GThreadPool defaulted to 0 for max_unused_threads (meaning thread-pool
threads would exit immediately if there was not already another task
waiting for them), and 0 for max_idle_time (meaning unused threads
would linger forever, though this is only relevant if you changed
max_unused_threads).
However, GIOScheduler changed the global defaults to 2 and 15*1000,
respectively, arguing that these were more useful defaults. And they
are, so let's use them.
https://bugzilla.gnome.org/show_bug.cgi?id=661767
And remove the 'joinable' argument from g_thread_new() and
g_thread_new_full().
Change the wording in the docs. Clarify expectations for
(deprecated) g_thread_create().
Deprecate both g_thread_create functions and add
g_thread_new() and g_thread_new_full(). The new functions
expect a name for the thread.
Change GThreadPool, GMainContext and GDBus to create named threads.
https://bugzilla.gnome.org/show_bug.cgi?id=660635
- remove all inline assembly versions
- implement the atomic operations using either GCC intrinsics, the
Windows interlocked API or a mutex-based fallback
- drop gatomic-gcc.c since these are now defined in the header file.
Adjust Makefile.am accordingly.
- expand the set of operations: support 'get', 'set', 'compare and
exchange', 'add', 'or', and 'xor' for both integers and pointers
- deprecate g_atomic_int_exchange_and_add since g_atomic_int_add (as
with all the new arithmetic operations) now returns the prior value
- unify the use of macros: all functions are now wrapped in macros that
perform the proper casts and checks
- remove G_GNUC_MAY_ALIAS use; it was never required for the integer
operations (since casting between pointers that only vary in
signedness of the target is explicitly permitted) and we avoid the
need for the pointer operations by using simple 'void *' instead of
'gpointer *' (which caused the 'type-punned pointer' warning)
- provide function implementations of g_atomic_int_inc and
g_atomic_int_dec_and_test: these were strictly macros before
- improve the documentation to make it very clear exactly which types
of pointers these operations may be used with
- remove a few uses of the now-deprecated g_atomic_int_exchange_and_add
- drop initialisation of gatomic from gthread (by using a GStaticMutex
instead of a GMutex)
- update glib.symbols and documentation sections files
Closes#650823 and #650935
* glib/gthreadpool.c (wakeup_thread_marker): make this a "const
gpointer" rather than a gconstpointer to avoid warnings later
* glib/pcre/pcre_ucp_searchfuncs.c:
* glib/pcre/pcre_valid_utf8.c: #include "config.h"
* glib/tests/printf.c (test_d): fool gcc into not warning about
some printf format strings that we know are dubious
svn path=/trunk/; revision=7552
* glib/gthreadpool.c: Make sure
g_thread_pool_stop_unused_threads() actually stops unused threads
and global limits (like max idle time and max unused threads) can
be set without creating a thread pool first. Fixed#335215 (patch
from Chris Wilson).
* tests/threadpool-test.c: Added two new tests, tests setting
global limits before creating a thread pool. The second test
makes sure unused threads are actually stopped when using the
g_thread_pool_stop_unused_threads().
* glib/gthreadpool.c: Updated the documentation to explain that
when the maximum threads is > 1 the sort functionality is not 100%
accurate due to the ramdom nature of the scheduler choosing which
threads to execute. Fixes bug #334943.
* tests/threadpool-test.c: Disabled the debugging by default and
fixed the sort test to set the maximum threads to 1 to guarantee
the thread entry function is called in order.
2006-02-15 Sebastian Wilhelmi <seppi@seppi.de>
* glib/gthreadpool.c: Fix deadlock when signalling the thread
which freed a thread pool (#331110, Chris Wilson).
2006-01-17 Sebastian Wilhelmi <seppi@seppi.de>
* glib/gthreadpool.c: To avoid deadlocks get rid of the settings
G_LOCK. Use the unused_thread_queue lock instead. Change
g_thread_pool_thread_proxy such that threads only wait on
non-exlusive pools for at most a 1/2 second. Do not reorder tasks
due to superfluous tasks. Global tasks wait at most for
max-idle-time milliseconds. Make sure, that no task is woken up
twice for the same event via a wakeup_serial. This fixes#324228.
* tests/threadpool-test.c: Adapt test accordingly. Do not pass
invalid NULL into the thread pools. This as well fixes#327290.
* docs/reference/glib/glib-sections.txt:
* glib/glib.symbols:
* glib/gthreadpool.[ch]:
- Added new API g_thread_pool_get_idle_time() and
g_thread_pool_set_idle_time(). (#324228).
* tests/threadpool-test.c:
- Updated test case to do thread pool sorting, thread pool with
no sorting and a thread pool with idle thread timeouts.
2005-12-20 Matthias Clasen <mclasen@redhat.com>
* glib/glib.symbols:
* glib/gthreadpool.h:
* glib/gthreadpool.c (g_thread_pool_set_sort_function): New function
to sort tasks pushed into a threadpool. (#324479, Martyn Russell)
* tests/threadpool-test.c: Test this.
2005-07-20 Matthias Clasen <mclasen@redhat.com>
* glib/gthreadpool.c (g_thread_pool_free): Don't get
stuck in here if immediate is TRUE. (#310954,
Hong Jen Yee)
* tests/threadpool-test.c (main): Test immediate == TRUE.
2005-03-13 Matthias Clasen <mclasen@redhat.com>
Make PLT-reduction work with gcc4, and don't include
everything in galias.h:
* glib/glib.symbols: Group symbols by header and source file.
* glib/makegalias.pl: Protect definitions by the same
preprocessor symbols used to guard the headers. Move
the alias declarations to a separate file which is
produced when calling makegalias.pl -def
* glib/Makefile.am (galiasdef.c): Add a rule to generate this
file.
* glib/*.c: Include galias.h after the other GLib headers,
include galiasdef.c at the bottom.
2002-04-13 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* glib/gthreadpool.c (g_thread_pool_thread_proxy): Unlock the pool
for all threads leaving it. (#78348)
2001-06-07 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Look for nanosleep function.
* gtimer.c: Use nanosleep for g_usleep, when found.
* gtimer.c, gtimer.h: Add g_time_val_add function. Closes#54271.
* gasyncqueue.c: Documentation updates.
* gthreadpool.c: Use g_time_val_add now that we have it.
* glib/glib-sections.txt, glib/tmpl/date.sgml: Add g_time_val_add.
* glib/tmpl/threads.sgml: Updated.
2001-05-18 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gthread.c, gthread.h: Renamed g_thread_create to
g_thread_create_full and added macro g_thread_create, which omits
'stack_size', 'bound' and 'priority' parameters. Also removed
'bound' from GThread struct.
* gthreadpool.h, gthreadpool.c: Adapted GThreadPool to the above
changes. GThreadPool lost the 'priority' and 'bound'
members. g_thread_pool_new the 'stack_size', 'bound' and
'priority' parameters.
* tests/mainloop-test.c, tests/thread-test.c,
tests/threadpool-test.c: Adapted to the above changes.
2001-05-09 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gthread.c, gthread.h: Renamed 'value' and 'arg' to 'data' and
'thread_func' to 'func' to make it more consistent with the rest
of GLib.
* gthreadpool.c, gthreadpool.h: Moved 'stack_size' from the public
members of GThreadPool to the private ones. Renamed 'thread_func'
to 'func' as above. Moved up 'user_data' in g_thead_pool_new
argument list and in GThreadPool struct.
2001-05-08 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gmain.c, gthread.c, gthread.h: Moved func and arg members from
GRealThread to GThread, such that they can be accessed by the
user.
* gthread.c, gthread.h: Due to popular demand (Tim being the
populus here ;-) threads now have a 'return value', which is
returned by g_thread_join and is either the return of the topmost
thread function or the value given to g_thread_exit.
* gthreadpool.c, tests/mainloop-test.c, tests/thread-test.c:
Adapted to the above change.
2001-04-17 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gthreadpool.c (g_thread_pool_thread_proxy): Until now every
thread pool always had at least one tread waiting to avoid
switching overhead in case a new task would be added soon after
one finished. This however means a big waste of threads, if many
mostly inactive thread pools are involved. Now such a waiting
thread will only wait for half a second (This value is of course
very randomly picked) and go to the global threadpool afterwards.
MCVS: ----------------------------------------------------------------------
2001-04-03 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gthreadpool.c: Added documentation.
* gthreadpool.c: The global thread pool now also is seperated for
bound and unbound threads. Only threads with standard stack size
go to the global pool. g_thread_pool_new now protects the global
setup of inform_mutex etc. with a lock. Fixed some typos. Unlock
the queue after g_thread_pool_wakeup_and_stop_all in the proxy.
2000-11-21 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Add a surrogate for thread priorities using PID
niceness for systems with no thread priorities and different PIDs
for threads of the same process (most notably: Linux). Define
G_THREAD_USE_PID_SURROGATE in that case, as used by
gthread-posix.c. Also make the system thread bigger by
sizeof (long) to contain the thread's PID.
* gfileutils.c: Include stdlib.h for mkstemp prototype.
* gthread.c: Add priority range checks to the affected functions.
* gthreadpool.c: Remove unused variable.
* gthread-impl.c, gthread-posix.c, gthread-solaris.c: Removed
g_thread_map_priority function in favour of the
g_thread_priority_map array. Initialize the array with
PRIORITY_{...}_VALUE, if available and interpolate beetween the
bounds if .._NORMAL_.. and .._HIGH_.. are not available.
* gthread-posix.c: If we should use the PID niceness as a
surrogate for thread priorities (G_THREAD_USE_PID_SURROGATE is
defined), then disable normal priority handling and use PIDs and
setpriority() instead. Depends on the thread to write its PID into
the place after the thread id right after thread creation.
2000-11-09 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gthreadpool.c: Don't take threads with other priorities into
account as changing the priority is highly unportable. (Actually
using it at all already is unportable, but even sometimes where
that works, changing priority is not possible).
