changelog

eCos1.31版

2000-03-21  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl:

	Fix thread operations URL.

2000-03-20  John Dallaway  <[e-mail]>

	* cdl/counters.cdl,
	  cdl/interrupts.cdl,
	  cdl/kernel.cdl,
	  cdl/scheduler.cdl,
	  cdl/synch.cdl,
	  cdl/thread.cdl:

	Adjust documentation URLs.

2000-03-07  Jesper Skov  <[e-mail]>

	* tests/mutex3.cxx: Minor tweak from Hugo to allow reduced runtime
	on sims.

2000-02-29  Jonathan Larmour  <[e-mail]>

	* src/debug/dbg-thread-demux.c:
	CYG_HAL_SPARCLITE_SLEB -> CYGPKG_HAL_SPARCLITE_SLEB
	Also fix a comment typo

2000-02-29  Jesper Skov  <[e-mail]>

	* tests/kcache1.c: Don't run last part of test on MIPS sim - it's
	very slow.

2000-02-25  Jesper Skov  <[e-mail]>

	* tests/dhrystone.c: 
	* tests/tm_basic.cxx: 
	Don't allow use of devices for diag output when running performace
	tests.

2000-02-17  Jonathan Larmour  <[e-mail]>

	* include/pkgconf/kernel.h: Make
	CYGSEM_KERNEL_SYNCH_MUTEX_PRIORITY_INHERITANCE_SIMPLE a cdl_component
	since it has children now

2000-02-17  Jesper Skov  <[e-mail]>
	CR101202, CR902078
	* include/mvarimpl.inl: Avoid unlinking the list from the head
	during coalescing.

2000-02-11  Jesper Skov  <[e-mail]>

	* tests/dhrystone.c (PASSES): Reduced for MPC8xx targets.

2000-02-10  Jesper Skov  <[e-mail]>

	* tests/except1.cxx (except0_main): 
	* tests/kexcept1.c (except0_main):
	Also reclaim DATA_TLB_MISS vector if available.

2000-02-03  Jesper Skov  <[e-mail]>

	* src/sched/lottery.cxx (schedule): CYG_HAL_POWERPC_x->CYGPKG_...

2000-02-01  Jonathan Larmour  <[e-mail]>

	* tests/tm_basic.cxx (NTEST_THREADS): Reduce further to fit on small
	targets
	(NMBOXES): Likewise
	(NCOUNTERS): Likewise

	* src/common/clock.cxx: Rework last change - change types of
	clock_dsr_read and delta to unsigned instead

2000-01-31  Jonathan Larmour  <[e-mail]>

	* src/common/clock.cxx (isr): Use HAL_CLOCK_LATENCY, not HAL_CLOCK_READ
	(dsr): Likewise
	
	* cdl/counters.cdl (CYGVAR_KERNEL_COUNTERS_CLOCK_DSR_LATENCY): 
	Copy configury for this option here from include/pkgconf/kernel.h - it
	was left behind by accident

2000-01-31 Simon FitzMaurice  <[e-mail]>
  * cdl/*.cdl:

  Adjust help URLs in line with new doc layout.
  
2000-01-31  Jesper Skov  <[e-mail]>

	* tests/tm_basic.cxx: Use CYG_TEST_NA insetad of PASS when not
	applicable.

2000-01-28 Simon FitzMaurice  <[e-mail]>
  * cdl/*.cdl:

  Adjust help URLs in line with new doc layout.
  
2000-01-26  Hugo Tyson  <[e-mail]>

	* include/clock.hxx (class Cyg_Clock): Add members to convert to
	and from this clock resolution and others.

	* include/clock.inl (convert): New Cyg_Clock member.

	* src/common/clock.cxx (construct_converter): New routine to
	factorize a rational (or very near neighbour) to help with clock
	tick conversions for a particular clock.
	(get_other_to_clock_converter):
	(get_clock_to_other_converter): Cyg_Clock member functions to
	construct converters to or from this clock, using above routine.

	* tests/clockcnv.cxx: New file: test the clock converters.

	* tests/PKGconf.mak (TESTS): Add clockcnv to the tests.

	* cdl/kernel.cdl: Add clockcnv to the tests.

2000-01-20  Hugo Tyson  <[e-mail]>

	* cdl/synch.cdl
	(CYGSEM_KERNEL_SYNCH_MUTEX_PRIORITY_INHERITANCE_SIMPLE_RELAY):
	Change this to an active_if on simple prio-inh.  Was broken.

2000-01-19  Hugo Tyson  <[e-mail]>

	* cdl/*.cdl: Add descriptions to a number of options &c which were
	lacking same, also tidied up other typos as noticed en passant.
	Also reorganized calculated opts about per-thread-data in
	thread.cdl; these are really for info only.

2000-01-19  Jesper Skov  <[e-mail]>
	CR 902054
	* tests/stress_threads.c: Don't assert requested priority =
	allocated. Allow some flexibility of handler priorities.

2000-01-13  Hugo Tyson  <[e-mail]>

	* tests/mutex3.cxx (control_thread): Add a CYG_TEST_INIT();
	Whoops!

2000-01-11  Hugo Tyson  <[e-mail]>

	* cdl/kernel.cdl: Add new kernel test mutex3, put in to test
	RELAY prio inheritance extension.

	* tests/mutex3.cxx: New test case.  It tests mutex prio
	inheritance - or not - in a multiple thread manner, depending only
	on delays and the clock, so it tests that scheduling does the
	right thing, rather than simply checking reported priorities.

	* tests/PKGconf.mak: Add new kernel test mutex3, put in to test
	RELAY prio inheritance extension.

2000-01-11  Hugo Tyson  <[e-mail]>

	* cdl/synch.cdl: New config option added,
	(CYGSEM_KERNEL_SYNCH_MUTEX_PRIORITY_INHERITANCE_SIMPLE_RELAY).

	* include/pkgconf/kernel.h: New config option added,
	(CYGSEM_KERNEL_SYNCH_MUTEX_PRIORITY_INHERITANCE_SIMPLE_RELAY).
	It is a child of ..._PRIORITY_INHERITANCE_SIMPLE and is really
	only there to turn off if you really want the old implementation.

	* src/sync/mutex.cxx (unlock): Add call to
	Cyg_SchedThread::relay_priority() when relinquishing the mutex, in
	order to relay any raised priority to the new owner.  Also count
	the mutex we are waiting for as well as those held for correct
	recovery in the case of inheriting priority then being killed, for
	example.

	* include/sched.hxx (class Cyg_SchedThread::relay_priority): New
	member function declared.

	* src/sched/sched.cxx (Cyg_SchedThread::relay_priority): New
	member function to implement pass-it-on or "discovered ceiling"
	priority inheritance extension to the simple algorithm.  Also use
	get_current_priority() in Cyg_SchedThread::inherit_priority() to
	let this work.

2000-01-11  Hugo Tyson  <[e-mail]>

	* include/clock.hxx (class Cyg_Counter::counter): make the counter
	volatile, otherwise a busy-wait loop for the real time clock to
	change never completes.

1999-12-23  Hugo Tyson  <[e-mail]>

	* src/sched/sched.cxx (inherit_priority): Fix bug; inheriting a
	2nd time would overwrite the saved initial_priority, so
	disinheriting had no effect.

1999-12-13  Jonathan Larmour  <[e-mail]>

	* tests/tm_basic.cxx (run_alarm_tests): Add new measurement of
	latency between alarm firing and thread resume

1999-12-13  Jonathan Larmour  <[e-mail]>

	* cdl/kernel.cdl: Fix some minor mistakes in the lists of tests to
	build conditionally

1999-12-08  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl:

	Conditionally build the tests 'stress_threads', 'kcache1' and
	'kcache2'.

1999-11-30  Hugo Tyson  <[e-mail]>

	* src/common/kapi.cxx (cyg_scheduler_lock, cyg_scheduler_unlock):
	assert good range for sched lock, and be defensive anyway.  This
	is in response to a long time taken to track down a "unlocked too
	many times" bug.

1999-11-24  Gary Thomas  <[e-mail]>

	* tests/tm_basic.cxx: Size tests based on available resources,
	using MLT files.  Fall back if not available.

1999-11-22  Nick Garnett  <[e-mail]>

	* src/sched/bitmap.cxx (Cyg_Scheduler_Implementation::add_thread):
	Fixed typo in assertion, and moved it to a more useful place.

1999-11-03  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl: Define all tests.

1999-10-29  Gary Thomas  <[e-mail]>

	* tests/PKGconf.mak: Enable cache tests for Cirrus Logic boards.

1999-10-27  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl:

	Specify CYGPKG_KERNEL_TESTS in terms of testing
	source files.

1999-10-26  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl:

	Define initial version of CYGPKG_KERNEL_TESTS for
	verification purposes.

1999-10-19  John Dallaway  <[e-mail]>

	* cdl/kernel.cdl:

	Define CDL options for package-specific CFLAGS.

1999-10-08  Jesper Skov  <[e-mail]>

	* tests/kcache1.c: Fix array size.

	* src/sched/bitmap.cxx (add_thread): 
	* src/sched/mlqueue.cxx (add_thread): 
	Check for valid thread priority.

1999-10-07  John Dallaway  <[e-mail]>

	* cdl/counters.cdl,
	  cdl/interrupts.cdl,
	  cdl/scheduler.cdl:

	Specify radio buttons using CDL interfaces.

1999-10-06  Bart Veer  <[e-mail]>

	* cdl/scheduler.cdl:
	Make the schedulers mutually exclusive via a cdl_interface

1999-10-05  Jesper Skov  <[e-mail]>

	* tests/kcache1.c: Reduced memory footprint.

1999-09-25  Jesper Skov  <[e-mail]>

	* tests/stress_threads.c: Added date header and flush() calls.

1999-09-16  Hugo Tyson  <[e-mail]>

	* tests/tm_basic.cxx (run_mutex_tests): Must unlock mutices before
	destroying, given the change below.

	* src/common/kapi.cxx (cyg_cond_destroy): Call the correct
	destructor.  As it happens, it didn't matter because all these
	destructors do is assert the queue is empty, and the queue is the
	2nd word in every case.
	(cyg_mutex_destroy): Call the destructor at all, so that
	assertions are made.

1999-09-13  Gary Thomas  <[e-mail]>

	* src/intr/intr.cxx (call_pending_DSRs_inner): Add assert to check
	for valid DSR (before trying to use it).

1999-09-13  Jesper Skov  <[e-mail]>

	* tests/stress_threads.c: Print out malloc system info.
	(print_statistics): Fix buglet.

1999-09-10  Jesper Skov  <[e-mail]>

	* tests/stress_threads.c: Print out more info. Reduce dump
	frequency as test runs.

1999-08-25  Jonathan Larmour  <[e-mail]>

	* include/pkgconf/kernel.h: Add missing close brace in
	CYGIMP_KERNEL_COUNTERS_CLOCK_LATENCY CDL.

1999-08-24  Nick Garnett  <[e-mail]>

	* tests/kcache1.c (entry0): Depending on architecture, set
	cyg_test_is_simulator for last two tests. Otherwise they take far
	too long and time out.

1999-08-23  Hugo Tyson  <[e-mail]>

	* tests/intr0.cxx (vsr0): 
	* tests/kintr0.c (vsr0): 
	Add a comment to the effect that vsr0() is NOT a valid VSR on any
	known platform; VSRs must be writ in assembler.  Customer
	requested this, the examples are rather confusing otherwise.

1999-08-23  Nick Garnett  <[e-mail]>

	* include/pkgconf/kernel.h:
	* src/common/clock.cxx:
	* tests/tm_basic.cxx:
	Added CYGVAR_KERNEL_COUNTERS_CLOCK_DSR_LATENCY to enable recording
	of DSR latency figures. Added code controlled by this option to
	clock and tm_basic. Also made interrupt latency measurement
	primarily dependent on CYGVAR_KERNEL_COUNTERS_CLOCK_LATENCY since
	whether HAL_CLOCK_LATENCY is defined is not sufficient.

1999-08-17  John Dallaway  <[e-mail]>

	* cdl/counters.cdl, cdl/interrupts.cdl, cdl/scheduler.cdl:

	Implement radio buttons using "FIXME radio" hack in
	description field for now.

1999-08-16  Jonathan Larmour  <[e-mail]>

	* src/sload/sload.c: Rename AM32 to AM31

1999-08-12  Nick Garnett  <[e-mail]>

	* src/debug/dbg_gdb.cxx (dbg_make_threadref): Extended test for
	uninitialized thread pointer.

1999-07-29  Jesper Skov  <[e-mail]>

	* src/sync/mutex.cxx (lock): Removed assertion again. Not possible
	to determine if a violation wil cause a deadlock.

1999-07-27  Jesper Skov  <[e-mail]>

	* src/sync/mutex.cxx (lock): Added simple assertion check for
	deadlocks.

	* tests/stress_threads.c: Only allow printf from main thread to
	prevent deadlocks.

1999-07-14  Hugo Tyson  <[e-mail]>

	* tests/kcache1.c (entry0): Also perform timing tests with
	HAL_DCACHE_INVALIDATE_ALL and/or HAL_ICACHE_INVALIDATE_ALL in the
	loop if provided by the HAL.  Invalidating the Data cache can
	spoil clock operation, so elapsed time appears to be 0.

	* tests/kcache2.c (entry0): Replace the "if (0)" for having a
	copy-back cache with a proper test on a HAL macro.  Even though I
	couldn't actually get those parts of the test to work on the
	platform I'm currently working on.

	* tests/PKGconf.mak (TESTS): Do build kcache1 and kcache2 for the
	ARM_EBSA285; it seems kosher.  kcache2 does absolutely nothing, of
	course.
	
1999-07-09  Jesper Skov  <[e-mail]>
        PR 20210, 20142
	* tests/stress_threads.c: Shifted thread priorities to make room
	for main() at priority 0.

	Fixed indentation.
	
1999-07-08  Jesper Skov  <[e-mail]>
        PR 20244
	* tests/stress_threads.c: Changed cyg_user_start() to main()
	ensuring a bigger stack (it calls sprintf).
	Increased STACK_SIZE by 2kB for printf calls.

1999-06-30  Jesper Skov  <[e-mail]>

	* tests/dhrystone.c: Also NA-quit if CYGPKG_INFRA_DEBUG or
	CYGPKG_KERNEL_INSTRUMENT are enabled.

1999-06-23  Jesper Skov  <[e-mail]>
        PR 20209
	* tests/dhrystone.c: Use fewer loops on ARM targets - they don't
	have a cache.

1999-06-22  Nick Garnett  <[e-mail]>

	* src/intr/intr.cxx (Cyg_Interrupt::chain_isr): Was testing the wrong
	macro for the default ISR. Fixed.

	* src/common/clock.cxx (Cyg_RealTimeClock::isr): Add HANDLED bit