread.h.in doesn't redefine the poll(2) environment under built-time because it's already defined by the included pth.h. [Kriton Kyrimis ] *) Add readv/writev environment fallback definitions to pthread.h.in. [Ralf S. Engelschall] *) Support for AmigaOS in pthreads.c: timespec attributes have (correct) ts_ prefix under AmigaOS and not the (bogus) POSIX tv_ prefix [Kriton Kyrimis ] *) Add `#include ' for struct timespec to pthread.h.in and removed timespec references in pth.h.in. [Ralf S. Engelschall] *) Fixed documentation for pth_spawn() and pth_wait(). [Ralf S. Engelschall, Eric Newton ] Changes between 1.0.1 and 1.0.2 (22-Jul-1999 to 28-Jul-1999) *) Upgraded to GNU shtool 1.4.5. [Ralf S. Engelschall] *) Added `uninstall' target to Makefile [Ralf S. Engelschall] *) Cleanup package to conform to the GNU standards: added AUTHORS document; renamed CHANGES to ChangeLog and added NEWS document; renamed LICENSE to COPYING; renamed CREDITS to THANKS; added hint to bug-pth@gnu.org to pth.pod manual. [Ralf S. Engelschall] *) Fixed HAVE_SIGSTACK alternative in pth_mctx.c [Jim Jagielski ] *) Removed internal "was_blocking" attribute in event structure because it was never used. [Ralf S. Engelschall] *) Fixed documentation of pth_xxx_init() synchronization functions. [Ralf S. Engelschall] Changes between 1.0.0 and 1.0.1 (16-Jul-1999 to 22-Jul-1999) *) Added missing documentation for pth_timeout(). [Ralf S. Engelschall] *) Moved pthread_p{read,write}() to pth_high.c as pth_p{read,write}() and added syscall wrapper support for it. [Ralf S. Engelschall] *) Added support for readv(2) and writev(2): There are now four new API functions pth_readv, pth_readv_ev, pth_writev and pth_writev_ev. Additionally on platforms where no native readv(2)/writev(2) exists, the struct iovec is faked and the internal implementation is based on read(2)/write(2). This way the functionality is available everywhere. Finally, the syscall support was extended to cover readv and writev now, too. [Ralf S. Engelschall] *) Added support for FNDELAY to pth_nonblocking. [Ralf S. Engelschall] *) Cleaned up pthread.h and pth.h and added extra forward definitions for struct timeval/timespec to avoid problems on some platforms (like SCO). [Ralf S. Engelschall] *) Added "make check" as an alias for "make test" [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.4 and use new `shtool tarball' [Ralf S. Engelschall] Changes between 1.0b8 and 1.0.0 (16-Jul-1999 to 16-Jul-1999) *) Fixed scheduler reinitialization. [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.4-dev. [Ralf S. Engelschall] *) Cleaned up syscall stuff: usleep() is not longer mapped because it's too unstandardized, sigwait() is again mapped because it is standardized and those platforms which fail are broken. [Ralf S. Engelschall] *) Fixed pthread-config.in: -lpth => -lpthread [Ralf S. Engelschall] *) Downgraded required Autoconf version to 2.12 [Ralf S. Engelschall] *) Moved pth_readline[_ev]() from the Pth API into test_common.c because this function is just a half-way solution (pth_read doesn't use the same buffer) and when one needs buffered I/O it's better to use a _real_ I/O library like Sfio, etc. [Ralf S. Engelschall] *) Cleanup to Makefile.in [Ralf S. Engelschall] Changes between 1.0b7 and 1.0b8 (14-Jul-1999 to 16-Jul-1999) *) Fixed result handling of stack growth detection. [Ralf S. Engelschall, Frank Carpenter ] *) Added explicit libtool --mode flags in Makefile.in to avoid compiler guessing problems on esoteric platforms. [Ralf S. Engelschall] *) Fixed broken condition variable handling: the scheduler incorrectly checked the condition variable flags. [Ralf S. Engelschall, Andrew Hunter ] *) Allow main thread to be joinable to fulfill POSIX. [Ralf S. Engelschall] *) Fixed a nasty bug in the scheduler: the write and exceptional filedescriptor sets were not passed to the select(): ARGL! [Ralf S. Engelschall] *) Added implicit initialization to pth_syscall.c stuff and pthread.c stuff. [Ralf S. Engelschall] *) Fixed again static initializations via PTHREAD_XXX_INITIALIZER [Ralf S. Engelschall] *) Fixed again PTHREAD_ONCE_INIT [Ralf S. Engelschall] *) Added PTHREAD_CANCELED to pthread.h. [Ralf S. Engelschall] *) Fixed POSIX pthread_testcancel(): it is defined to have void return type and not int. [Ralf S. Engelschall] Changes between 1.0b6 and 1.0b7 (14-Jul-1999 to 14-Jul-1999) *) Replaced FALSE value for PTHREAD_ONCE_INIT in pthread.h with 0 to be not dependent on such defines. [Ralf S. Engelschall, Rick Brownrigg] *) Disabled syscall soft mapping for usleep because it causes prototype conflicts with unistd.h on some platforms due to the fact that usleep(3) is bad standardized. [Ralf S. Engelschall, Rick Brownrigg] *) Added pthread_cancel() to POSIX emulation. [Ralf S. Engelschall, Rick Brownrigg] *) Fixed poll(2)/poll.h detection in Autoconf. [Ralf S. Engelschall, Igor A. Minyukoff ] Changes between 1.0b5 and 1.0b6 (11-Jul-1999 to 14-Jul-1999) *) Fixed pth_attr_ctrl(): it failed with EACCES because of incorrect comparisons [Igor A. Minyukoff ] *) Replaced PTHREAD_{MUTEX,COND,RWLOCK}_INITIALIZER with an alternative implementation to allow the assignment to `static' variables, too. [Ralf S. Engelschall, Rick Brownrigg ] Changes between 1.0b4 and 1.0b5 (08-Jul-1999 to 11-Jul-1999) *) Fixed pth_waitpid() semantics. [Ralf S. Engelschall] *) Added pth_poll() and pth_poll_ev() with the following trick: it is always present (independent whether the platform has poll(2) or not), because it is (AND HAS TO BE) internally based on pth_select(). The poll function wrapping support was added, too. [Ralf S. Engelschall] *) Moved pth_init() to the top of all test applications. [Ralf S. Engelschall] *) Added direct system call mapping support: --enable-syscall-soft enables #define's in pth.h and pthread.h which do a soft-mapping of e.g. read to pth_read. This is _always_ possible. --enable-syscall-hard enables the exportation of syscall wrappers (e.g. read) by the Pth library and the calling of the real system calls via syscall(2), hence this variant is only available when the platform supports syscall(2) [e.g. yes: *BSD, Linux, Solaris, HP/UX, IRIX, UnixWare; no: AIX, SCO5] For the soft variant the whole application sources have to include pth[read].h and existing libraries (like stdio!) can still block the whole process, while for the hard variant the application sources do not all have to include pth[read].h an existing libraries (including stdio!) magically use Pth's system call wrappers. [Ralf S. Engelschall] *) Speeded up `configure --help' by not creating the header [Ralf S. Engelschall] *) Added pth_select_ev() and pth_select() based on the new PTH_EVENT_SELECT and the old PTH_EVENT_TIME events which emulate 4.2BSD's select(2), but suspend only the current thread. Additionally added a select() wrapper to the POSIX wrapper API. [Ralf S. Engelschall] *) Added PTH_EVENT_SELECT which can be used to suspend the current thread until an event occured in one of three fd _sets_ similar to select(2) but without the timeout facility. [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.3. This especially fixes the shared library generation under GNU/Linux now. [Ralf S. Engelschall] *) Added pth_abort() and a corresponding pthread_abort() for cruel ways to cancel a thread. [Ralf S. Engelschall] *) Fixed POSIX pread()/pwrite(): Their mutex variables have to be static variables and not local variables. [Ralf S. Engelschall] *) Added #define for POSIX sched_yield() to pthread.h.in [Ralf S. Engelschall] Changes between 1.0b3 and 1.0b4 (07-Jul-1999 to 08-Jul-1999) *) Added POSIX pread() and pwrite() [parallel I/O] emulation functions to pthread.{ch]. [Ralf S. Engelschall] *) Removed double-definition of the fallback types in pthread.h.in. [Ralf S. Engelschall] *) Added pthread_atfork() to pthread.c [Ralf S. Engelschall] *) Added pth_atfork_{push,pop}() functions and moved them together with pth_fork() to a new pth_fork.c source files. Updated also the documentation to describe the new at-fork handlers. [Ralf S. Engelschall] *) Write still missing documentation for pth_cleanup_{push,pop}() [Ralf S. Engelschall] *) Fixed again pth_event_concat(): The last attempt was to inefficient and still had a bug. [Igor A. Minyukoff , Ralf S. Engelschall] *) Cleaned up all name references: The official long name is `GNU Portable Threads', the official short name is `GNU Pth' and the common prefix is `pth_' and `PTH_'. [Ralf S. Engelschall] Changes between 1.0b2 and 1.0b3 (04-Jul-1999 to 07-Jul-1999) *) Remove a trailing comma inside an enum in pthread.h.in [Martin Kraemer ] *) Added two new API event-related functions which can be used to extract the contents of existing events: pth_event_typeof() and pth_event_extract(). [Ralf S. Engelschall] *) Fixed pth_event_concat(): It was broken because it handled real/non-single-event event rings not correctly. [Igor A. Minyukoff , Ralf S. Engelschall] *) Fixed pth_event_walk(): It was broken for PTH_WALK_NEXT [Igor A. Minyukoff , Ralf S. Engelschall] *) Fixed memory leak in pth_join() [Igor A. Minyukoff ] *) Fixed manual page pthread-config.pod [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.4.1 and GNU libtool 1.3.3 [Ralf S. Engelschall] *) Fixed test_pthread.c: sleep(3) requires unistd.h [Ralf S. Engelschall, Martin Kraemer ] *) Cleaned up --enable-batch/--enable-pthread [Ralf S. Engelschall] Changes between 1.0b1 and 1.0b2 (28-Jun-1999 to 04-Jul-1999) *) Upgraded to GNU shtool 1.4.0 [Ralf S. Engelschall] *) Changed return value type of pth_version() from "int" to "long" to avoid problems with too large numbers. [Ralf S. Engelschall] *) Simplified internal representation of pth_once_t [Ralf S. Engelschall] *) !!ATTENTION!! Renamed _anything_ from the old coding-name nps/NPS to the new official name pth/PTH. [Ralf S. Engelschall] *) Added the new POSIX.1c pthread emulation library (pthread* files). It has to be enabled explicitly via --enable-pthread because it's disabled per default) [Ralf S. Engelschall] *) Enhanced pthread.pod manual page. [Ralf S. Engelschall] Changes between 0.9.21 and 1.0b1 (25-Jun-1999 to 28-Jun-1999) *) Added internal pth_util_cpystrn() to pth_util.c [Ralf S. Engelschall] *) removed pth_attr(), pth_priority(), pth_detach() [Ralf S. Engelschall] *) added new pth_attr_*() functions for manipulating pth_attr_t objects; this is now more flexible because those objects now can be also bound to threads when retrieved via pth_attr_of(). THIS IS A HEAVY BUT IMPORTANT API CHANGE! READ THE DOCS AND THE TEST PROGRAMS FOR DETAILS. SORRY, BUT IS HAS TO BE DONE BEFORE WE ENTER THE STABLE 1.0 VERSIONS. [Ralf S. Engelschall] *) Added --enable-profile and --enable-batch to Autoconf [Ralf S. Engelschall] *) Updated pth.pod to reflect latest changes. [Ralf S. Engelschall] *) Renamed pth_state_* to upper case names. [Ralf S. Engelschall] *) Moved pth_sigmask() to pth_high.c [Ralf S. Engelschall] *) Renamed pth_sigraise() to pth_raise() and added support for per-thread signal delivery via pth_raise(). [Ralf S. Engelschall] *) Renamed pth.c to pth_lib.c [Ralf S. Engelschall] *) Upgraded to GNU shtool 1.3.1 [Ralf S. Engelschall] *) Added pth_version() function [Ralf S. Engelschall] *) Added a real test Makefile target via new test_std.c [Ralf S. Engelschall] *) Added more references to pth.pod [Ralf S. Engelschall] ___ ___ / _ \ / _ \ | | | | (_) | | |_| |\__, | ___\___(_) /_/__________________________________________________________ Changes between 0.9.20 and 0.9.21 (24-Jun-1999 to 25-Jun-1999) *) Upgraded to final shtool 1.3.0 release version [Ralf S. Engelschall] *) Removed all variables names from prototypes to avoid conflicts [Ralf S. Engelschall] *) Removed unnecessary casts for malloc() calls. [Ralf S. Engelschall] *) Added pth_sigraise() function for later implementing pthread_kill() and currently at least for testing whether a thread still exists in the system (by sending it a signal 0). [Ralf S. Engelschall] *) Renamed remaining XXX_INITIALIZER to XXX_INIT to be consistent with other init values. [Ralf S. Engelschall] *) Prefixed all pth_attr_t attributes with "at_" [Ralf S. Engelschall] *) Removed pth_equal(): we're not such crazy than POSIX... [Ralf S. Engelschall] Changes between 0.9.19 and 0.9.20 (21-Jun-1999 to 24-Jun-1999) *) Upgraded to latest shtool 1.3.0-dev [Ralf S. Engelschall] *) Changed pth_yield(void) to pth_yield(pth_t) to allow the specification of a thread which should be favored. This allows the usage of the old concept of co-routines where a thread/routine switches to a particular target thread. [Ralf S. Engelschall] *) Added hint about async-safety to pth.pod [Ralf S. Engelschall] *) Added POSIX-style consistency wrapper pth_sigmask() for sigprocmask(). [Ralf S. Engelschall] *) Added try argument to pth_mutex_acquire() to allow us later to emulate POSIX pthread_mutex_trylock() [Ralf S. Engelschall] *) Moved errno stuff to a new dedicated pth_errno.c source file [Ralf S. Engelschall] *) Changed function signatures to support return codes with errno [Ralf S. Engelschall] *) pth_exit(val) in the main() thread now really just terminated the thread and not immediately the process, i.e. the thread goes sleeping until all other threads are gone and then does an exit(val). [Ralf S. Engelschall] Changes between 0.9.18 and 0.9.19 (20-Jun-1999 to 21-Jun-1999) *) Added a POSIX style pth_detach() [Ralf S. Engelschall] *) Added pth_sync.c with POSIX style mutual exclusion locks (mutex), read-write locks (rwlock) and condition variable (cond) support [Ralf S. Engelschall] *) Added a trivial mutex demo to test_misc.c [Ralf S. Engelschall] Changes between 0.9.17 and 0.9.18 (18-Jun-1999 to 20-Jun-1999) *) Greatly extended the manual page pth.pod [Ralf S. Engelschall] *) Added inclusion hack to pthread.h.in [Ralf S. Engelschall] *) Added per thread cleanup functions pth_cleanup_{push,pop}() modeled directly after the POSIX pthread_cleanup_{push,pop}() functions [Ralf S. Engelschall] *) Fixed a memory leak related to pth_key_xxx(). [Ralf S. Engelschall] *) Added internal pth_pqueue_contains() function [Ralf S. Engelschall] *) Added public API function pth_cancel(), pth_cancel_state() and pth_cancel_point() for thread cancellation support modeled after the POSIX thread cancellation facility. [Ralf S. Engelschall] *) Removed pthread.* stuff. It will be re-integrated later for Pth 1.1.x [Ralf S. Engelschall] *) Added cancellation flags to pth_attr(). [Ralf S. Engelschall] *) Added cancellation demo support to test_sig.c [Ralf S. Engelschall] Changes between 0.9.16 and 0.9.17 (09-Jun-1999 to 18-Jun-1999) *) Removed pthread_p.h [Ralf S. Engelschall] *) Upgraded to latest shtool 1.3.0-dev [Ralf S. Engelschall] *) Added stack overflow detection: a SIGSEGV is created when a guardian memory address at the bottom of the stack was overridden. This SIGSEGV can be even catched by the application. [Ralf S. Engelschall] Changes between 0.9.15 and 0.9.16 (04-Jun-1999 to 09-Jun-1999) *) Some Autoconf related cleanups. [Ralf S. Engelschall] *) Added configure --enable-batch for FreeBSD port [Ralf S. Engelschall] *) Removed TODO file [Ralf S. Engelschall] *) Moved sigdelete stuff to new pth_util.c source [Ralf S. Engelschall] *) Added pth_sigwait_ev() [Ralf S. Engelschall] *) Updated pth.pod for recent changes [Ralf S. Engelschall] *) Fixed GNU/Linux libc v5 pth_mctx_set() variant [Ralf S. Engelschall, Felix von Leitner ] Changes between 0.9.14 and 0.9.15 (01-Jun-1999 to 04-Jun-1999) *) Upgraded to shtool 1.2.9 [Ralf S. Engelschall] *) Avoid race conditions in emulated sigsetjmp() switching [Ralf S. Engelschall] *) Added test_sig.c [Ralf S. Engelschall] *) Fixed getsockopt() call: errlen wasn't initialized [Ralf S. Engelschall, Anton Umnikov ] *) Allow pth_event_walk() also to use PTH_UNTIL_OCCURRED to walk to the next/prev. occurred event in a ring [Ralf S. Engelschall] *) Completely overhauled the signal processing in the scheduler [Ralf S. Engelschall] *) Added pth_sigwait() with POSIX semantics [Ralf S. Engelschall] *) Fixed pth_join(): It returns -1 when only one thread is left [Ralf S. Engelschall] *) Autoconf now no longer adds -g for gcc unless --enable-debug [Ralf S. Engelschall] Changes between 0.9.13 and 0.9.14 (01-Jun-1999 to 01-Jun-1999) *) Simplified the event construction by removing unnecessary PTH_UNTIL_XX flags and renaming the remaining ones to PTH_UNTIL_YY_XXX. [Ralf S. Engelschall] *) Renamed PTH_EVENT_IRQ to PTH_EVENT_SIG and implemented it the first time via sigpending()/sigismember(). [Ralf S. Engelschall] *) Documented the event facility in the manual page [Ralf S. Engelschall] *) Switched TCB's name attribute from "char *" to "char[40]" and copy in the value when spawning. This allows applications to generate the name temporarily only. [Ralf S. Engelschall] *) Cleaned up source code at lots of edges... [Ralf S. Engelschall] Changes between 0.9.12 and 0.9.13 (30-May-1999 to 01-Jun-1999) *) Cleaned up Makefile.in even more [Ralf S. Engelschall] *) Add stackaddr argument to pth_attr() to allow the application to specificy a particular stack [Ralf S. Engelschall] *) Added pthread* files which will contain the POSIX.1c threading ("pthread") API wrapper for PTH. This is still work in progress and not enabled per default (one has to use --enable-pthread). [Ralf S. Engelschall] *) Added SunSoft's pthread_June95.ps document ("PThreads Summary") as pthreads.ps [Ralf S. Engelschall] *) Update pth.pod to reflect recent changes. [Ralf S. Engelschall] *) Optimized pth_time_cmp() [Ralf S. Engelschall] *) Added support for per-thread signal masks [Ralf S. Engelschall] *) Made pth_mctx_set() more robust: It now uses sigsuspend for waiting, correctly blocks signals and restores SIGUSR1 stuff [Ralf S. Engelschall] *) Fixed license messages in sources files [Ralf S. Engelschall] *) Removed const from pth_connect and pth_connect_ev to avoid problems [Ralf S. Engelschall] *) Make sure Autoconf doesn't add -lnsl twice [Ralf S. Engelschall] *) Cleaned up pth_mctx.c even more [Ralf S. Engelschall] Changes between 0.9.11 and 0.9.12 (28-May-1999 to 30-May-1999) *) Some fixes to test_httpd.c: init of addrlen, REQ_MAX handling [Ralf S. Engelschall] *) Allow pth_join(NULL) to join any available terminated thread [Ralf S. Engelschall] *) Fixed incorrect pth_event_concat() usages: terminated NULL was missing [Ralf S. Engelschall] *) Added pth_connect, pth_connect_ev and pth_accept_ev functions [Ralf S. Engelschall] *) Removed unnecessary b_extra stuff from pth_read/pth_write [Ralf S. Engelschall] *) Removed "intern" on pth_time(), it's a public API function [Ralf S. Engelschall] *) Added pth_timeout() constructor function [Ralf S. Engelschall] *) Fixed event handling for all pth_xxx_ev() functions [Ralf S. Engelschall] *) Use pth_readline_ev() in test_mp to show how pth_xxx_ev() works [Ralf S. Engelschall] *) Converted pth_nap() and pth_join() to use PTH_MODE_STATIC [Ralf S. Engelschall] *) Fixed event initialization in event rings [Ralf S. Engelschall] *) Kicked out the whole "occurred event ring" stuff because it's not really necessary (one can walk through a single ring and check for occurred events easily) and this way two nasty bugs can be fixed and the event manager simplified a lot. pth_wait() not has only one argument and this is a read-only one, i.e. only the occured flag is set in the events in this ring, but the ring is no longer changed. [Ralf S. Engelschall] Changes between 0.9.10 and 0.9.11 (28-May-1999 to 28-May-1999) *) Fix warnings in test_httpd.c [Ralf S. Engelschall] *) Upgraded to final shtool 1.2.8 [Ralf S. Engelschall] *) Added PTH_KEY_INIT [Ralf S. Engelschall] *) Added PTH_MODE_STATIC for statically allocating a per-thread event. This is now used by the various high-level functions to avoid unnecessary alloc/free sequences. [Ralf S. Engelschall] Changes between 0.9.9 and 0.9.10 (25-May-1999 to 28-May-1999) *) Cleaned up Makefile.in [Ralf S. Engelschall] *) Upgraded to GNU libtool 1.3.2 [Ralf S. Engelschall] *) Upgraded to new shtool 1.2.8 which includes scpp [Ralf S. Engelschall] *) Merged all pth_foo.h's into pth_foo.c's with the help of shtool scpp [Ralf S. Engelschall] *) Updated dependencies after pth_p.h overhauling [Ralf S. Engelschall] *) Added TODO file [Ralf S. Engelschall] Changes between 0.9.8 and 0.9.9 (24-May-1999 to 25-May-1999) *) Finally converted library generation to use GNU libtool [Ralf S. Engelschall] *) Add hints to BIND 8 and adns to manual page [Ralf S. Engelschall] *) Fixed memory leaks in test_httpd and test_mp [Ralf S. Engelschall, Flux ] *) Fixed typos in manual page [Ralf S. Engelschall] *) Fix -lsocket -lnsl tests for Siemens platforms [Ralf S. Engelschall] *) Fix chmod in Autoconf stuff [Ralf S. Engelschall] *) Removed unneccessary pth_ring_linked() [Ralf S. Engelschall] Changes between 0.9.7 and 0.9.8 (23-May-1999 to 24-May-1999) *) Fixed bug in priority queue element counting (q_num) [Ralf S. Engelschall] *) pth_spawn() now inherits the attributes from the parent thread [Ralf S. Engelschall] *) Merged pth_stat() and pth_load() into a generic extensible pth_ctrl() [Ralf S. Engelschall] *) Added new pth_fork() function [Ralf S. Engelschall] *) Added depend target to Makefile.in and generated dependencies [Ralf S. Engelschall] *) Moved joinable argument from pth_attr() to the flags argument [Ralf S. Engelschall] *) Added PTH_CTRL_GETPRIO and PTH_CTRL_GETNAME. [Ralf S. Engelschall] *) Typo was everywhere: preemtive => preemptive [Ralf S. Engelschall] *) Typo was everywhere: occured => occurred [Ralf S. Engelschall] *) Finished manual page [Ralf S. Engelschall] *) Removed useless PTH_UNTIL_RUNNING [Ralf S. Engelschall] *) Replaced errno support with a better variant [Ralf S. Engelschall] *) Lots of small fixes and portability cleanups [Ralf S. Engelschall] Changes between 0.9.6 and 0.9.7 (22-May-1999 to 23-May-1999) *) Added PORTING document. [Ralf S. Engelschall] *) Added errno support (pth_errno.[ch]). [Ralf S. Engelschall] *) Added exponential average load calculations for scheduler. The current load value can be retrieved by the application through pth_load(). [Ralf S. Engelschall] Changes between 0.9.5 and 0.9.6 (21-May-1999 to 22-May-1999) *) Added more documentation. [Ralf S. Engelschall] *) Autoconf now determines the direction of stack grow and pth_mctx_set() uses this information for sigstack() now. [Ralf S. Engelschall] *) Fixed acconfig.h [Ralf S. Engelschall] *) Increased stacksizes (was too small for Solaris and friends) [Ralf S. Engelschall] *) Switched from BSD-style license to LGPL license [Ralf S. Engelschall] *) Fixed a nasty bug in scheduler: when a thread was moved from the wait to the ready queue, it's state wasn't changed to pth_state_ready. This way the next time it was handled it was incorrectly filed to the waiting queue (which caused core dumps because the ev_waiting/ev_occurred pointers didn't exist). This way now the stuff works great also under Solaris! [Ralf S. Engelschall] Changes between 0.9.4 and 0.9.5 (21-May-1999 to 21-May-1999) *) Renamed pth_{get,set}specific() to pth_key_{get,set}data(). [Ralf S. Engelschall] *) Renamed remaining pth_message_xxx() to pth_msgport_xxx() [Ralf S. Engelschall] *) Added pth_{readline,read,write}_ev() functions which are similar to pth_{readline,read,write}() but have an additional event argument which can be used for timeouts and other parallel events. [Ralf S. Engelschall] Changes between 0.9.3 and 0.9.4 (14-May-1999 to 21-May-1999) *) Added ring datastructures (pth_ring.[ch]). [Ralf S. Engelschall] *) Add an AmigaOS-style message port facility (pth_msg.[ch]). [Ralf S. Engelschall] *) Added a message port related test program (test_mp.c). [Ralf S. Engelschall] *) Fix scheduler event management bug: when no I/O events existed plus a next timer but other events (e.g. message ports) already occurred, the scheduler blocked instead of handling the already occurred events first. [Ralf S. Engelschall] *) Fix bug in event creating: the event occurrence flag has to be cleared in pth_wait() to avoid spinning events which were already occurred. [Ralf S. Engelschall] Changes between 0.9.2 and 0.9.3 (14-May-1999 to 14-May-1999) *) Added pth_once() for later (pthreads!) [Ralf S. Engelschall] *) Splitted pth.c into pth.c and pth_high.c [Ralf S. Engelschall] *) Added pth_data.[ch] with pthread style data storage stuff [Ralf S. Engelschall] *) Moved readline stuff from test_httpd to pth_high.c as pth_readline() [Ralf S. Engelschall] *) Removed debugging fprintf from test_httpd to speed it up [Ralf S. Engelschall] *) Fixed listen() call: now uses REQ_MAX for backlog and tests for -1 [Ralf S. Engelschall] *) Fix a polling-related timer bug in the event manager [Ralf S. Engelschall] Changes between 0.9.1 and 0.9.2 (13-May-1999 to 14-May-1999) *) renamed pth_asynchronize() to pth_nonblocking() [Ralf S. Engelschall] *) removed implicit pth_nonblocking() calls [Ralf S. Engelschall] *) speeded up pth_accept by optimizing event alloc/free [Ralf S. Engelschall] *) cleaned up pth_waitpid [Ralf S. Engelschall] *) Changed semantics of pth_wait(var, NULL): it now means that the events in var are not removed (although their chanining might be changed) [Ralf S. Engelschall] *) Renamed internal pth_usleep() to pth_time_usleep() [Ralf S. Engelschall] *) Replaced pth_sleep() with pth_nap() and added two replacement functions: pth_sleep() for a variant of sleep(3) and pth_usleep() for a variant of usleep(3). [Ralf S. Engelschall] *) Optimized scheduler a little bit by inlining pth_time_{set,add,sub} [Ralf S. Engelschall] *) Added pth_pqueue_{head,walk} functions and used it in eventmanager [Ralf S. Engelschall] *) Rewrote pth_sched_eventmanager() without a surrounding loop [Ralf S. Engelschall] *) Implemented a readline() function to speedup test_httpd [Ralf S. Engelschall] Changes between 0.9.0 and 0.9.1 (12-May-1999 to 13-May-1999) *) Imported into CVS repository [Ralf S. Engelschall] Changes between GENESIS and 0.9.0 (Feb-1999 to 12-May-1999) *) Created initial version on FreeBSD [Ralf S. Engelschall] *) Ported to Linux [Ralf S. Engelschall]