diff options
| author | rpj <rpj> | 2001-02-09 06:51:30 +0000 |
|---|---|---|
| committer | rpj <rpj> | 2001-02-09 06:51:30 +0000 |
| commit | 1648c7a97f27d10ad302c6141562ece01065e1d7 (patch) | |
| tree | c16f939acf7ac98db38039d747f5804d04c33fe1 /ANNOUNCE | |
| parent | 2b3eede0b834a82c7dce5ec328f3929c0effc536 (diff) | |
Remodeled mutex routines again to eliminate critical sections.
Diffstat (limited to 'ANNOUNCE')
| -rw-r--r-- | ANNOUNCE | 43 |
1 files changed, 33 insertions, 10 deletions
@@ -55,24 +55,47 @@ New: be different to some other implementations, eg
Solaris.
- PTHREAD_MUTEX_NORMAL will simulate thread deadlock
+ PTHREAD_MUTEX_NORMAL will cause thread deadlock
if the owner of a mutex tries to relock it without
- first unlocking it, however the lock will be released
- if the owner thread is async-canceled.
+ first unlocking it. It has slightly less overhead.
-- Pthreads-win32 mutexes are now always based on
- Win32 critical sections. We no longer use Win32
- mutexes when TryEnterCriticalSection isn't
- supported.
+ The behaviour of PTHREAD_MUTEX_DEFAULT can be
+ remapped to any other type through the new
+ non-portable function
+
+ pthread_mutex_setdefaulttype_np()
+
+ (see README.NONPORTABLE)
+
+ Remapping only effects newly initialised mutexes.
+ So if you want behaviour more like Solaris
+ POSIX mutexes, which are non-recursive by default,
+ you can call this at the start of your application:
+
+ int previousType;
+ pthread_mutex_setdefaulttype_np(PTHREAD_MUTEX_ERRORCHECK,
+ &previousType);
+
+- Pthreads-win32 mutexes are now based on Win32
+ critical sections for all Windows versions. We no longer
+ depend on TryEnterCriticalSection.
- Thomas Pfaff <tpfaff@gmx.net>
+ This may change again before the next snapshot
+ to rely solely on Win32's Interlocked* routines.
+
Bugs fixed:
+- Pthread_mutex_trylock() now properly returns EBUSY
+ even when the current thread owns the mutex.
+ Consequently, pthread_mutex_destroy() will no longer
+ destroy a locked mutex (it will return EBUSY).
+ - Thomas Pfaff <tpfaff@gmx.net>
+
- The condition variable and read-write lock functions
- have been improved. For details re the [fixed] problems
- in the CV implementation see the file README.CV
+ have been improved. For discussion re the [fixed]
+ problems in the CV implementation see the file README.CV
- Alexander Terekhov <TEREKHOV@de.ibm.com>
-
Known bugs in this snapshot
---------------------------
|