Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Annotation of /sml/trunk/NOTES/HISTORY
ViewVC logotype

Annotation of /sml/trunk/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1011 - (view) (download)
Original Path: sml/trunk/HISTORY

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 : george 1003 Date: yyyy/mm/dd
12 : dbm 570 Tag: <post-commit CVS tag>
13 :     Description:
14 : george 1003 ----------------------------------------------------------------------
15 : blume 1011 Name: Matthias Blume
16 :     Date: 2002/01/10
17 :     Tag: blume-20020110-newffigen
18 :     Description:
19 :    
20 :     1. Updates to portable graph code.
21 :    
22 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
23 :     more scalable now so that even huge interfaces such as the one
24 :     for GTK compile in finite time and space. :-)
25 :     See src/ml-nlffigen/README for details on what's new.
26 :    
27 :     ----------------------------------------------------------------------
28 : george 1003 Name: Lal George
29 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
30 :     Tag: george-20011206-rm-native-copy
31 :     Description:
32 :    
33 :     Removed the native COPY and FCOPY instructions
34 :     from all the architectures and replaced it with the
35 :     explicit COPY instruction from the previous commit.
36 :    
37 :     It is now possible to simplify many of the optimizations
38 :     modules that manipulate copies. This has not been
39 :     done in this change.
40 :    
41 :     ----------------------------------------------------------------------
42 :     Name: Lal George
43 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
44 :     Tag: george-20011206-mlrisc-instruction
45 :     Description:
46 : george 824
47 : george 1003 Changed the representation of instructions from being fully abstract
48 :     to being partially concrete. That is to say:
49 :    
50 :     from
51 :     type instruction
52 :    
53 :     to
54 :     type instr (* machine instruction *)
55 :    
56 :     datatype instruction =
57 :     LIVE of {regs: C.cellset, spilled: C.cellset}
58 :     | KILL of {regs: C.cellset, spilled: C.cellset}
59 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
60 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
61 :     | INSTR of instr
62 :    
63 :     This makes the handling of certain special instructions that appear on
64 :     all architectures easier and uniform.
65 :    
66 :     LIVE and KILL say that a list of registers are live or killed at the
67 :     program point where they appear. No spill code is generated when an
68 :     element of the 'regs' field is spilled, but the register is moved to
69 :     the 'spilled' (which is present, more for debugging than anything else).
70 :    
71 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
72 :     We used to generate:
73 :    
74 :     DEFFREG f1
75 :     f1 := f2 + f3
76 :     trapb
77 :    
78 :     but now generate:
79 :    
80 :     f1 := f2 + f3
81 :     trapb
82 :     LIVE {regs=[f1,f2,f3], spilled=[]}
83 :    
84 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
85 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
86 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
87 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
88 :     this architecture).
89 :    
90 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
91 :     all the architectures. This will result in further simplification of the
92 :     register allocator that must be aware of them for coalescing purposes, and
93 :     will also simplify certain aspects of the machine description that provides
94 :     callbacks related to parallel copies.
95 :    
96 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
97 :     machine instruction set!
98 :    
99 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
100 :     functions for making porting easier -- essentially converting upper case
101 :     to lower case. All machine instructions (of type instr) are in upper case,
102 :     and the lower case form generates an MLRISC instruction. For example on
103 :     the alpha we have:
104 :    
105 :     datatype instr =
106 :     LDA of {r:cell, b:cell, d:operand}
107 :     | ...
108 :    
109 :     val lda : {r:cell, b:cell, d:operand} -> instruction
110 :     ...
111 :    
112 :     where lda is just (INSTR o LDA), etc.
113 :    
114 : leunga 815 ----------------------------------------------------------------------
115 : blume 986 Name: Matthias Blume
116 : blume 993 Date: 2001/11/22 21:40:00 EST
117 :     Tag: Release_110_37
118 :     Description:
119 :    
120 :     Release 110.37. This time for real.
121 :    
122 :     ----------------------------------------------------------------------
123 :     Name: Matthias Blume
124 : blume 988 Date: 2001/11/21 16:35:00 EST
125 :     Tag: blume-20011121-foot-in-mouth
126 :     Description:
127 :    
128 :     Removed the "Release_110_37" tag because of a serious bug.
129 :     This will be re-tagged once the bug is fixed.
130 :    
131 :     ----------------------------------------------------------------------
132 :     Name: Matthias Blume
133 : blume 987 Date: 2001/11/21 16:14:00 EST
134 :     Tag: blume-20011121-forgottenfile
135 :     Description:
136 :    
137 :     Forgot to add a file. (Just a .tex-file -- part of
138 :     the CM manual source.)
139 :    
140 :     ----------------------------------------------------------------------
141 :     Name: Matthias Blume
142 : blume 986 Date: 2001/11/21 16:10:00 EST
143 : blume 988 Tag: blume-20011121-invalid_110_37
144 : blume 986 Description:
145 :    
146 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
147 :     because we found a serious bug in all non-x86 backends.
148 :     - Matthias
149 :    
150 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
151 :     so that code object name strings are directly inserted into code
152 :     objects at code generation time. The only business the runtime system
153 :     has with this is now to read the name strings on occasions.
154 :     (The encoding of the name string has also changed somewhat.)
155 :    
156 :     2. CM now implements a simple "set calculus" for specifying export lists.
157 :     In particular, it is now possible to refer to the export lists of
158 :     other libraries/groups/sources and form unions as well as differences.
159 :     See the latest CM manual for details.
160 :    
161 :     3. An separate notion of "proxy" libraries has again be eliminated from
162 :     CM's model. (Proxy libraries are now simply a special case of using
163 :     the export list calculus.)
164 :    
165 :     4. Some of the existing libraries now take advantage of the new set
166 :     calculus.
167 :     (Notice that not all libraries have been converted because some
168 :     of the existing .cm-files are supposed to be backward compatible
169 :     with 110.0.x.)
170 :    
171 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
172 :     "General.exnMessage"! The former relies on a certain hook to be
173 :     initialized, and that often does not happen in the stand-alone case.)
174 :    
175 :     ----------------------------------------------------------------------
176 : george 984 Name: Lal George
177 :     Date: 2001/11/21 13:56:18 EST
178 :     Tag: george-2001121-pseudo-ops
179 :     Description:
180 :    
181 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
182 :     ought to never be any question of incompatabilities with
183 :     pseudo-op syntax expected by host assemblers.
184 :    
185 :     For now, only modules supporting GAS syntax are implemented
186 :     but more should follow, such as MASM, and vendor assembler
187 :     syntax, e.g. IBM as, Sun as, etc.
188 :    
189 :     ----------------------------------------------------------------------
190 : blume 977 Name: Matthias Blume
191 :     Date: 2001/11/14 11:52:00 EST
192 :     Tag: blume-20011114-srcname
193 :     Description:
194 :    
195 :     1. Routed the name of the current source file to mlriscgen where it
196 :     should be directly emitted into the code object. (This last part
197 :     is yet to be done.)
198 :    
199 :     2. Some cleanup of the pgraph code to make it match the proposal that
200 :     I put out the other day. (The proposal notwithstanding, things are
201 :     still in flux here.)
202 :    
203 :     ----------------------------------------------------------------------
204 : george 976 Name: Lal George
205 :     Date: 2001/11/14 09:44:04 EST
206 :     Tag:
207 :     Description:
208 :    
209 :     Fix for a backpatching bug reported by Allen.
210 :    
211 :     Because the boundary between short and long span-dependent
212 :     instructions is +/- 128, there are an astounding number of
213 :     span-dependent instructions whose size is over estimated.
214 :    
215 :     Allen came up with the idea of letting the size of span
216 :     dependent instructions be non-monotonic, for a maxIter
217 :     number of times, after which the size must be monotonically
218 :     increasing.
219 :    
220 :     This table shows the number of span-dependent instructions
221 :     whose size was over-estimated as a function of maxIter, for the
222 :     file Parse/parse/ml.grm.sml:
223 :    
224 :     maxIter # of instructions:
225 :     10 687
226 :     20 438
227 :     30 198
228 :     40 0
229 :    
230 :     In compiling the compiler, there is no significant difference in
231 :     compilation speed between maxIter=10 and maxIter=40. Actually,
232 :     my measurements showed that maxIter=40 was a tad faster than
233 :     maxIter=10! Also 96% of the files in the compiler reach a fix
234 :     point within 13 iterations, so fixing maxIter at 40, while high,
235 :     is okay.
236 :    
237 :     ----------------------------------------------------------------------
238 : blume 975 Name: Matthias Blume
239 :     Date: 2001/10/31 15:25:00 EST
240 :     Tag: blume-20011031-pgraph
241 :     Description:
242 :    
243 :     CKIT:
244 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
245 :     argument identifiers.
246 :     * Changed the return type of TypeUtil.getFunction accordingly.
247 :     * Type equality ignores the argument names.
248 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
249 :     if there is a mismatch.
250 :    
251 :     installation script:
252 :     * attempts to use "curl" if available (unless "wget" is available as well)
253 :    
254 :     CM:
255 :     * has an experimental implementation of "portable graphs" which I will
256 :     soon propose as an implementation-independent library format
257 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
258 :    
259 :     NLFFI-LIB:
260 :     * some cleanup (all cosmetic)
261 :    
262 :     NLFFIGEN:
263 :     * temporarily disabled the mechanism that suppresses ML output for
264 :     C definitions whose identifiers start with an underscore character
265 :     * generate val bindings for enum constants
266 :     * user can request that only one style (light or heavy) is being used;
267 :     default is to use both (command-line arguments: -heavy and -light)
268 :     * fixed bug in handling of function types involving incomplete pointers
269 :     * generate ML entry points that take record arguments (i.e., using
270 :     named arguments) for C functions that have a prototype with named
271 :     arguments
272 :     (see changes to CKIT)
273 :    
274 :     ----------------------------------------------------------------------
275 : leunga 972 Name: Allen Leung
276 :     Date: 2001/10/27 20:34:00 EDT
277 :     Tag: leunga-20011027-x86-fast-fp-call
278 :     Description:
279 :    
280 :     Fixed the bug described in blume-20010920-slowfp.
281 :    
282 :     The fix involves
283 :     1. generating FCOPYs in FSTP in ia32-svid
284 :     2. marking a CALL with the appropriate annotation
285 :    
286 :     ----------------------------------------------------------------------
287 : blume 961 Name: Matthias Blume
288 :     Date: 2001/10/16 11:32:00 EDT
289 :     Tag: blume-20011016-netbsd
290 :     Description:
291 :    
292 :     Underscore patch from Chris Richards (fixing problem with compiling
293 :     runtime system under recent NetBSD).
294 :    
295 :     ----------------------------------------------------------------------
296 : george 959 Name: Allen Leung
297 :     Date: 2001/10/12 17:18:32 EDT 2001
298 :     Tag: leung-20011012-x86-printflowgraph
299 :     Description:
300 :    
301 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
302 :    
303 :     ----------------------------------------------------------------------
304 : george 952 Name: Lal George
305 : george 958 Date: 2001/10/11 23:51:34 EDT
306 :     Tag: george-20011011-too-many-instrs
307 :     Description:
308 :    
309 :     The representation of a program point never expected to see more
310 :     than 65536 instructions in a basic block!
311 :    
312 :     ----------------------------------------------------------------------
313 :     Name: Lal George
314 : george 952 Date: 2001/10/09 09:41:37 EDT
315 :     Tag: george-20011008-mlrisc-labels
316 :     Description:
317 :    
318 :     Changed the machine description files to support printing of
319 :     local and global labels in assembly code, based on host assembler
320 :     conventions.
321 :    
322 :     ----------------------------------------------------------------------
323 : blume 916 Name: Matthias Blume
324 : blume 939 Date: 2001/09/25 15:25:00 EDT
325 :     Tag: blume-20010925-exninfo
326 :     Description:
327 :    
328 :     I provided a non-hook implementation of exnName (at the toplevel) and
329 :     made the "dummy" implementation of exnMessage (at the toplevel) more
330 :     useful: if nothing gets "hooked in", then at least you are going to
331 :     see the exception name and a message indicating why you don't see more.
332 :    
333 :     [For the time being, programs that need exnMessage and want to use
334 :     ml-build should either use General.exnMessage (strongly recommended) or
335 :     refer to structure General at some other point so that CM sees a
336 :     static dependency.]
337 :    
338 :     [Similar remarks go for "print" and "use": If you want to use their
339 :     functionality in stand-alone programs generated by ml-build, then use
340 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
341 :    
342 :     ----------------------------------------------------------------------
343 :     Name: Matthias Blume
344 : blume 938 Date: 2001/09/20 17:28:00 EDT
345 :     Tag: blume-20010920-slowfp
346 :     Description:
347 :    
348 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
349 :    
350 :     ----------------------------------------------------------------------
351 :     Name: Matthias Blume
352 : blume 937 Date: 2001/09/20 17:20:00 EDT
353 :     Tag: blume-20010920-canonicalpaths
354 :     Description:
355 :    
356 :     0. Updated the BOOT file (something that I forgot to do earlier).
357 :    
358 :     1. Small internal change to CM so that it avoids "/../" in filenames
359 :     as much as possible (but only where it is safe).
360 :    
361 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
362 :     that arguments that contain delimiters are passed through correctly.
363 :     This change also means that all "special" arguments of the form
364 :     @SMLxxx... must come first.
365 :    
366 :     3. Changed install script to put relative anchor names for tool commands
367 :     into pathconfig.
368 :    
369 :     ----------------------------------------------------------------------
370 :     Name: Matthias Blume
371 : george 952 >>>>>>> 1.169
372 : blume 932 Date: 2001/09/18 15:35:00 EDT
373 :     Tag: blume-20010918-readme11036
374 :     Description:
375 :    
376 :     Added README files.
377 :    
378 :     ----------------------------------------------------------------------
379 :     Name: Matthias Blume
380 : blume 929 Date: 2001/09/18 11:45:00 EDT
381 :     Tag: Release_110_36 (retag)
382 :     Description:
383 :    
384 :     Fixed mistake in config/preloads. Retagged as 110.36.
385 :    
386 :     ----------------------------------------------------------------------
387 :     Name: Matthias Blume
388 : blume 928 Date: 2001/09/18 09:40:00 EDT
389 : blume 929 Tag: Release_110_36_orig (tag changed)
390 : blume 928 Description:
391 :    
392 :     New version (110.36). New bootfiles.
393 :    
394 :     ----------------------------------------------------------------------
395 :     Name: Matthias Blume
396 : blume 927 Date: 2001/09/14 16:15:00 EDT
397 :     Tag: blume-20010914-x86fastfp
398 :     Description:
399 :    
400 :     John committed some changes that Allen made, in particular a (hopefully)
401 :     correctly working version of the x86-fp module.
402 :    
403 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
404 :     flag to "true". Everything seems to compile to a fixpoint ok, and
405 :     "mandelbrot" speeds up by about 15%.
406 :    
407 :     ----------------------------------------------------------------------
408 :     Name: Matthias Blume
409 : blume 923 Date: 2001/09/13 11:20:00 EDT
410 :     Tag: blume-20010913-minimal
411 :     Description:
412 :    
413 :     1. Stefan Monnier's patch to fix a miscompilation problem that
414 :     was brought to light by John Reppy's work on Moby.
415 :    
416 :     2. Implemented a minimal "structure Compiler" that contains just
417 :     "version" and "architecture". The minimal version will be
418 :     available when the full version is not. This is for backward-
419 :     compatibility with code that wants to test Compiler.version.
420 :    
421 :     ----------------------------------------------------------------------
422 :     Name: Matthias Blume
423 : blume 920 Date: 2001/08/28 14:03:00 EDT
424 :     Tag: blume-20010828-ml-lex
425 :     Description:
426 :    
427 :     Fix for bug 1581, received from Neophytos Michael.
428 :    
429 :     ----------------------------------------------------------------------
430 :     Name: Matthias Blume
431 : blume 918 Date: 2001/08/27 11:20:00 EDT
432 :     Tag: blume-20010827-readme11035
433 :     Description:
434 :    
435 :     Fleshed out the README file for 110.35.
436 :    
437 :     ----------------------------------------------------------------------
438 :     Name: Matthias Blume
439 : blume 916 Date: 2001/08/24 17:10:00 EDT
440 :     Tag: Release_110_35
441 :     Description:
442 :    
443 :     New version number (110.35). New bootfiles.
444 :    
445 :     ----------------------------------------------------------------------
446 : george 915 Name: Lal George
447 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
448 :     Tag: george-20010824-MLRISC-graphs
449 :     Description:
450 :    
451 :     removed clusters from MLRISC completely and replaced with graphs.
452 :    
453 :     ----------------------------------------------------------------------
454 :     Name: Matthias Blume
455 : blume 905 Date: 2001/08/23 17:50:00 EDT
456 :     Tag: blume-20010823-toplevel
457 :     Description:
458 :    
459 :     - some reorganization of the code that implements various kinds of
460 :     environments in the compiler (static, dynamic, symbolic, combined)
461 :     - re-implemented the EnvRef module so that evalStream works properly
462 :     (if the stream contains references to "use", "CM.make", etc.)
463 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
464 :    
465 :     ----------------------------------------------------------------------
466 :     Name: Matthias Blume
467 : blume 904 Date: 2001/08/20 15:50 EDT
468 :     Tag: blume20010820-slipup
469 :     Description:
470 :    
471 :     I forgot to commit a few files. Here they are...
472 :    
473 :     ----------------------------------------------------------------------
474 :     Name: Matthias Blume
475 : blume 903 Date: 2001/08/20 15:35:00 EDT
476 :     Tag: blume-20010820-debugprof
477 :     Description:
478 :    
479 :     !!!! NEW BOOTFILES !!!!
480 :    
481 :     This is another round of reorganizing the compiler sources. This
482 :     time the main goal was to factor out all the "instrumentation"
483 :     passes (for profiling and backtracing) into their own library.
484 :     The difficulty was to do it in such a way that it does not depend
485 :     on elaborate.cm but only on elabdata.cm.
486 :    
487 :     Therefore there have been further changes to both elaborate.cm and
488 :     elabdata.cm -- more "generic" things have been moved from the former
489 :     to the latter. As a result, I was forced to split the assignment
490 :     of numbers indicating "primtyc"s into two portions: SML-generic and
491 :     SML/NJ-specific. Since it would have been awkward to maintain,
492 :     I bit the bullet and actually _changed_ the mapping between these
493 :     numbers and primtycs. The bottom line of this is that you need
494 :     a new set of bin- and bootfiles.
495 :    
496 :     I have built new bootfiles for all architectures, so doing a fresh
497 :     checkout and config/install.sh should be all you need.
498 :    
499 :     The newly created library's name is
500 :    
501 :     $smlnj/viscomp/debugprof.cm
502 :    
503 :     and its sources live under
504 :    
505 :     src/compiler/DebugProf
506 :    
507 :     ----------------------------------------------------------------------
508 :     Name: Matthias Blume
509 : blume 902 Date: 2001/08/15 17:15:00 EDT
510 :     Tag: blume-20010815-compreorg
511 :     Description:
512 :    
513 :     This is a first cut at reorganizing the CM libraries that make up the
514 :     core of the compiler. The idea is to separate out pieces that could
515 :     be used independently by tools, e.g., the parser, the typechecker, etc.
516 :    
517 :     The current status is a step in this direction, but it is not quite
518 :     satisfactory yet. Expect more changes in the future.
519 :    
520 :     Here is the current (new) organization...
521 :    
522 :     What used to be $smlnj/viscomp/core.cm is now divided into
523 :     six CM libraries:
524 :    
525 :     $smlnj/viscomp/basics.cm
526 :     /parser.cm
527 :     /elabdata.cm
528 :     /elaborate.cm
529 :     /execute.cm
530 :     /core.cm
531 :    
532 :     The CM files for these libraries live under src/system/smlnj/viscomp.
533 :     All these libraries are proxy libraries that contain precisely
534 :     one CM library component. Here are the locations of the components
535 :     (all within the src/compiler tree):
536 :    
537 :     Basics/basics.cm
538 :     Parse/parser.cm
539 :     ElabData/elabdata.cm
540 :     Elaborator/elaborate.cm
541 :     Execution/execute.cm
542 :     core.cm
543 :    
544 :     [This organization is the same that has been used already
545 :     for a while for the architecture-specific parts of the visible
546 :     compiler and for the old version of core.cm.]
547 :    
548 :     As you will notice, many source files have been moved from their
549 :     respective original locations to a new home in one of the above
550 :     subtrees.
551 :    
552 :     The division of labor between the new libraries is the following:
553 :    
554 :     basics.cm:
555 :     - Simple, basic definitions that pertain to many (or all) of
556 :     the other libraries.
557 :     parser.cm:
558 :     - The SML parser, producing output of type Ast.dec.
559 :     - The type family for Ast is also defined and exported here.
560 :     elabdata.cm:
561 :     - The datatypes that describe input and output of the elaborator.
562 :     This includes types, absyn, and static environments.
563 :     elaborator.cm:
564 :     - The SML/NJ type checker and elaborator.
565 :     This maps an Ast.dec (with a given static environment) to
566 :     an Absyn.dec (with a new static environment).
567 :     - This libraries implements certain modules that used to be
568 :     structures as functors (to remove dependencies on FLINT).
569 :     execute.cm:
570 :     - Everything having to do with executing binary code objects.
571 :     - Dynamic environments.
572 :     core.cm:
573 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
574 :     - Top-level modules.
575 :     - FLINT (this should eventually become its own library)
576 :    
577 :     Notes:
578 :    
579 :     I am not 100% happy with the way I separated the elaborator (and its
580 :     data structures) from FLINT. Two instances of the same problem:
581 :    
582 :     1. Data structures contain certain fields that carry FLINT-specific
583 :     information. I hacked around this using exn and the property list
584 :     module from smlnj-lib. But the fact that there are middle-end
585 :     specific fields around at all is a bit annoying.
586 :    
587 :     2. The elaborator calculates certain FLINT-related information. I tried
588 :     to make this as abstract as I could using functorization, but, again,
589 :     the fact that the elaborator has to perform calculations on behalf
590 :     of the middle-end at all is not nice.
591 :    
592 :     3. Having to used exn and property lists is unfortunate because it
593 :     weakens type checking. The other alternative (parameterizing
594 :     nearly *everything*) is not appealing, though.
595 :    
596 :     I removed the "rebinding =" warning hack because due to the new organization
597 :     it was awkward to maintain it. As a result, the compiler now issues some of
598 :     these warnings when compiling init.cmi during bootstrap compilation. On
599 :     the plus side, you also get a warning when you do, for example:
600 :     val op = = Int32.+
601 :     which was not the case up to now.
602 :    
603 :     I placed "assign" and "deref" into the _Core structure so that the
604 :     code that deals with the "lazy" keyword can find them there. This
605 :     removes the need for having access to the primitive environment
606 :     during elaboration.
607 :    
608 :     ----------------------------------------------------------------------
609 :     Name: Matthias Blume
610 : blume 898 Date: 2001/08/13
611 :     Tag: blume-20010813-closures
612 :     Description:
613 :    
614 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
615 :     performance of certain loops by avoiding needless closure allocation.
616 :    
617 :     ----------------------------------------------------------------------
618 : george 890 Name: Lal George
619 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
620 :     Tag: george-20010731-x86-fmalloc
621 :     Description: Fixed bug in x86 calls
622 :    
623 :     There was a bug where call instructions would mysteriously
624 :     vanish. The call instruction had to be one that returned
625 :     a floating point value.
626 :    
627 :     ----------------------------------------------------------------------
628 :     Name: Lal George
629 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
630 :     Tag: george-20010719-simple-cells
631 :     Description:
632 :    
633 :     I have dramatically simplified the interface for CELLS in MLRISC.
634 :    
635 :     In summary, the cells interface is broken up into three parts:
636 :    
637 :     1. CellsBasis : CELLS_BASIS
638 :    
639 :     CellsBasis is a top level structure and common for all
640 :     architectures. it contains the definitions of basic datatypes
641 :     and utility functions over these types.
642 :    
643 :     2. functor Cells() : CELLS
644 :    
645 :     Cells generates an interface for CELLS that incorporates the
646 :     specific resources on the target architecture, such as the
647 :     presence of special register classes, their number and size,
648 :     and various useful substructures.
649 :    
650 :     3. <ARCH>CELLS
651 :    
652 :     e.g. SparcCells: SPARCCELLS
653 :    
654 :     <ARCH>CELLS usually contains additional bindings for special
655 :     registers on the architecture, such as:
656 :    
657 :     val r0 : cell (* register zero *)
658 :     val y : cell (* Y register *)
659 :     val psr : cell (* processor status register *)
660 :     ...
661 :    
662 :     The structure returned by applying the Cells functor is opened
663 :     in this interface.
664 :    
665 :     The main implication of all this is that the datatypes for cells is
666 :     split between CellsBasis and CELLS -- a fairly simple change for user
667 :     code.
668 :    
669 :     In the old scheme the CELLS interface had a definitional binding of
670 :     the form:
671 :    
672 :     signature CELLS = sig
673 :    
674 :     structure CellsBasis = CellsBasis
675 :    
676 :     ...
677 :    
678 :     end
679 :    
680 :     With all the sharing constraints that goes on in MLRISC, this old
681 :     design quickly leads to errors such as:
682 :    
683 :     "structure definition spec inside of sharing ... "
684 :    
685 :    
686 :     and appears to require an unacceptable amount of sharing and where
687 :     constraint hackery.
688 :    
689 :     I think this error message (the interaction of definitional specs and
690 :     sharing) requires more explanation on our web page.
691 :    
692 :     ----------------------------------------------------------------------
693 : blume 826 Name: Matthias Blume
694 : blume 879 Date: 2001/07/19 15:00:00 EDT
695 :     Tag: blume-20010719-libreorg
696 :     Description:
697 :    
698 :     This update puts together a fairly extensive but straightforward change
699 :     to the way the libraries that implement the interactive system are
700 :     organized:
701 :    
702 :     The biggest change is the elimination of structure Compiler. As a
703 :     replacement for this structure, there is now a CM library
704 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
705 :     that exports all the substructures of the original structure Compiler
706 :     directly. So instead of saying Compiler.Foo.bar one now simply
707 :     says Foo.bar. (The CM libraries actually export a collection of
708 :     structures that is richer than the collection of substructures of
709 :     structure Compiler.)
710 :    
711 :     To make the transition smooth, there is a separate library called
712 :     $smlnj/compiler/compiler.cm which puts together and exports the
713 :     original structure Compiler (or at least something very close to it).
714 :    
715 :     There are five members of the original structure Compiler
716 :     that are not exported directly but which instead became members
717 :     of a new structure Backend (described by signature BACKEND). These are:
718 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
719 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
720 :     architecture (: string).
721 :    
722 :     Structure Compiler.Version has become structure CompilerVersion.
723 :    
724 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
725 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
726 :     or x86, respectively.
727 :     Each of these exports the same frontend structures that
728 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
729 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
730 :     Hppa, PPC, Sparc, or X86, respectively.
731 :    
732 :     Library $smlnj/compiler/all.cm exports the union of the exports of
733 :     $smlnj/compiler/<arch>.cm
734 :    
735 :     There are no structures <Arch>Compiler anymore, use
736 :     $smlnj/compiler/<arch>.cm instead.
737 :    
738 :     Library host-compiler-0.cm is gone. Instead, the internal library
739 :     that instantiates CM is now called cm0.cm. Selection of the host
740 :     compiler (backend) is no longer done here but. (Responsibility for it
741 :     now lies with $smlnj/compiler/current.cm. This seems to be more
742 :     logical.)
743 :    
744 :     Many individual files have been moved or renamed. Some files have
745 :     been split into multiple files, and some "dead" files have been deleted.
746 :    
747 :     Aside from these changes to library organization, there are also changes
748 :     to the way the code itself is organized:
749 :    
750 :     Structure Binfile has been re-implemented in such a way that it no
751 :     longer needs any knowledge of the compiler. It exclusively deals
752 :     with the details of binfile layout. It no longer invokes the
753 :     compiler (for the purpose of creating new prospective binfile
754 :     content), and it no longer has any knowledge of how to interpret
755 :     pickles.
756 :    
757 :     Structure Compile (: COMPILE) has been stripped down to the bare
758 :     essentials of compilation. It no longer deals with linking/execution.
759 :     The interface has been cleaned up considerably.
760 :    
761 :     Utility routines for dealing with linking and execution have been
762 :     moved into their own substructures.
763 :    
764 :     (The ultimate goal of these changes is to provide a light-weight
765 :     binfile loader/linker (at least for, e.g., stable libraries) that
766 :     does not require CM or the compiler to be present.)
767 :    
768 :     CM documentation has been updated to reflect the changes to library
769 :     organization.
770 :    
771 :     ----------------------------------------------------------------------
772 :     Name: Matthias Blume
773 : blume 875 Date: 2001/07/10 17:30:00 EDT
774 :     Tag: Release_110_34
775 :     Description:
776 :    
777 : blume 876 Minor tweak to 110.34 (re-tagged):
778 :    
779 :     - README.html file added to CVS repository
780 :     - runtime compiles properly under FreeBSD 3.X and 4.X
781 :    
782 :     ----------------------------------------------------------------------
783 :     Name: Matthias Blume
784 :     Date: 2001/07/10 17:30:00 EDT
785 :     Tag: Release_110_34
786 :     Description:
787 :    
788 : blume 875 New version number (110.34). New bootfiles.
789 :    
790 :     ----------------------------------------------------------------------
791 :     Name: Matthias Blume
792 : blume 874 Date: 2001/07/09 16:00:00 EDT
793 :     Tag: blume-20010709-more-varargs
794 :     Description:
795 :    
796 :     I changed the handling of varargs in ml-nlffigen again:
797 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
798 :    
799 :     The immediate effect is that you can actually call a varargs function
800 :     from ML -- but you can't actually supply any arguments beyond the ones
801 :     specified explicitly. (For example, you can call printf with its format
802 :     string, but you cannot pass additional arguments.)
803 :    
804 :     This behavior is only marginally more useful than the one before, but
805 :     it has the advantage that a function or, more importantly, a function
806 :     type never gets dropped on the floor, thus avoiding follow-up problems with
807 :     other types that refer to the offending one.
808 :    
809 :     ----------------------------------------------------------------------
810 :     Name: Matthias Blume
811 : blume 873 Date: 2001/07/09 11:25:00 EDT
812 :     Tag: blume-20010709-varargs
813 :     Description:
814 :    
815 :     1. ckit-lib.cm now exports structure Error
816 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
817 :     with a warning accompanied by a source location. Moreover, it
818 :     merely skips the offending function or type and proceeds with the
819 :     rest of its work.u As a result, one can safely feed C code containing
820 :     "..." to ml-nlffigen.
821 :     3. There are some internal improvements to CM, providing slightly
822 :     more general string substitutions in the tools subsystem.
823 :    
824 :     ----------------------------------------------------------------------
825 :     Name: Matthias Blume
826 : blume 854 Date: 2001/06/27 15:10:00 EDT
827 :     Tag: blume-20010627-concur
828 :     Description:
829 :    
830 :     Fixed a small bug in CM's handling of parallel compilation.
831 :     (You could observe the bug by Control-C-interrupting an ordinary
832 :     CMB.make or CM.stabilize and then attaching some compile servers.
833 :     The result was that all of a sudden the previously interrupted
834 :     compilation would continue on its own. This was because of
835 :     an over-optimization: CM did not bother to clean out certain queues
836 :     when no servers were attached "anyway", resulting in the contents
837 :     of these queues to grab control when new servers did get attached.)
838 :    
839 :     There is also another minor update to the CM manual.
840 :    
841 :     ----------------------------------------------------------------------
842 :     Name: Matthias Blume
843 : blume 853 Date: 2001/06/26 16:15:00 EDT
844 :     Tag: blume-20010626-cmdoc
845 :     Description:
846 :    
847 :     Minor typo fixed in CM manual (syntax diagram for libraries).
848 :    
849 :     ----------------------------------------------------------------------
850 :     Name: Matthias Blume
851 : blume 852 Date: 2001/06/25 22:55:00 EDT
852 :     Tag: blume-20010625-x86pc
853 :     Description:
854 :    
855 :     Fixed a nasty bug in the X86 assembly code that caused signal
856 :     handlers to fail (crash) randomly.
857 :    
858 :     ----------------------------------------------------------------------
859 :     Name: Matthias Blume
860 : blume 846 Date: 2001/06/25 12:05:00 EDT
861 :     Tag: blume-20010625-nlffigen
862 :     Description:
863 :    
864 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
865 :     Nick Carter <nbc@andrew.cmu.edu>.
866 :    
867 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
868 :     2. Default names for generated files are now derived from the name of
869 :     the C file *without its directory*. In particular, this causes generated
870 :     files to be placed locally even if the C file is in some system directory.
871 :     3. Default names for generated signatures and structures are also derived
872 :     from the C file name without its directory. This avoids silly things
873 :     like "structure GL/GL".
874 :     (Other silly names are still possible because ml-nlffigen does not do
875 : blume 847 a thorough check of whether generated names are legal ML identifiers.
876 :     When in doubt, use command line arguments to force particular names.)
877 : blume 846
878 :     ----------------------------------------------------------------------
879 :     Name: Matthias Blume
880 : blume 845 Date: 2001/06/21 12:25:00 EDT
881 :     Tag: blume-20010621-eXene
882 :     Description:
883 :    
884 :     eXene now compiles and (sort of) works again.
885 :    
886 :     The library name (for version > 110.33) is $/eXene.cm.
887 :    
888 :     I also added an new example in src/eXene/examples/nbody. See the
889 :     README file there for details.
890 :    
891 :     ----------------------------------------------------------------------
892 :     Name: Matthias Blume
893 : blume 844 Date: 2001/06/20 16:40:00 EDT
894 :     Tag: blume-20010620-cml
895 :     Description:
896 :    
897 :     CML now compiles and works again.
898 :    
899 :     Libraries (for version > 110.33):
900 :    
901 :     $cml/cml.cm Main CML library.
902 :     $cml/basis.cm CML's version of $/basis.cm.
903 :     $cml/cml-internal.cm Internal helper library.
904 :     $cml/core-cml.cm Internal helper library.
905 :     $cml-lib/trace-cml.cm Tracing facility.
906 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
907 :    
908 :     The installer (config/install.sh) has been taught how to properly
909 :     install this stuff.
910 :    
911 :     ----------------------------------------------------------------------
912 :     Name: Matthias Blume
913 : blume 843 Date: 2001/06/19 17:55:00 EDT
914 :     Tag: blume-20010619-instantiate
915 :     Description:
916 :    
917 :     This un-breaks the fix for bug 1432.
918 :     (The bug was originally fixed in 110.9 but I broke it again some
919 :     time after that.)
920 :    
921 :     ----------------------------------------------------------------------
922 :     Name: Matthias Blume
923 : blume 842 Date: 2001/06/19 17:25:00 EDT
924 :     Tag: blume-20010619-signals
925 :     Description:
926 :    
927 :     This should (hopefully) fix the long-standing signal handling bug.
928 :     (The runtime system was constructing a continuation record with an
929 :     incorrect descriptor which would cause the GC to drop data on the floor...)
930 :    
931 :     ----------------------------------------------------------------------
932 :     Name: Matthias Blume
933 : blume 840 Date: 2001/06/15 15:05:00 EDT
934 : blume 841 Tag: blume-20010615-moresparc
935 :     Description:
936 :    
937 :     Here is a short late-hour update related to Sparc c-calls:
938 :    
939 :     -- made handling of double-word arguments a bit smarter
940 :    
941 :     -- instruction selection phase tries to collapse certain clumsily
942 :     constructed ML-Trees; typical example:
943 :    
944 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
945 :    
946 :     This currently has no further impact on SML/NJ since mlriscGen does
947 :     not seem to generate such patterns in the first place, and c-calls
948 :     (which did generate them in the beginning) has meanwhile been fixed
949 :     so as to avoid them as well.
950 :    
951 :     ----------------------------------------------------------------------
952 :     Name: Matthias Blume
953 :     Date: 2001/06/15 15:05:00 EDT
954 : blume 840 Tag: blume-20010615-sparc
955 :     Description:
956 :    
957 :     The purpose of this update is to provide an implementation of NLFFI
958 :     on Sparc machines.
959 :    
960 :     Here are the changes in detail:
961 :    
962 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
963 :     the Sparc implementation of the c-calls API.
964 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
965 :     for accessing the ML frame. Thus, we have a real frame pointer and
966 :     can freely modify %sp without need for an omit-frame-ptr phase.
967 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
968 :     this case.
969 :     * ml-nlffigen has been taught to produce code for different architectures
970 :     and calling conventions.
971 :     * In a way similar to what was done in the x86 case, the Sparc
972 :     backend uses its own specific extension to mltree. (For example,
973 :     it needs to be able to generate UNIMP instructions which are part
974 :     of the calling convention.)
975 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
976 :     to make it easier to plug in new machine- and os-dependent parts).
977 :    
978 :     There are some other fairly unrelated bug fixes and cleanups as well:
979 :    
980 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
981 :     that they properly share their libraries with existing SML/NJ libraries.
982 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
983 :     follow-up error messages.
984 :     * Updates to CM/CMB documentation.
985 :    
986 :     TODO items:
987 :    
988 :     * MLRISC should use a different register as its asmTemp on the Sparc.
989 :     (The current %o2 is a really bad choice because it is part of the
990 :     calling conventions, so things might interfere in unexpected ways.)
991 :    
992 :     ----------------------------------------------------------------------
993 :     Name: Matthias Blume
994 : blume 839 Date: 2001/06/07
995 :     Tag: blume-20010607-calls
996 :     Description:
997 :    
998 :     A number of internal changes related to C calls and calling conventions:
999 :    
1000 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
1001 :     number of bytes popped implicitly (by the callee). In most cases
1002 :     this field is 0 but on x86/win32 it is some non-zero value. This
1003 :     is information provided for the benefit of the "omit-frameptr" pass.
1004 :     2. The CALL instruction on the x86 carries a similar "pops" field.
1005 :     The instruction selection phase copies its value from the ML-Tree
1006 :     CALL statement.
1007 :     3. On all other architectures, the instruction selection phase checks
1008 :     whether "pops=0" and complains if not.
1009 :     4. The c-calls implementation for x86 now accepts two calling conventions:
1010 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
1011 :     up after the call and pops is set to 0. For "stdcall", the caller
1012 :     does nothing, leaving the cleanup to the callee; pops is set to
1013 :     the number of bytes that were pushed onto the stack.
1014 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1015 :     distinguish between "ccall" and "stdcall".
1016 :     6. The UNIMP instruction has been added to the supported Sparc instruction
1017 :     set. (This is needed for implementing the official C calling convention
1018 :     on this architecture.)
1019 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1020 :     work with the latest CM.
1021 :    
1022 :     ----------------------------------------------------------------------
1023 :     Name: Matthias Blume
1024 : blume 838 Date: 2001/06/05 15:10:00 EDT
1025 :     Tag: blume-20010605-cm-index
1026 :     Description:
1027 :    
1028 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1029 :    
1030 :     1. CM can now generate "index files". These are human-readable files
1031 :     that list on a per-.cm-file basis each toplevel symbol defined or
1032 :     imported. The location of the index file for
1033 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1034 :     To enable index-file generation, set CM.Control.generate_index to true
1035 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
1036 :    
1037 :     The CM manual has been updated accordingly.
1038 :    
1039 :     2. I made some slight modifications to the c-calls API in MLRISC.
1040 :    
1041 :     a) There is now a callback to support saving/restoring of
1042 :     dedicated but caller-save registers around the actual call
1043 :     instruction.
1044 :     b) One can optionally specify a comment-annotation for the
1045 :     call instruction.
1046 :    
1047 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1048 :     (For example, the comment annotation shows the C prototype of
1049 :     the function being called.)
1050 :    
1051 :     ----------------------------------------------------------------------
1052 :     Name: Matthias Blume
1053 : blume 837 Date: 2001/06/01 13:30:00 EDT
1054 :     Tag: blume-20010601-nlffi-cleanup
1055 :     Description:
1056 :    
1057 :     This is mostly a cleanup of MLFFI stuff:
1058 :    
1059 :     - some signature files have been put into a more exposed place
1060 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
1061 :     - ml-nlffigen changed accordingly
1062 :     - tutorial updated
1063 :    
1064 :     Other changes:
1065 :    
1066 :     - author's affiliation in CM manual(s) updated
1067 :     - some more recognized keywords added to Allen's sml.sty
1068 :    
1069 :     ----------------------------------------------------------------------
1070 :     Name: Matthias Blume
1071 : blume 836 Date: 2001/05/25 15:30:00 EDT
1072 :     Tag: blume-20010525-iptr
1073 :     Description:
1074 :    
1075 :     - put the official 110.33-README (as it appears on the ftp server) under
1076 :     CVS
1077 :     - fixed a small bug related to incomplete pointer types in
1078 :     ml-nlffigen
1079 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1080 :     (it does not need the 'f type parameter)
1081 :    
1082 :     ----------------------------------------------------------------------
1083 :     Name: Matthias Blume
1084 : blume 834 Date: 2001/05/23 14:30:00 EDT
1085 :     Tag: Release_110_33
1086 :     Description:
1087 :    
1088 :     New version number (110.33). New bootfiles.
1089 :    
1090 :     ----------------------------------------------------------------------
1091 :     Name: Matthias Blume
1092 : blume 833 Date: 2001/05/22 18:06:00 EDT
1093 :     Tag: blume-20010522-targets
1094 :     Description:
1095 :    
1096 :     Made install.sh use file config/targets.customized if it exists, falling
1097 :     back to config/targets if it doesn't. This way one can have a customized
1098 :     version of the targets file without touching the "real thing", thus
1099 :     eliminating the constant fear of accidentally checking something bogus
1100 :     back into the CVS repository... (File config/targets.customized must
1101 :     not be added to the repository!)
1102 :    
1103 :     ----------------------------------------------------------------------
1104 :     Name: Matthias Blume
1105 : blume 832 Date: 2001/05/22 16:30:00 EDT
1106 :     Tag: blume-20010522-minitut
1107 :     Description:
1108 :    
1109 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1110 :     struct returns.
1111 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
1112 :     incomplete, preliminary documentation for NLFFI.
1113 :    
1114 :     ----------------------------------------------------------------------
1115 :     Name: Matthias Blume
1116 : blume 829 Date: 2001/05/14 11:30:00 EDT
1117 :     Tag: blume-20010514-script
1118 :     Description:
1119 :    
1120 :     Some bugs in install script fixed.
1121 :    
1122 :     In addition to that I also made a slight change to the NLFFI API:
1123 :     Functors generated by ml-nlffigen now take the dynamic library as a
1124 :     straight functor argument, not as a suspended one. (The original
1125 :     functor code used to force the suspension right away anyway, so there
1126 :     was nothing gained by this complication of the interface.)
1127 :    
1128 :     ----------------------------------------------------------------------
1129 :     Name: Matthias Blume
1130 : blume 828 Date: 2001/05/11 14:35:00 EDT
1131 :     Tag: blume-20010511-ml-nlffi
1132 :     Description:
1133 :    
1134 :     I finally took the plunge and added my new FFI code to the main
1135 :     repository. For x86-linux it is now ready for prime-time.
1136 :    
1137 :     There are two new subdirectories of "src":
1138 :    
1139 :     - ml-nlffi-lib:
1140 :     The utility library for programs using the FFI interface.
1141 :     Here is the implementation of $/c.cm and its associated low-level
1142 :     partners $/c-int.cm and $/memory.cm.
1143 :     - ml-nlffigen:
1144 :     A stand-alone program for generating ML glue code from C source
1145 :     code.
1146 :    
1147 :     Building ml-nlffigen requires $/ckit-lib.cm.
1148 :    
1149 :     The config/install.sh script has been updates to do the Right Thing
1150 :     (hopefully).
1151 :    
1152 :     Notice that the source tree for the C-Kit will not be put under "src"
1153 :     but directly under the installation root directory. (This is the
1154 :     structure that currently exists on the CVS server when you check out
1155 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
1156 :    
1157 :     Bugs: No documentation yet.
1158 :    
1159 :     ----------------------------------------------------------------------
1160 :     Name: Matthias Blume
1161 : blume 826 Date: 2001/05/09 16:35:00 EDT
1162 :     Tag: blume-20010509-cpscontract
1163 :     Description:
1164 :    
1165 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
1166 :     wrapper/unwrapper elimination did not decrement usage counts and some
1167 :     dead variables got overlooked by the dead-up logic.)
1168 :    
1169 :     ----------------------------------------------------------------------
1170 : george 824 Name: Lal George
1171 :     Date: 2001/05/08 17:26:09 EDT
1172 :     Tag: george-20010508-omit-frameptr
1173 :     Description:
1174 :    
1175 :     Changes to implement the omit-frame-pointer optimization to support
1176 :     raw C calls. For now, there is only support on the Intel x86, but
1177 :     other architectures will follow as more experience is gained with this.
1178 :    
1179 :    
1180 :     ----------------------------------------------------------------------
1181 : blume 818 Name: Matthias Blume
1182 : blume 819 Date: 2001/05/07 14:40:00 EDT
1183 :     Tag: blume-20010507-proxies
1184 :     Description:
1185 :    
1186 :     I made into "proxy libraries" all libraries that qualify for such a
1187 :     change. (A qualifying library is a library that has another library or
1188 :     groups as its sole member and repeats that member's export list
1189 :     verbatim. A proxy library avoids this repetition by omitting its export
1190 :     list, effectively inheriting the list that its (only) member exports.
1191 :     See the CM manual for more explanation.)
1192 :     The main effect is that explicit export lists for these libraries
1193 :     do not have to be kepts in sync, making maintenance a bit easier.
1194 :    
1195 :     I also added copyright notices to many .cm-files.
1196 :    
1197 :     Last but not least, I made a new set of bootfiles.
1198 :    
1199 :     ----------------------------------------------------------------------
1200 :     Name: Matthias Blume
1201 : blume 818 Date: 2001/05/04 17:00:00 EDT
1202 :     Tag: blume-20010504-cm-lsplit
1203 :     Description:
1204 :    
1205 :     0. John merged pending changes to $/smlnj-lib.cm
1206 :    
1207 :     1. Allen's previous change accidentally backed out of one of Lal's
1208 :     earlier changes. I undid this mistake (re-introducing Lal's change).
1209 :    
1210 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1211 :     within the compiler where applicable. There is some code simplification
1212 :     because of that.
1213 :    
1214 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
1215 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
1216 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
1217 :     it can now also be controlled on a per-source basis: CM has been taught
1218 :     a new tool parameter applicable to ML source files.
1219 :    
1220 :     - To turn lambda-splitting off completely:
1221 :     local open Compiler.Control.LambdaSplitting in
1222 :     val _ = set Off
1223 :     end
1224 :     - To make "no lambda-splitting" the global default (but allow per-source
1225 :     overriding); this is the initial setting:
1226 :     local open Compiler.Control.LambdaSplitting in
1227 :     val _ = set (Default NONE)
1228 :     end
1229 :     - To make "lambda-splitting with aggressiveness a" the global default
1230 :     (and allow per-source overriding):
1231 :     local open Compiler.Control.LambdaSplitting in
1232 :     val _ = set (Default (SOME a))
1233 :     end
1234 :    
1235 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
1236 :     write (in the respective .cm-file):
1237 :     a.sml (lambdasplitting:off)
1238 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
1239 :     a.sml (lambdasplitting:on)
1240 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1241 :     <a> is a decimal non-negative integer):
1242 :     a.sml (lambdasplitting:<a>)
1243 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
1244 :     a.sml (lambdasplitting:infinity)
1245 :     - To use the global default for a.sml:
1246 :     a.sml (lambdasplitting:default)
1247 :     or simply
1248 :     a.sml
1249 :    
1250 :     ----------------------------------------------------------------------
1251 : leunga 815 Name: Allen Leung
1252 :     Date: 2001/05/04 01:57:00 EDT
1253 :     Tag: leunga-20010504-sync
1254 :     Description:
1255 : leunga 788
1256 : leunga 815 MLRISC features.
1257 :    
1258 :     1. Fix to CMPXCHG instructions.
1259 :     2. Changed RA interface to allow annotations in callbacks.
1260 :     3. Added a new method to the stream interface to allow annotations updates.
1261 :    
1262 : george 749 ----------------------------------------------------------------------
1263 : blume 812 Name: Matthias Blume
1264 : blume 814 Date: 2001/05/01 11:45:00 EDT
1265 :     Tag: blume-20010501-pcedittmp
1266 :     Description:
1267 :    
1268 :     Changed install.sh to use the current working directory instead of
1269 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
1270 :     of /usr/tmp caused trouble with MacOS X because of file premission
1271 :     problems.
1272 :    
1273 :     ----------------------------------------------------------------------
1274 :     Name: Matthias Blume
1275 : blume 812 Date: 2001/04/20 11:10:00 EDT
1276 :     Tag: blume-20010420-inMLflag
1277 :     Description:
1278 :    
1279 :     - added vp_limitPtrMask to vproc-state.h
1280 :     (for use by the raw-C-calls mechanism to implement proper interrupt
1281 :     handling)
1282 :     - made the ML compiler aware of various data-structure offsets so it
1283 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
1284 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
1285 :     raw C-calls
1286 :    
1287 :     ----------------------------------------------------------------------
1288 : george 810 Name: Lal George
1289 :     Date: 2001/04/20 09:15:28 EDT
1290 :     Tag: george-20010420-macosX
1291 :     Description:
1292 :    
1293 :     - Changes to port to Mac OS X; Darwin.
1294 :    
1295 :     - In the process I found that sqrt was broken on the PPC, because the
1296 :     fsqrt instruction is not implemented.
1297 :    
1298 :     ----------------------------------------------------------------------
1299 : blume 799 Name: Matthias Blume
1300 : blume 808 Date: 2001/04/18 12:45:00 EDT
1301 :     Tag: blume-20010418-ccalls
1302 :     Description:
1303 :    
1304 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
1305 :     (this bug prevented structure arguments containing pointers from being
1306 :     passed correctly)
1307 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
1308 :     structure arguments are represented as a pointer to the beginning
1309 :     of the structure (instead of having a series of synthesized arguments,
1310 :     one for each structure member)
1311 :    
1312 :     - made makeml script's verbosity level configurable via environment
1313 :     variable (MAKEML_VERBOSITY)
1314 :    
1315 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1316 : blume 808 in rawmem-x86.sml; we are now using the real thing
1317 :    
1318 :     ----------------------------------------------------------------------
1319 :     Name: Matthias Blume
1320 : blume 806 Date: 2001/03/22 16:25:00 EST
1321 :     Tag: blume-20010322-bootfiles
1322 :     Description:
1323 :    
1324 :     Created a new set of bootfiles (for your automatic installation convenience).
1325 :    
1326 :     ----------------------------------------------------------------------
1327 :     Name: Matthias Blume
1328 : blume 805 Date: 2001/03/22 15:10:00 EST
1329 :     Tag: blume-20010322-rawmem-parcm
1330 :     Description:
1331 :    
1332 :     1. All "raw memory access" primitives for the new FFI are implemented now
1333 :     (at least on the x86).
1334 :     2. Some further cleanup of CM's parallel make mechanism.
1335 :    
1336 :     ----------------------------------------------------------------------
1337 :     Name: Matthias Blume
1338 : blume 801 Date: 2001/03/19 17:53:00 EST
1339 :     Tag: blume-20010319-parallel
1340 :     Description:
1341 :    
1342 :     Parallel make (using compile servers) now works again.
1343 :    
1344 :     To this end, CM.stabilize and CMB.make have been modified to work in
1345 :     two passes when compile servers are attached:
1346 :     1. Compile everything, do not perform stabilization; this pass
1347 :     uses compile servers
1348 :     2. Stabilize everything; this pass does not use compile servers
1349 :     If there are no compile servers, the two passes are combined into one
1350 :     (as before). Splitting the passes increases the inherent parallelism
1351 :     in the dependency graph because the entire graph including all
1352 :     libraries is available at the same time. This, in turn, improves
1353 :     server utilization. The downside is that the master process will
1354 :     have to do some extra work after compilation is done (because for
1355 :     technical reasons it must re-read all the binfiles during stabilization).
1356 :    
1357 :     ----------------------------------------------------------------------
1358 :     Name: Matthias Blume
1359 : blume 800 Date: 2001/03/16 12:22:00 EST
1360 :     Tag: blume-20010316-bootfiles
1361 :     Description:
1362 :    
1363 :     Created a new set of bootfiles (for your automatic installation convenience).
1364 :    
1365 :     ----------------------------------------------------------------------
1366 :     Name: Matthias Blume
1367 : blume 799 Date: 2001/03/16 11:00:00 EST
1368 :     Tag: blume-20010316-MLTREE-fixup
1369 :     Description:
1370 :    
1371 :     This is a minor fixup for an (untagged) earlier commit by Allen.
1372 :     (A file was missing).
1373 :    
1374 :     ----------------------------------------------------------------------
1375 : leunga 788 Name: Allen Leung
1376 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
1377 :     Tag: leunga-20010305-cut-support
1378 :    
1379 :     1. New support for alternative control-flow in MLTREE.
1380 :     Currently we support
1381 :    
1382 :     FLOW_TO(CALL ...., [k1,...,kn])
1383 :    
1384 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
1385 :     in Moby
1386 :    
1387 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
1388 :    
1389 :     2. Register Allocator
1390 :    
1391 :     Changes in interface [from Fermin, John]
1392 :    
1393 :     3. Alpha 8-bit SLL support [Fermin]
1394 :    
1395 :     4. All architectures
1396 :    
1397 :     A new module (ClusterExpandCopies) for expanding parallel copies.
1398 :    
1399 :     ----------------------------------------------------------------------
1400 :     Name: Allen Leung
1401 : leunga 788 Date: 2001/02/27 23:07:00 EST
1402 :     Tag: leunga-20010227-minor-stuff
1403 :    
1404 :     1. Alpha bug fix for CMOVNE
1405 :     2. Handle mltree COND(..,FCMP ...,...)
1406 :     3. Bug fix in simplifier
1407 :    
1408 :     ----------------------------------------------------------------------
1409 : blume 777 Name: Matthias Blume
1410 : blume 781 Date: 2001/01/30 17:50:00 EST
1411 :     Tag: blume-20010130-sync
1412 :     Description:
1413 :    
1414 :     This is just a minor update to sync my devel branch with the main brach.
1415 :     The only visible change is the addition of some README files.
1416 :    
1417 :     ----------------------------------------------------------------------
1418 :     Name: Matthias Blume
1419 : blume 778 Date: 2001/01/12 23:30:00 JST
1420 :     Tag: blume-20010112-bootfiles
1421 :     Description:
1422 :    
1423 :     Made a new set of bootfiles that goes with the current state of the
1424 :     repository.
1425 :    
1426 :     ----------------------------------------------------------------------
1427 :     Name: Matthias Blume
1428 : blume 777 Date: 2001/01/12 21:20:00 JST
1429 :     Tag: blume-20010112-sync
1430 :     Description:
1431 :    
1432 :     I am just flushing out some minor changes that had accumulated in
1433 :     my private branch in order to sync with the main tree. (This is
1434 :     mainly because I had CVS trouble when trying to merge _into_ my
1435 :     private branch.)
1436 :    
1437 :     Most people should be completely unaffected by this.
1438 :    
1439 :     ----------------------------------------------------------------------
1440 : leunga 775 Name: Allen Leung
1441 :     Date: Thu Jan 11 21:03:00 EST 2001
1442 :     Tag: leunga-20010111-labexp=mltree
1443 :     Description:
1444 :    
1445 :     1. Removed the type LabelExp and replace it by MLTree.
1446 :     2. Rewritten mltree-simplify with the pattern matcher tool.
1447 :     3. There were some bugs in alpha code generator which would break
1448 :     64-bit code generation.
1449 :     4. Redo the tools to generate code with the
1450 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
1451 :     are now generated by perl scripts.
1452 :    
1453 :     ----------------------------------------------------------------------
1454 : blume 771 Name: Matthias Blume
1455 : blume 774 Date: 2001/01/10 21:55:00 JST
1456 :     Tag: blume-20010110-rcc
1457 :     Description:
1458 :    
1459 :     The RCC stuff now seems to work (but only on the x86).
1460 :     This required hacking of the c-calls interface (and -implementation) in
1461 :     MLRISC.
1462 :    
1463 :     Normal compiler users should be unaffected.
1464 :    
1465 :     ----------------------------------------------------------------------
1466 :     Name: Matthias Blume
1467 : blume 773 Date: 2001/01/09 01:20:00 JST
1468 :     Tag: blume-20010109-rcc
1469 :     Description:
1470 :    
1471 :     This is a fairly big patch, flushing out a large number of pending
1472 :     changes that I made to my development copy over the last couple of days.
1473 :    
1474 :     Of practical relevance at this moment is a workaround for a pickling
1475 :     bug that Allen ran into the other day. The cause of the bug itself is
1476 :     still unknown and it might be hard to fix it properly, but the
1477 :     workaround has some merits of its own (namely somewhat reducing pickling
1478 :     overhead for certain libraries). Therefore, I think this solution should
1479 :     be satisfactory at this time.
1480 :    
1481 :     The rest of the changes (i.e., the vast majority) has to do with my
1482 :     ongoing efforts of providing direct support for C function calls from
1483 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
1484 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
1485 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
1486 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
1487 :     handle RCC.
1488 :    
1489 :     The last part (the changes to mlriscGen.sml) are still known to be
1490 :     wrong on the x86 and not implemented on all other architectures. But
1491 :     the infrastructure is in place. I had to change a few functor
1492 :     signatures in the backend to be able to route the CCalls interface
1493 :     from MLRISC there, and I had to specialize the mltree type (on the
1494 :     x86) to include the necessary extensions. (The extensions themselves
1495 :     were already there and redy to go in MLRISC/x86).
1496 :    
1497 :     Everything should be very happy as soon as someone helps me with
1498 :     mlriscGen.sml...
1499 :    
1500 :     In any case, nothing of this should matter to anyone as long as the
1501 :     new primop is not being used (which is going to be the case unless you
1502 :     find it where I hid it :). The rest of the compiler is completely
1503 :     unaffected.
1504 :    
1505 :     ----------------------------------------------------------------------
1506 :     Name: Matthias Blume
1507 : blume 772 Date: 2001/01/05 00:30:00 JST
1508 :     Tag: blume-20010105-primops
1509 :     Description:
1510 :    
1511 :     Added some experimental support for work that I am doing right now.
1512 :     These changes mostly concern added primops, but there is also a new
1513 :     experimental C library in the runtime system (but currently not enabled
1514 :     anywhere except on Linux/X86).
1515 :    
1516 :     In the course of adding primops (and playing with them), I discovered that
1517 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
1518 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
1519 :     To recover, I made type information in INL_PRIM mandatory and changed
1520 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
1521 :     has complete, correct type information (i.e., no bottom types).
1522 :    
1523 :     Since all these changes mean that we need new binfiles, I also bumped the
1524 :     version number to 110.32.1.
1525 :    
1526 :     ----------------------------------------------------------------------
1527 :     Name: Matthias Blume
1528 : blume 771 Date: 2000/12/30 22:10:00 JST
1529 :     Tag: blume-20001230-various
1530 :     Description:
1531 :    
1532 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
1533 :     to each other using path anchors. (See CM manual for explanation.)
1534 :    
1535 :     Updated CM documentation.
1536 :    
1537 :     Fixed some bugs in CM.
1538 :    
1539 :     Implemented "proxy" libraries (= syntactic sugar for CM).
1540 :    
1541 :     Added "-quiet" option to makeml and changed runtime system accordingly.
1542 :    
1543 :     Added cleanup handler for exportML to reset timers and compiler stats.
1544 :    
1545 :     ----------------------------------------------------------------------
1546 : george 761 Name: Lal George
1547 :     Date: 2000/12/22 22:22:58 EST 2000
1548 :     Tag: Release_110_32
1549 :     Description:
1550 :    
1551 :     Infinite precision used throughout MLRISC.
1552 :     see MLRISC/mltree/machine-int.sig
1553 :    
1554 :     ----------------------------------------------------------------------
1555 : blume 756 Name: Matthias Blume
1556 : blume 760 Date: 2000/12/22 23:16:00 JST
1557 :     Tag: blume-20001222-warn
1558 :     Description:
1559 : blume 759
1560 : blume 760 Corrected wording and formatting of some CM warning message which I
1561 :     broke in my previous patch.
1562 : blume 759
1563 :     ----------------------------------------------------------------------
1564 :     Name: Matthias Blume
1565 : blume 763 Date: 2000/12/22 21:20:00 JST
1566 :     Tag: blume-20001222-anchorenv
1567 :     Description:
1568 :    
1569 :     Fixed CM's handling of anchor environments in connection with CMB.make.
1570 :    
1571 :     ----------------------------------------------------------------------
1572 :     Name: Matthias Blume
1573 : blume 758 Date: 2000/12/22 13:15:00 JST
1574 :     Tag: blume-20001222-cleanup
1575 :     Description:
1576 :    
1577 :     Removed src/cm/ffi which does not (and did not) belong here.
1578 :    
1579 :     ----------------------------------------------------------------------
1580 :     Name: Matthias Blume
1581 : blume 757 Date: 2000/12/21 23:55:00 JST
1582 :     Tag: blume-20001221-exn
1583 :     Description:
1584 :    
1585 :     Probably most important: CM no longer silently swallows all exceptions
1586 :     in the compiler.
1587 :     Plus: some other minor CM changes. For example, CM now reports some
1588 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
1589 :    
1590 :     ----------------------------------------------------------------------
1591 :     Name: Matthias Blume
1592 : blume 756 Date: 2000/12/15 00:01:05 JST
1593 :     Tag: blume-20001215-dirtool
1594 :     Description:
1595 :    
1596 :     - "dir" tool added.
1597 :     - improvements and cleanup to Tools structure
1598 :     - documentation updates
1599 :    
1600 :     ----------------------------------------------------------------------
1601 : leunga 752 Name: Allen Leung
1602 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
1603 :     Description:
1604 :     Tag: leunga-20001214-int-inf
1605 :     Description:
1606 :    
1607 :     In IntInf, added these standard functions, which are missing from our
1608 :     implementation:
1609 :    
1610 :     andb : int * int -> int
1611 :     xorb : int * int -> int
1612 :     orb : int * int -> int
1613 :     notb : int -> int
1614 :     << : int * word -> int
1615 :     ~>> : int * word -> int
1616 :    
1617 :     Not tested, I hope they are correct.
1618 :    
1619 :     ----------------------------------------------------------------------
1620 :     Name: Allen Leung
1621 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
1622 :     Description:
1623 :     Tag: leunga-20001208-nowhere
1624 :     Description:
1625 :    
1626 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
1627 :     to generate better error messages etc. Plus a brief manual.
1628 :    
1629 :     ----------------------------------------------------------------------
1630 : george 749 Name: Lal George
1631 :     Date: 2000/12/08 09:54:02 EST 2000
1632 :     Tag: Release_110_31
1633 :     Description:
1634 : george 705
1635 : george 749 - Version 110.31
1636 : leunga 591 ----------------------------------------------------------------------
1637 : leunga 744 Name: Allen Leung
1638 :     Date: Thu Dec 7 22:01:04 EST 2000
1639 :     Tag: leunga-20001207-cell-monster-hack
1640 :     Description:
1641 :    
1642 :     Major MLRISC internal changes. Affect all clients.
1643 :     Summary:
1644 :    
1645 :     1. Type CELLS.cell = int is now replaced by a datatype.
1646 :     As a result, the old regmap is now gone. Almost all interfaces
1647 :     in MLRISC change as a consequence.
1648 :    
1649 :     2. A new brand version of machine description tool (v3.0) that generates
1650 :     modules expecting the new interface. The old version is removed.
1651 :    
1652 :     3. The RA interface has been further abstracted into two new functors.
1653 :     RISC_RA and X86RA. These functors have much simpler interfaces.
1654 :     [See also directory MLRISC/demo.]
1655 :    
1656 :     4. Some other new source->source code generation tools are available:
1657 :    
1658 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1659 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1660 :     I use this tool to generate the peephole optimizers---with the new
1661 :     cell type changes, peephole rules are becoming difficult to write
1662 :     without conditional pattern matching.
1663 :    
1664 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
1665 :     cover the entire MLRISC source tree so many things broke.
1666 :    
1667 :     6. CM files have been moved to the subdirectory MLRISC/cm.
1668 :     They are moved because there are a lot of them and they clutter up the
1669 :     root dir.
1670 :    
1671 :     7. More detailed documentation to come...
1672 :    
1673 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1674 :     a makeml -rebuild first. This is because of other other
1675 :     changes that Matthias has made (see below).
1676 :    
1677 :    
1678 :     ----------------------------------------------------------------------
1679 : blume 733 Name: Matthias Blume
1680 : blume 742 Date: 2000/11/30 23:12:00 JST
1681 :     Tag: blume-20001130-filereorg
1682 :     Description:
1683 :    
1684 :     Some manual updates and some file reorganizations in CM.
1685 :    
1686 :     ----------------------------------------------------------------------
1687 :     Name: Matthias Blume
1688 : blume 737 Date: 2000/11/24 17:45:00 JST
1689 :     Tag: blume-20001124-link
1690 :     Description:
1691 :    
1692 :     Drastically improved link traversal code for the case that the dynamic
1693 :     value was already loaded at bootstrap time. As a result, CM and CMB
1694 :     now both load blazingly fast -- even on a very slow machine. Also,
1695 :     memory consumption has been further reduced by this.
1696 :    
1697 :     Warning: The format of the PIDMAP file has changed. THerefore, to
1698 :     bootstrap you have to do this:
1699 :    
1700 :     1. Run CMB.make
1701 :     2. Make a symbolic link for the boot directory:
1702 :     ln -s sml.boot.ARCH-OS xxx
1703 :     3. "Rebuild" the boot directory:
1704 :     ./makeml -boot xxx -rebuild sml ; rm xxx
1705 :     4. Boot normally:
1706 :     ./makeml
1707 :    
1708 :     ----------------------------------------------------------------------
1709 :     Name: Matthias Blume
1710 : blume 735 Date: 2000/11/21 21:20:00 JST
1711 :     Tag: blume-20001121-tools
1712 :     Description:
1713 :    
1714 :     Continued hacking on autoloading problem -- with success this time.
1715 :     Also changed tool-plugin mechanism. See new CM manual.
1716 :    
1717 :     ----------------------------------------------------------------------
1718 :     Name: Matthias Blume
1719 : blume 734 Date: 2000/11/19 14:30:00 JST
1720 :     Tag: blume-20001119-autoload
1721 :     Description:
1722 :    
1723 :     Some hacking to make autoloading faster. Success for CMB, no success
1724 :     so far for CM. There is a reduced structure CM' that autoloads faster.
1725 :     (This is a temporary, non-documented hack to be eliminated again when
1726 :     the general problem is solved.)
1727 :    
1728 :     ----------------------------------------------------------------------
1729 :     Name: Matthias Blume
1730 : blume 733 Date: 2000/11/17 14:10:00 JST
1731 :     Tag: blume-20001117-pickle-lib
1732 :     Description:
1733 :    
1734 :     1. Eliminated comp-lib.cm
1735 :     2. Made pickle-lib.cm
1736 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
1737 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1738 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
1739 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
1740 :     Duplicated intmap modules thrown out.
1741 :     6. Hunted down all duplicated SCC code and replaced it with
1742 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1743 :     7. Rewrote Feedback module.
1744 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
1745 :     should be thrown out and equivalent modules from smlnj-lib.cm
1746 :     should be used (IntRedBlackSet, IntListSet, ...).
1747 :    
1748 :     Confirmed that compiler compiles to fixpoint.
1749 :    
1750 :     ----------------------------------------------------------------------
1751 : leunga 731 Name: Allen Leung
1752 :     Date: 2000/11/10 18:00:00
1753 :     Tag: leunga-20001110-new-x86-fp
1754 :    
1755 :     A new x86 floating point code generator has been added.
1756 :     By default this is turned off. To turn this on, do:
1757 :    
1758 :     CM.autoload "$smlnj/compiler.cm";
1759 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1760 :    
1761 :     Changes:
1762 :    
1763 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
1764 :     2. Changed the extension callback for FTANGENT to generate:
1765 :    
1766 :     fptan
1767 :     fstp %st(0)
1768 :     instead of
1769 :     fptan
1770 :     fstpl ftempmem
1771 :    
1772 :     3. Numerous assembly fixes for x86.
1773 :    
1774 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
1775 :     support for a whole bunch of instructions and addressing modes:
1776 :    
1777 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
1778 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
1779 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
1780 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1781 :     fxch %st(n)
1782 :     fld %st(n)
1783 :     fst %st(n)
1784 :     fst mem
1785 :     fstp %st(n)
1786 :     fucom %st(n)
1787 :     fucomp %st(n)
1788 :    
1789 :     All these are now generated when the fast fp mode is turned on.
1790 :    
1791 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1792 :    
1793 :     ----------------------------------------------------------------------
1794 : blume 729 Name: Matthias Blume
1795 :     Date: 2000/11/09 11:20:00 JST
1796 :     Tag: blume-20001109-scc
1797 :     Description:
1798 :    
1799 :     Eliminated some code duplication:
1800 :    
1801 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1802 :     (Otherwise the functor is useless.)
1803 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
1804 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
1805 :    
1806 :     ----------------------------------------------------------------------
1807 : george 721 Name: Lal George
1808 :     Date: 2000/11/06 09:02:21 EST 2000
1809 :     Tag: Release_110_30
1810 :     Description:
1811 :    
1812 :     - Version 110.30
1813 :     ----------------------------------------------------------------------
1814 : blume 715 Name: Matthias Blume
1815 : blume 716 Date: 2000/11/04 14:45:00
1816 :     Tag: blume-20001104-mlbuild
1817 :     Description:
1818 :    
1819 :     - Made ml-build faster on startup.
1820 :     - Documentation fixes.
1821 :    
1822 :     ----------------------------------------------------------------------
1823 :     Name: Matthias Blume
1824 : blume 715 Date: 2000/11/02 17:00:00 JST
1825 :     Tag: blume-20001102-condcomp
1826 :     Description:
1827 :    
1828 :     - Small tweaks to pickler -- new BOOTFILES!
1829 :     - Version bumped to 110.29.2.
1830 :     - Added conditional compilation facility to init.cmi (see comment there).
1831 :     ----------------------------------------------------------------------
1832 : leunga 713 Name: Allen Leung
1833 :     Date: 2000/10/23 19:31:00
1834 :     Tag: leunga-20001023-demo-ra
1835 :    
1836 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1837 :     2. Test programs for the graph library updated
1838 :     3. Some new MLRISC demo programs added
1839 :    
1840 :     ----------------------------------------------------------------------
1841 : blume 710 Name: Matthias Blume
1842 : blume 711 Date: 2000/08/31 22:15:00 JST
1843 :     Tag: blume-20001017-errmsg
1844 :     Description:
1845 :    
1846 :     More error message grief: Where there used to be no messages, there
1847 :     now were some that had bogus error regions. Fixed.
1848 :    
1849 :     ----------------------------------------------------------------------
1850 :     Name: Matthias Blume
1851 : blume 710 Date: 2000/08/31 17:30:00 JST
1852 :     Tag: blume-20001017-v110p29p1
1853 :     Description:
1854 :    
1855 :     I made a version 110.29.1 with new bootfiles.
1856 :    
1857 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
1858 :     CM documentation changes and a small bugfix in CM's error reporting.
1859 :    
1860 :     ----------------------------------------------------------------------
1861 : george 705 Name: Lal George
1862 :     Date: 2000/09/27 14:42:35 EDT
1863 :     Tag: george-20000927-nodestatus
1864 :     Description:
1865 :    
1866 :     Changed the type of the nodestatus, so that:
1867 :    
1868 :     SPILLED(~1) is now SPILLED
1869 :     SPILLED(m) where m>=0 is now MEMREG(m)
1870 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
1871 :    
1872 :     ----------------------------------------------------------------------
1873 : blume 697 Name: Matthias Blume
1874 : blume 703 Date: 2000/09/07 14:45:00 JST
1875 :     Tag: blume-20000907-cmerrmsg
1876 :     Description:
1877 :    
1878 :     Small tweak to CM to avoid getting ML syntax error messages twice.
1879 :    
1880 :     ----------------------------------------------------------------------
1881 :     Name: Matthias Blume
1882 : blume 701 Date: 2000/08/31 18:00:00 JST
1883 :     Tag: blume-20000831-cvsbootfiles
1884 :     Description:
1885 :    
1886 :     New URL for boot files (because the 110.29 files on the BL server do
1887 :     now work correctly with my updated install scripts for yacc and lex).
1888 :    
1889 :     ----------------------------------------------------------------------
1890 :     Name: Matthias Blume
1891 : blume 697 Date: 2000/08/08 12:33:00 JST
1892 :     Tag: blume-20000808-manual
1893 :     Description:
1894 :    
1895 :     Tiny update to CM manual.
1896 :    
1897 :     ----------------------------------------------------------------------
1898 : leunga 695 Name: Allen Leung
1899 :     Date: 2000/08/7 19:31:00
1900 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
1901 :    
1902 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
1903 :     and Mobdy.
1904 :    
1905 :     1. x86
1906 :    
1907 :     a. Fixes to peephole module by John and Dan.
1908 :     b. Assembly fix to SETcc by Allen.
1909 :     c. Fix to c-call by John.
1910 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
1911 :     e. Instruction selection optimization to SETcc as suggested by John.
1912 :    
1913 :     For example,
1914 :    
1915 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1916 :    
1917 :     should generate:
1918 :    
1919 :     MOVL a, x
1920 :     SUBL b, x
1921 :     SHRL 31, x
1922 :    
1923 :     2. IR stuff
1924 :    
1925 :     A bunch of new DJ-graph related algorithms added. These
1926 :     speed up SSA construction.
1927 :    
1928 :     3. SSA + Scheduling
1929 :    
1930 :     Added code for SSA and scheduling to the repository
1931 :    
1932 :     ----------------------------------------------------------------------
1933 : george 694 Name: Lal George
1934 :     Date: 2000/07/27 11:53:14 EDT
1935 :    
1936 :     Tag: lal-20000727-linux-ppc
1937 :     Description:
1938 :    
1939 :     Made changes to support Linux PPC.
1940 :     p.s. I have confirmation that the 110.29 boot files work fine.
1941 :    
1942 :     ----------------------------------------------------------------------
1943 : blume 684 Name: Matthias Blume
1944 : blume 692 Date: 2000/07/27 17:40:00 JST
1945 :     Tag: blume-20000727-scripts
1946 :     Description:
1947 :    
1948 :     !!!! WARNING !!!!
1949 :     You must recompile the runtime system!
1950 :     !!!! WARNING !!!!
1951 :    
1952 :     This is basically another round of script-enhancements:
1953 :    
1954 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1955 :     and undefine CM preprocessor symbols.
1956 :    
1957 :     2. ml-build avoids generating a new heap image if it finds that the
1958 :     existing one is still ok. (The condition is that no ML file had to
1959 :     be recompiled and all ML files are found to be older that the heap
1960 :     file.)
1961 :    
1962 :     To make this work smoothly, I also hacked the runtime system as
1963 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
1964 :     (.sparc-solaris, ...) that is currently being used.
1965 :    
1966 :     Moreover, the signature of CM.mk_standalone has changed. See the
1967 :     CM manual.
1968 :    
1969 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
1970 :     CM manual for details.)
1971 :    
1972 :     4. More CM manual updates:
1973 :     - all of the above has been documented.
1974 :     - there is now a section describing the (CM-related) command line
1975 :     arguments that are accepted by the "sml" command
1976 :    
1977 :     ----------------------------------------------------------------------
1978 :     Name: Matthias Blume
1979 : blume 691 Date: 2000/07/25 16:20:00 JST
1980 :     Tag: blume-20000725-makedepend
1981 :     Description:
1982 :    
1983 :     Added a script called ml-makedepend. This can be used in makefiles
1984 :     for Unix' make in a way very similar to the "makedepend" command for
1985 :     C.
1986 :    
1987 :     The script internally uses function CM.sources.
1988 :    
1989 :     Synopsis:
1990 :    
1991 :     ml-makedepend [-f makefile] cmfile targetname
1992 :    
1993 :     The default for the makefile is "makefile" (or "Makefile" should
1994 :     "makefile" not exist).
1995 :    
1996 :     ml-makedepend adds a cmfile/targetname-specific section to this
1997 :     makefile (after removing the previous version of this section). The
1998 :     section contains a single dependency specification with targetname on
1999 :     the LHS (targetname is an arbitrary name), and a list of files derived
2000 :     from the cmfile on the RHS. Some of the files on the RHS are
2001 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
2002 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
2003 :     path names. The makefile writer is responsible for making sure that
2004 :     these variables have correct at the time "make" is invoked.
2005 :    
2006 :     ----------------------------------------------------------------------
2007 :     Name: Matthias Blume
2008 : blume 690 Date: 2000/07/22 23:30:00 JST
2009 :     Tag: blume-20000722-urlupdate
2010 :     Description:
2011 :    
2012 :     Changed BOOT and config/srcarchiveurl to point to BL server:
2013 :    
2014 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
2015 :    
2016 :     ----------------------------------------------------------------------
2017 :     Name: Matthias Blume
2018 : blume 686 Date: 2000/07/18 18:00:00 JST
2019 :     Tag: blume-20000718-Version_110_29
2020 :     Description:
2021 :    
2022 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
2023 :    
2024 :     2. Updated config/version to 110.29
2025 :    
2026 :     3. Updated config/srcarchiveurl
2027 :    
2028 :     3. New boot files!
2029 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
2030 :    
2031 :     ----------------------------------------------------------------------
2032 :     Name: Matthias Blume
2033 : blume 684 Date: 2000/07/11 13:58:00 JST
2034 :     Tag: blume-20000711-doctypo
2035 :     Description:
2036 :    
2037 :     Fixed a few typos in CM manual.
2038 :    
2039 :     ----------------------------------------------------------------------
2040 : leunga 683 Name: Allen Leung
2041 :     Date: 2000/06/15 00:38:00
2042 :     Tag: leunga-20000704-sparc-x86
2043 :    
2044 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
2045 :     2. fix to x86 RET bug [found by Dan Grossman]
2046 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
2047 :    
2048 :     Affects c-- and moby only
2049 :    
2050 :     ----------------------------------------------------------------------
2051 : blume 680 Name: Matthias Blume
2052 : blume 682 Date: 2000/07/04 15:26:00
2053 :     Tag: blume-20000704-trigger
2054 :     Description:
2055 :    
2056 :     1. Improvements to CM manual.
2057 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
2058 :     of getting a back-trace. The function, when called, raises an
2059 :     internal exception which explicitly carries the full back-trace history,
2060 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
2061 :     or not). The interactive loop will print that history once it arrives
2062 :     at top level.
2063 :     Short of having all exceptions implicitly carry the full history, the
2064 :     recommended way of using this facility is:
2065 :     - compile your program with instrumentation "on"
2066 :     - run it, when it raises an exception, look at the history
2067 :     - if the history is "cut off" because of some handler, go and modify
2068 :     your program so that it explicitly calls BTrace.trigger
2069 :     - recompile (still instrumented), and rerun; look at the full history
2070 :    
2071 :     ----------------------------------------------------------------------
2072 :     Name: Matthias Blume
2073 :     Date: 2000/07/03 15:36:00 JST
2074 : blume 680 Tag: blume-20000702-manual
2075 :     Description:
2076 :    
2077 :     Small corrections and updates to CM manual.
2078 :    
2079 :     ----------------------------------------------------------------------
2080 : blume 682 Name: Matthias Blume
2081 : blume 679 Date: 2000/06/29 16:04:00 JST
2082 :     Tag: blume-20000629-yacctool
2083 :     Description:
2084 :    
2085 :     Changes:
2086 :    
2087 :     1. Class "mlyacc" now takes separate arguments to pass options to
2088 :     generated .sml- and .sig-files independently.
2089 :     2. Corresponding CM manual updates.
2090 :     3. BTrace module now also reports call sites. (However, for loop clusters
2091 :     it only shows from where the cluster was entered.) There are associated
2092 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
2093 :    
2094 :     ----------------------------------------------------------------------
2095 : blume 675 Name: Matthias Blume
2096 : blume 678 Date: 2000/06/27 16:51:00 JST
2097 :     Tag: blume-20000627-noweb
2098 :     Description:
2099 :    
2100 :     Changes:
2101 :    
2102 :     1. Implemented "subdir" and "witness" options for noweb tool.
2103 :     This caused some slight internal changes in CM's tool implementation.
2104 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
2105 :     some remaining issues from earlier path anchor changes.
2106 :     3. Updated CM manual accordingly.
2107 :    
2108 :     4. Changed implementation of back-tracing so that I now consider it
2109 :     ready for prime-time.
2110 :    
2111 :     In particular, you don't have to explicitly trigger the back-trace
2112 :     anymore. Instead, if you are running BTrace-instrumented code and
2113 :     there is an uncaught exception (regardless of whether or not it was
2114 :     raised in instrumented code), the top-level evalloop will print
2115 :     the back-trace.
2116 :    
2117 :     Features:
2118 :    
2119 :     - Instrumented and uninstrumented code work together seemlessly.
2120 :     (Of course, uninstrumented code is never mentioned in actual
2121 :     back-traces.)
2122 :    
2123 :     - Asymptotic time- and space-complexity of instrumented code is
2124 :     equal to that of uninstrumented code. (This means that
2125 :     tail-recursion is preserved by the instrumentation phase.)
2126 :    
2127 :     - Modules whose code has been instrumented in different sessions
2128 :     work together without problem.
2129 :    
2130 :     - There is no penalty whatsoever on uninstrumented code.
2131 :    
2132 :     - There is no penalty on "raise" expressions, even in
2133 :     instrumented code.
2134 :    
2135 :     A potential bug (or perhaps it is a feature, too):
2136 :    
2137 :     A back-trace reaches no further than the outermost instrumented
2138 :     non-trivial "raise". Here, a "trivial" raise is one that is the
2139 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
2140 :    
2141 :     <exp> handle e => raise e
2142 :    
2143 :     and even
2144 :    
2145 :     <exp> handle Foo => raise Bar
2146 :    
2147 :     and, of course, through
2148 :    
2149 :     <exp> handle Foo => ...
2150 :    
2151 :     if the exception was not Foo.
2152 :    
2153 :     Back-traces always reach right through any un-instrumented code
2154 :     including any of its "handle" expressions, trivial or not.
2155 :    
2156 :     To try this out, do the following:
2157 :    
2158 :     - Erase all existing binfiles for your program.
2159 :     (You may keep binfiles for those modules where you think you
2160 :     definitely don't need back-tracing.)
2161 :     - Turn on back-trace instrumentation:
2162 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2163 :     - Recompile your program. (I.e., run "CM.make" or "use".)
2164 :     - You may now turn instrumentation off again (if you want):
2165 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2166 :     - Run your program as usual. If it raises an exception that
2167 :     reaches the interactive toplevel, then a back-trace will
2168 :     automatically be printed. After that, the toplevel loop
2169 :     will print the exception history as usual.
2170 :    
2171 :     ----------------------------------------------------------------------
2172 :     Name: Matthias Blume
2173 : blume 677 Date: 2000/06/26 09:56:46 JST
2174 :     Tag: blume-20000626-setup
2175 :     Description:
2176 :    
2177 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2178 :     ML code before and after compiling a file (e.g., to set compiler
2179 :     flags)
2180 :    
2181 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2182 :     - associated changes to btrace.sml (BTrace instrumentation pass)
2183 :     - cleaner implementation of btimp.sml (BTrace tracing and report
2184 :     module)
2185 :    
2186 :     CM manual: * new path encoding documented
2187 :     * description of setup-parameter to "sml" added
2188 :    
2189 :     The biggest user-visible change to back-tracing is that it is no
2190 :     longer necessary to compile all traced modules within the same
2191 :     session. (This was a real limitation.)
2192 :    
2193 :     ----------------------------------------------------------------------
2194 :     Name: Matthias Blume
2195 : blume 676 Date: 2000/06/24 12:40:00 JST
2196 :     Tag: blume-20000624-startup
2197 :     Description:
2198 :    
2199 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
2200 :     bit too often -- to put it mildly. :)
2201 :    
2202 :     ----------------------------------------------------------------------
2203 :     Name: Matthias Blume
2204 : blume 675 Date: 2000/06/23 18:20:00 JST
2205 :     Tag: blume-20000623-btrace
2206 :     Description:
2207 :    
2208 :     This updates adds a backtrace facility to aid programmers in debugging
2209 :     their programs. This involves the following changes:
2210 :    
2211 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
2212 :     keeping track of the current call stack. When programs are compiled
2213 :     in a special mode, the compiler will insert calls to these hooks
2214 :     into the user program.
2215 :     "Hook" means that it is possible for different implementations of
2216 :     back-tracing to register themselves (at different times).
2217 :    
2218 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
2219 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
2220 :     It can be turned on using this call:
2221 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2222 :     Turning it off again:
2223 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2224 :     Querying the current status:
2225 :     SMLofNJ.Internals.BTrace.mode NONE;
2226 :     Annotated programs are about twice as big as normal ones, and they
2227 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
2228 :     where all hooks do nothing). The slowdown with a plugin that is
2229 :     actually useful (such as the one supplied by default) is even greater,
2230 :     but in the case of the default plugin it is still only an constant
2231 :     factor (amortized).
2232 :    
2233 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
2234 :     with a sub-structure BTrace for controlling back-tracing. In particular,
2235 :     the above-mentioned function "mode" controls whether the annotation
2236 :     phase is invoked by the compiler. Another important function is
2237 :     "trigger": when called it aborts the current execution and causes
2238 :     the top-level loop to print a full back-trace.
2239 :    
2240 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
2241 :     for back-tracing. It keeps track of the dynamic call stack and in
2242 :     addition to that it keeps a partial history at each "level" of that
2243 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
2244 :     d and b (at separate times, dynamically), then the report will show:
2245 :    
2246 :     GOTO d
2247 :     /c
2248 :     GOTO \b
2249 :     CALL a
2250 :    
2251 :     This shows that there was an initial non-tail call of a, then a
2252 :     tail-call to b or c, looping behavior in a cluster of functions that
2253 :     consist of b and c, and then a goto from that cluster (i.e., either from
2254 :     b or from c) to d.
2255 :    
2256 :     Note that (depending on the user program) the amount of information
2257 :     that the back-trace module has to keep track of at each level is bounded
2258 :     by a constant. Thus, the whole implementation has the same asymptotical
2259 :     complexity as the original program (both in space and in time).
2260 :    
2261 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
2262 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
2263 :     which is raised by the "trigger" function mentioned above.
2264 :    
2265 :     Notes on usage:
2266 :    
2267 :     - Annotated code works well together with unannotated code:
2268 :     Unannotated calls simply do not show up at all in the backtrace.
2269 :    
2270 :     - It is not a good idea to let modules that were annotated during
2271 :     different sessions run at the same time. This is because the compiler
2272 :     chooses small integers to identify individual functions, and there
2273 :     will be clashes if different modules were compiled in separate sessions.
2274 :     (Nothing will crash, and you will even be told about the clashes, but
2275 :     back-trace information will in general not be useful.)
2276 :    
2277 :     - Back-tracing can be confused by callcc and capture.
2278 :    
2279 :     - The only way of getting a back-trace right now is to explicitly
2280 :     invoke the "trigger" function from your user program. Eventually, we
2281 :     should make every exception carry back-trace information (if
2282 :     available). But since this creates more overhead at "raise"-time
2283 :     (similar to the current exnHistory overhead), I have not yet
2284 :     implemented this. (The implementation will be rather easy.) With
2285 :     exceptions carrying back-trace information, this facility will be even
2286 :     more useful because users don't need to modify their programs...
2287 :    
2288 :     - While it is possible to compile the compiler with back-trace
2289 :     annotations turned on (I did it to get some confidence in
2290 :     correctness), you must make absolutely sure that core.sml and
2291 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
2292 :     be compiled with annotation because there is no core access yet, but
2293 :     if you compile btimp.sml with annotation, then the system will go into
2294 :     an infinite recursion and crash.)
2295 :     Since CM currently does not know about BTrace, the only way to turn
2296 :     annotations on and off for different modules of the compiler is to
2297 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
2298 :     this is awkward and clumsy.
2299 :    
2300 :     Sample sessions:
2301 :    
2302 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2303 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
2304 :     [autoloading]
2305 :     [autoloading done]
2306 :     val it = false : bool
2307 :     - structure X = struct
2308 :     - fun main n = let
2309 :     - fun a (x, 0) = d x
2310 :     - | a (x, n) = b (x, n - 1)
2311 :     - and b (x, n) = c (x, n)
2312 :     - and c (x, n) = a (x, n)
2313 :     - and d x = e (x, 3)
2314 :     - and e (x, 0) = f x
2315 :     - | e (x, n) = e (x, n - 1)
2316 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
2317 :     - | f n = n * g (n - 1)
2318 :     - and g n = a (n, 3)
2319 :     - in
2320 :     - f n
2321 :     - end
2322 :     - end;
2323 :     structure X : sig val main : int -> int end
2324 :     - X.main 3;
2325 :     *** BACK-TRACE ***
2326 :     GOTO stdIn:4.2-13.20: X.main[2].f
2327 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2328 :     GOTO stdIn:4.2-13.20: X.main[2].d
2329 :     / stdIn:4.2-13.20: X.main[2].a
2330 :     | stdIn:4.2-13.20: X.main[2].b
2331 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2332 :     CALL stdIn:4.2-13.20: X.main[2].g
2333 :     GOTO stdIn:4.2-13.20: X.main[2].f
2334 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2335 :     GOTO stdIn:4.2-13.20: X.main[2].d
2336 :     / stdIn:4.2-13.20: X.main[2].a
2337 :     | stdIn:4.2-13.20: X.main[2].b
2338 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2339 :     CALL stdIn:4.2-13.20: X.main[2].g
2340 :     GOTO stdIn:4.2-13.20: X.main[2].f
2341 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2342 :     GOTO stdIn:4.2-13.20: X.main[2].d
2343 :     / stdIn:4.2-13.20: X.main[2].a
2344 :     | stdIn:4.2-13.20: X.main[2].b
2345 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2346 :     CALL stdIn:4.2-13.20: X.main[2].g
2347 :     GOTO stdIn:4.2-13.20: X.main[2].f
2348 :     CALL stdIn:2.15-17.4: X.main[2]
2349 :     -
2350 :    
2351 :     (Note that because of a FLINt bug the above code currently does not
2352 :     compile without BTrace turned on.)
2353 :    
2354 :     Here is another example, using my modified Tiger compiler:
2355 :    
2356 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2357 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
2358 :     [autoloading]
2359 :     [autoloading done]
2360 :     val it = false : bool
2361 :     - CM.make "sources.cm";
2362 :     [autoloading]
2363 :     ...
2364 :     [autoloading done]
2365 :     [scanning sources.cm]
2366 :     [parsing (sources.cm):parse.sml]
2367 :     [creating directory CM/SKEL ...]
2368 :     [parsing (sources.cm):tiger.lex.sml]
2369 :     ...
2370 :     [wrote CM/sparc-unix/semant.sml]
2371 :     [compiling (sources.cm):main.sml]
2372 :     [wrote CM/sparc-unix/main.sml]
2373 :     [New bindings added.]
2374 :     val it = true : bool
2375 :     - Main.compile ("../testcases/merge.tig", "foo.out");
2376 :     *** BACK-TRACE ***
2377 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
2378 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2379 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2380 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2381 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2382 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2383 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
2384 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
2385 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
2386 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2387 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
2388 :     CALL main.sml:1.18-118.4: Main.compile[2]
2389 :     -
2390 :    
2391 :     ----------------------------------------------------------------------
2392 : blume 673 Name: Matthias Blumen
2393 :     Date: 2000/06/21 18:00:00 JST
2394 :     Tag: blume-20000621-manual
2395 :     Description:
2396 :    
2397 :     CM manual update: Path environments documented.
2398 :    
2399 :     ----------------------------------------------------------------------
2400 : blume 666 Name: Matthias Blume
2401 : blume 672 Date: 2000/06/19 13:40:00
2402 :     Tag: blume-20000619-manual
2403 :     Description:
2404 :    
2405 :     CM manual and system/README update. This only covers the fact that
2406 :     there are no more implicit anchors. (Path environments and the "bind"
2407 :     option to "cm" have yet to be documented.)
2408 :    
2409 :     ----------------------------------------------------------------------
2410 :     Name: Matthias Blume
2411 : blume 671 Date: 2000/06/19 11:05:00 JST
2412 :     Tag: blume-20000619-chdir-bugfix
2413 :     Description:
2414 :    
2415 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
2416 :    
2417 :     ----------------------------------------------------------------------
2418 :     Name: Matthias Blume
2419 : blume 670 Date: 2000/06/18 22:00:10 JST
2420 :     Tag: blume-20000618-implicit-anchors-really-gone
2421 :     Description:
2422 :    
2423 :     I updates the previous HISTORY entry where I forgot to mention that
2424 :     implicit anchors are no longer with us.
2425 :    
2426 :     The current update also gets rid of the (now useless) controller
2427 :     CM.Control.implicit_anchors.
2428 :    
2429 :     ----------------------------------------------------------------------
2430 :     Name: Matthias Blume
2431 : blume 666 Date: 2000/06/16 17:30:00 JST
2432 :     Tag: blume-20000616-anchorenv
2433 :     Description:
2434 :    
2435 :     This patch implements the long anticipated (just kidding :) "anchor
2436 :     environment" mechanism. In the course of doing this, I also
2437 :     re-implemented CM's internal "SrcPath" module from scratch. The new
2438 :     one should be more robust in certain boundary cases. In any case, it
2439 :     is a lot cleaner than its predecessor (IMHO).
2440 :    
2441 :     This time, although there is yet another boot file format change, I
2442 :     kept the unpickler backward-compatible. As a result, no new bootfiles
2443 :     are necessary and bootstrapping is straightforward. (You cannot read
2444 :     new bootfiles into an old system, but the other way around is no
2445 :     problem.)
2446 :    
2447 :     Visible changes:
2448 :    
2449 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
2450 :     longer recognized at all. This means that such path names are not
2451 :     illegal either. For example, the name basis.cm simply refers to a
2452 :     local file called "basis.cm" (i.e, the name is an ordinary path
2453 :     relative to .cm-files directory). Or, to put it differently, only
2454 :     names that start with $ are anchored paths.
2455 : blume 666
2456 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
2457 :     vanished.
2458 :    
2459 :     John (Reppy) had critizised this as soon as I originally proposed and
2460 : blume 666 implemented it, but at that time I did not really deeply believe
2461 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
2462 :     in another place where it cannot be seen as an abbreviation for
2463 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
2464 :     seems to mean (i.e., it "expands" into the corresponding anchor
2465 :     value).
2466 :    
2467 :     However, when paths are used as members in CM description files, it
2468 :     continues to be true that there must be at least another arc after the
2469 :     anchor. This is now enforced separately during semantic analysis
2470 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
2471 : blume 666
2472 :     ** 2. The "cm" class now accepts an option "bind". The option's value
2473 :     is a sub-option list of precisely two items -- one labeled "anchor"
2474 :     and the other one labeled "value". As you might expect, "anchor" is
2475 :     used to specify an anchor name to be bound, and "value" specifies what
2476 :     the anchor is being bound to.
2477 :    
2478 :     The value must be a directory name and can be given in either standard
2479 :     syntax (including the possibility that it is itself an anchored path)
2480 :     or native syntax.
2481 :    
2482 :     Examples:
2483 :    
2484 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
2485 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
2486 :    
2487 :     and so on.
2488 :    
2489 :     The meaning of this is that the .cm-file will be processed with an
2490 :     augmented anchor environment where the given anchor(s) is/are bound to
2491 :     the given values(s).
2492 :    
2493 :     The rationale for having this feature is this: Suppose you are trying
2494 :     to use two different (already stable) libraries a.cm and b.cm (that
2495 :     you perhaps didn't write yourself). Further, suppose each of these
2496 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
2497 :     Normally you would now have a problem because the anchor "lib" can not
2498 :     be bound to more than one value globally. Therefore, the project that
2499 :     uses both a.cm and b.cm must locally redirect the anchor to some other
2500 :     place:
2501 :    
2502 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
2503 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
2504 :    
2505 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
2506 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
2507 :    
2508 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
2509 :     you when you do so at the time of CM.stabilize). Therefore, you can
2510 :     also use an anchored path as the value:
2511 :    
2512 :     a.cm (bind:(anchor:lib value:$a-lib))
2513 :     b.cm (bind:(anchor:lib value:$b-lib))
2514 :    
2515 :     Now you can globally configure (using the usual CM.Anchor.anchor or
2516 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
2517 :     itself is always locally bound, setting it globally is no longer
2518 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
2519 :     can still be used as a global anchor for separate purposes. As a
2520 :     matter of fact, one can locally define "lib" in terms of a global
2521 :     "lib":
2522 :    
2523 :     a.cm (bind:(anchor:lib value:$lib/a))
2524 :     b.cm (bind:(anchor:lib value:$lib/b))
2525 :    
2526 :     ** 3: The encoding of path names has changed. This affects the way
2527 :     path names are shown in CM's progress report and also the internal
2528 :     protocol encoding used for parallel make.
2529 :    
2530 :     The encoding now uses one or more ':'-separated segments. Each
2531 :     segments corresponds to a file that has been specified relative to the
2532 :     file given by its preceding segment. The first segment is either
2533 :     relative to the CWD, absolute, or anchored. Each segment itself is
2534 :     basically a Unix pathname; all segments but the first are relative.
2535 :    
2536 :     Example:
2537 :    
2538 :     $foo/bar/baz.cm:a/b/c.sml
2539 :    
2540 :     This path denotes the file bar/a/b/c.sml relative to the directory
2541 :     denoted by anchor "foo". Notice that the encoding also includes
2542 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
2543 :     paths are resolved relative to the .cm-files directory, so baz.cm must
2544 :     be ignored to get the "real" pathname.
2545 :    
2546 :     To make this fact more obvious, CM puts the names of such "virtual
2547 :     arcs" into parentheses when they appear in progress reports. (No
2548 :     parentheses will appear in the internal protocol encoding.) Thus,
2549 :     what you really see is:
2550 :    
2551 :     $foo/bar/(baz.cm):a/b/c.sml
2552 :    
2553 :     I find this notation to be much more informative than before.
2554 :    
2555 :     Another new feature of the encoding is that special characters
2556 :     including parentheses, colons, (back)slashes, and white space are
2557 :     written as \ddd (where ddd is the decimal encoding of the character).
2558 :    
2559 : blume 670 *** The CM manual still needs to be updated.
2560 :    
2561 : blume 666 ----------------------------------------------------------------------
2562 : leunga 660 Name: Allen Leung
2563 :     Date: 2000/06/15 00:38:00
2564 :     Tag: leunga-20000615-x86-peephole
2565 :    
2566 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
2567 :    
2568 :     ----------------------------------------------------------------------
2569 : blume 658 Name: Matthias Blume
2570 :     Date: 2000/06/12 11:40:00
2571 :     Tag: blume-20000612-parmakefix
2572 :     Description:
2573 :    
2574 :     More cleanup after changing the file naming scheme: This time I
2575 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
2576 :    
2577 :     ----------------------------------------------------------------------
2578 : leunga 657 Name: Allen Leung
2579 :     Date: 2000/06/09 01:25:00
2580 :     Tag: leunga-20000609-various
2581 :    
2582 :     None of these things should affect normal SML/NJ operations
2583 :    
2584 :     1. Peephole improvements provided by Fermin (c--)
2585 :     2. New annotation DEFUSE for adding extra dependence (moby)
2586 :     3. New X86 LOCK instructions (moby)
2587 :     4. New machine description language for reservation tables (scheduling)
2588 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
2589 :     trees etc.)
2590 :     6. I've changed the CM files so that they can work with versions
2591 :     110.0.6, 110.25 and 110.28
2592 :    
2593 :     ----------------------------------------------------------------------
2594 : blume 652 Name: Matthias Blume
2595 : blume 656 Date: 2000/06/09 12:40:00
2596 :     Tag: blume-20000609-log
2597 :     Description:
2598 :    
2599 :     - Removed all(?) remaining RCS Log entries from sources.
2600 :    
2601 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
2602 :     anchored paths).
2603 :    
2604 :     ----------------------------------------------------------------------
2605 :     Name: Matthias Blume
2606 : blume 653 Date: 2000/06/07 17:00:00 JST
2607 :     Tag: blume-20000607-no-implicit-anchors
2608 :     Description:
2609 :    
2610 :     1. This update changes the default setting for
2611 :     CM.Control.implicit_anchors from true to false. This means that
2612 :     implicit anchors are no longer permitted by default. I also tried to
2613 :     make sure that nothing else still relies on implicit anchors.
2614 :     (This is the next step on the schedule towards a CM that does not even
2615 :     have the notion of implicit anchors anymore.)
2616 :    
2617 :     2. More CM manual updates.
2618 :    
2619 :     3. I managed to track down and fix the pickling bug I mentioned last
2620 :     time. Because of the previously existing workaround, this entails no
2621 :     immediate practical changes.
2622 :    
2623 :     ----------------------------------------------------------------------
2624 :     Name: Matthias Blume
2625 : blume 652 Date: 2000/06/06 11:15:00 JST
2626 :     Tag: blume-20000606-lazierpickle
2627 :     Description:
2628 :    
2629 :     !!!! NEW BOOT FILES !!!!
2630 :    
2631 :     * The main purpose of this update is to make library pickles lazier in
2632 :     order to reduce the initial space penalty for autoloading a library.
2633 :     As a result, it is now possible to have $smlnj/compiler.cm
2634 :     pre-registered. This should take care of the many complaints or
2635 :     inquiries about missing structure Compiler. This required changes to
2636 :     CM's internal data structures and small tweaks to some algorithms.
2637 :    
2638 :     As a neat additional effect, it is no longer necessary (for the sake
2639 :     of lean heap image files) to distinguish between a "minimal" CM and a
2640 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
2641 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
2642 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
2643 :    
2644 :     To make the life of compiler-hackers easier, "makeml" now also
2645 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
2646 :     words, after you bootstrap a new sml for the first time, you will not
2647 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
2648 :     around you will still have to do it, though.)
2649 :    
2650 :     * A second change consists of major updates to the CM manual. There
2651 :     are now several appendices with summary information and also a full
2652 :     specification of the CM description file syntax.
2653 :    
2654 :     * In directory src/system I added the script "allcross". This script
2655 :     invokes sml and cross-compiles the compiler for all supported
2656 :     architectures. (Useful when providing a new set of boot files.)
2657 :    
2658 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
2659 :     added a small tweak to pickle-util.sml to work around this problem,
2660 :     but it is not a proper fix yet. I will investigate further. (The
2661 :     effect of the bug was an inflation of library pickle size.)
2662 :    
2663 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
2664 :    
2665 :     ----------------------------------------------------------------------
2666 : leunga 646 Name: Allen Leung
2667 : leunga 648 Date: 2000/05/25 17:28 EDT
2668 :     Tag: leunga-20000525-ra
2669 :     Description:
2670 :    
2671 :     Fixed a bug in freezing phase of the register allocator.
2672 :    
2673 :     ----------------------------------------------------------------------
2674 :     Name: Allen Leung
2675 : leunga 647 Date: 2000/05/15 22:53 EDT
2676 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
2677 :     Description:
2678 :    
2679 :     1. Alpha
2680 :    
2681 :     Slight cleanup. Removed the instruction SGNXL
2682 :    
2683 :     2. X86
2684 :    
2685 :     Added the following instructions to the instruction set:
2686 :    
2687 :     ROLx, RORx,
2688 :     BTx, BTSx, BTLx, BTRx,
2689 :     XCHGx, and variants with the LOCK prefix
2690 :    
2691 :     3. Register Allocation
2692 :    
2693 :     The module ra-rewrite-with-renaming has been improved.
2694 :    
2695 : leunga 647 These have no effect on SML/NJ.
2696 :    
2697 : leunga 646 ----------------------------------------------------------------------
2698 : blume 642 Name: Matthias Blume
2699 : blume 645 Date: 2000/05/15 16:20:00 JST
2700 :     Tag: blume-20000515-lightrebuild
2701 :     Description:
2702 :    
2703 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
2704 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
2705 :     defined. In effect, the command will not build any cross-compiler
2706 :     backends and therefore finish more quickly.
2707 :    
2708 :     The "fixpt" script also takes a "-light" switch to be able to use
2709 :     this new facility while compiling for a fixpoint.
2710 :    
2711 :     2. I replaced all mentions of anchored paths in group owner specifications
2712 :     with simple relative paths (usually starting with "..").
2713 :     The rationale is that a library's internal workings should not be
2714 :     compromised by the lack of some anchor. (An anchor is necessary
2715 :     for someone who wants to refer to the library by an anchored path,
2716 :     but it should not be necessary to build the same library in the first
2717 :     place.)
2718 :    
2719 :     3. I changed the way CM's tool mechanism determines the shell command
2720 :     string used for things like ml-yacc etc. so that it does not break
2721 :     when CM.Control.implicit_anchors is turned off.
2722 :    
2723 :     ----------------------------------------------------------------------
2724 :     Name: Matthias Blume
2725 : blume 644 Date: 2000/05/12 18:20:00 JST
2726 :     Tag: blume-20000512-ml-build
2727 :     Description:
2728 :    
2729 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
2730 :     getting installed properly (by config/install.sh).
2731 :    
2732 :     ----------------------------------------------------------------------
2733 :     Name: Matthias Blume
2734 : blume 643 Date: 2000/05/12 17:30:00 JST
2735 :     Tag: blume-20000512-anchors
2736 :     Description:
2737 :    
2738 :     !!! NEW BOOT FILES !!!
2739 :    
2740 :     This change is in preparation of fading out support for "implicitly
2741 :     anchored path names". I went through all sources and used the
2742 :     explicit (and relatively new) $-notation. See system/README and the
2743 :     CM manual for more info on this.
2744 :    
2745 :     I also modified the anchoring scheme for some things such as "smlnj",
2746 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
2747 :     anchors are more expressive: anchor name and first arc do not have to
2748 :     coincide. This entails the following user-visible change:
2749 :    
2750 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
2751 :     particular, when you fire up sml with a command-line argument, say,
2752 :     e.g.:
2753 :    
2754 :     sml '$smlnj/cmb.cm'
2755 :    
2756 :     At the ML toplevel prompt:
2757 :    
2758 :     CM.autoload "$smlnj/cmb.cm";
2759 :    
2760 :     There is also a new controller in CM.Control that can be used to turn
2761 :     off all remaining support for implicit anchors by saying:
2762 :    
2763 :     CM.autoload "$smlnj/
2764 :     #set CM.Control.implicit_anchors false;
2765 :    
2766 :     This causes CM to reject implicitly anchored paths. This is (for the
2767 :     time being) less permissive than the "final" version where there will
2768 :     be no more such implicit anchors and relative paths will be just that:
2769 :     relative.
2770 :    
2771 :     The next step (version after next version?) will be to make the
2772 :     default for CM.Control.implicit_anchors false. After the dust has
2773 :     settled, I can then produce the "final" version of this...
2774 :    
2775 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
2776 :    
2777 :     ----------------------------------------------------------------------
2778 :     Name: Matthias Blume
2779 : blume 642 Date: 2000/05/11 16:30:00 JST
2780 :     Tag: blume-20000511-sources
2781 :     Description:
2782 :    
2783 :     The main change is that I added function CM.sources as a generalized
2784 :     version of the earlier CM.makedepend. This entails the following
2785 :     additional changes:
2786 :    
2787 :     - CM.makedepend has been dropped.
2788 :    
2789 :     - CM manual has been updated.
2790 :    
2791 :     - TOOLS signature and API have been changed.
2792 :    
2793 :     ----------------------------------------------------------------------
2794 : leunga 640 Name: Allen Leung
2795 : leunga 641 Date: 2000/05/10 21:17 EDT
2796 :     Tag: leunga-20000510-moby-c--ssa
2797 :     Description:
2798 :    
2799 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2800 :     None of these affect SML/NJ.
2801 :    
2802 :     1. Register Allocation
2803 :    
2804 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2805 :     This module tries to remove local (i.e. basic block level) redundancies
2806 :     during spilling.
2807 :    
2808 :     b. A new framework for performing region based register allocation.
2809 :     Not yet entirely functional.
2810 :    
2811 :     2. X86
2812 :    
2813 :     a. DefUse for POP was missing the stack pointer [found by Lal]
2814 :     b. Reload for CALL was incorrect in X86Spill [found by John]
2815 :     c. Various fixes in X86Spill so that it can be used correctly for
2816 :     the new spilling module.
2817 :    
2818 :     3. SSA/IR
2819 :    
2820 :     a. New module ir/dj-dataflow.sml implements elimination based
2821 :     data flow analysis.
2822 :    
2823 :     4. MLRiscGen
2824 :    
2825 :     a. Fix for gc type annotation
2826 :    
2827 :     5. MDGen
2828 :    
2829 :     Various fixes for machine description -> ml code translation. For ssa
2830 :     only.
2831 :    
2832 :     ----------------------------------------------------------------------
2833 :     Name: Allen Leung
2834 : leunga 640 Date: 2000/05/08 22:17 EDT
2835 :     Tag: leunga-20000508-labexp
2836 :     Description:
2837 :    
2838 :     Fermin has found a few assembly problems with constant expressions
2839 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
2840 :     which choke on dumb assemblers. This is his fix.
2841 :    
2842 :     ----------------------------------------------------------------------
2843 : dbm 635 Name: Dave MacQueen
2844 :     Date: 2000/04/09 14:00 EDT
2845 :     Tag: dbm-20000502-Version_110_28
2846 :     Description:
2847 :    
2848 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2849 :    
2850 :     2. Updated config/version to 110.28
2851 :    
2852 :     3. Updated config/srcarchiveurl
2853 :    
2854 :     3. New boot files!
2855 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2856 :    
2857 :     ----------------------------------------------------------------------
2858 : blume 634 Name: Matthias Blume
2859 :     Date: 2000/05/01 19:05:00 JST
2860 :     Tag: blume-20000501-noweb
2861 :     Description:
2862 :    
2863 :     A new noweb tool has been added. The existing system is entirely
2864 :     unaffected by this, but some CM users have asked for renewed noweb
2865 :     support. Everything is documented in the CM manual.
2866 :    
2867 :     New (plugin) libraries:
2868 :    
2869 :     noweb-tool.cm
2870 :     nw-ext.cm
2871 :    
2872 :     ----------------------------------------------------------------------
2873 : dbm 633 Name: Dave MacQueen
2874 :     Date: 2000/04/30 12:40PM EDT
2875 :     Tag: dbm-20000430-bug_fixes
2876 :     Description:
2877 :    
2878 :     1. Fix for bug 1498
2879 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2880 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2881 :     added toRealArray function
2882 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
2883 :     added check for tag Obj.RealArray to array printing case in ppObj
2884 :    
2885 :     2. Fix for bug 1510
2886 :     smlnj/src/compiler/Semant/types/typesutil.sml
2887 :     fixed definition of dummyargs (used by equalTycon) so that
2888 :     dummy args are distinct types
2889 :    
2890 :     ----------------------------------------------------------------------
2891 : blume 629 Name: Matthias Blume
2892 : blume 632 Date: 2000/04/30 01:00:00 JST
2893 :     Tag: blume-20000430-versions
2894 :     Description:
2895 :    
2896 :     1. CM version numbering added. This is an implementation of Lal's
2897 :     proposal for adding version numbers and version checking to .cm
2898 :     files. Lal said that his proposal was just that -- a proposal.
2899 :     For the time being I went ahead and implemented it so that people
2900 :     can comment on it. Everything is completely backward-compatible
2901 :     (except for the stable library format, i.e., new bootfiles!).
2902 :    
2903 :     As usual, see the CM manual for details.
2904 :    
2905 :     2. An alternative syntax for anchored paths has been implemented.
2906 :     Dave has recently voiced the same concerns that I had when I did
2907 :     this, so there should be some support. My take is that eventually
2908 :     I will let support for the current syntax (where anchors are
2909 :     "implicit") fade out in favor of the new, explicit syntax.
2910 :     In order to be backward-compatible, both old and new syntax are
2911 :     currently supported.
2912 :    
2913 :     Again, see the CM manual for details.
2914 :    
2915 :     3. Parallel make is trying to be slightly smarter: When the master
2916 :     process finds a "bottleneck", i.e., when there is only one
2917 :     compilation unit that can be compiled and everybody else is
2918 :     waiting on it, then it will simply compile it directly instead
2919 :     of clumsily telling one of the slaves to do it.
2920 :    
2921 :     4. Support for "unsharing" added. This is necessary in order to be
2922 :     able to have two different versions of the same library running
2923 :     at the same time (e.g., for trying out a new MLRISC while still
2924 :     having the old MLRISC linked into the current compiler, etc.)
2925 :     See the CM manual.
2926 :    
2927 :     5. Simple "makedepend" functionality added for generating Makefile
2928 :     dependency information. (This is rather crude at the moment.
2929 :     Expect some changes here in the future.)
2930 :    
2931 :     6. ".fun" added as a recognized suffix for ML files. Also documented
2932 :     explicitly in the manual that the fallback behavior (unknown suffix
2933 :     -> ML file) is not an official feature!
2934 :    
2935 :     7. Small changes to the pickler for stable libraries.
2936 :    
2937 :     8. Several internal changes to CM (for cleanup/improvement).
2938 :    
2939 :    
2940 :     !!!! NEW BINFILES !!!!
2941 :    
2942 :     ----------------------------------------------------------------------
2943 :     Name: Matthias Blume
2944 : blume 631 Date: 2000/04/28 17:30:00 JST
2945 :     Tag: blume-20000428-pathconfig
2946 :     Description:
2947 :    
2948 :     1. I changed config/install.sh to remove duplicate entries from the
2949 :     lib/pathconfig file at the end. Moreover, the final version of
2950 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
2951 :     in src/system/installml.
2952 :    
2953 :     2. The config/install.sh script now consistently uses relative
2954 :     pathnames in lib/pathconfig whenever the anchor is in the lib
2955 :     directory. (So far this was true for the libraries that come
2956 :     pre-compiled and bundled as part of the bootfiles but not for
2957 :     libraries that are compiled by the script itself.)
2958 :    
2959 :     ----------------------------------------------------------------------
2960 :     Name: Matthias Blume
2961 : blume 629 Date: 2000/04/26 13:10:00 JST
2962 :     Tag: blume-20000426-fun_suffix
2963 :     Description:
2964 :    
2965 :     Added ".fun" as a recognized file name suffix (for ML code).
2966 :    
2967 :     ----------------------------------------------------------------------
2968 : leunga 628 Name: Allen Leung
2969 :     Date: 2000/04/25 17:00:00 EST
2970 :     Tag: leunga-20000425-alpha-ra
2971 :     Description:
2972 :    
2973 :     1. Alpha
2974 :    
2975 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
2976 :     in C--.
2977 :    
2978 :     2. RA
2979 :    
2980 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
2981 :     and the interference graph.
2982 :    
2983 :     ----------------------------------------------------------------------
2984 : dbm 627 Name: Dave MacQueen
2985 :     Date: 2000/04/25/
2986 :     Tag: dbm-20000425-mlyacc_doc_examples
2987 :     Description:
2988 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2989 :     Updated all three examples in src/ml-yacc/examples to run
2990 :     under 110.* using CM.make.
2991 :    
2992 :     ----------------------------------------------------------------------
2993 : leunga 624 Name: Allen Leung
2994 :     Date: 2000/04/20 23:04:00 EST
2995 :     Tag: leunga-20000420-ssa-c---stuff
2996 :     Description:
2997 :    
2998 :     This update synchronizes my repository with Yale's. Most of these
2999 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
3000 :    
3001 :     1. Register Allocator
3002 :    
3003 :     a. An improvement in the interference graph construction:
3004 :     Given a copy
3005 :    
3006 :     s <- t
3007 :    
3008 : leunga 628 no interference edge between s and t is added for this definition of s.
3009 : leunga 624
3010 :     b. I've added two new spill heuristic modules that Fermin and I developed
3011 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
3012 :     useful for others (Moby?)
3013 :    
3014 :     2. X86
3015 :    
3016 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
3017 :    
3018 :     3. Alpha
3019 :    
3020 :     a. Added the BSR instruction and code generation that goes with it [C--]
3021 :     b. Other fixes too numerous to recount provided by Fermin [C--]
3022 :    
3023 :     4. Regmaps
3024 :    
3025 :     a. The regmaps are not initialized with the identity physical bindings
3026 :     at creation time. This is unneeded.
3027 :    
3028 :     5. MLRISC Optimizations
3029 :    
3030 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
3031 :     intersects with liveness incrementally in linear time! Woohoo!
3032 :     This is now used in my new SSA construction algorithm.
3033 :    
3034 :     b. THe branch reorganization module is now smarter about linear chains of
3035 :     basic blocks.
3036 :    
3037 :    
3038 :     ----------------------------------------------------------------------
3039 : blume 615 Name: Matthias Blume
3040 :     Date: 2000/04/12 13:52:00 JST
3041 :     Tag: blume_main_v110p27_1
3042 :     Description:
3043 :    
3044 :     Changed install.sh script to handle archive files without version number
3045 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
3046 :     name of the boot file archive.
3047 :    
3048 :     ----------------------------------------------------------------------
3049 : dbm 607 Name: Dave MacQueen
3050 :     Date: 2000/04/09 14:00 EDT
3051 :     Tag: dbm-20000410-Version_110_27
3052 :     Description:
3053 :    
3054 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
3055 :    
3056 :     2. Updated src/config/version to 110.27
3057 :    
3058 :     3. New boot files!
3059 :    
3060 :     ----------------------------------------------------------------------
3061 : leunga 606 Name: Allen Leung
3062 :     Date: 2000/04/09 19:09:00 EST
3063 :     Tag: leunga-20000409-misc
3064 :     Description:
3065 :    
3066 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
3067 :    
3068 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
3069 :    
3070 :     ----------------------------------------------------------------------
3071 : monnier 605 Name: Stefan
3072 :     Date: 2000/04/07 10:00:00 EDT
3073 :     Tag: monnier-20000406-branch-handling
3074 :     Description:
3075 :    
3076 :     Improved handling of branches (mostly those generated from
3077 :     polymorphic equality), removed switchoff and changed the
3078 :     default optimization settings (more cpsopt and less flintopt).
3079 :    
3080 :     ----------------------------------------------------------------------
3081 : leunga 601 Name: Allen Leung
3082 : leunga 602 Date: 2000/04/06 01:30:00 EST
3083 :     Tag: leunga-20000406-peephole-x86-SSA-2
3084 :     Description:
3085 :    
3086 :     Forgot a few files.
3087 :    
3088 :     ----------------------------------------------------------------------
3089 :     Name: Allen Leung
3090 : leunga 601 Date: 2000/04/06 00:36:00 EST
3091 :     Tag: leunga-20000406-peephole-x86-SSA
3092 :     Description:
3093 :    
3094 :     1. New Peephole code
3095 :    
3096 :     2. Minor improvement to X86 instruction selection
3097 :    
3098 :     3. Various fixes to SSA and machine description -> code translator
3099 :    
3100 :     ----------------------------------------------------------------------
3101 : blume 595 Name: Matthias Blume
3102 :     Date: 2000/04/05 12:30:00 JST
3103 :     Tag: blume_main_v110p26p2_3
3104 :     Description:
3105 :    
3106 :     This update just merges three minor cosmetic updates to CM's sources
3107 :     to get ready for the 110.27 code freeze on Friday. No functionality
3108 :     has changed.
3109 :    
3110 :     ----------------------------------------------------------------------
3111 : leunga 593 Name: Allen Leung
3112 : leunga 594 Date: 2000/04/04 19:39:00 EST
3113 :     Tag: leunga-20000404-x86-asm
3114 :     Description:
3115 :    
3116 :     1. Fixed a problem in X86 assembly.
3117 :    
3118 :     Things like
3119 :    
3120 :     jmp %eax
3121 :     jmp (%eax)
3122 :    
3123 :     should be output as
3124 :    
3125 :     jmp *%eax
3126 :     jmp *(%eax)
3127 :    
3128 :     2. Assembly output
3129 :    
3130 :     Added a new flag
3131 :    
3132 :     "asm-indent-copies" (default to false)
3133 :    
3134 :     When this flag is on, parallel copies will be indented an extra level.
3135 :    
3136 :     ----------------------------------------------------------------------
3137 :     Name: Allen Leung
3138 : leunga 593 Date: 2000/04/04 03:18:00 EST
3139 :     Tag: leunga-20000404-C--Moby
3140 :     Description:
3141 :    
3142 :     All of these fixes are related to C--, Moby, and my own optimization
3143 :     stuff; so they shouldn't affect SML/NJ.
3144 :    
3145 :     1. X86
3146 :    
3147 :     Various fixes related floating point, and extensions.
3148 :    
3149 :     2. Alpha
3150 :    
3151 :     Some extra patterns related to loads with signed/zero extension
3152 :     provided by Fermin.
3153 :    
3154 :     3. Assembly
3155 :    
3156 : monnier 659 When generating assembly, resolve the value of client defined constants,
3157 : leunga 593 instead of generating symbolic values. This is controlled by the
3158 :     new flag "asm-resolve-constants", which is default to true.
3159 :    
3160 :     4. Machine Descriptions
3161 :    
3162 :     a. The precedence parser was slightly broken when parsing infixr symbols.
3163 :     b. The type generalizing code had the bound variables reversed, resulting
3164 :     in a problem during arity raising.
3165 :     c. Various fixes in machine descriptions.
3166 :    
3167 :     ----------------------------------------------------------------------
3168 : blume 592 Name: Matthias Blume
3169 :     Date: 2000/04/03 16:05:00 JST
3170 :     Tag: blume_main_v110p26p2_2
3171 :     Description:
3172 :    
3173 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
3174 :     now done via the ordinary static environment that is context to each
3175 :     compilation unit.
3176 :    
3177 :     To this end, I arranged that instead of "structure Core" as "structure
3178 :     _Core" is bound in the pervasive environment. Core access is done via
3179 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
3180 : blume 592 legal surface-syntax symbol).
3181 :    
3182 :     The current solution is much cleaner because the core environment is
3183 :     now simply part of the pervasive environment which is part of every
3184 :     compilation unit's context anyway. In particular, this eliminates all
3185 :     special-case handling that was necessary until now in order to deal
3186 :     with dynamic and symbolic parts of the core environment.
3187 :    
3188 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3189 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
3190 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
3191 :     more details.
3192 :    
3193 :     I also tried to track down all mentions of "Core" (as string argument
3194 :     to Symbol.strSymbol) in the compiler and replaced them with a
3195 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
3196 :     name appears in one place only.
3197 :    
3198 :     Binfile and bootfile format have not changed, but the switchover from
3199 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
3200 :     bootfiles anyway.
3201 :    
3202 :     ----------------------------------------------------------------------
3203 : leunga 591 Name: Allen Leung
3204 :     Date: 2000/04/02 21:17:00 EST
3205 :     Tag: leunga-20000402-mltree
3206 :     Description:
3207 : leunga 585
3208 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
3209 :     2. Added a bunch of files from my repository. These are currently
3210 :     used by other non-SMLNJ backends.
3211 :    
3212 : leunga 576 ----------------------------------------------------------------------
3213 : leunga 590 Name: Allen Leung
3214 :     Date: 2000/03/31 21:15:00 EST
3215 :     Tag: leunga-20000331-aliasing
3216 :     Description:
3217 :    
3218 :     This update contains a rewritten (and hopefully more correct) module
3219 :     for extracting aliasing information from CPS.
3220 :    
3221 :     To turn on this feature:
3222 :    
3223 :     Compiler.Control.CG.memDisambiguate := true
3224 :    
3225 :     To pretty print the region information with assembly
3226 :    
3227 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
3228 :    
3229 :     To control how many levels of aliasing information are printed, use:
3230 :    
3231 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
3232 :    
3233 :     The default of n is 3.
3234 :    
3235 :     ----------------------------------------------------------------------
3236 : dbm 589 Name: David MacQueen
3237 :     Date: 2000/03/31 11:15:00 EST
3238 :     Tag: dbm-20000331-runtime_fix
3239 :     Description:
3240 :    
3241 :     This update contains:
3242 :    
3243 :     1. runtime/c-lib/c-libraries.c
3244 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
3245 :    
3246 :     2. fix for bug 1556
3247 :     system/Basis/Implementation/NJ/internal-signals.sml
3248 :    
3249 :     ----------------------------------------------------------------------
3250 : blume 587 Name: Matthias Blume
3251 : blume 588 Date: 2000/03/31 18:00:00 JST
3252 :     Tag: blume_main_v110p26p2_1
3253 :     Description:
3254 :    
3255 :     This update contains:
3256 :    
3257 :     1. A small change to CM's handling of stable libraries:
3258 :     CM now maintains one "global" modmap that is used for all stable
3259 :     libraries. The use of such a global modmap maximizes sharing and
3260 :     minimizes the need for re-traversing parts of environments during
3261 :     modmap construction. (However, this has minor impact since modmap
3262 :     construction seems to account for just one percent or less of total
3263 :     compile time.)
3264 :    
3265 :     2. I added a "genmap" phase to the statistics. This is where I got the
3266 :     "one percent" number (see above).
3267 :    
3268 :     3. CM's new tool parameter mechanism just became _even_ better. :)
3269 :     - The parser understands named parameters and recursive options.
3270 :     - The "make" and "shell" tools use these new features.
3271 :     (This makes it a lot easier to cascade these tools.)
3272 :     - There is a small syntax change: named parameters use a
3273 :    
3274 :     <name> : ( <option> ... ) or
3275 :     <name> : <string>
3276 :    
3277 :     syntax. Previously, named parameters were implemented in an
3278 :     ad-hoc fashion by each tool individually (by parsing strings)
3279 :     and had the form
3280 :    
3281 :     <name>=<string>
3282 :    
3283 :     See the CM manual for a full description of these issues.
3284 :    
3285 :     ----------------------------------------------------------------------
3286 :     Name: Matthias Blume
3287 : blume 587 Date: 2000/03/30 18:00:00 JST
3288 :     Tag: blume_main_v110p26p2_0
3289 :     Description:
3290 :    
3291 :     !!!!! WARNING !!!!!!
3292 :     !! New binfiles !!
3293 :     !!!!!!!!!!!!!!!!!!!!
3294 :    
3295 :     This update contains:
3296 :    
3297 :     1. Moderate changes to CM:
3298 :    
3299 :     - Changes to CM's tools mechanism. In particular, it is now possible
3300 :     to have tools that accept additional "command line" parameters
3301 :     (specified in the .cm file at each instance where the tool's class is
3302 :     used).
3303 :    
3304 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
3305 :     facilitate fairly seamless hookup to portions of code managed using
3306 : blume 587 Makefiles or Shell scripts.
3307 :    
3308 :     There are no classes "shared" or "private" anymore. Instead, the
3309 :     sharing annotation is now a parameter to the "sml" class.
3310 :    
3311 :     There is a bit of generic machinery for implementing one's own
3312 :     tools that accept command-line parameters. However, I am not yet fully
3313 :     satisfied with that part, so expect changes here in the future.
3314 :    
3315 :     All existing tools are described in the CM manual.
3316 :    
3317 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
3318 : blume 587 error messages that tended to be more annoying than helpful.)
3319 :    
3320 :     2. Major changes to the compiler's static environment data structures.
3321 :    
3322 :     - no CMStaticEnv anymore.
3323 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
3324 :     but it is called Environment), no conversions between different
3325 :     kinds of static environments
3326 :    
3327 :     - There is still a notion of a "modmap", but such modmaps are generated
3328 :     on demand at the time when they are needed. This sounds slow, but I
3329 :     sped up the code that generates modmaps enough for this not to lead to
3330 :     a slowdown of the compiler (at least I didn't detect any).
3331 :    
3332 :     - To facilitate rapid modmap generation, static environments now
3333 :     contain an (optional) "modtree" structure. Modtree annotations are
3334 :     constructed by the unpickler during unpickling. (This means that
3335 :     the elaborator does not have to worry about modtrees at all.)
3336 :     Modtrees have the advantage that they are compositional in the same
3337 :     way as the environment data structure itself is compositional.
3338 :     As a result, modtrees never hang on to parts of an environment that
3339 :     has already been rendered "stale" by filtering or rebinding.
3340 :    
3341 :     - I went through many, many trials and errors before arriving at the
3342 :     current solution. (The initial idea of "linkpaths" did not work.)
3343 :     But the result of all this is that I have touched a lot of files that
3344 :     depend on the "modules" and "types" data structures (most of the
3345 :     elaborator). There were a lot of changes during my "linkpath" trials
3346 :     that could have been reverted to their original state but weren't.
3347 :     Please, don't be too harsh on me for messing with this code a bit more
3348 : monnier 921 than what was strictly necessary... (I _did_ resist the temptation
3349 : blume 587 of doing any "global reformatting" to avoid an untimely death at
3350 :     Dave's hands. :)
3351 :    
3352 :     - One positive aspect of the previous point: At least I made sure that
3353 :     all files that I touched now compile without warnings (other than
3354 :     "polyEqual").
3355 :    
3356 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
3357 :     redundant modmaps)
3358 :    
3359 :     ----------------------------------------------------------------------
3360 : leunga 580 Name: Allen Leung
3361 : leunga 585 Date: 2000/03/29 18:00:00
3362 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
3363 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
3364 : leunga 585 Description:
3365 :    
3366 :     This update contains *MAJOR* changes to the way code is generated from CPS
3367 :     in the module mlriscGen, and in various backend modules.
3368 :    
3369 :     CHANGES
3370 :     =======
3371 :    
3372 :     1. MLRiscGen: forward propagation fix.
3373 :    
3374 :     There was a bug in forward propagation introduced at about the same time
3375 :     as the MLRISC x86 backend, which prohibits coalescing to be
3376 :     performed effectively in loops.
3377 :    
3378 :     Effect: speed up of loops in RISC architectures.
3379 :     By itself, this actually slowed down certain benchmarks on the x86.
3380 :    
3381 :     2. MLRiscGen: forward propagating addresses from consing.
3382 :    
3383 :     I've changed the way consing code is generated. Basically I separated
3384 :     out the initialization part:
3385 :    
3386 :     store tag, offset(allocptr)
3387 :     store elem1, offset+4(allocptr)
3388 :     store elem2, offset+8(allocptr)
3389 :     ...
3390 :     store elemn, offset+4n(allocptr)
3391 :    
3392 :     and the address computation part:
3393 :    
3394 :     celladdr <- offset+4+alloctpr
3395 :    
3396 :     and move the address computation part
3397 :    
3398 :     Effect: register pressure is generally lower as a result. This
3399 :     makes compilation of certain expressions much faster, such as
3400 :     long lists with non-trivial elements.
3401 :    
3402 :     [(0,0), (0,0), .... (0,0)]
3403 :    
3404 :     3. MLRiscGen: base pointer elimination.
3405 :    
3406 :     As part of the linkage mechanism, we generate the sequence:
3407 :    
3408 :     L: ... <- start of the code fragment
3409 :    
3410 :     L1:
3411 :     base pointer <- linkreg - L1 + L
3412 :    
3413 :     The base pointer was then used for computing relocatable addresses
3414 :     in the code fragment. Frequently (such as in lots of continuations)
3415 :     this is not needed. We now eliminate this sequence whenever possible.
3416 :    
3417 :     For compile time efficiency, I'm using a very stupid local heuristic.
3418 :     But in general, this should be done as a control flow analysis.
3419 :    
3420 :     Effect: Smaller code size. Speed up of most programs.
3421 :    
3422 :     4. Hppa back end
3423 :    
3424 :     Long jumps in span dependence resolution used to depend on the existence
3425 :     of the base pointer.
3426 :    
3427 :     A jump to a long label L was expanded into the following sequence:
3428 :    
3429 :     LDIL %hi(L-8192), %r29
3430 :     LDO %lo(L-8192)(%r29), %r29
3431 :     ADD %r29, baseptr, %r29
3432 :     BV,n %r0(%r29)
3433 :    
3434 :     In the presence of change (3) above, this will not work. I've changed
3435 :     it so that the following sequence of instructions are generated, which
3436 :     doesn't mention the base pointer at all:
3437 :    
3438 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
3439 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
3440 :     BV,n %r0(%r29) /* Jump */
3441 :    
3442 :     5. Alpha back end
3443 :    
3444 :     New alpha instructions LDB/LDW have been added, as per Fermin's
3445 :     suggestions. This is unrelated to all other changes.
3446 :    
3447 :     6. X86 back end
3448 :    
3449 :     I've changed andl to testl in the floating point test sequence
3450 :     whenever appropriate. The Intel optimization guide states that
3451 : monnier 659 testl is preferable to andl.
3452 : leunga 585
3453 :     7. RA (x86 only)
3454 :    
3455 :     I've improved the spill propagation algorithm, using an approximation
3456 :     of maximal weighted independent sets. This seems to be necessary to
3457 :     alleviate the negative effect in light of the slow down in (1).
3458 :    
3459 :     I'll write down the algorithm one of these days.
3460 :    
3461 :     8. MLRiscGen: frequencies
3462 :    
3463 :     I've added an annotation that states that all call gc blocks have zero
3464 :     execution frequencies. This improves register allocation on the x86.
3465 :    
3466 :     BENCHMARKS
3467 :     ==========
3468 :    
3469 :     I've only perform the comparison on 110.25.
3470 :    
3471 :     The platforms are:
3472 :    
3473 :     HPPA A four processor HP machine (E9000) with 5G of memory.
3474 :     X86 A 300Hhz Pentium II with 128M of memory, and
3475 :     SPARC An Ultra sparc 2 with 512M of memory.
3476 :    
3477 :     I used the following parameters for the SML benchmarks:
3478 :    
3479 :     @SMLalloc
3480 :     HPPA 256k
3481 :     SPARC 512k
3482 :     X86 256k
3483 :    
3484 :     COMPILATION TIME
3485 :     ----------------
3486 :     Here are the numbers comparing the compilation times of the compilers.
3487 :     I've only compared 110.25 compiling the new sources versus
3488 :     a fixpoint version of the new compiler compiling the same.
3489 :    
3490 :     110.25 New
3491 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
3492 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
3493 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
3494 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
3495 :    
3496 :     110.25 New
3497 :     Code Size Code Size
3498 :     HPPA 8596736 8561421
3499 :     SPARC 8974299 8785143
3500 :     X86 9029180 8716783
3501 :    
3502 :     So in summary, things are at least as good as before. Dramatic
3503 :     reduction in compilation is obtained on the Sparc; I can't explain it,
3504 :     but it is reproducible. Perhaps someone should try to reproduce this
3505 :     on their own machines.
3506 :    
3507 :     SML BENCHMARKS
3508 :     --------------
3509 :    
3510 :     On the average, all benchmarks perform at least as well as before.
3511 :    
3512 :     HPPA Compilation Time Spill+Reload Run Time
3513 :     110.25 New 110.25 New 110.25 New
3514 :    
3515 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
3516 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
3517 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
3518 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
3519 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
3520 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
3521 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
3522 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
3523 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
3524 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
3525 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
3526 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
3527 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
3528 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
3529 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
3530 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
3531 :     --------------------------------------------------------------------------
3532 :     Average 6.12% 4.09%
3533 :    
3534 :     SPARC Compilation Time Spill+Reload Run Time
3535 :     110.25 New 110.25 New 110.25 New
3536 :    
3537 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
3538 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
3539 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
3540 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
3541 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
3542 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
3543 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
3544 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
3545 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
3546 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
3547 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
3548 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
3549 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
3550 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
3551 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
3552 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
3553 :     ----------------------------------------------------------------------------
3554 :     Average 6.94% 2.64%
3555 :    
3556 :     X86 Compilation Time Spill+Reload Run Time
3557 :     110.25 New 110.25 New 110.25 New
3558 :    
3559 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
3560 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
3561 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
3562 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
3563 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
3564 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
3565 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
3566 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
3567 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
3568 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
3569 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
3570 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
3571 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
3572 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
3573 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
3574 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
3575 :     ----------------------------------------------------------------------------
3576 :     Average 1.22% 3.36%
3577 :    
3578 :     ----------------------------------------------------------------------
3579 :     Name: Allen Leung
3580 : leunga 583 Date: 2000/03/23 16:25:00
3581 :     Tag: leunga-20000323-fix_x86_alpha
3582 :     Description:
3583 :    
3584 :     1. X86 fixes/changes
3585 :    
3586 :     a. The old code generated for SETcc was completely wrong.
3587 :     The Intel optimization guide is VERY misleading.
3588 :    
3589 :     2. ALPHA fixes/changes
3590 :    
3591 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
3592 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
3593 :     c. Added reassociation code for address computation.
3594 :    
3595 :     ----------------------------------------------------------------------
3596 :     Name: Allen Leung
3597 : leunga 580 Date: 2000/03/22 01:23:00
3598 :     Tag: leunga-20000322-fix_x86_hppa_ra
3599 :     Description:
3600 :    
3601 :     1. X86 fixes/changes
3602 :    
3603 :     a. x86Rewrite bug with MUL3 (found by Lal)
3604 :     b. Added the instructions FSTS, FSTL
3605 :    
3606 :     2. PA-RISC fixes/changes
3607 :    
3608 :     a. B label should not be a delay slot candidate! Why did this work?
3609 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
3610 :     as it should be.
3611 :     c. The assembly syntax for fstds and fstdd was wrong.
3612 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
3613 :     operand variant of COMCLR/LDO.
3614 :    
3615 :     3. Generic MLRISC
3616 :    
3617 :     a. shuffle.sml rewritten to be slightly more efficient
3618 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
3619 :    
3620 :     4. Register Allocator
3621 :    
3622 :     a. I now release the interference graph earlier during spilling.
3623 :     May improve memory usage.
3624 :    
3625 :     ----------------------------------------------------------------------
3626 : blume 577 Name: Matthias Blume
3627 : blume 578 Date: 2000/03/14 14:15:32
3628 :     Tag: blume_main_v110p26p1_2
3629 :     Description:
3630 :    
3631 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
3632 :     additional argument called "template" which is an optional string that
3633 : monnier 659 specifies the layout of the tool command line. See the CM manual for
3634 : blume 578 explanation.
3635 :    
3636 : monnier 659 2. A special-purpose tool can be "registered" by simply dropping the
3637 : blume 578 corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
3638 :     directory where the .cm file lives that uses this tool. (The
3639 :     behavior/misfeature until now was to look for the tool description
3640 :     files in the current working directory.) As before, tool description
3641 :     files could also be anchored -- in which case they can live anywhere
3642 :     they like. Following the recent e-mail discussion, this change should
3643 :     make it easier to have special-purpose tools that are shipped together
3644 :     with the sources of the program that uses them.
3645 :    
3646 :     ----------------------------------------------------------------------
3647 :     Name: Matthias Blume
3648 : blume 577 Date: 2000/03/10 07:48:34
3649 :     Tag: blume_main_v110p26p1_1
3650 :     Description:
3651 :    
3652 :     I added a re-written version of Dave's fixpt script to src/system.
3653 :     Changes relative to the original version:
3654 :     - sh-ified (not everybody has ksh)
3655 :     - automatically figures out which architecture it runs on
3656 :     - uses ./makeml a bit more cleverly
3657 :     - never invokes ./installml (and, thus, does not clobber your
3658 :     good and working installation of sml in case something goes wrong)
3659 :     - accepts max iteration count using option "-iter <n>"
3660 :     - accepts a "base" name using option "-base <base>"
3661 :    
3662 :     It does not build any extraneous heap images but directly rebuilds
3663 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
3664 :     it can incorporate existing bin- and boot- hierarchies. For example,
3665 :     suppose the base is set to "sml" (which is the default). Then it
3666 :     successively builds
3667 :    
3668 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
3669 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
3670 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
3671 :     ...
3672 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
3673 :    
3674 :     and so on. If any of these already exist, it will just use what's
3675 :     there. In particular, many people will have the initial set of bin
3676 :     and boot files around, so this saves time for at least one full
3677 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
3678 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
3679 :     they might be the result of an earlier partial run of fixpt (which
3680 : monnier 659 perhaps got accidentally killed). In this case, fixpt will quickly
3681 : blume 577 move through what exists before continuing where it left off earlier,
3682 :     and, thus, saves a lot of time.
3683 :    
3684 :     ----------------------------------------------------------------------
3685 : leunga 576 Name: Allen Leung
3686 :     Date: 00/03/10 02:20:00
3687 :     Tag: leunga-20000310-fix_x86_asm_ra
3688 :     Description:
3689 : dbm 570
3690 : leunga 576 More assembly output problems involving the indexed addressing mode
3691 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
3692 :     fix.
3693 :    
3694 :     The interface and implementation of the register allocator have been changed
3695 :     slightly to accommodate the possibility to skip the register allocation
3696 :     phases completely and go directly to memory allocation. This is needed
3697 :     for C-- use.
3698 :    
3699 : dbm 570 ----------------------------------------------------------------------
3700 : blume 572 Name: Matthias Blume
3701 : blume 575 Date: 00/03/09 10:23:53
3702 :     Tag: blume_main_v110p26p1_0
3703 :     Description:
3704 :    
3705 :     * Complete re-organization of library names. Many libraries have been
3706 :     consolidated so that they share the same path anchor. For example,
3707 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
3708 :     are SML/NJ-specific are under "smlnj". Notice that names like
3709 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
3710 :     for a complete description of the new naming scheme. Quick reference:
3711 :    
3712 :     host-cmb.cm -> smlnj/cmb.cm
3713 :     host-compiler.cm -> smlnj/compiler.cm
3714 :     full-cm.cm -> smlnj/cm.cm
3715 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
3716 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
3717 :    
3718 :     * Bug fixes in CM.
3719 :     - exceptions in user code are being passed through (i.e., reach top level)
3720 :     - more bugs in paranoia mode fixed
3721 :     - bug related to checking group owners fixed
3722 :    
3723 :     * New install.sh script that automagically fetches archive files:
3724 :     The new file config/srcarchiveurl must contain the URL of the
3725 :     (remote) directory that contains bin files (or other source archives).
3726 :     If install.sh does not find the archive locally, it tries to get
3727 :     it from that remote directory.
3728 :     This should simplify installation further: For machines that have
3729 :     access to the internet, just fetch <version>-config.tgz, unpack it,
3730 : monnier 659 edit config/targets, and go (run config/install.sh). The script will
3731 : blume 575 fetch everything else that it might need all by itself.
3732 :    
3733 :     For CVS users, this mechanism is not relevant for source archives, but
3734 :     it is convenient for getting new sets of binfiles.
3735 :    
3736 :     Archives should be tar files compressed with either gzip, compress, or
3737 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
3738 :    
3739 :     ----------------------------------------------------------------------
3740 :     Name: Matthias Blume
3741 : blume 572 Date: 2000/03/07 04:01:04
3742 :     Tag: blume_main_v110_26_2
3743 : dbm 570 Description:
3744 : blume 572 - size info in BOOTLIST
3745 :     * no fixed upper limits for number of bootfiles or length of
3746 :     bootfile names in runtime
3747 :     * falling back to old behavior if no BOOTLIST size info found
3748 :     - allocation size heuristics in .run-sml
3749 :     * tries to read cache size from /proc/cpuinfo (this is important for
3750 :     small-cache Celeron systems!)
3751 :     - install.sh robustified
3752 :     - CM manual updates
3753 :     - paranoid mode
3754 :     * no more CMB.deliver() (i.e., all done by CMB.make())
3755 :     * can re-use existing sml.boot.* files
3756 :     * init.cmi now treated as library
3757 :     * library stamps for consistency checks
3758 :     - sml.boot.<arch>-<os>/PIDMAP file
3759 :     * This file is read by the CM startup code. This is used to minimize
3760 :     the amount of dynamic state that needs to be stowed away for the
3761 :     purpose of sharing between interactive system and user code.
3762 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
3763 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
3764 :     like most other CM state variables.
3765 :     - Compiler.CMSA eliminated
3766 :     * No longer supported by CM anyway.
3767 :     - fixed bugs in pickler that kept biting Stefan
3768 :     * past refs to past refs (was caused by the possibility that
3769 :     ad-hoc sharing is more discriminating than hash-cons sharing)
3770 :     * integer overflow on LargeInt.minInt
3771 :     - ml-{lex,yacc} build scripts now use new mechanism
3772 :     for building standalone programs
3773 :     - fixed several gcc -Wall warnings that were caused by missing header
3774 :     files, missing initializations, etc., in runtime (not all warnings
3775 :     eliminated, though)

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