changelog

ecos为实时嵌入式操作系统

	* src/common/kapi.cxx: Changed cyg_exception_call_handler
	definition to be consistent with declaration.

	* include/kapi.h: Change cyg_code_t to signed type.

	* tests/except1.cxx (except0_main): 
	* tests/kexcept1.c (except0_main): 
	Init diag before making first output.

1999-02-26  Jesper Skov  <[e-mail]>
        PR 19284
	* tests/mutex2.cxx (mutex2_main): Set priorities in a different
	order.

1999-02-25  Nick Garnett  <[e-mail]>

	* include/sched.hxx (class Cyg_Scheduler_Base):
	Added CYGBLD_ATTRIB_ASM_ALIAS(cyg_scheduler_sched_lock) to
	declaration of sched_lock member. This makes it have the given
	name in the object and assembler files. This in turn eliminates
	the need to have a C++ mangled name in the HAL.

	* src/intr/intr.cxx:
	Cyg_Interrupt::detach() was not translating the vector to a table
	index correctly in the chained case. Fixed.

1999-02-24  Nick Garnett  <[e-mail]>

	* include/intr.hxx: 
	* src/intr/intr.cxx:
	Added an interrupt disable counter to
	Cyg_Interrupt::disable_interrupts() and
	Cyg_Interrupt::enable_interrupts().

1999-02-23  Nick Garnett  <[e-mail]>

	* include/pkgconf/kernel.h:
	Added support for 66MHz part.

	* include/instrmnt.h:
	Added CYG_INSTRUMENT_EVENT_MUTEX_RELEASE and
	CYG_INSTRUMENT_EVENT_MUTEX_RELEASED. This should have been part of
	the 1999-02-22 checkin.

1999-02-23  Jonathan Larmour  <[e-mail]>

	* tests/clock0.cxx, tests/clock1.cxx, tests/except1.cxx,
	  tests/kcache1.c, tests/kcache2.c, tests/kclock0.c, tests/kclock1.c,
	  tests/kexcept1.c, tests/kflag0.c, tests/kflag1.c, tests/kill.cxx,
	  tests/kintr0.c, tests/kmbox1.c, tests/kmemfix1.c, tests/kmemvar1.c,
	  tests/kmutex0.c, tests/kmutex1.c, tests/ksched1.c, tests/ksem0.c,
	  tests/ksem1.c, tests/kthread0.c, tests/kthread1.c, tests/thread1.cxx:
	Use CYG_TEST_NA() rather than CYG_TEST_PASS_FINISH() for tests that
	are not applicable for this configuration

1999-02-22  Nick Garnett  <[e-mail]>

	* include/kapi.h: 
	* include/mutex.hxx: 
	* src/sync/mutex.cxx: 
	* src/common/kapi.cxx: 
	* tests/mutex2.cxx:
	* tests/PKGconf.mak: 
	Added Kernel API support for thread release, which allows a thread
	to be broken out of any wait. Also added support for releasing all
	threads waiting to lock a mutex, both at C++ and Kernel API
	levels. Added a test program, mutex2, to test this fuctionality
	out.
	
1999-02-20  Jonathan Larmour  <[e-mail]>

	* include/except.hxx: 
	* include/kapidata.h
	Rename CYG_EXCEPTION_COUNT -> CYGNUM_HAL_EXCEPTION_COUNT in line
	with HAL changes
	Rename deliver_exception() -> cyg_hal_deliver_exception()
	QA improvements

	* src/common/clock.cxx:
	Rename CYG_VECTOR_RTC -> CYGNUM_HAL_INTERRUPT_RTC in line with
	HAL changes

	* src/common/except.cxx:
	Rename CYG_EXCEPTION_* -> CYGNUM_HAL_EXCEPTION_* in line
	with HAL changes
	Allow for CYGNUM_HAL_EXCEPTION_MIN != 0
	Rename deliver_exception() -> cyg_hal_deliver_exception()
	Add more tracing, and fix some existing
	QA improvements

	* src/intr/intr.cxx:
	Rename CYG_ISR_* -> CYGNUM_HAL_ISR_* in line with HAL changes
	Add more tracing, and fix some existing
	QA improvements

	* src/sched/bitmap.cxx:
	* src/sched/mlqueue.cxx:
	* src/sync/mutex.cxx:
	Add more tracing and fix some existing
	QA improvements

	* tests/except1.cxx:
	* tests/kexcept1.c
	Rename CYG_EXCEPTION_* -> CYGNUM_HAL_EXCEPTION_* in line
	with HAL changes
	QA improvements
	Use new CYG_TEST_GDBCMD to tell GDB not to stop
	Remove special simulator case as it should now work

	* tests/intr0.cxx:
	* tests/kintr0.c:
	Rename CYG_ISR/VSR_* -> CYGNUM_HAL_ISR/VSR_* in line
	with HAL changes
	QA improvements		

1999-02-16  Jonathan Larmour  <[e-mail]>

	* tests/kexcept1.c (except0_main): Tell GDB to not stop on
	SIGBUS or SIGSEGV since we know we're going to cause an exception

1999-02-12  Hugo Tyson  <[e-mail]>

	* src/common/thread.cxx (idle_thread_stack): 
	Override CYGNUM_KERNEL_THREADS_IDLE_STACK_SIZE if
	CYGNUM_HAL_MINIMUM_STACK_SIZE demands it.

	* include/pkgconf/kernel.h (CYGNUM_KERNEL_THREADS_IDLE_STACK_SIZE):
	Document that this option can be overridden by HALs demands.

1999-02-11  Jesper Skov  <[e-mail]>

	* src/debug/dbg-thread-syscall.h: 
	* src/debug/dbg-thread-demux.c: 
	Added handling of dbg_scheduler_func calls.

1999-02-11  Nick Garnett  <[e-mail]>

	* include/kapi.h: 
	* src/common/kapi.cxx: 
	Added API support for per-thread data.

	* include/kapi.h: 
	* src/common/kapi.cxx:
	* tests/ksem1.c: 
	* tests/tm_basic.cxx: 
	Changed arguments to Kernel C API semaphore functions to take
	cyg_count32 rather than cyg_ucount32. This makes them consistent
	with the underlying implementation and allows for negative counts
	in the future. Changed some tests to match.
	Fixes PR 17877.

1999-02-08  Jesper Skov  <[e-mail]>

	* tests/kexcept1.c (entry0): 
	* tests/except1.cxx (entry0): 
	Don't call cause_exception on SIMs.

1999-02-05  Jesper Skov  <[e-mail]>

	* src/common/clock.cxx (isr): Ignore a latency of 0 when finding
	min_clock_latency.

1999-02-04  Jesper Skov  <[e-mail]>
        PR 19075
	* include/instrmnt.h: Slightly overdid the use of CYG_UNUSED_PARAM
	macros.

1999-02-04  Jesper Skov  <[e-mail]>

	* src/intr/intr.cxx:
	Make instrumentation macros always use the parameters to avoid
	compiler warnings.

1999-02-03  Jesper Skov  <[e-mail]>
        PR 18075
	* tests/memfix1.cxx:
	* tests/kmemfix1.c:
	Make timing less sensitive.
	
1999-02-02  Jesper Skov  <[e-mail]>

	* tests/bin_sem2.cxx: Fixed compiler warning.

1999-02-02  Jesper Skov  <[e-mail]>
        PR 18971
	* tests/bin_sem2.cxx (bin_sem2_main): Reduce run time in SIM.

1999-02-01  Nick Garnett  <[e-mail]>

	* tests/kcache2.c: Added calls to HAL_DCACHE_SYNC() before all
	relevant calls to HAL_DCACHE_DISABLE(). This should have been done
	on 1999-01-25 where is was only done to one instance.

1999-02-01  Gary Thomas  <[e-mail]>

	* tests/tm_basic.cxx: Adjust stack sizes for platforms with
	limited memory.

1999-01-28  Nick Garnett  <[e-mail]>

	* src/intr/intr.cxx: When using chained interrupts,
	interrupt_end() is passed NULL as the intr argument. If
	instrumentation is also enabled, the CYG_INSTRUMENT_INTR() calls
	will try to indirect through zero. Added ifdefs to avoid
	this. Also test intr for NULL before trying to post the DSR.
	Fixes PR 18771.

1999-01-28  Jesper Skov  <[e-mail]>

	* tests/tm_basic.cxx: Fixed strings.

1999-01-26  Hugo Tyson  <[e-mail]>

	* tests/kcache1.c (time0): Waste much less time if running in a
	simulator.  Do only 40 loops instead of 4000.
	In consequence the nasty meddling with MAX_STRIDE depending on
	HAL_xxx_SIM package definitions can go.

1999-01-26  Jesper Skov  <[e-mail]>
        PR 18902
	* src/debug/dbg-thread-demux.c: Oops. Undid my change of 1999-01-21.

1999-01-25  Jonathan Larmour  <[e-mail]>

	* src/common/except.cxx: Make global exception initialisation object
	be constructed at INTERRUPT priority

1999-01-25  Jesper Skov  <[e-mail]>
        PR 18002
	* tests/thread_gdb.c: Ensure that there are enough priorities.

1999-01-25  Jesper Skov  <[e-mail]>
        PR 18875
	* tests/tm_basic.cxx: Ensure that there are enough priorities.

1999-01-25  Nick Garnett  <[e-mail]>

	* tests/kcache2.c (test_dstore): Added call to HAL_DCACHE_SYNC()
	before disabling cache. On some architectures disabling the cache
	would cause the following call to do nothing.
	Fixes PR 18849.

1999-01-22  Jesper Skov  <[e-mail]>

	* tests/tm_basic.cxx: Don't run on Linux target.

1999-01-21  Hugo Tyson  <[e-mail]>
	These changes are to get tests working with the SPARClite port; it
	doesn't do 8-byte register access unless it's properly aligned.
	
	* tests/philo.cxx (cyg_start): 
	Use tests/testaux.hxx to get new threads, to get the alignment
	required; see below.

	* tests/thread2.cxx (thread2_main):
	Use tests/testaux.hxx to get new threads, to get the alignment
	required; see below.

	* tests/tm_basic.cxx: 
	Larger stack for greedy processor.

	* tests/testaux.hxx: 
	thread_obj[][] and stack[][] are now 8-byte aligned like certain
	processors require; Cyg_Thread contains cyg_tick_count which is
	64-bit so any "standalone" C++ object would be so aligned.  These
	dynamically allocated ones should be too.

1999-01-21  Jesper Skov  <[e-mail]>
        PR 18747
	* tests/flag1.cxx: Increase delays/timeouts to prevent failure
	when configured with full trace logging.

1999-01-21  Jesper Skov  <[e-mail]>

	* src/debug/dbg-thread-demux.c: Removed PR 17327 workarounds.

1999-01-19  Jonathan Larmour  <[e-mail]>

	* src/sched/sched.cxx (start): 
	CYG_REFERENCE_SYMBOL -> CYG_REFERENCE_OBJECT in line with
	cyg_type.h change

1999-01-19  Jesper Skov  <[e-mail]>
        PR 18824
	* src/sync/mutex.cxx: 
	* include/pkgconf/kernel.h: Let
	CYGMFN_KERNEL_SYNCH_CONDVAR_TIMED_WAIT require
	CYGFUN_KERNEL_THREADS_TIMER.

1999-01-18  Nick Garnett  <[e-mail]>

	* src/debug/dbg-thread-demux.c (patch_dbg_syscalls): Added test
	for CYGDBG_KERNEL_DEBUG_GDB_THREAD_SUPPORT in patch_dbg_syscalls().

1999-01-15  Gary Thomas  <[e-mail]>

	* include/mvarimpl.inl (free): Memory list sort was incorrect.

	* tests/PKGconf.mak (TESTS): Don't build cache tests on
	some platforms.

1999-01-15  Jesper Skov  <[e-mail]>
        PR 18768
	
	* tests/kmbox1.c:
	* tests/mbox1.cxx:
	Increase delays used so the overhead of GDB packets doesn't make
	the tests fail.

1999-01-14  Jesper Skov  <[e-mail]>

	* src/sched/sched.cxx (start): Use new CYG_REFERENCE_SYMBOL macro.

1999-01-14  Nick Garnett  <[e-mail]>

	* src/debug/dbg_gdb.cxx: Extended return types to include success,
	fail and caller-do-it as required by msnyder.

	* src/debug/dbg-thread-demux.c: Moved prototype of dbg_scheduler()
	to dbg-threads-api.h.

1999-01-13  Jesper Skov  <[e-mail]>

	* src/sched/sched.cxx (start): Better implementation of the
	real_time_clock reference.

1999-01-13  Jesper Skov  <[e-mail]>

	* include/pkgconf/kernel.h: Disable CYGSEM_KERNEL_SCHED_TIMESLICE
	in CYG_HAL_ROM_MONITOR magic.

1999-01-07  Jesper Skov  <[e-mail]>

	* include/pkgconf/kernel.h: Added RTC values for i386/Linux.


1999-01-12  Nick Garnett  <[e-mail]>

	* src/debug/dbg-thread-demux.c:
	Changed API of patch_dbg_syscalls() to take a pointer to the start
	of the vector and insert several vectors.

	* src/debug/dbg_gdb.cxx:
	Added dbg_scheduler() function to allow debuggers to lock and
	unlock the scheduler. This helps when stepping multithreaded code.

	* include/sched.hxx:
	* include/sched.inl:
	Added Cyg_Scheduler::unlock_simple() to decrement the lock without
	rescheduling.

1999-01-05  Jesper Skov  <[e-mail]>
        PR 18574

	* src/sched/sched.cxx (start): Reference the real time clock to
	ensure it is included when linking.

1999-01-04  Jesper Skov  <[e-mail]>

	* tests/thread1.cxx:
	* tests/mbox1.cxx:
	* tests/kmbox1.c:
	* tests/kill.cxx:
	* tests/kflag1.c:
	* tests/kclock1.c:
	* tests/flag1.cxx:
	* src/common/thread.cxx:
	CYGFUN_KERNEL_THREADS_TIMER requires CYGVAR_KERNEL_COUNTERS_CLOCK
	so don't check for the latter explicitly.

1999-01-04  Jesper Skov  <[e-mail]>
	PR 18573

	* tests/thread1.cxx:
	* tests/mbox1.cxx:
	* tests/kmbox1.c:
	* tests/kflag1.c:
	* tests/flag1.cxx:
        Don't run tests that rely on delay() when delay() is only defined
	as an empty function.

1998-12-24  Bart Veer  <[e-mail]>

	* src/sync/mutex.cxx: 
	* src/sync/mbox.cxx: 
	* src/sync/flag.cxx: 
	* src/sched/sched.cxx: 
	* src/mem/memvar.cxx: 
	* src/mem/memfixed.cxx: 
	* src/common/thread.cxx: 
	* src/common/clock.cxx:
	* include/mempoolt.inl: 
	* include/mempolt2.inl: 
	* include/mboxt2.inl: 
	* include/mboxt.inl:
	check_this() member functions should now be const

1998-12-22  Nick Garnett  <[e-mail]>