Updated to latest CVS version of pthreads

1 files changed, 326 insertions, 26 deletions

diff --git a/pthreads/ChangeLog b/pthreads/ChangeLog
index b2396ebf4..fcd364b8b 100644
--- a/pthreads/ChangeLog
+++ b/pthreads/ChangeLog
@@ -1,3 +1,303 @@
+2012-10-16  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* Makefile: Remove SDK environment setting; now needs to be done

+	explicitly before running any nmake.

+	* GNUmakefile: Move per command-line ARCH setting to TESTS_ENV.

+

+2012-10-04  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* pthread_tryjoin_np.c: New API

+	* pthread.c (pthread_try_join_np): Added.

+	* common.mk (pthread_tryjoin_np): Added.

+	* pthread.h (pthread_tryjoin_np): Added.

+	* NEWS: Updated.

+	* README: Updated.

+	* ANNOUNCE: Updated.

+

+2012-10-02  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* sched.h (cpu_set_t): Redefined.

+	* implement.h (_sched_cpu_set_vector_): Created as the private equivalent

+	of cpu_set.

+	(pthread_thread_t_.cpuset): Type change. 

+	* sched_setaffinity.c: Reflect changes to cpu_set_t and _sched_cpu_set_vector_.

+	* pthread_setaffinity.c: Likewise.

+	* create.c: Likewise.

+	* pthread_self.c: Likewise.

+	* ptw32_new.c: Likewise.

+

+2012-09-28  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* pthread.c: pulls all individual source modules into a single

+	translation unit.

+	* common.mk: Remove everything related to the non-inlined dll targets.

+	This removes all the intermediate .c files that #include other .c files

+	except for pthread.c.

+	* Makefile: Remove and rename targets; remove or edit variables. All

+	of this is to remove the intermediate translation unit aggregation

+	source files.

+	* GNUmakefile: Likewise.

+	* attr.c: Removed.

+	* barrier.c: Removed.

+	* cancel.c: Removed.

+	* condvar.c: Removed.

+	* exit.c: Removed.

+	* fork.c: Removed.

+	* misc.c: Removed.

+	* mutex.c: Removed.

+	* nonportable.c: Removed.

+	* private.c: Removed.

+	* rwlock.c: Removed.

+	* sched.c: Removed.

+	* spin.c: Removed.

+	* sync.c: Removed.

+	* tsd.c: Removed.

+

+2012-09-28  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* Makefile: expand on rudimentary install target; add DEST_LIB_NAME

+	variable defaulting to "pthread.lib".

+	* GNUmakefile: Add install target similar to Makefile with DEST_LIB_NAME

+	defaulting to "libpthread.a".

+

+2012-09-28  Daniel Richard. G <danielg at teragram dot com>

+

+	* all: #include<config.h>; renamed HAVE_PTW32_CONFIG_H define in

+	build files to HAVE_CONFIG_H since we no longer need a

+	uniquely-named symbol for this.

+	* Bmakefile: Removed _WIN32_WINNT assignment from build files since

+	this is now handled in source.

+	* Wmakefile: Likewise.

+	* Makefile: Added mkdir invocations to "install" target.

+	* common.mk: Elaborated the pthread.$(OBJEXT) dependency list.

+	* pthread.h: Removed the #include"config.h" bit.

+

+2012-09-23  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* GNUmakefile: Modify "all-tests" to use new targets in tests

+	GNUmakefile.

+	* Makefile: Similarly.

+

+2012-09-22  Daniel Richard. G <danielg at teragram dot com>

+

+	* GNUmakefile: Reordered the command lines in the "help" target

+	to match the ordering of the targets in the makefile, which IMO

+	is nicer to the eye; tweaked some of the parentheticals for better

+	clarity; delete *.manifest files, in case the user just finished

+	doing an MSVC build.

+	* Makefile: Use *.static_stamp for the static targets instead of

+	*.static; added a note re VC++6 and /EHs vs. /EHa; reordered the

+	targets, and added a number of new ones (e.g. VSE-small-static);

+	added a note recommending *-inlined and *-small-static to make

+	things a little easier for users bewildered by the large number of

+	targets; reworked the all-tests[-cflags] targets so that (1) more

+	useful targets are built first (the small-static targets make it

+	easier to track down compilation errors), (2) *-debug build and

+	test targets can be used, (3) less-useful build/test permutations

+	are enabled only if EXHAUSTIVE and/or MORE_EXHAUSTIVE is defined,

+	and (4) /MDd and /MTd are covered too; "nmake all-tests-cflags

+	EXHAUSTIVE=1 MORE_EXHAUSTIVE=1" takes a few hours to run; moved

+	some of the VCE targets up, since the pattern in the file was

+	already to list VCE targets first, then VSE, then VC; actually

+	touch the stamp file in the stamp targets.

+	* README.NONPORTABLE: It's "DllMain", not "dllMain".

+	* common.mk: Start of an attempt to define dependencies for

+	pthread.$(OBJEXT).

+	* implement.h: Generalized PTW32_BROKEN_ERRNO into

+	PTW32_USES_SEPARATE_CRT; don't do the autostatic-anchoring thing

+	if we're not building the library!

+	* pthread.h: Moved the PTW32_CDECL bit into sched.h. pthread.h

+	already #includes sched.h, so the latter is a good place to put

+	definitions that need to be shared in common; severely simplified

+	the errno declaration for Open Watcom, made it applicable only to

+	Open Watcom, and made the comment less ambiguous; updated the long

+	comment describing PTW32_BROK^WPTW32_USES_SEPARATE_CRT; added

+	(conditional) declaration of pthread_win32_set_terminate_np(), as

+	well as ptw32_terminate_handler (so that eh.h doesn't have to get

+	involved).

+	* pthread_cond_wait.c: Missed a couple of errno conversions.

+	* pthread_mutex_consistent.c: Visual Studio (either 2010 or 2008

+	Express, don't recall now) actually errored out due to charset

+	issues in this file, so I've replaced non-ASCII characters with

+	ASCII approximations.

+	* ptw32_threadStart.c: Big rewrite of ptw32_threadStart().

+	Everything passes with this, except for GCE (and I can't figure

+	out why).

+	* sched.h: Moved the PTW32_CDECL section here (and made it

+	idempotent); need to #include <stdlib.h> for size_t (one of the test

+	programs #includes sched.h as the very first thing); moved the

+	DWORD_PTR definition up, since it groups better with the pid_t

+	definition; also need ULONG_PTR, don't need PDWORD_PTR; can't use

+	PTW32_CONFIG_MSVC6, because if you only #include sched.h you

+	don't get that bit in pthread.h; use a cpp symbol

+	(PTW32_HAVE_DWORD_PTR) to inhibit defining *_PTR if needed. Note

+	that this isn't #defined inside the conditional, because there are

+	no other instances of these typedefs that need to be disabled, and

+	sched.h itself is already protected against multiple inclusion;

+	DWORD_PTR can't be size_t, because (on MSVC6) the former is "unsigned

+	long" and the latter is "unsigned int" and C++ doesn't see them as

+	interchangeable; minor edit to the comment... I don't like saying

+	"VC++" without the "Microsoft" qualifier; use PTW32_CDECL instead of

+	a literal __cdecl (this was why I moved the PTW32_CDECL bit into this

+	file).

+	* semaphore.h: Put in another idempotentized PTW32_CDECL bit here;

+	use PTW32_CDECL instead of __cdecl, and fixed indentation of function

+	formal parameters.

+

+2012-09-21  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* create.c: Major changes to incorporate CPU affinity inheritance.

+	* pthread_self.c: Likewise.

+	* ptw32_new.c (cpuset): Initialise new pthread_thread_t element.

+	* pthread.h (DWORD_PTR): Conditional definition moved to sched.h.

+	* sched.h (DWORD_PTR): As above; other changes.

+	* sem_post.c: Fix errno handling and restructure.

+	* sem_getvalue.c: Fix return value and restructure.

+	

+2012-09-18  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* sched_setaffinity.c: New API to set process CPU affinity in POSIX

+	context; compatibility with Linux.

+	* pthread_setaffinity.c: Likewise.

+	* implement.h (pthread_t_): Added cpuset element.

+	* sched.h: Added new prototypes. 

+	* sched.h (cpu_set_t): Support for new process and thread affinity API.

+	* pthread.h: Added new prototypes.

+

+2012-09-16  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* README (Version numbering): Changes back to major.minor.micro.

+	* README.NONPORTABLE: Updated description around process/thread

+	attach/detach routines.

+

+2012-09-05  Daniel Richard. G <danielg at teragram dot com>

+

+	* implement.h: whitespace adjustment.

+	* dll.c: Facilitate PTW32_STATIC_LIB being defined in a header file.

+

+2012-09-04  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* Makefile (VCEFLAGS): Changed from /EHsc to /EHs which fixed a problem

+	in tests/once3.c which was causing it to hang.

+

+2012-09-03  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* Makefile: Remove descriptive info from help target, just list the

+	available targets. Output tends to be poorly formatted and cluttered

+	otherwise.

+	(VCE-static): Add VC++ static build target.

+	(VCE-small-static): Likewise.

+	(VCE-small-static-debug): Likewise.

+	(VCE-small-static-debug): Likewise.

+

+2012-09-02  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* All: correct spelling to 'cancellation'.

+

+2012-08-31  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* pthread_attr_getschedpolicy.c: Remove pedantic arg check.

+	* pthread_getschedparam.c: Likewise.

+	* pthread_mutex_timelock.c: Restructure to address unreached final

+	return statement.

+

+2012-08-31  Daniel Richard. G <danielg at teragram dot com>

+

+	* implement.h (INLINE): only define if building the inlined make targets. G++

+	complained about undefined reference to ptw32_robust_mutex_remove() because it

+	appears in separate compilation units for "make GCE".

+

+2012-08-29  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* ptw32_MCS_lock.c (ptw32_mcs_flag_wait): Fix cast in first 'if' statement.

+	* pthread_mutex_consistent.c (comment): Fix awkward grammar.

+	* pthread_mutexattr_init.c: Initialize robustness element.

+

+2012-08-29  Daniel Richard. G <danielg at teragram dot com>

+

+	* implement.h (PTW32_INTERLOCKED_SIZE): Define as long or LONGLONG.

+	(PTW32_INTERLOCKED_SIZEPTR): Define as long* or LONGLONG*.

+	* pthread_attr_getschedpolicy.c (SCHED_MAX): Fix cast.

+	* ptw32_mutex_check_need_init.c: Fix static mutexattr_t struct initializations.

+	* ptw32_threadStart.c (ExceptionFilter): Add cast.

+	* ptw32_throw.c: Add cast.

+

+2012-08-18  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* pthread_timedjoin_np.c: New non-portable function.

+	* common.mk (pthread_timedjoin_np): Add new function.

+	* nonportable.c (pthread_timedjoin_np): Likewise.

+

+2012-08-16  Daniel Richard. G <danielg at teragram dot com>

+

+	* pthread.h (PTW32_CONFIG_MINGW): Defined to simplify complex macro combination.

+	* (PTW32_CONFIG_MSVC6): Likewise.

+	* (PTW32_CONFIG_MSVC8): Likewise.

+	* autostatic.c: Substitute new macros.

+	* create.c: Likewise.

+	* pthread_cond_wait.c: Likewise.

+	* pthread_exit.c: Likewise.

+	* pthread_once.c: Likewise.

+	* pthread_rwlock_timedwrlock.c: Likewise.

+	* pthread_rwlock_wrlock.c: Likewise.

+	* pthread_win32_attach_detach_np.c: Likewise.

+	* ptw32_relmillisecs.c: Likewise.

+	* ptw32_threadDestroy.c: Likewise.

+	* ptw32_threadStart.c: Likewise.

+	* ptw32_throw.c: Likewise.

+	* sem_timedwait.c: Likewise.

+	* sem_wait.c: Likewise.

+	* implement.h: Likewise.

+	* sched.h: Likewise.

+

+2012-08-11  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* common.mk (default_target): restore previous behaviour of outputing

+	useful help when "make" is run without a target argument.

+

+2012-08-11  Daniel Richard. G <danielg at teragram dot com>

+

+	* autostatic.c (ptw32_autostatic_anchor): new function; other

+	changes aimed at de-abstracting functionality.

+	* impliment.h (ptw32_autostatic_anchor): dummy reference to

+	ensure that autostatic.o is always linked into static applications.

+	* GNUmakefile: Various improvements.

+	* Makefile: Likewise.

+

+2012-03-19  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* implement.h: Fix interlocked pointer casting under VC++ x64.

+

+2012-03-19  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* implement.h (ptw32_spinlock_check_need_init): added missing

+	forward declaration.

+	

+2012-07-19  Daniel Richard. G <danielg at teragram dot com>

+

+	* common.mk: New; macros common to all build environment makefiles.

+	* Bmakefile: Include new common.mk

+	* Makefile: Likewise; various fixes; added normal and small objects

+	static build.

+	* GNUmakefile: Likewise.

+

+2012-03-18  Ross Johnson <ross dot johnson at homemail dot com dot au>

+

+	* create.c (pthread_create): add __cdecl attribute to thread routine

+	arg

+	* implement.h (pthread_key_t): add __cdecl attribute to destructor

+	element

+	(ThreadParms): likewise for start element

+	* pthread.h (pthread_create): add __cdecl to prototype start arg

+	(pthread_once): likewise for init_routine arg

+	(pthread_key_create): likewise for destructor arg

+	(ptw32_cleanup_push): replace type of routine arg with previously

+	defined ptw32_cleanup_callback_t

+	* pthread_key_create.c: add __cdecl attribute to destructor arg

+	* pthread_once.c: add __cdecl attribute to init_routine arg

+	* ptw32_threadStart.c (start): add __cdecl to start variable type

+

+

 2011-07-06  Ross Johnson <ross dot johnson at homemail dot com dot au>

 

 	* pthread_cond_wait.c (pragma inline_depth): this is almost redundant

@@ -345,7 +645,7 @@
 2007-01-06  Romano Paolo Tenca <rotenca at telvia dot it>

 

 	* pthread_cond_destroy.c: Replace sem_wait() with non-cancelable

-	ptw32_semwait() since pthread_cond_destroy() is not a cancelation

+	ptw32_semwait() since pthread_cond_destroy() is not a cancellation

 	point.

 	* implement.h (ptw32_spinlock_check_need_init): Add prototype.

 	* ptw32_MCS_lock.c: Reverse order of includes.

@@ -627,11 +927,11 @@
 	This uses late initialisation similar to the technique already used for

 	pthreads-win32 mutexes and semaphores (from Alexander Terekhov).

 

-	The subsequent cancelation cleanup additions (by rpj) could not be implemented

+	The subsequent cancellation cleanup additions (by rpj) could not be implemented

 	without sacrificing some of the efficiency in Gottlob's design. In particular,

 	although each once_control uses it's own event to block on, a global CS is

 	required to manage it - since the event must be either re-usable or

-	re-creatable under cancelation. This is not needed in the non-cancelable

+	re-creatable under cancellation. This is not needed in the non-cancelable

 	design because it is able to mark the event as closed (forever).

 

 	When uncontested, a CS operation is equivalent to an Interlocked operation

@@ -746,7 +1046,7 @@
 	'PThreadStateCancelPending' state as short cut to avoid entering

 	kernel space via WaitForSingleObject() call. This was obviated

 	by user space sema acquisition in sem_wait() and sem_timedwait(),

-	which are also cancelation points. A call to pthread_testcancel()

+	which are also cancellation points. A call to pthread_testcancel()

 	was required, which introduced a kernel call, effectively nullifying

 	any gains made by the user space sem acquisition checks.

 	* pthread_cancel.c (pthread_cancel): Set new thread

@@ -768,7 +1068,7 @@
 	being the address of the thread struct. The new pthread_t retains

 	the reuse stack but provides virtually unique thread IDs.

 	* sem_wait.c (ptw32_sem_wait_cleanup): New routine used for

-	cancelation cleanup.

+	cancellation cleanup.

 	* sem_timedwait.c (ptw32_sem_timedwait_cleanup): Likewise.

 

 2004-10-22  Ross Johnson  <rpj at callisto.canberra.edu.au>

@@ -808,12 +1108,12 @@
 	keeps it's own counter. Although the motivation to do this has existed

 	for a long time, credit goes to Alexander Terekhov for providing

 	the logic. I have deviated slightly from AT's logic to add the waiters

-	count, which has made the code more complicated by adding cancelation

+	count, which has made the code more complicated by adding cancellation

 	cleanup. This also appears to have broken the VCE (C++ EH) version of

 	the library (the same problem as previously reported - see BUGS #2),

 	only apparently not fixable using the usual workaround, nor by turning

 	all optimisation off. The GCE version works fine, so it is presumed to

-	be a bug in MSVC++ 6.0. The cancelation exception is thrown and caught

+	be a bug in MSVC++ 6.0. The cancellation exception is thrown and caught

 	correctly, but the cleanup class destructor is never called. The failing

 	test is tests\semaphore4.c.

 	* sem_wait.c (sem_wait): Implemented user space check model.

@@ -1016,15 +1316,15 @@
 

 	* pthread_cancel.c (pthread_cancel): Adapted to use auto-detected

 	QueueUserAPCEx features at run-time.

-	(ptw32_RegisterCancelation): Drop in replacement for QueueUserAPCEx()

+	(ptw32_Registercancellation): Drop in replacement for QueueUserAPCEx()

 	if it can't be used. Provides older style non-preemptive async

-	cancelation.

+	cancellation.

 	* pthread_win32_attach_detach_np.c (pthread_win32_attach_np):

 	Auto-detect quserex.dll and the availability of alertdrv.sys;

 	initialise and close on process attach/detach.

-	* global.c (ptw32_register_cancelation): Pointer to either

-	QueueUserAPCEx() or ptw32_RegisterCancelation() depending on

-	availability. QueueUserAPCEx makes pre-emptive async cancelation

+	* global.c (ptw32_register_cancellation): Pointer to either

+	QueueUserAPCEx() or ptw32_Registercancellation() depending on

+	availability. QueueUserAPCEx makes pre-emptive async cancellation

 	possible.

 	* implement.h: Add definitions and prototypes related to QueueUserAPC.

 

@@ -1034,7 +1334,7 @@
 	APC feature.

 	* pthread_cancel.c (pthread_cancel): Initial integration of

 	QueueUserAPCEx into pthreads-win32 to provide true pre-emptive

-	async cancelation of threads, including blocked threads.

+	async cancellation of threads, including blocked threads.

 

 2004-05-06  Makoto Kato  <raven at oldskool.jp>

 

@@ -1141,7 +1441,7 @@
 

 2003-09-03  Ross Johnson  <rpj at callisto.canberra.edu.au>

 

-	* w32_cancelableWait.c (ptw32_cancelable_wait): Allow cancelation

+	* w32_cancelableWait.c (ptw32_cancelable_wait): Allow cancellation

 	of implicit POSIX threads as well.

 

 2003-09-02  Ross Johnson  <rpj at callisto.canberra.edu.au>

@@ -1557,7 +1857,7 @@
 2002-02-07  Alexander Terekhov<TEREKHOV at de.ibm.com>

 

 	* nonportable.c (pthread_delay_np): Make a true

-	cancelation point. Deferred cancels will interrupt the

+	cancellation point. Deferred cancels will interrupt the

 	wait.

 

 2002-02-07  Ross Johnson  <rpj at setup1.ise.canberra.edu.au

@@ -1666,8 +1966,8 @@
 2002-02-02  Ross Johnson  <rpj at setup1.ise.canberra.edu.au>

 

 	* cancel.c: Rearranged some code and introduced checks

-	to disable cancelation at the start of a thread's cancelation

-	run to prevent double cancelation. The main problem

+	to disable cancellation at the start of a thread's cancellation

+	run to prevent double cancellation. The main problem

 	arises if a thread is canceling and then receives a subsequent

 	async cancel request.

 	* private.c: Likewise.

@@ -1732,7 +2032,7 @@
 	* pthread.h: Unless the build explicitly defines __CLEANUP_SEH,

 	__CLEANUP_CXX, or __CLEANUP_C, then the build defaults to

 	__CLEANUP_C style cleanup. This style uses setjmp/longjmp

-	in the cancelation and thread exit implementations and therefore

+	in the cancellation and thread exit implementations and therefore

 	won't do stack unwinding if linked to applications that have it

 	(e.g. C++ apps). This is currently consistent with most/all

 	commercial Unix POSIX threads implementations.

@@ -1919,7 +2219,7 @@
 2001-07-15  Ross Johnson  <rpj at setup1.ise.canberra.edu.au>

 

 	* rwlock.c (pthread_rwlock_wrlock): Is allowed to be

-	a cancelation point; re-enable deferred cancelability

+	a cancellation point; re-enable deferred cancelability

 	around the CV call.

 

 2001-07-10  Ross Johnson  <rpj at setup1.ise.canberra.edu.au>

@@ -1982,8 +2282,8 @@
 	for this object.

 	* pthread.h (PTHREAD_BARRIER_INITIALIZER): Removed.

 	* rwlock.c (pthread_rwlock_wrlock): This routine is

-	not a cancelation point - disable deferred

-	cancelation around call to pthread_cond_wait().

+	not a cancellation point - disable deferred

+	cancellation around call to pthread_cond_wait().

 

 2001-07-05  Ross Johnson  <rpj at setup1.ise.canberra.edu.au>

 

@@ -2235,7 +2535,7 @@
 	* (pthread_mutex_lock): Likewise.

 	* (pthread_mutex_trylock): Likewise.

 	* (pthread_mutex_unlock): Likewise.

-	* Use longjmp/setjmp to implement cancelation when building the library

+	* Use longjmp/setjmp to implement cancellation when building the library

 	using a C compiler which doesn't support exceptions, e.g. gcc -x c (note

 	that gcc -x c++ uses exceptions).

 	* Also fixed some of the same typos and eliminated PT_STDCALL as

@@ -2563,7 +2863,7 @@
 	* pthread.h: Add compile-time message when using

 	MSC_VER compiler and C++ EH to warn application

 	programmers to use PtW32Catch instead of catch(...)

-	if they want cancelation and pthread_exit to work.

+	if they want cancellation and pthread_exit to work.

 

 	* implement.h: Remove #include <semaphore.h>; we

 	use our own local semaphore.h.

@@ -2611,7 +2911,7 @@
 	* pthread.h (PtW32CatchAll): Add macro. When compiling

 	applications using VC++ with C++ EH rather than SEH

 	'PtW32CatchAll' must be used in place of any 'catch( ... )'

-	if the application wants pthread cancelation or

+	if the application wants pthread cancellation or

 	pthread_exit() to work.

 

 2000-08-03  Ross Johnson  <rpj at special.ise.canberra.edu.au>

@@ -2845,7 +3145,7 @@
 

 1999-11-10  Ross Johnson  <rpj at ixobrychus.canberra.edu.au>

 

-	* sync.c (pthread_join): Make it a deferred cancelation point.

+	* sync.c (pthread_join): Make it a deferred cancellation point.

 

 	* misc.c (pthread_self): Explicitly initialise implicitly

 	created thread state to default values.

@@ -4574,7 +4874,7 @@ Sun Jul 26 00:09:59 1998  Ross Johnson  <rpj at ixobrychus.canberra.edu.au>
  	a part of making the eventual pthreads DLL binary compatible

  	through version changes.

 

-	* condvar.c (cond_wait): Add cancelation point. This applies the

+	* condvar.c (cond_wait): Add cancellation point. This applies the

 	point to both pthread_cond_wait() and pthread_cond_timedwait().

 

 	* exit.c (pthread_exit): Rename "this" to "us".