Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 694, Thu Jul 27 16:00:25 2000 UTC	revision 847, Mon Jun 25 16:11:22 2001 UTC
#	Line 11	Line 11
11	Date:	Date:
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14
15		----------------------------------------------------------------------
16		Name: Matthias Blume
17		Date: 2001/06/25 12:05:00 EDT
18		Tag: blume-20010625-nlffigen
19		Description:
20
21		This update fixes a number of minor bugs in ml-nlffigen as reported by
22		Nick Carter <nbc@andrew.cmu.edu>.
23
24		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
25		2. Default names for generated files are now derived from the name of
26		the C file *without its directory*.  In particular, this causes generated
27		files to be placed locally even if the C file is in some system directory.
28		3. Default names for generated signatures and structures are also derived
29		from the C file name without its directory.  This avoids silly things
30		like "structure GL/GL".
31		(Other silly names are still possible because ml-nlffigen does not do
32		a thorough check of whether generated names are legal ML identifiers.
33		When in doubt, use command line arguments to force particular names.)
34
35		----------------------------------------------------------------------
36		Name: Matthias Blume
37		Date: 2001/06/21 12:25:00 EDT
38		Tag: blume-20010621-eXene
39		Description:
40
41		eXene now compiles and (sort of) works again.
42
43		The library name (for version > 110.33) is $/eXene.cm.
44
45		I also added an new example in src/eXene/examples/nbody.  See the
46		README file there for details.
47
48		----------------------------------------------------------------------
49		Name: Matthias Blume
50		Date: 2001/06/20 16:40:00 EDT
51		Tag: blume-20010620-cml
52		Description:
53
54		CML now compiles and works again.
55
56		Libraries (for version > 110.33):
57
58		$cml/cml.cm            Main CML library.
59		$cml/basis.cm          CML's version of $/basis.cm.
60		$cml/cml-internal.cm   Internal helper library.
61		$cml/core-cml.cm       Internal helper library.
62		$cml-lib/trace-cml.cm  Tracing facility.
63		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
64
65		The installer (config/install.sh) has been taught how to properly
66		install this stuff.
67
68		----------------------------------------------------------------------
69		Name: Matthias Blume
70		Date: 2001/06/19 17:55:00 EDT
71		Tag: blume-20010619-instantiate
72		Description:
73
74		This un-breaks the fix for bug 1432.
75		(The bug was originally fixed in 110.9 but I broke it again some
76		time after that.)
77
78		----------------------------------------------------------------------
79		Name: Matthias Blume
80		Date: 2001/06/19 17:25:00 EDT
81		Tag: blume-20010619-signals
82		Description:
83
84		This should (hopefully) fix the long-standing signal handling bug.
85		(The runtime system was constructing a continuation record with an
86		incorrect descriptor which would cause the GC to drop data on the floor...)
87
88		----------------------------------------------------------------------
89		Name: Matthias Blume
90		Date: 2001/06/15 15:05:00 EDT
91		Tag: blume-20010615-moresparc
92		Description:
93
94		Here is a short late-hour update related to Sparc c-calls:
95
96		-- made handling of double-word arguments a bit smarter
97
98		-- instruction selection phase tries to collapse certain clumsily
99		constructed ML-Trees; typical example:
100
101		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
102
103		This currently has no further impact on SML/NJ since mlriscGen does
104		not seem to generate such patterns in the first place, and c-calls
105		(which did generate them in the beginning) has meanwhile been fixed
106		so as to avoid them as well.
107
108		----------------------------------------------------------------------
109		Name: Matthias Blume
110		Date: 2001/06/15 15:05:00 EDT
111		Tag: blume-20010615-sparc
112		Description:
113
114		The purpose of this update is to provide an implementation of NLFFI
115		on Sparc machines.
116
117		Here are the changes in detail:
118
119		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
120		the Sparc implementation of the c-calls API.
121		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
122		for accessing the ML frame.  Thus, we have a real frame pointer and
123		can freely modify %sp without need for an omit-frame-ptr phase.
124		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
125		this case.
126		* ml-nlffigen has been taught to produce code for different architectures
127		and calling conventions.
128		* In a way similar to what was done in the x86 case, the Sparc
129		backend uses its own specific extension to mltree.  (For example,
130		it needs to be able to generate UNIMP instructions which are part
131		of the calling convention.)
132		* ml-nlffi-lib was reorganized to make it more modular (in particular,
133		to make it easier to plug in new machine- and os-dependent parts).
134
135		There are some other fairly unrelated bug fixes and cleanups as well:
136
137		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
138		that they properly share their libraries with existing SML/NJ libraries.
139		* I fixed a minor cosmetic bug in CM, supressing certain spurious
140		follow-up error messages.
141		* Updates to CM/CMB documentation.
142
143		TODO items:
144
145		* MLRISC should use a different register as its asmTemp on the Sparc.
146		(The current %o2 is a really bad choice because it is part of the
147		calling conventions, so things might interfere in unexpected ways.)
148
149		----------------------------------------------------------------------
150		Name: Matthias Blume
151		Date: 2001/06/07
152		Tag: blume-20010607-calls
153		Description:
154
155		A number of internal changes related to C calls and calling conventions:
156
157		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
158		number of bytes popped implicitly (by the callee).  In most cases
159		this field is 0 but on x86/win32 it is some non-zero value.  This
160		is information provided for the benefit of the "omit-frameptr" pass.
161		2. The CALL instruction on the x86 carries a similar "pops" field.
162		The instruction selection phase copies its value from the ML-Tree
163		CALL statement.
164		3. On all other architectures, the instruction selection phase checks
165		whether "pops=0" and complains if not.
166		4. The c-calls implementation for x86 now accepts two calling conventions:
167		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
168		up after the call and pops is set to 0.  For "stdcall", the caller
169		does nothing, leaving the cleanup to the callee; pops is set to
170		the number of bytes that were pushed onto the stack.
171		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
172		distinguish between "ccall" and "stdcall".
173		6. The UNIMP instruction has been added to the supported Sparc instruction
174		set. (This is needed for implementing the official C calling convention
175		on this architecture.)
176		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
177		work with the latest CM.
178
179		----------------------------------------------------------------------
180		Name: Matthias Blume
181		Date: 2001/06/05 15:10:00 EDT
182		Tag: blume-20010605-cm-index
183		Description:
184
185		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
186
187		1. CM can now generate "index files".  These are human-readable files
188		that list on a per-.cm-file basis each toplevel symbol defined or
189		imported.  The location of the index file for
190		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
191		To enable index-file generation, set CM.Control.generate_index to true
192		or export an environment-symbol: export CM_GENERATE_INDEX=true.
193
194		The CM manual has been updated accordingly.
195
196		2. I made some slight modifications to the c-calls API in MLRISC.
197
198		a) There is now a callback to support saving/restoring of
199		dedicated but caller-save registers around the actual call
200		instruction.
201		b) One can optionally specify a comment-annotation for the
202		call instruction.
203
204		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
205		(For example, the comment annotation shows the C prototype of
206		the function being called.)
207
208		----------------------------------------------------------------------
209		Name: Matthias Blume
210		Date: 2001/06/01 13:30:00 EDT
211		Tag: blume-20010601-nlffi-cleanup
212		Description:
213
214		This is mostly a cleanup of MLFFI stuff:
215
216		- some signature files have been put into a more exposed place
217		- the ugly 'f type parameter is gone (simplifies types tremendously!)
218		- ml-nlffigen changed accordingly
219		- tutorial updated
220
221		Other changes:
222
223		- author's affiliation in CM manual(s) updated
224		- some more recognized keywords added to Allen's sml.sty
225
226		----------------------------------------------------------------------
227		Name: Matthias Blume
228		Date: 2001/05/25 15:30:00 EDT
229		Tag: blume-20010525-iptr
230		Description:
231
232		- put the official 110.33-README (as it appears on the ftp server) under
233		CVS
234		- fixed a small bug related to incomplete pointer types in
235		ml-nlffigen
236		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
237		(it does not need the 'f type parameter)
238
239		----------------------------------------------------------------------
240		Name: Matthias Blume
241		Date: 2001/05/23 14:30:00 EDT
242		Tag: Release_110_33
243		Description:
244
245		New version number (110.33).  New bootfiles.
246
247		----------------------------------------------------------------------
248		Name: Matthias Blume
249		Date: 2001/05/22 18:06:00 EDT
250		Tag: blume-20010522-targets
251		Description:
252
253		Made install.sh use file config/targets.customized if it exists, falling
254		back to config/targets if it doesn't.  This way one can have a customized
255		version of the targets file without touching the "real thing", thus
256		eliminating the constant fear of accidentally checking something bogus
257		back into the CVS repository...  (File config/targets.customized must
258		not be added to the repository!)
259
260		----------------------------------------------------------------------
261		Name: Matthias Blume
262		Date: 2001/05/22 16:30:00 EDT
263		Tag: blume-20010522-minitut
264		Description:
265
266		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
267		struct returns.
268		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
269		incomplete, preliminary documentation for NLFFI.
270
271		----------------------------------------------------------------------
272		Name: Matthias Blume
273		Date: 2001/05/14 11:30:00 EDT
274		Tag: blume-20010514-script
275		Description:
276
277		Some bugs in install script fixed.
278
279		In addition to that I also made a slight change to the NLFFI API:
280		Functors generated by ml-nlffigen now take the dynamic library as a
281		straight functor argument, not as a suspended one.  (The original
282		functor code used to force the suspension right away anyway, so there
283		was nothing gained by this complication of the interface.)
284
285		----------------------------------------------------------------------
286		Name: Matthias Blume
287		Date: 2001/05/11 14:35:00 EDT
288		Tag: blume-20010511-ml-nlffi
289		Description:
290
291		I finally took the plunge and added my new FFI code to the main
292		repository.  For x86-linux it is now ready for prime-time.
293
294		There are two new subdirectories of "src":
295
296		- ml-nlffi-lib:
297		The utility library for programs using the FFI interface.
298		Here is the implementation of $/c.cm and its associated low-level
299		partners $/c-int.cm and $/memory.cm.
300		- ml-nlffigen:
301		A stand-alone program for generating ML glue code from C source
302		code.
303
304		Building ml-nlffigen requires $/ckit-lib.cm.
305
306		The config/install.sh script has been updates to do the Right Thing
307		(hopefully).
308
309		Notice that the source tree for the C-Kit will not be put under "src"
310		but directly under the installation root directory.  (This is the
311		structure that currently exists on the CVS server when you check out
312		module "sml".)  Fortunately, config/install.sh knows about this oddity.
313
314		Bugs: No documentation yet.
315
316		----------------------------------------------------------------------
317		Name: Matthias Blume
318		Date: 2001/05/09 16:35:00 EDT
319		Tag: blume-20010509-cpscontract
320		Description:
321
322		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
323		wrapper/unwrapper elimination did not decrement usage counts and some
324		dead variables got overlooked by the dead-up logic.)
325
326		----------------------------------------------------------------------
327		Name: Lal George
328		Date: 2001/05/08  17:26:09 EDT
329		Tag: george-20010508-omit-frameptr
330		Description:
331
332		Changes to implement the omit-frame-pointer optimization to support
333		raw C calls. For now, there is only support on the Intel x86, but
334		other architectures will follow as more experience is gained with this.
335
336
337		----------------------------------------------------------------------
338		Name: Matthias Blume
339		Date: 2001/05/07 14:40:00 EDT
340		Tag: blume-20010507-proxies
341		Description:
342
343		I made into "proxy libraries" all libraries that qualify for such a
344		change.  (A qualifying library is a library that has another library or
345		groups as its sole member and repeats that member's export list
346		verbatim.  A proxy library avoids this repetition by omitting its export
347		list, effectively inheriting the list that its (only) member exports.
348		See the CM manual for more explanation.)
349		The main effect is that explicit export lists for these libraries
350		do not have to be kepts in sync, making maintenance a bit easier.
351
352		I also added copyright notices to many .cm-files.
353
354		Last but not least, I made a new set of bootfiles.
355
356		----------------------------------------------------------------------
357		Name: Matthias Blume
358		Date: 2001/05/04 17:00:00 EDT
359		Tag: blume-20010504-cm-lsplit
360		Description:
361
362		0. John merged pending changes to $/smlnj-lib.cm
363
364		1. Allen's previous change accidentally backed out of one of Lal's
365		earlier changes.  I undid this mistake (re-introducing Lal's change).
366
367		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
368		within the compiler where applicable.  There is some code simplification
369		because of that.
370
371		3. The "split" phase (in FLINT) is now part of the default list of phases.
372		Compiler.Control.LambdaSplitting.* can be used to globally control the
373		lambda-splitting (cross-module-inlining) engine.  In addition to that,
374		it can now also be controlled on a per-source basis: CM has been taught
375		a new tool parameter applicable to ML source files.
376
377		- To turn lambda-splitting off completely:
378		local open Compiler.Control.LambdaSplitting in
379		val _ = set Off
380		end
381		- To make "no lambda-splitting" the global default (but allow per-source
382		overriding); this is the initial setting:
383		local open Compiler.Control.LambdaSplitting in
384		val _ = set (Default NONE)
385		end
386		- To make "lambda-splitting with aggressiveness a" the global default
387		(and allow per-source overriding):
388		local open Compiler.Control.LambdaSplitting in
389		val _ = set (Default (SOME a))
390		end
391
392		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
393		write (in the respective .cm-file):
394		a.sml (lambdasplitting:off)
395		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
396		a.sml (lambdasplitting:on)
397		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
398		<a> is a decimal non-negative integer):
399		a.sml (lambdasplitting:<a>)
400		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
401		a.sml (lambdasplitting:infinity)
402		- To use the global default for a.sml:
403		a.sml (lambdasplitting:default)
404		or simply
405		a.sml
406
407		----------------------------------------------------------------------
408		Name: Allen Leung
409		Date: 2001/05/04 01:57:00 EDT
410		Tag: leunga-20010504-sync
411		Description:
412
413		MLRISC features.
414
415		1. Fix to CMPXCHG instructions.
416		2. Changed RA interface to allow annotations in callbacks.
417		3. Added a new method to the stream interface to allow annotations updates.
418
419		----------------------------------------------------------------------
420		Name: Matthias Blume
421		Date: 2001/05/01 11:45:00 EDT
422		Tag: blume-20010501-pcedittmp
423		Description:
424
425		Changed install.sh to use the current working directory instead of
426		/usr/tmp for a temporary file (pcedittmp).  The previous choice
427		of /usr/tmp caused trouble with MacOS X because of file premission
428		problems.
429
430		----------------------------------------------------------------------
431		Name: Matthias Blume
432		Date: 2001/04/20 11:10:00 EDT
433		Tag: blume-20010420-inMLflag
434		Description:
435
436		- added vp_limitPtrMask to vproc-state.h
437		(for use by the raw-C-calls mechanism to implement proper interrupt
438		handling)
439		- made the ML compiler aware of various data-structure offsets so it
440		can generate code for accessing the vp_inML flag and vp_limitPtrMask
441		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
442		raw C-calls
443
444		----------------------------------------------------------------------
445		Name: Lal George
446		Date: 2001/04/20 09:15:28 EDT
447		Tag: george-20010420-macosX
448		Description:
449
450		- Changes to port to Mac OS X; Darwin.
451
452		- In the process I found that sqrt was broken on the PPC, because the
453		fsqrt instruction is not implemented.
454
455		----------------------------------------------------------------------
456		Name: Matthias Blume
457		Date: 2001/04/18 12:45:00 EDT
458		Tag: blume-20010418-ccalls
459		Description:
460
461		- fixed two off-by-4 errors in the x86-specific c-calls implementation
462		(this bug prevented structure arguments containing pointers from being
463		passed correctly)
464		- changed the raw-C-call code in mlriscGen.sml in such a way that
465		structure arguments are represented as a pointer to the beginning
466		of the structure (instead of having a series of synthesized arguments,
467		one for each structure member)
468
469		- made makeml script's verbosity level configurable via environment
470		variable (MAKEML_VERBOSITY)
471
472		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
473		in rawmem-x86.sml; we are now using the real thing
474
475		----------------------------------------------------------------------
476		Name: Matthias Blume
477		Date: 2001/03/22 16:25:00 EST
478		Tag: blume-20010322-bootfiles
479		Description:
480
481		Created a new set of bootfiles (for your automatic installation convenience).
482
483		----------------------------------------------------------------------
484		Name: Matthias Blume
485		Date: 2001/03/22 15:10:00 EST
486		Tag: blume-20010322-rawmem-parcm
487		Description:
488
489		1. All "raw memory access" primitives for the new FFI are implemented now
490		(at least on the x86).
491		2. Some further cleanup of CM's parallel make mechanism.
492
493		----------------------------------------------------------------------
494		Name: Matthias Blume
495		Date: 2001/03/19 17:53:00 EST
496		Tag: blume-20010319-parallel
497		Description:
498
499		Parallel make (using compile servers) now works again.
500
501		To this end, CM.stabilize and CMB.make have been modified to work in
502		two passes when compile servers are attached:
503		1. Compile everything, do not perform stabilization; this pass
504		uses compile servers
505		2. Stabilize everything; this pass does not use compile servers
506		If there are no compile servers, the two passes are combined into one
507		(as before).  Splitting the passes increases the inherent parallelism
508		in the dependency graph because the entire graph including all
509		libraries is available at the same time.  This, in turn, improves
510		server utilization.  The downside is that the master process will
511		have to do some extra work after compilation is done (because for
512		technical reasons it must re-read all the binfiles during stabilization).
513
514		----------------------------------------------------------------------
515		Name: Matthias Blume
516		Date: 2001/03/16 12:22:00 EST
517		Tag: blume-20010316-bootfiles
518		Description:
519
520		Created a new set of bootfiles (for your automatic installation convenience).
521
522		----------------------------------------------------------------------
523		Name: Matthias Blume
524		Date: 2001/03/16 11:00:00 EST
525		Tag: blume-20010316-MLTREE-fixup
526		Description:
527
528		This is a minor fixup for an (untagged) earlier commit by Allen.
529		(A file was missing).
530
531		----------------------------------------------------------------------
532		Name: Allen Leung
533		Date: Mon Mar  5 18:54:57 EST 2001
534		Tag: leunga-20010305-cut-support
535
536		1. New support for alternative control-flow in MLTREE.
537		Currently we support
538
539		FLOW_TO(CALL ...., [k1,...,kn])
540
541		This is needed for 'cuts to' in C-- and try/handle-like constructs
542		in Moby
543
544		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
545
546		2. Register Allocator
547
548		Changes in interface [from Fermin, John]
549
550		3. Alpha 8-bit SLL support [Fermin]
551
552		4. All architectures
553
554		A new module (ClusterExpandCopies) for expanding parallel copies.
555
556		----------------------------------------------------------------------
557		Name: Allen Leung
558		Date: 2001/02/27 23:07:00 EST
559		Tag: leunga-20010227-minor-stuff
560
561		1. Alpha bug fix for CMOVNE
562		2. Handle mltree COND(..,FCMP ...,...)
563		3. Bug fix in simplifier
564
565		----------------------------------------------------------------------
566		Name: Matthias Blume
567		Date: 2001/01/30 17:50:00 EST
568		Tag: blume-20010130-sync
569		Description:
570
571		This is just a minor update to sync my devel branch with the main brach.
572		The only visible change is the addition of some README files.
573
574		----------------------------------------------------------------------
575		Name: Matthias Blume
576		Date: 2001/01/12 23:30:00 JST
577		Tag: blume-20010112-bootfiles
578		Description:
579
580		Made a new set of bootfiles that goes with the current state of the
581		repository.
582
583		----------------------------------------------------------------------
584		Name: Matthias Blume
585		Date: 2001/01/12 21:20:00 JST
586		Tag: blume-20010112-sync
587		Description:
588
589		I am just flushing out some minor changes that had accumulated in
590		my private branch in order to sync with the main tree.  (This is
591		mainly because I had CVS trouble when trying to merge _into_ my
592		private branch.)
593
594		Most people should be completely unaffected by this.
595
596		----------------------------------------------------------------------
597		Name: Allen Leung
598		Date: Thu Jan 11 21:03:00 EST 2001
599		Tag: leunga-20010111-labexp=mltree
600		Description:
601
602		1.  Removed the type LabelExp and replace it by MLTree.
603		2.  Rewritten mltree-simplify with the pattern matcher tool.
604		3.  There were some bugs in alpha code generator which would break
605		64-bit code generation.
606		4.  Redo the tools to generate code with the
607		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
608		are now generated by perl scripts.
609
610		----------------------------------------------------------------------
611		Name: Matthias Blume
612		Date: 2001/01/10 21:55:00 JST
613		Tag: blume-20010110-rcc
614		Description:
615
616		The RCC stuff now seems to work (but only on the x86).
617		This required hacking of the c-calls interface (and -implementation) in
618		MLRISC.
619
620		Normal compiler users should be unaffected.
621
622		----------------------------------------------------------------------
623		Name: Matthias Blume
624		Date: 2001/01/09 01:20:00 JST
625		Tag: blume-20010109-rcc
626		Description:
627
628		This is a fairly big patch, flushing out a large number of pending
629		changes that I made to my development copy over the last couple of days.
630
631		Of practical relevance at this moment is a workaround for a pickling
632		bug that Allen ran into the other day.  The cause of the bug itself is
633		still unknown and it might be hard to fix it properly, but the
634		workaround has some merits of its own (namely somewhat reducing pickling
635		overhead for certain libraries).  Therefore, I think this solution should
636		be satisfactory at this time.
637
638		The rest of the changes (i.e., the vast majority) has to do with my
639		ongoing efforts of providing direct support for C function calls from
640		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
641		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
642		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
643		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
644		handle RCC.
645
646		The last part (the changes to mlriscGen.sml) are still known to be
647		wrong on the x86 and not implemented on all other architectures.  But
648		the infrastructure is in place. I had to change a few functor
649		signatures in the backend to be able to route the CCalls interface
650		from MLRISC there, and I had to specialize the mltree type (on the
651		x86) to include the necessary extensions. (The extensions themselves
652		were already there and redy to go in MLRISC/x86).
653
654		Everything should be very happy as soon as someone helps me with
655		mlriscGen.sml...
656
657		In any case, nothing of this should matter to anyone as long as the
658		new primop is not being used (which is going to be the case unless you
659		find it where I hid it :). The rest of the compiler is completely
660		unaffected.
661
662		----------------------------------------------------------------------
663		Name: Matthias Blume
664		Date: 2001/01/05 00:30:00 JST
665		Tag: blume-20010105-primops
666		Description:
667
668		Added some experimental support for work that I am doing right now.
669		These changes mostly concern added primops, but there is also a new
670		experimental C library in the runtime system (but currently not enabled
671		anywhere except on Linux/X86).
672
673		In the course of adding primops (and playing with them), I discovered that
674		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
675		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
676		To recover, I made type information in INL_PRIM mandatory and changed
677		prim.sml as well as built-in.sml accordingly.  The InLine structure now
678		has complete, correct type information (i.e., no bottom types).
679
680		Since all these changes mean that we need new binfiles, I also bumped the
681		version number to 110.32.1.
682
683		----------------------------------------------------------------------
684		Name: Matthias Blume
685		Date: 2000/12/30 22:10:00 JST
686		Tag: blume-20001230-various
687		Description:
688
689		Added proxy libraries for MLRISC and let MLRISC libraries refer
690		to each other using path anchors.  (See CM manual for explanation.)
691
692		Updated CM documentation.
693
694		Fixed some bugs in CM.
695
696		Implemented "proxy" libraries (= syntactic sugar for CM).
697
698		Added "-quiet" option to makeml and changed runtime system accordingly.
699
700		Added cleanup handler for exportML to reset timers and compiler stats.
701
702		----------------------------------------------------------------------
703		Name: Lal George
704		Date: 2000/12/22 22:22:58 EST 2000
705		Tag: Release_110_32
706		Description:
707
708		Infinite precision used throughout MLRISC.
709		see MLRISC/mltree/machine-int.sig
710
711		----------------------------------------------------------------------
712		Name: Matthias Blume
713		Date: 2000/12/22 23:16:00 JST
714		Tag: blume-20001222-warn
715		Description:
716
717		Corrected wording and formatting of some CM warning message which I
718		broke in my previous patch.
719
720		----------------------------------------------------------------------
721		Name: Matthias Blume
722		Date: 2000/12/22 21:20:00 JST
723		Tag: blume-20001222-anchorenv
724		Description:
725
726		Fixed CM's handling of anchor environments in connection with CMB.make.
727
728		----------------------------------------------------------------------
729		Name: Matthias Blume
730		Date: 2000/12/22 13:15:00 JST
731		Tag: blume-20001222-cleanup
732		Description:
733
734		Removed src/cm/ffi which does not (and did not) belong here.
735
736		----------------------------------------------------------------------
737		Name: Matthias Blume
738		Date: 2000/12/21 23:55:00 JST
739		Tag: blume-20001221-exn
740		Description:
741
742		Probably most important: CM no longer silently swallows all exceptions
743		in the compiler.
744		Plus: some other minor CM changes.  For example, CM now reports some
745		sizes for generated binfiles (code, data, envpickle, lambdapickle).
746
747		----------------------------------------------------------------------
748		Name: Matthias Blume
749		Date: 2000/12/15 00:01:05 JST
750		Tag: blume-20001215-dirtool
751		Description:
752
753		- "dir" tool added.
754		- improvements and cleanup to Tools structure
755		- documentation updates
756
757		----------------------------------------------------------------------
758		Name: Allen Leung
759		Date: Thu Dec 14 03:45:24 EST 2000
760		Description:
761		Tag:  leunga-20001214-int-inf
762		Description:
763
764		In IntInf, added these standard functions, which are missing from our
765		implementation:
766
767		andb : int * int -> int
768		xorb : int * int -> int
769		orb  : int * int -> int
770		notb : int -> int
771		<<   : int * word -> int
772		~>>  : int * word -> int
773
774		Not tested, I hope they are correct.
775
776		----------------------------------------------------------------------
777		Name: Allen Leung
778		Date: Fri Dec  8 19:23:26 EST 2000
779		Description:
780		Tag:  leunga-20001208-nowhere
781		Description:
782
783		Slight improvements to the 'nowhere' tool to handle OR-patterns,
784		to generate better error messages etc.  Plus a brief manual.
785
786		----------------------------------------------------------------------
787		Name: Lal George
788		Date: 2000/12/08 09:54:02 EST 2000
789		Tag: Release_110_31
790		Description:
791
792		- Version 110.31
793		----------------------------------------------------------------------
794		Name: Allen Leung
795		Date: Thu Dec  7 22:01:04 EST 2000
796		Tag:  leunga-20001207-cell-monster-hack
797		Description:
798
799		Major MLRISC internal changes.  Affect all clients.
800		Summary:
801
802		1.  Type CELLS.cell = int is now replaced by a datatype.
803		As a result, the old regmap is now gone.  Almost all interfaces
804		in MLRISC change as a consequence.
805
806		2.  A new brand version of machine description tool (v3.0) that generates
807		modules expecting the new interface.  The old version is removed.
808
809		3.  The RA interface has been further abstracted into two new functors.
810		RISC_RA and X86RA.  These functors have much simpler interfaces.
811		[See also directory MLRISC/demo.]
812
813		4.  Some other new source->source code generation tools are available:
814
815		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
816		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
817		I use this tool to generate the peephole optimizers---with the new
818		cell type changes, peephole rules are becoming difficult to write
819		without conditional pattern matching.
820
821		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
822		cover the entire MLRISC source tree so many things broke.
823
824		6.  CM files have been moved to the subdirectory MLRISC/cm.
825		They are moved because there are a lot of them and they clutter up the
826		root dir.
827
828		7.  More detailed documentation to come...
829
830		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
831		a makeml -rebuild first.  This is because of other other
832		changes that Matthias has made (see below).
833
834
835		----------------------------------------------------------------------
836		Name: Matthias Blume
837		Date: 2000/11/30 23:12:00 JST
838		Tag: blume-20001130-filereorg
839		Description:
840
841		Some manual updates and some file reorganizations in CM.
842
843		----------------------------------------------------------------------
844		Name: Matthias Blume
845		Date: 2000/11/24 17:45:00 JST
846		Tag: blume-20001124-link
847		Description:
848
849		Drastically improved link traversal code for the case that the dynamic
850		value was already loaded at bootstrap time.  As a result, CM and CMB
851		now both load blazingly fast -- even on a very slow machine.  Also,
852		memory consumption has been further reduced by this.
853
854		Warning: The format of the PIDMAP file has changed.  THerefore, to
855		bootstrap you have to do this:
856
857		1. Run CMB.make
858		2. Make a symbolic link for the boot directory:
859		ln -s sml.boot.ARCH-OS xxx
860		3. "Rebuild" the boot directory:
861		./makeml -boot xxx -rebuild sml ; rm xxx
862		4. Boot normally:
863		./makeml
864
865		----------------------------------------------------------------------
866		Name: Matthias Blume
867		Date: 2000/11/21 21:20:00 JST
868		Tag: blume-20001121-tools
869		Description:
870
871		Continued hacking on autoloading problem -- with success this time.
872		Also changed tool-plugin mechanism.  See new CM manual.
873
874		----------------------------------------------------------------------
875		Name: Matthias Blume
876		Date: 2000/11/19 14:30:00 JST
877		Tag:  blume-20001119-autoload
878		Description:
879
880		Some hacking to make autoloading faster.  Success for CMB, no success
881		so far for CM.  There is a reduced structure CM' that autoloads faster.
882		(This is a temporary, non-documented hack to be eliminated again when
883		the general problem is solved.)
884
885		----------------------------------------------------------------------
886		Name: Matthias Blume
887		Date: 2000/11/17 14:10:00 JST
888		Tag: blume-20001117-pickle-lib
889		Description:
890
891		1. Eliminated comp-lib.cm
892		2. Made pickle-lib.cm
893		3. Eliminated all uses of intset.sml (from comp-lib.cm)
894		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
895		equivalent constructs from smlnj-lib.cm (INtHashTable).
896		5. Point 4. also goes for those uses of intmap.* in MLRISC.
897		Duplicated intmap modules thrown out.
898		6. Hunted down all duplicated SCC code and replaced it with
899		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
900		7. Rewrote Feedback module.
901		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
902		should be thrown out and equivalent modules from smlnj-lib.cm
903		should be used (IntRedBlackSet, IntListSet, ...).
904
905		Confirmed that compiler compiles to fixpoint.
906
907		----------------------------------------------------------------------
908		Name: Allen Leung
909		Date: 2000/11/10 18:00:00
910		Tag: leunga-20001110-new-x86-fp
911
912		A new x86 floating point code generator has been added.
913		By default this is turned off.  To turn this on, do:
914
915		CM.autoload "$smlnj/compiler.cm";
916		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
917
918		Changes:
919
920		1.  Changed FTAN to FPTAN so that the assembly output is correct.
921		2.  Changed the extension callback for FTANGENT to generate:
922
923		fptan
924		fstp  %st(0)
925		instead of
926		fptan
927		fstpl ftempmem
928
929		3.  Numerous assembly fixes for x86.
930
931		5.  Cleaned up the machine code output module x86/x86MC.sml and added
932		support for a whole bunch of instructions and addressing modes:
933
934		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
935		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
936		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
937		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
938		fxch %st(n)
939		fld %st(n)
940		fst %st(n)
941		fst mem
942		fstp %st(n)
943		fucom %st(n)
944		fucomp %st(n)
945
946		All these are now generated when the fast fp mode is turned on.
947
948		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
949
950		----------------------------------------------------------------------
951		Name: Matthias Blume
952		Date: 2000/11/09 11:20:00 JST
953		Tag: blume-20001109-scc
954		Description:
955
956		Eliminated some code duplication:
957
958		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
959		(Otherwise the functor is useless.)
960		2. Used GraphSCCFn where SCCUtilFun was used previously.
961		3. Got rid of SCCUtilFun (in comp-lib.cm).
962
963		----------------------------------------------------------------------
964		Name: Lal George
965		Date: 2000/11/06 09:02:21 EST 2000
966		Tag: Release_110_30
967		Description:
968
969		- Version 110.30
970		----------------------------------------------------------------------
971		Name: Matthias Blume
972		Date: 2000/11/04 14:45:00
973		Tag: blume-20001104-mlbuild
974		Description:
975
976		- Made ml-build faster on startup.
977		- Documentation fixes.
978
979		----------------------------------------------------------------------
980		Name: Matthias Blume
981		Date: 2000/11/02 17:00:00 JST
982		Tag: blume-20001102-condcomp
983		Description:
984
985		- Small tweaks to pickler -- new BOOTFILES!
986		- Version bumped to 110.29.2.
987		- Added conditional compilation facility to init.cmi (see comment there).
988		----------------------------------------------------------------------
989		Name: Allen Leung
990		Date: 2000/10/23 19:31:00
991		Tag: leunga-20001023-demo-ra
992
993		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
994		2. Test programs for the graph library updated
995		3. Some new MLRISC demo programs added
996
997		----------------------------------------------------------------------
998		Name: Matthias Blume
999		Date: 2000/08/31 22:15:00 JST
1000		Tag: blume-20001017-errmsg
1001		Description:
1002
1003		More error message grief: Where there used to be no messages, there
1004		now were some that had bogus error regions.  Fixed.
1005
1006		----------------------------------------------------------------------
1007		Name: Matthias Blume
1008		Date: 2000/08/31 17:30:00 JST
1009		Tag: blume-20001017-v110p29p1
1010		Description:
1011
1012		I made a version 110.29.1 with new bootfiles.
1013
1014		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1015		CM documentation changes and a small bugfix in CM's error reporting.
1016
1017		----------------------------------------------------------------------
1018		Name: Lal George
1019		Date: 2000/09/27 14:42:35 EDT
1020		Tag: george-20000927-nodestatus
1021		Description:
1022
1023		Changed the type of the nodestatus, so that:
1024
1025		SPILLED(~1)             is now SPILLED
1026		SPILLED(m) where m>=0   is now MEMREG(m)
1027		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1028
1029		----------------------------------------------------------------------
1030		Name: Matthias Blume
1031		Date: 2000/09/07 14:45:00 JST
1032		Tag: blume-20000907-cmerrmsg
1033		Description:
1034
1035		Small tweak to CM to avoid getting ML syntax error messages twice.
1036
1037		----------------------------------------------------------------------
1038		Name: Matthias Blume
1039		Date: 2000/08/31 18:00:00 JST
1040		Tag: blume-20000831-cvsbootfiles
1041		Description:
1042
1043		New URL for boot files (because the 110.29 files on the BL server do
1044		now work correctly with my updated install scripts for yacc and lex).
1045
1046		----------------------------------------------------------------------
1047		Name: Matthias Blume
1048		Date: 2000/08/08 12:33:00 JST
1049		Tag: blume-20000808-manual
1050		Description:
1051
1052		Tiny update to CM manual.
1053
1054		----------------------------------------------------------------------
1055		Name: Allen Leung
1056		Date: 2000/08/7 19:31:00
1057		Tag: leunga-20000807-a-whole-bunch-of-stuff
1058
1059		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1060		and Mobdy.
1061
1062		1.  x86
1063
1064		a.  Fixes to peephole module by John and Dan.
1065		b.  Assembly fix to SETcc by Allen.
1066		c.  Fix to c-call by John.
1067		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1068		e.  Instruction selection optimization to SETcc as suggested by John.
1069
1070		For example,
1071
1072		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1073
1074		should generate:
1075
1076		MOVL a, x
1077		SUBL b, x
1078		SHRL 31, x
1079
1080		2.  IR stuff
1081
1082		A bunch of new DJ-graph related algorithms added.  These
1083		speed up SSA construction.
1084
1085		3.  SSA + Scheduling
1086
1087		Added code for SSA and scheduling to the repository
1088
1089	----------------------------------------------------------------------	----------------------------------------------------------------------
1090	Name: Lal George	Name: Lal George
1091	Date: 2000/07/27 11:53:14 EDT	Date: 2000/07/27 11:53:14 EDT

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.694   changed lines   Added in v.847

---

root@smlnj-gforge.cs.uchicago.edu	ViewVC Help
Powered by ViewVC 1.0.0