changelog

eCos1.31版

2000-02-16  Jesper Skov  <[e-mail]>

	* cdl/hal_sparclite_sleb.cdl: removed fix me

2000-02-07  Jesper Skov  <[e-mail]>

	* cdl/hal_sparclite_sleb.cdl: use cpu options when linking.

2000-02-04  Hugo Tyson  <[e-mail]>

	* cdl/hal_sparclite_sleb.cdl (define_proc): Add output of a
	#define for CYGHWR_HAL_SPARCLITE_HAS_ASR17 - because a Fujitsu
	SPARClite does.

2000-01-24  John Dallaway  <[e-mail]>

	* cdl/*.cdl:

	Remove obsolete option CYGTST_TESTING_IDENTIFIER.

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.

1999-12-20  Gary Thomas  <[e-mail]>

	* cdl/hal_sparclite_sleb.cdl: Add -Wl for linker options.

1999-11-25  Gary Thomas  <[e-mail]>

	* include/pkgconf/mlt_sparclite_sleb_rom.h: 
	* include/pkgconf/mlt_sparclite_sleb_ram.h: New file(s).

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

	* cdl/hal_sparclite_sleb.cdl: Define build options.

1999-11-01  Jesper Skov  <[e-mail]>

	* cdl/hal_sparclite_sleb.cdl: Added.
	 Use define_proc for const header defs.
	
1999-10-14  John Dallaway  <[e-mail]>

	* include/pkgconf/mlt_sparclite_sleb_ram.mlt,
	  include/pkgconf/mlt_sparclite_sleb_rom.mlt:

	Fix .mlt files to match the latest .ldi files. ROM startup
	requires latest config tool which contains fix for PR 19799.

1999-10-13  Hugo Tyson  <[e-mail]>

	* include/pkgconf/hal_sparclite_sleb.h: Add CDL and
	[un]definitions for handling optional Multiple Vector Trapping and
	optional copy of vectors into RAM - when it is optional.  Also
	define CYGHWR_HAL_SPARCLITE_HAS_ASR17 so that the arch HAL
	self-configures correctly.

	* include/halboot.si: Copy trampoline code into section
	.ram_vectors at __ram_vectors_start if so configured.

	* include/pkgconf/mlt_sparclite_sleb_ram.ldi: 
	* include/pkgconf/mlt_sparclite_sleb_ram.mlt: 
	* include/pkgconf/mlt_sparclite_sleb_rom.ldi: 
	* include/pkgconf/mlt_sparclite_sleb_rom.mlt: 
	Link in new variable size, sometimes 0 size, ram_vectors section;
	this does away with the need to fib about the size of RAM, it was
	previously defined to start at 0x4000200 to reserve a bit for
	trampoline code.
	(comment from the example .ldi files...)
	The ram_vectors section is to allow some free space to copy
	vectors into from the ROM.  This is required to be variable size
	to accomodate SVT or MVT; 80 bytes vs 4kB.  Copying is not
	necessary with MVT, but it is optional because it may offer
	performance gains.  Copying is required for SVT because the
	(aligned) start of ROM contains initialization instructions.  RAM
	copy is used rather than leave a big gap in the ROM to get an
	aligned address for the trampoline code.  For RAM startup,
	ram_vectors will usually be of size zero, unless MVT and copying
	are enabled for memory estimation reasons.

1999-06-14  Hugo Tyson  <[e-mail]>

	* tests/slebtime.cxx (entry2): Fix a dumb fencepost prob in some
	string poking code for output without printf...

1999-04-23  Jesper Skov  <[e-mail]>

	* src/hal_diag.c (hal_bsp_console_write): Added a workaround for PR
	19926.

1999-04-09  Hugo Tyson  <[e-mail]>

	* include/hal_cache.h (HAL_[DI]CACHE_IS_ENABLED): 
	Implement these macros.

1999-04-09  Hugo Tyson  <[e-mail]>

	* include/halboot.si: Use 10 wait states for access to the 86940
	on CS1; it seems to make the Reliability Difference(tm).

1999-04-08  Hugo Tyson  <[e-mail]>

	* tests/slebtime.cxx: Add CYG_INIT_PRIORITY( APPLICATION ) to the
	two static thread objects so that the test works even when libc
	steals away the default constructors.  Unfortunately this also
	means switching to an assignment constructor instead of a call.

1999-04-08  John Dallaway  <[e-mail]>

	* include/pkgconf/*.ldi: Revised SECTION_* macro arguments to
	avoid padded output sections (PR 19787)

1999-04-07  Hugo Tyson  <[e-mail]>

	* include/hal_hwio.h: New file: define accessors for 86940 via
	alternate address space 4, register names and all that... just a
	paste out of previous hal_diag.h - it was the wrong place, but
	that's how it evolved.

	* include/hal_diag.h: Include hal_hwio.h to get accessors for
	86940 et al.  Do not define them here, it was the wrong place.

	* include/hal_clock.h: 
	* include/hal_xpic.h:
	Include hal_hwio.h instead of hal_diag.h	

1999-04-07  Hugo Tyson  <[e-mail]>

	* tests/slebintr.c (checkallbut): This is a more informative
	version of the test when it fails: this is still under
	investigation somewhat, and running the more verbose version
	nightly will be useful.	

1999-03-30  Hugo Tyson  <[e-mail]>

	* tests/slebintr.c (start): After some thought, the timing code in
	here could cause the test to take arbitrarily long; so some safety
	features added.  Normally the test loop is very fast, but with
	cache effects and instrumentation and -O0 it could take ages; this
	could explain the occasional timeouts we have seen for this test.

1999-03-26  Hugo Tyson  <[e-mail]>

	* include/hal_cache.h: 
	Implement the cache control macros: at least
	for x in ( D I )
	    HAL_xCACHE_ENABLE/DISABLE
	    HAL_xCACHE_INVALIDATE_ALL, HAL_xCACHE_SYNC
	in the first instance.  kcache1 reports sensible numbers too!
	
1999-03-26  Hugo Tyson  <[e-mail]>

	* tests/slebintr.c (HAL_CLOCK_READ): If there is no kernel, the
	clock is not initialized, so looking at it is useless for a clue.
	So provide a dummy so we perform a "few" loops when there's no
	kernel. If there is no kernel, cyg_user_start() is never called,
	main gets in and steals the CPU.  So we hang.  So if there is no
	kernel, define cyg_start() instead to wrest control.  Also very
	much weaken the looping count check; kernel instrumentation, for
	example, is more than enough to slow the world down too much.

	* tests/slebstak.c (cyg_[user_]start):
	If there is no kernel, cyg_user_start() is never called, main gets
	in and steals the CPU.  So we hang.  So if there is no kernel,
	define cyg_start() instead to wrest control.

	* tests/slebtime.cxx: 
	Make this build when no kernel present; include of testcase &c
	was the wrong side of the ifdef.  Doh!

1999-03-25  Hugo Tyson  <[e-mail]>

	* src/hal_priv.c (hal_clock_initialize): Oh dear, the clock was
	only running half the speed it should have; this will make the
	tm_basic figures look bad, maybe.  It _appeared_ right because of
	spurious interrupts, making two ISRs per tick, which covered the
	error beautifully.  So, the initialization is now corrected, now
	that the interrupt management is better dealt with.

	* include/hal_xpic.h (HAL_INTERRUPT_ACKNOWLEDGE): Loop until
	either the interrupt has stopped being asserted by the
	prioritization latch (CLIRL) or the source has actually re-flagged
	the interrupt, so that repeated, fast interrupts will re-interrupt
	rather than causing a loop here.  This is to prevent spurious
	double interrupts caused by HAL_INTERRUPT_ACKNOWLEDGE() not taking
	effect before the CPU re-enabled interrupts; this had been
	covering the clock bug as well as confusing the serial system.

	Also placed a HAL_INTERRUPT_ACKNOWLEDGE() in
	HAL_INTERRUPT_CONFIGURE() since the doc implies it is necessary
	when changing that stuff; a spurious interrupt may be latched.
	
1999-03-24  Hugo Tyson  <[e-mail]>

	* include/hal_xpic.h (HAL_INTERRUPT_CONFIGURE): 
	Implement this functionality, since Gary wanted to experiment with
	it.  Also add HAL_INTERRUPT_QUERY_INFO (platform specific) to read
	back the settings of a particular interrupt source.

	* tests/slebintr.c: NEW FILE: test case for the above new macros.
	* tests/slebstak.c: NEW FILE: tests the register save/restore code
		by carrying out lots of factorial calculations.
	* tests/slebtime.c: NEW FILE: tests HAL clocks for sanity; prints
		useful diagnostics.  Most useful when a human observes, if
		it goes wrong this test is more likely to report a timeout
		than a failure.

	* tests/sleb.c: FILE DELETED - moved to slebstak.c
	
	* tests/PKGconf.mak (TESTS): Reflect above changes.

1999-03-23  John Dallaway  <[e-mail]>

	* include/pkgconf/hal_sparclite_sleb.h: Modify display string
	for consistency.

1999-03-23  Hugo Tyson  <[e-mail]>

	* src/hal_diag.c (hal_diag_write_char): Precisely undo the
	previous change: we disable interrupts to prevent the clock
	advancing during the time taken to output to GDB.

1999-03-18  Hugo Tyson  <[e-mail]>

	* src/hal_diag.c (hal_diag_write_char_serial): Do not disable
	interrupts while calling the CygMon write-line routine; CygMon is
	better now and can cope with interrupts there.

1999-03-17  John Dallaway  <[e-mail]>

	* include/pkgconf/mlt_sparclite_sleb_rom.mlt: Regenerate using
	the eCos configuration tool to resolve errors.

1999-03-17  Hugo Tyson  <[e-mail]>