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/HISTORY
ViewVC logotype

Annotation of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1282 - (view) (download)

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 : leunga 1142
15 : george 1136 ----------------------------------------------------------------------
16 : blume 1229 Name: Matthias Blume
17 : blume 1282 Date: 2002/10/17 09:10:00 EDT
18 : blume 1279 Tag: Release_110_42
19 :     Description:
20 :    
21 : blume 1282 In good old tradition, there has been a slight hiccup so that we have
22 :     to patch 110.42 after the fact. The old release tag has been replaced
23 :     (see below).
24 :    
25 :     The change solves a problem with two competing approaches the
26 :     configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
27 :     each other's way.
28 :    
29 :     This change only affects the runtime system code and the installer script.
30 :     (No new bootfiles.)
31 :    
32 :     ----------------------------------------------------------------------
33 :     Name: Matthias Blume
34 :     Date: 2002/10/16 12:00:00 EDT
35 :     Tag: Release_110_42_removed
36 :     Description:
37 :    
38 : blume 1279 New working release. New bootfiles.
39 :    
40 :     ----------------------------------------------------------------------
41 :     Name: Matthias Blume
42 : blume 1273 Date: 2002/10/10 13:10:00 EDT
43 :     Tag: blume-20021010-ppc-divs
44 :     Description:
45 :    
46 :     The mltree operator DIVS must be implemented with an overflow check on
47 :     the PPC because the hardware indicates divide-by-zero using "overflow" as
48 :     well.
49 :    
50 :     ----------------------------------------------------------------------
51 :     Name: Matthias Blume
52 : blume 1261 Date: 2002/07/23 11:45:00
53 :     Tag: blume-20020723-smlnj-home
54 :     Description:
55 :    
56 :     Sml now senses the SMLNJ_HOME environment variable. If this is set,
57 :     then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
58 :     CM_PATHCONFIG is also set), the path configuration file is assumed
59 :     to be in $SMLNJ_HOME/lib/pathconfig. This way one can easily move
60 :     the entire tree to some other place and everything will "just work".
61 :    
62 :     (Companion commands such as ml-build and ml-makedepend also sense this
63 :     variable.)
64 :    
65 :     ----------------------------------------------------------------------
66 :     Name: Matthias Blume
67 : blume 1259 Date: 2002/07/12 21:19:00 EDT
68 :     Tag: blume-20020712-liveness
69 :     Description:
70 :    
71 :     Exported two useful "step" functions from liveness module (MLRISC).
72 :    
73 :     ----------------------------------------------------------------------
74 :     Name: Matthias Blume
75 : blume 1253 Date: 2002/07/05 16:00 EDT
76 :     Tag: Release_110_41
77 :     Description:
78 :    
79 :     New working release. New bootfiles.
80 :    
81 :     ----------------------------------------------------------------------
82 :     Name: Matthias Blume
83 : blume 1250 Date: 2002/07/05 10:25:00 EDT
84 :     Tag: blume-20020705-btimp
85 :     Description:
86 :    
87 :     Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
88 :     other clients can set up backtracing support.
89 :    
90 :     ----------------------------------------------------------------------
91 :     Name: Matthias Blume
92 : blume 1249 Date: 2002/06/25 17:23:00 EDT
93 :     Tag: blume-20020625-fpmax
94 :     Description:
95 :    
96 :     Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
97 :     case. (The sense of the isNaN test was reversed -- which made min and
98 :     max always return their first argument.)
99 :    
100 :     ----------------------------------------------------------------------
101 :     Name: Matthias Blume
102 : blume 1240 Date: 2002/06/11
103 :     Tag: blume-20020611-unixpath
104 :     Description:
105 :    
106 :     Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
107 :    
108 :     ----------------------------------------------------------------------
109 :     Name: Matthias Blume
110 : blume 1238 Date: 2002/06/10 16:35:00 EDT
111 :     Tag: blume-20020610-ieeereal
112 :     Description:
113 :    
114 :     I back-ported my implementation of IEEEReal.fromString from the
115 :     idlbasis-devel branch so that we can test it.
116 :    
117 :     Another small change is that ppDec tries to give more information
118 :     than just "<sig>" in the case of functors. However, this code is
119 :     broken in some mysterious way if the functor's body's signature
120 :     has not been declared by ascription but gets inferred from the
121 :     implementation. This needs fixing...
122 :    
123 :     ----------------------------------------------------------------------
124 :     Name: Matthias Blume
125 : blume 1229 Date: 2002/05/31
126 :     Tag: blume-20020531-btrace-mode
127 :     Description:
128 :    
129 :     Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by
130 :     the wayside when I switched over to using Controls everywhere.)
131 :    
132 :     ----------------------------------------------------------------------
133 : george 1222 Name: Lal George
134 :     Date: 2002/05/23 12:21:40 EDT
135 :     Tag: george-20020523-visual-labels
136 :     Description:
137 :    
138 :     Labels are now displayed in the graphical output to make
139 :     the fall-through and target blocks obvious.
140 :    
141 :     ----------------------------------------------------------------------
142 : blume 1204 Name: Matthias Blume
143 : blume 1220 Date: 2002/05/22 11:03:00 EDT
144 :     Tag: blume-20020522-shrink
145 :     Description:
146 :    
147 :     John tweaked yesterday's fix for 1131 to handle an out-of-memory
148 :     situation that comes up when allocating huge arrays.
149 :    
150 :     ----------------------------------------------------------------------
151 :     Name: Matthias Blume
152 : blume 1215 Date: 2002/05/21 16:00:00 EDT
153 :     Tag: Release_110_40
154 :     Description:
155 :    
156 :     New working release (110.40). New bootfiles.
157 :    
158 :     [Also: John Reppy fixed GC bug 1131.]
159 :    
160 :     ----------------------------------------------------------------------
161 :     Name: Matthias Blume
162 : blume 1212 Date: 2002/05/21 12:35:00 EDT
163 :     Tag: blume-20020521-cmdoc
164 :     Description:
165 :    
166 :     CM documentation update.
167 :    
168 :     ----------------------------------------------------------------------
169 :     Name: Matthias Blume
170 : blume 1211 Date: 2002/05/21 10:55:00 EDT
171 :     Tag: blume-20020521-misc
172 :     Description:
173 :    
174 :     - John tweaked runtime to be silent on heap export (except when
175 :     GC messages are on).
176 :     - I added a few more things (cross-compiling versions of CMB) to
177 :     config/preloads (as suggestions).
178 :    
179 :     ----------------------------------------------------------------------
180 :     Name: Matthias Blume
181 : blume 1208 Date: 2002/05/20 22:25:00 EDT
182 :     Tag: blume-20020520-controls
183 :     Description:
184 :    
185 :     - Added ControlUtil structure to control-lib.cm.
186 :     - Use it throughout.
187 :     - Used Controls facility to define MLRISC controls (as opposed to
188 :     registering MLRISC control ref cells with Controls after the fact)
189 :     - Fixed messed-up controls priorities.
190 :    
191 :     * Removed again all the stuff from config/preloads that one wouldn't
192 :     be able to preload at the time the initial heap image is built.
193 :     (Many libraries, e.g., CML, do not exist yet at this time. The
194 :     only libraries that can be preloaded via config/preloads are those
195 :     that come bundled with the bootfiles.)
196 :    
197 :     ----------------------------------------------------------------------
198 :     Name: Matthias Blume
199 : blume 1204 Date: 2002/05/20 10:59:00 EDT
200 :     Tag: blume-20020520-preloads
201 :     Description:
202 :    
203 :     Added a lot of commented-out suggestions for things to be included
204 :     in config/preloads.
205 :    
206 :     ----------------------------------------------------------------------
207 : leunga 1203 Name: Allen Leung
208 :     Date: 2002/05/18 14:20:00 EDT
209 :     Tag: leunga-20020518-mdl
210 :     Description:
211 :    
212 :     o Made the mdl tool stuff compile and run again.
213 :     o I've disabled all the stuff that depends on RTL specifications; they
214 :     are all badly broken anyway.
215 :    
216 :     ----------------------------------------------------------------------
217 : blume 1201 Name: Matthias Blume
218 :     Date: 2002/05/17 16:49:00 EDT
219 :     Tag: blume-20020517-controls
220 :     Description:
221 :    
222 :     0. John Reppy made several modifications to the SML/NJ library.
223 :     In particular, there is a shiny new controls-lib.cm.
224 :    
225 :     1. Pushed new controls interface through compiler so that everything
226 :     compiles again.
227 :    
228 :     2. Added FormatComb and FORMAT_COMB to the CML version of the
229 :     SML/NJ library (so that CML compiles again).
230 :    
231 :     3. Modified init scripts because XXX_DEFAULT environment variables
232 :     are no longer with us. (Boot-time initialization is now done
233 :     using the same environment variables that are also used for
234 :     startup-time initialization of controls.)
235 :    
236 :     ----------------------------------------------------------------------
237 : george 1192 Name: Lal George
238 :     Date: 2002/05/15 09:20:10 EDT
239 :     Tag: george-20020515-pseudo-op-decls
240 :     Description:
241 :    
242 :     All pseudo-ops emitted before the first segment declaration
243 :     such as TEXT, DATA, and BSS directives are assumed to be global
244 :     declarations and are emitted first in the assembly file. This is
245 :     useful in a number of situations where one has pseudo-ops that are not
246 :     specific to any segment, and also works around the constraint that one
247 :     cannot have client pseudo-ops in the TEXT segment.
248 :    
249 :     Because no segment is associated with these declarations it is
250 :     an error to allocate any space or objects before the first segment
251 :     directive and an exception will be raised. However, we cannot make
252 :     this check for client pseudo-ops.
253 :    
254 :     These top level declarations are a field in the CFG graph_info.
255 :     In theory you can continue to add to this field after the CFG has been
256 :     built -- provided you know what you are doing;-)
257 :    
258 :     ----------------------------------------------------------------------
259 : blume 1175 Name: Matthias Blume
260 : blume 1190 Date: 2002/05/13 16:40:00 EDT
261 :     Tag: blume-20020513-pp-etc
262 :     Description:
263 :    
264 :     A few minor bugfixes:
265 :    
266 :     - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
267 :     (Bogus printouts for redefined bindings still occur. Compiler
268 :     bug should no longer occur now. We need to redo the prettyprinter
269 :     from scratch.)
270 :    
271 :     - CM pathname printer now also adds escape sequences for ( and )
272 :    
273 :     - commend and docu fixes for ml-nlffi
274 :    
275 :     ----------------------------------------------------------------------
276 :     Name: Matthias Blume
277 : blume 1189 Date: 2002/05/10 16:40:00 EDT
278 :     Tag: blume-20020510-erg-textio
279 :     Description:
280 :    
281 :     Applied the following bugfix provided by Emden Gansner:
282 :    
283 :     Output is corrupted when outputSubstr is used rather than output.
284 :    
285 :     The problem occurs when a substring
286 :    
287 :     ss = (s, dataStart, dataLen)
288 :    
289 :     where dataStart > 0, fills a stream buffer with avail bytes left.
290 :     avail bytes of s, starting at index dataStart, are copied into the
291 :     buffer, the buffer is flushed, and then the remaining dataLen-avail
292 :     bytes of ss are copied into the beginning of the buffer. Instead of
293 :     starting this copy at index dataStart+avail in s, the current code
294 :     starts the copy at index avail.
295 :    
296 :     Fix:
297 :     In text-io-fn.sml, change line 695 from
298 :     val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
299 :     to
300 :     val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
301 :    
302 :     ----------------------------------------------------------------------
303 :     Name: Matthias Blume
304 : blume 1186 Date: 2002/04/12 13:55:00 EDT
305 :     Tag: blume-20020412-assyntax
306 :     Description:
307 :    
308 :     1. Grabbed newer assyntax.h from the XFree86 project.
309 :     2. Fiddled with how to compile X86.prim.asm without warnings.
310 :     3. (Very) Minor cleanup in CM.
311 :    
312 :     ----------------------------------------------------------------------
313 :     Name: Matthias Blume
314 : blume 1185 Date: 2002/04/01 (no joke!) 17:07:00 EST
315 :     Tag: blume-20020401-x86div
316 :     Description:
317 :    
318 :     Added full support for div/mod/rem/quot on the x86, using the machine
319 :     instruction's two results (without clumsily recomputing the remainder)
320 :     directly where appropriate.
321 :    
322 :     Some more extensive power-of-two support was added to the x86 instruction
323 :     selector (avoiding expensive divs, mods, and muls where they can be
324 :     replaced with cheaper shifts and masks). However, this sort of thing
325 :     ought to be done earlier, e.g., within the CPS optimizer so that
326 :     all architectures benefit from it.
327 :    
328 :     The compiler compiles to a fixed point, but changes might be somewhat
329 :     fragile nevertheless. Please, report any strange things that you might
330 :     see wrt. div/mod/quot/rem...
331 :    
332 :     ----------------------------------------------------------------------
333 :     Name: Matthias Blume
334 : blume 1184 Date: 2002/03/29 17:22:00
335 :     Tag: blume-20020329-div
336 :     Description:
337 :    
338 :     Fixed my broken div/mod logic. Unfortunately, this means that the
339 :     inline code for div/mod now has one more comparison than before.
340 :     Fast paths (quotient > 0 or remainder = 0) are not affected, though.
341 :     The problem was with quotient = 0, because that alone does not tell
342 :     us which way the rounding went. One then has to look at whether
343 :     remainder and divisor have the same sign... :(
344 :    
345 :     Anyway, I replaced the bootfiles with fresh ones...
346 :    
347 :     ----------------------------------------------------------------------
348 :     Name: Matthias Blume
349 : blume 1183 Date: 2002/03/29 14:10:00 EST
350 :     Tag: blume-20020329-inlprims
351 :     Description:
352 :    
353 :     NEW BOOTFILES!!! Version number bumped to 110.39.3.
354 :    
355 :     Primops have changed. This means that the bin/boot-file formats have
356 :     changed as well.
357 :    
358 :     To make sure that there is no confusion, I made a new version.
359 :    
360 :    
361 :     CHANGES:
362 :    
363 :     * removed REMT from mltree (remainder should never overflow).
364 :    
365 :     * added primops to deal with divisions of all flavors to the frontend
366 :    
367 :     * handled these primops all the way through so they map to their respective
368 :     MLRISC support
369 :    
370 :     * used these primops in the implementation of Int, Int32, Word, Word32
371 :    
372 :     * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
373 :    
374 :     * parameterized INLMIN, INLMAX, and INLABS by a numkind
375 :    
376 :     * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
377 :     floating point
378 :    
379 :     * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
380 :     and Real (but Real.abs maps to a separate floating-point-only primop)
381 :    
382 :    
383 :     TODO items:
384 :    
385 :     * Hacked Alpha32 instruction selection, disabling the selection of REMx
386 :     instructions because the machine instruction encoder cannot handle
387 :     them. (Hppa, PPC, and Sparc instruction selection did not handle
388 :     REM in the first place, and REM is supported by the x86 machine coder.)
389 :    
390 :     * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
391 :     selection phase. (The two can be streamlined because the hardware
392 :     delivers both quotient and remainder at the same time anyway.)
393 :    
394 :     * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
395 :     (Currently the behavior is inconsistent both across architectures and
396 :     wrt. the draft Basis spec.)
397 :    
398 :     * Word8 should eventually be handled natively, too.
399 :    
400 :     * There seems to be one serious bug in mltree-gen.sml. It appears, though,
401 :     as if there currently is no execution path that could trigger it in
402 :     SML/NJ. (The assumptions underlying functions arith and promotable do not
403 :     hold for things like multiplication and division.)
404 :    
405 :     ----------------------------------------------------------------------
406 :     Name: Matthias Blume
407 : blume 1181 Date: 2002/03/27 16:27:00 EST
408 :     Tag: blume-20020327-mlrisc-divisions
409 :     Description:
410 :    
411 :     Added support for all four division operations (ML's div, mod, quot,
412 :     and rem) to MLRISC. In the course of doing so, I also rationalized
413 :     the naming (no more annoying switch-around of DIV and QUOT), by
414 :     parameterizing the operation by div_rounding_mode (which can be either
415 :     DIV_TO_ZERO or DIV_TO_NEGINF).
416 :    
417 :     The generic MLTreeGen functor takes care of compiling all four
418 :     operations down to only round-to-zero div.
419 :    
420 :     Missing pieces:
421 :    
422 :     * Doing something smarter than relying on MLTreeGen on architectures
423 :     like, e.g., the x86 where hardware division delivers both quotient and
424 :     remainder at the same time. With this, the implementation of the
425 :     round-to-neginf operations could be further streamlined.
426 :    
427 :     * Remove inlining support for div/mod/rem from the frontend and replace it
428 :     with primops that get carried through to the backend. Do this for all
429 :     int and word types.
430 :    
431 :     ----------------------------------------------------------------------
432 :     Name: Matthias Blume
433 : blume 1180 Date: 2002/03/25 17:25:00 EST
434 :     Tag: blume-20020325-divmod
435 :     Description:
436 :    
437 :     I improved (hopefully without breaking them) the implementation of Int.div,
438 :     Int.mod, and Int.rem. For this, the code in translate.sml now takes
439 :     advantage of the following observations:
440 :    
441 :     Let q = x quot y r = x rem y
442 :     d = x div y m = x mod y
443 :    
444 :     where "quot" is the round-to-zero version of integer division that
445 :     hardware usually provides. Then we have:
446 :    
447 :     r = x - q * y where neither the * nor the - will overflow
448 :     d = if q >= 0 orelse x = q * y then q else q - 1
449 :     where neither the * nor the - will overflow
450 :     m = if q >= 0 orelse r = 0 then r else r + y
451 :     where the + will not overflow
452 :    
453 :     This results in substantial simplification of the generated code.
454 :     The following table shows the number of CFG nodes and edges generated
455 :     for
456 :     fun f (x, y) = x OPER y
457 :     (* with OPER \in div, mod, quot, rem *)
458 :    
459 :    
460 :     OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
461 :     --------------------------------------------------------
462 :     div | 24 | 39 | 12 | 16
463 :     mod | 41 | 71 | 12 | 16
464 :     quot | 8 | 10 | 8 | 10
465 :     rem | 10 | 14 | 8 | 10
466 :    
467 :    
468 :     ----------------------------------------------------------------------
469 :     Name: Matthias Blume
470 : blume 1179 Date: 2002/03/25 22:06:00 EST
471 :     Tag: blume-20020325-cprotobug
472 :     Description:
473 :    
474 :     Fixed a bug in cproto (c prototype decoder).
475 :    
476 :     ----------------------------------------------------------------------
477 :     Name: Matthias Blume
478 : blume 1178 Date: 2002/03/25 16:00:00 EST
479 :     Tag: blume-20020325-raw-primops
480 :     Description:
481 :    
482 :     I did some cleanup to Allen's new primop code and
483 :     replaced yesterday's bootfiles with new ones.
484 :     (But they are stored in the same place.)
485 :    
486 :     ----------------------------------------------------------------------
487 :     Name: Matthias Blume
488 : blume 1175 Date: 2002/03/24 22:40:00 EST
489 :     Tag: blume-20020324-bootfiles
490 :     Description:
491 :    
492 :     Made the bootfiles that Allen asked for.
493 :    
494 :     ----------------------------------------------------------------------
495 : leunga 1174 Name: Allen Leung
496 :     Date: 2002/03/23 15:50:00 EST
497 :     Tag: leunga-20020323-flint-cps-rcc-primops
498 :     Description:
499 :    
500 :     1. Changes to FLINT primops:
501 :    
502 :     (* make a call to a C-function;
503 :     * The primop carries C function prototype information and specifies
504 :     * which of its (ML-) arguments are floating point. C prototype
505 :     * information is for use by the backend, ML information is for
506 :     * use by the CPS converter. *)
507 :     | RAW_CCALL of { c_proto: CTypes.c_proto,
508 :     ml_args: ccall_type list,
509 :     ml_res_opt: ccall_type option,
510 :     reentrant : bool
511 :     } option
512 :     (* Allocate uninitialized storage on the heap.
513 :     * The record is meant to hold short-lived C objects, i.e., they
514 :     * are not ML pointers. With the tag, the representation is
515 :     * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
516 :     *)
517 :     | RAW_RECORD of {tag:bool,sz:int}
518 :     and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
519 :    
520 :     2. These CPS primops are now overloaded:
521 :    
522 :     rawload of {kind:numkind}
523 :     rawstore of {kind:numkind}
524 :    
525 :     The one argument form is:
526 :    
527 :     rawload {kind} address
528 :    
529 :     The two argument form is:
530 :    
531 :     rawload {kind} [ml object, byte-offset]
532 :    
533 :     3. RAW_CCALL/RCC now takes two extra arguments:
534 :    
535 :     a. The first is whether the C call is reentrant, i.e., whether
536 :     ML state should be saved and restored.
537 :     b. The second argument is a string argument specifying the name of
538 :     library and the C function.
539 :    
540 :     These things are currently not handled in the code generator, yet.
541 :    
542 :     4. In CProto,
543 :    
544 :     An encoding type of "bool" means "ml object" and is mapped into
545 :     C prototype of PTR. Note that "bool" is different than "string",
546 :     even though "string" is also mapped into PTR, because "bool"
547 :     is assigned an CPS type of BOGt, while "string" is assigned INT32t.
548 :    
549 :     5. Pickler/unpicker
550 :    
551 :     Changed to handle RAW_RECORD and newest RAW_CCALL
552 :    
553 :     6. MLRiscGen,
554 :    
555 :     1. Changed to handle the new rawload/rawstore/rawrecord operators.
556 :     2. Code for handling C Calls has been moved to a new module CPSCCalls,
557 :     in the file CodeGen/cpscompile/cps-c-calls.sml
558 :    
559 :     7. Added the conditional move operator
560 :    
561 :     condmove of branch
562 :    
563 :     to cps. Generation of this is still buggy so it is currently
564 :     disabled.
565 :    
566 :     ----------------------------------------------------------------------
567 : george 1168 Name: Lal George
568 :     Date: 2002/03/22 14:18:25 EST
569 : george 1171 Tag: george-20020322-cps-branch-prob
570 : george 1168 Description:
571 :    
572 : george 1171 Implemented the Ball-Larus branch prediction-heuristics, and
573 : george 1168 incorporated graphical viewers for control flow graphs.
574 :    
575 : george 1171 Ball-Larus Heuristics:
576 : george 1168 ---------------------
577 :     See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
578 :    
579 :     By design it uses the Dempster-Shafer theory for combining
580 :     probabilities. For example, in the function:
581 :    
582 :     fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
583 :    
584 :     the ball-larus heuristics predicts that the n=0 is unlikely
585 :     (OH-heuristic), and the 'then' branch is unlikely because of the
586 :     RH-heuristic -- giving the 'then' branch an even lower combined
587 : george 1171 probability using the Dempster-Shafer theory.
588 : george 1168
589 :     Finally, John Reppy's loop analysis in MLRISC, further lowers the
590 :     probability of the 'then' branch because of the loop in the else
591 :     branch.
592 :    
593 :    
594 :     Graphical Viewing:
595 :     ------------------
596 :     I merely plugged in Allen's graphical viewers into the compiler. The
597 :     additional code is not much. At the top level, saying:
598 :    
599 :     Control.MLRISC.getFlag "cfg-graphical-view" := true;
600 :    
601 :     will display the graphical view of the control flow graph just before
602 :     back-patching. daVinci must be in your path for this to work. If
603 :     daVinci is not available, then the default viewer can be changed
604 :     using:
605 :    
606 :     Control.MLRISC.getString "viewer"
607 :    
608 :     which can be set to "dot" or "vcg" for the corresponding viewers. Of
609 :     course, these viewers must be in your path.
610 :    
611 :     The above will display the compilation unit at the level of clusters,
612 :     many of which are small, boring, and un-interesting. Also setting:
613 :    
614 :     Control.MLRISC.getInt "cfg-graphical-view_size"
615 :    
616 :     will display clusters that are larger than the value set by the above.
617 :    
618 :    
619 :     ----------------------------------------------------------------------
620 : blume 1159 Name: Matthias Blume
621 :     Date: 2002/03/21 22:20:00 EST
622 :     Tag: blume-20020321-kmp-bugfix
623 :     Description:
624 :    
625 :     Changed the interface to the KMP routine in PreString and fixed
626 :     a minor bug in one place where it was used.
627 :    
628 :     ----------------------------------------------------------------------
629 : leunga 1156 Name: Allen Leung
630 : leunga 1158 Date: 2002/03/21 20:30:00 EST
631 :     Tag: leunga-20020321-cfg
632 :     Description:
633 :    
634 :     Fixed a potential problem in cfg edge splitting.
635 :    
636 :     ----------------------------------------------------------------------
637 :     Name: Allen Leung
638 : leunga 1156 Date: 2002/03/21 17:15:00 EST
639 :     Tag: leunga-20020321-x86-fp-cfg
640 :     Description:
641 :    
642 :     1. Recoded the buggy parts of x86-fp.
643 :    
644 :     a. All the block reordering code has been removed.
645 :     We now depend on the block placement phases to do this work.
646 :    
647 :     b. Critical edge splitting code has been simplified and moved into the
648 :     CFG modules, as where they belong.
649 :    
650 :     Both of these were quite buggy and complex. The code is now much, much
651 :     simpler.
652 :    
653 :     2. X86 backend.
654 :    
655 :     a. Added instructions for 64-bit support. Instruction selection for
656 :     64-bit has not been committed, however, since that
657 :     requires changes to MLTREE which haven't been approved by
658 :     Lal and John.
659 :    
660 :     b. Added support for FUCOMI and FUCOMIP when generating code for
661 :     PentiumPro and above. We only generate these instructions in
662 :     the fast-fp mode.
663 :    
664 :     c. Added cases for JP and JNP in X86FreqProps.
665 :    
666 :     3. CFG
667 :    
668 :     CFG now has a bunch of methods for edge splitting and merging.
669 :    
670 :     4. Machine description.
671 :    
672 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
673 :     description things:
674 :    
675 :     rtl-build.{sig,sml} and hppa.mdl fixed.
676 :    
677 :     NOTE: the machine description stuff in the repository is still broken.
678 :     Again, I can't put my fixes in because that involves
679 :     changes to MLTREE.
680 :    
681 :     ----------------------------------------------------------------------
682 : blume 1145 Name: Matthias Blume
683 : blume 1155 Date: 2002/03/20 15:55:00 EST
684 :     Tag: blume-20020320-kmp
685 :     Description:
686 :    
687 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
688 :     it for String.isSubstring, Substring.isSubstring, and
689 :     Substring.position.
690 :    
691 :     (Might need some stress-testing. Simple examples worked fine.)
692 :    
693 :     ----------------------------------------------------------------------
694 :     Name: Matthias Blume
695 : blume 1152 Date: 2002/03/19 16:37:00 EST
696 :     Tag: blume-20020319-witnesses
697 :     Description:
698 :    
699 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
700 :    
701 :     This implements a generic mechanism for changing constness qualifiers
702 :     anywhere within big C types without resorting to outright "casts".
703 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
704 :     modify the constness at the outermost level.)
705 :     The implementation of "convert" is based on the idea of "witness"
706 :     values -- values that are not used by the operation but whose types
707 :     "testify" to their applicability. On the implementation side, "convert"
708 :     is simply a projection (returning its second curried argument). With
709 :     cross-module inlining, it should not result in any machine code being
710 :     generated.
711 :    
712 :     ----------------------------------------------------------------------
713 :     Name: Matthias Blume
714 : blume 1148 Date: 2002/03/15 16:40:00 EST
715 :     Tag: blume-20020315-basis
716 :     Description:
717 :    
718 :     Provided (preliminary?) implementations for
719 :    
720 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
721 :    
722 :     and
723 :    
724 :     Substring.full
725 :    
726 :     Those are in the Basis spec but they were missing in SML/NJ.
727 :    
728 :     ----------------------------------------------------------------------
729 :     Name: Matthias Blume
730 : blume 1145 Date: 2002/03/14 21:30:00 EST
731 :     Tag: blume-20020314-controls
732 :     Description:
733 :    
734 :     Controls:
735 :     ---------
736 :    
737 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
738 :     it into its own library $/controls-lib.cm. The source tree for
739 :     this is under src/smlnj-lib/Controls.
740 :    
741 :     2. Changed the names of types and functions in this interface, so they
742 :     make a bit more "sense":
743 :    
744 :     module -> registry
745 :     'a registry -> 'a group
746 :    
747 :     3. The interface now deals in ref cells only. The getter/setter interface
748 :     is (mostly) gone.
749 :    
750 :     4. Added a function that lets one register an already-existing ref cell.
751 :    
752 :     5. Made the corresponding modifications to the rest of the code so that
753 :     everything compiles again.
754 :    
755 :     6. Changed the implementation of Controls.MLRISC back to something closer
756 :     to the original. In particular, this module (and therefore MLRISC)
757 :     does not depend on Controls. There now is some link-time code in
758 :     int-sys.sml that registers the MLRISC controls with the Controls
759 :     module.
760 :    
761 :     CM:
762 :     ---
763 :    
764 :     * One can now specify the lambda-split aggressiveness in init.cmi.
765 :    
766 :     ----------------------------------------------------------------------
767 : leunga 1142 Name: Allen Leung
768 :     Date: 2002/03/13 17:30:00 EST
769 :     Tag: leunga-20020313-x86-fp-unary
770 :     Description:
771 :    
772 :     Bug fix for:
773 :    
774 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
775 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
776 :     > - fun f(x,(y,z)) = Real.~ y;
777 :     > [autoloading]
778 :     > [autoloading done]
779 :     > fchsl (%eax), 184(%esp)
780 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
781 :     >
782 :     > uncaught exception Error
783 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
784 :    
785 :     The problem was that the code generator did not generate any fp registers
786 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
787 :     translate the pseudo fp instruction. This only happened with unary fp
788 :     operators in certain situations.
789 :    
790 :     ----------------------------------------------------------------------
791 : blume 1137 Name: Matthias Blume
792 : blume 1140 Date: 2002/03/13 14:00:00 EST
793 :     Tag: blume-20020313-overload-etc
794 :     Description:
795 :    
796 :     1. Added _overload as a synonym for overload for backward compatibility.
797 :     (Control.overloadKW must be true for either version to be accepted.)
798 :    
799 :     2. Fixed bug in install script that caused more things to be installed
800 :     than what was requested in config/targets.
801 :    
802 :     3. Made CM aware of the (_)overload construct so that autoloading
803 :     works.
804 :    
805 :     ----------------------------------------------------------------------
806 :     Name: Matthias Blume
807 : blume 1138 Date: 2002/03/12 22:03:00 EST
808 :     Tag: blume-20020312-url
809 :     Description:
810 :    
811 :     Forgot to update BOOT and srcarchiveurl.
812 :    
813 :     ----------------------------------------------------------------------
814 :     Name: Matthias Blume
815 : blume 1137 Date: 2002/03/12 17:30:00 EST
816 :     Tag: blume-20020312-version110392
817 :     Description:
818 :    
819 :     Yet another version number bump (because of small changes to the
820 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
821 :    
822 :     Changes:
823 :    
824 :     The new pid generation scheme described a few weeks ago was overly
825 :     complicated. I implemented a new mechanism that is simpler and
826 :     provides a bit more "stability": Once CM has seen a compilation
827 :     unit, it keeps its identity constant (as long as you do not delete
828 :     those crucial CM/GUID/* files). This means that when you change
829 : blume 1145 an interface, compile, then go back to the old interface, and
830 : blume 1137 compile again, you arrive at the original pid.
831 :    
832 :     There now also is a mechanism that instructs CM to use the plain
833 :     environment hash as a module's pid (effectively making its GUID
834 :     the empty string). For this, "noguid" must be specified as an
835 :     option to the .sml file in question within its .cm file.
836 :     This is most useful for code that is being generated by tools such
837 :     as ml-nlffigen (because during development programmers tend to
838 :     erase the tool's entire output directory tree including CM's cached
839 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
840 :     revert to the old, broken behavior of SML/NJ, but in specific cases
841 :     where there is no danger of interface confusion, its use is ok
842 :     (I think).
843 :    
844 :     ml-nlffigen by default generates "noguid" annotations. They can be
845 :     turned off by specifying -guid in its command line.
846 :    
847 :     ----------------------------------------------------------------------
848 : george 1136 Name: Lal George
849 :     Date: 2002/03/12 12 14:42:36 EST
850 :     Tag: george-20020312-frequency-computation
851 :     Description:
852 :    
853 :     Integrated jump chaining and static block frequency into the
854 :     compiler. More details and numbers later.
855 : leunga 1025
856 : george 1003 ----------------------------------------------------------------------
857 : george 1133 Name: Lal George
858 :     Date: 2002/03/11 11 22:38:53 EST
859 :     Tag: george-20020311-jump-chain-elim
860 :     Description:
861 :    
862 :     Tested the jump chain elimination on all architectures (except the
863 :     hppa). This is on by default right now and is profitable for the
864 :     alpha and x86, however, it may not be profitable for the sparc and ppc
865 :     when compiling the compiler.
866 :    
867 :     The gc test will typically jump to a label at the end of the cluster,
868 :     where there is another jump to an external cluster containing the actual
869 :     code to invoke gc. This is to allow factoring of common gc invocation
870 :     sequences. That is to say, we generate:
871 :    
872 :     f:
873 :     testgc
874 :     ja L1 % jump if above to L1
875 :    
876 :     L1:
877 :     jmp L2
878 :    
879 :    
880 :     After jump chain elimination the 'ja L1' instructions is converted to
881 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
882 :     up being implemented in their long form (if L2 is far away) using:
883 :    
884 :     jbe L3 % jump if below or equal to L3
885 :     jmp L2
886 :     L3:
887 :     ...
888 :    
889 :    
890 :     For large compilation units L2 may be far away.
891 :    
892 :    
893 :     ----------------------------------------------------------------------
894 : blume 1132 Name: Matthias Blume
895 :     Date: 2002/03/11 13:30:00 EST
896 :     Tag: blume-20020311-mltreeeval
897 :     Description:
898 :    
899 :     A functor parameter was missing.
900 :    
901 :     ----------------------------------------------------------------------
902 : leunga 1130 Name: Allen Leung
903 : leunga 1131 Date: 2002/03/11 10:30:00 EST
904 : leunga 1142 Tag: leunga-20020311-runtime-string0
905 : leunga 1131 Description:
906 :    
907 :     The representation of the empty string now points to a
908 :     legal null terminated C string instead of unit. It is now possible
909 :     to convert an ML string into C string with InlineT.CharVector.getData.
910 :     This compiles into one single machine instruction.
911 :    
912 :     ----------------------------------------------------------------------
913 :     Name: Allen Leung
914 : leunga 1130 Date: 2002/03/10 23:55:00 EST
915 :     Tag: leunga-20020310-x86-call
916 :     Description:
917 :    
918 :     Added machine generation for CALL instruction (relative displacement mode)
919 :    
920 :     ----------------------------------------------------------------------
921 : blume 1128 Name: Matthias Blume
922 :     Date: 2002/03/08 16:05:00
923 :     Tag: blume-20020308-entrypoints
924 :     Description:
925 :    
926 :     Version number bumped to 110.39.1. NEW BOOTFILES!
927 :    
928 :     Entrypoints: non-zero offset into a code object where execution should begin.
929 :    
930 :     - Added the notion of an entrypoint to CodeObj.
931 :     - Added reading/writing of entrypoint info to Binfile.
932 :     - Made runtime system bootloader aware of entrypoints.
933 :     - Use the address of the label of the first function given to mlriscGen
934 :     as the entrypoint. This address is currently always 0, but it will
935 :     not be 0 once we turn on block placement.
936 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
937 :     with entry points) from mlriscGen.
938 :    
939 :     ----------------------------------------------------------------------
940 : leunga 1127 Name: Allen Leung
941 :     Date: 2002/03/07 20:45:00 EST
942 :     Tag: leunga-20020307-x86-cmov
943 :     Description:
944 :    
945 :     Bug fixes for CMOVcc on x86.
946 :    
947 :     1. Added machine code generation for CMOVcc
948 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
949 :     3. CMOVcc cannot have an immediate operand as argument.
950 :    
951 :     ----------------------------------------------------------------------
952 : blume 1126 Name: Matthias Blume
953 :     Date: 2002/03/07 16:15:00 EST
954 :     Tag: blume-20020307-controls
955 :     Description:
956 :    
957 :     This is a very large but mostly boring patch which makes (almost)
958 :     every tuneable compiler knob (i.e., pretty much everything under
959 :     Control.* plus a few other things) configurable via both the command
960 :     line and environment variables in the style CM did its configuration
961 :     until now.
962 :    
963 :     Try starting sml with '-h' (or, if you are brave, '-H')
964 :    
965 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
966 :     implements the underlying generic mechanism.
967 :    
968 :     The interface to some of the existing such facilities has changed somewhat.
969 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
970 :     (The getFoo interface is still there for backward-compatibility, but its
971 :     use is deprecated.)
972 :    
973 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
974 :     that one can now twiddle the compiler settings when using this "batch"
975 :     compiler.
976 :    
977 :     TODO items:
978 :    
979 :     We should go through and throw out all controls that are no longer
980 :     connected to anything. Moreover, we should go through and provide
981 :     meaningful (and correct!) documentation strings for those controls
982 :     that still are connected.
983 :    
984 :     Currently, multiple calls to Controls.new are accepted (only the first
985 :     has any effect). Eventually we should make sure that every control
986 :     is being made (via Controls.new) exactly once. Future access can then
987 :     be done using Controls.acc.
988 :    
989 :     Finally, it would probably be a good idea to use the getter-setter
990 :     interface to controls rather than ref cells. For the time being, both
991 :     styles are provided by the Controls module, but getter-setter pairs are
992 :     better if thread-safety is of any concern because they can be wrapped.
993 :    
994 :     *****************************************
995 :    
996 :     One bug fix: The function blockPlacement in three of the MLRISC
997 :     backpatch files used to be hard-wired to one of two possibilities at
998 :     link time (according to the value of the placementFlag). But (I
999 :     think) it should rather sense the flag every time.
1000 :    
1001 :     *****************************************
1002 :    
1003 :     Other assorted changes (by other people who did not supply a HISTORY entry):
1004 :    
1005 :     1. the cross-module inliner now works much better (Monnier)
1006 :     2. representation of weights, frequencies, and probabilities in MLRISC
1007 :     changed in preparation of using those for weighted block placement
1008 :     (Reppy, George)
1009 :    
1010 :     ----------------------------------------------------------------------
1011 : george 1116 Name: Lal George
1012 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
1013 :     Tag: george-20020307-weighted-block-placement
1014 :    
1015 :     Tested the weighted block placement optimization on all architectures
1016 :     (except the hppa) using AMPL to generate the block and edge frequencies.
1017 :     Changes were required in the machine properties to correctly
1018 :     categorize trap instructions. There is an MLRISC flag
1019 :     "weighted-block-placement" that can be used to enable weighted block
1020 :     placement, but this will be ineffective without block/edge
1021 :     frequencies (coming soon).
1022 :    
1023 :    
1024 :     ----------------------------------------------------------------------
1025 :     Name: Lal George
1026 : george 1116 Date: 2002/03/05 17:24:48 EST
1027 :     Tag: george-20020305-linkage-cluster
1028 :    
1029 : george 1124 In order to support the block placement optimization, a new cluster
1030 :     is generated as the very first cluster (called the linkage cluster).
1031 :     It contains a single jump to the 'real' entry point for the compilation
1032 :     unit. Block placement has no effect on the linkage cluster itself, but
1033 :     all the other clusters have full freedom in the manner in which they
1034 :     reorder blocks or functions.
1035 : george 1116
1036 :     On the x86 the typical linkage code that is generated is:
1037 :     ----------------------
1038 :     .align 2
1039 :     L0:
1040 :     addl $L1-L0, 72(%esp)
1041 : george 1124 jmp L1
1042 : george 1116
1043 :    
1044 :     .align 2
1045 :     L1:
1046 :     ----------------------
1047 :    
1048 :     72(%esp) is the memory location for the stdlink register. This
1049 :     must contain the address of the CPS function being called. In the
1050 : george 1124 above example, it contains the address of L0; before
1051 : george 1116 calling L1 (the real entry point for the compilation unit), it
1052 :     must contain the address for L1, and hence
1053 :    
1054 :     addl $L1-L0, 72(%esp)
1055 :    
1056 : george 1124 I have tested this on all architectures except the hppa.The increase
1057 :     in code size is of course negligible
1058 : george 1116
1059 :     ----------------------------------------------------------------------
1060 : leunga 1115 Name: Allen Leung
1061 :     Date: 2002/03/03 13:20:00 EST
1062 :     Tag: leunga-20020303-mlrisc-tools
1063 :    
1064 :     Added #[ ... ] expressions to mlrisc tools
1065 :    
1066 :     ----------------------------------------------------------------------
1067 : blume 1096 Name: Matthias Blume
1068 : blume 1098 Date: 2002/02/27 12:29:00 EST
1069 :     Tag: blume-20020227-cdebug
1070 :     Description:
1071 :    
1072 :     - made types in structure C and C_Debug to be equal
1073 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1074 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1075 :    
1076 :     ----------------------------------------------------------------------
1077 :     Name: Matthias Blume
1078 : blume 1096 Date: 2002/02/26 12:00:00 EST
1079 :     Tag: blume-20020226-ffi
1080 :     Description:
1081 :    
1082 :     1. Fixed a minor bug in CM's "noweb" tool:
1083 :     If numbering is turned off, then truly don't number (i.e., do not
1084 :     supply the -L option to noweb). The previous behavior was to supply
1085 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
1086 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
1087 :    
1088 :     2. Once again, I reworked some aspects of the FFI:
1089 :    
1090 :     A. The incomplete/complete type business:
1091 :    
1092 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1093 :     gone!
1094 :     - ML types representing an incomplete type are now *equal* to
1095 :     ML types representing their corresponding complete types (just like
1096 :     in C). This is still safe because ml-nlffigen will not generate
1097 :     RTTI for incomplete types, nor will it generate functions that
1098 :     require access to such RTTI. But when ML code generated from both
1099 :     incomplete and complete versions of the C type meet, the ML types
1100 :     are trivially interoperable.
1101 :    
1102 :     NOTE: These changes restore the full generality of the translation
1103 :     (which was previously lost when I eliminated functorization)!
1104 :    
1105 :     B. Enum types:
1106 :    
1107 :     - Structure C now has a type constructor "enum" that is similar to
1108 :     how the "su" constructor works. However, "enum" is not a phantom
1109 :     type because each "T enum" has values (and is isomorphic to
1110 :     MLRep.Signed.int).
1111 :     - There are generic access operations for enum objects (using
1112 :     MLRep.Signed.int).
1113 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
1114 :     * The structure contains the definition of type "mlrep" (the ML-side
1115 :     representation type of the enum). Normally, mlrep is the same
1116 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1117 :     then mlrep will be defined as a datatype -- thus facilitating
1118 :     pattern matching on mlrep values.
1119 :     ("-ec" will be suppressed if there are duplicate values in an
1120 :     enumeration.)
1121 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1122 :     will be generated for each C enum constant xxx.
1123 :     * Conversion functions m2i and i2m convert between mlrep and
1124 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
1125 :     * Coversion functions c and ml convert between mlrep and "tag enum".
1126 :     * Access functions (get/set) fetch and store mlrep values.
1127 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1128 :     enumerations are merged into one single enumeration represented by
1129 :     structure E_'.
1130 :    
1131 :     ----------------------------------------------------------------------
1132 : leunga 1080 Name: Allen Leung
1133 : leunga 1094 Date: 2002/02/25 04:45:00 EST
1134 :     Tag: leunga-20020225-cps-spill
1135 :    
1136 :     This is a new implementation of the CPS spill phase.
1137 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1138 :     In case of problems, replace it with the old file spill.sml
1139 :    
1140 :     The current compiler runs into some serious performance problems when
1141 :     constructing a large record. This can happen when we try to compile a
1142 :     structure with many items. Even a very simple structure like the following
1143 :     makes the compiler slow down.
1144 :    
1145 :     structure Foo = struct
1146 :     val x_1 = 0w1 : Word32.int
1147 :     val x_2 = 0w2 : Word32.int
1148 :     val x_3 = 0w3 : Word32.int
1149 :     ...
1150 :     val x_N = 0wN : Word32.int
1151 :     end
1152 :    
1153 :     The following table shows the compile time, from N=1000 to N=4000,
1154 :     with the old compiler:
1155 :    
1156 :     N
1157 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
1158 :     MLRISC ra 0.06u 0.00s 0.05g
1159 :     (spills = 0 reloads = 0)
1160 :     TOTAL 0.63u 0.07s 0.21g
1161 :    
1162 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
1163 :     MLRISC ra 5.68u 0.59s 3.93g
1164 :     (spills = 0 reloads = 0)
1165 :     TOTAL 14.71u 0.99s 4.81g
1166 :    
1167 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
1168 :     MLRISC ra 5.54u 0.65s 3.91g
1169 :     (spills = 543 reloads = 1082)
1170 :     TOTAL 65.40u 3.13s 6.00g
1171 :    
1172 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
1173 :     MLRISC ra 0.80u 0.10s 0.55g
1174 :     (spills = 42 reloads = 84)
1175 :     TOTAL 129.42u 5.10s 4.13g
1176 :    
1177 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
1178 :     MLRISC ra 2.69u 0.27s 1.38g
1179 :     (spills = 62 reloads = 124)
1180 :     TOTAL 682.48u 19.61s 13.99g
1181 :    
1182 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
1183 :     MLRISC ra 4.96u 0.27s 2.72g
1184 :     (spills = 85 reloads = 170)
1185 :     TOTAL 2375.26u 57.21s 48.00g
1186 :    
1187 :     As you can see the old cps spill module suffers from some serious
1188 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
1189 :     instead of patching the problems up, I'm reimplementing it
1190 : leunga 1094 with a different algorithm. The new code is more modular,
1191 :     smaller when compiled, and substantially faster
1192 :     (O(n log n) time and O(n) space). Timing of the new spill module:
1193 :    
1194 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
1195 :     MLRISC ra 0.25u 0.02s 0.15g
1196 :     (spills=1 reloads=3)
1197 :     TOTAL 7.74u 0.34s 1.62g
1198 :    
1199 :     Implementation details:
1200 :    
1201 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
1202 :     number of live variables at any program point (the bandwidth)
1203 :     does not exceed a certain limit, which is determined by the
1204 :     size of the spill area.
1205 :    
1206 :     When the bandwidth is too large, we decrease the register pressure by
1207 :     packing live variables into spill records. How we achieve this is
1208 :     completely different than what we did in the old code.
1209 :    
1210 :     First, there is something about the MLRiscGen code generator
1211 :     that we should be aware of:
1212 :    
1213 :     o MLRiscGen performs code motion!
1214 :    
1215 :     In particular, it will move floating point computations and
1216 :     address computations involving only the heap pointer to
1217 :     their use sites (if there is only a single use).
1218 :     What this means is that if we have a CPS record construction
1219 :     statement
1220 :    
1221 :     RECORD(k,vl,w,e)
1222 :    
1223 :     we should never count the new record address w as live if w
1224 :     has only one use (which is often the case).
1225 :    
1226 :     We should do something similar to floating point, but the transformation
1227 :     there is much more complex, so I won't deal with that.
1228 :    
1229 :     Secondly, there are now two new cps primops at our disposal:
1230 :    
1231 :     1. rawrecord of record_kind option
1232 :     This pure operator allocates some uninitialized storage from the heap.
1233 :     There are two forms:
1234 :    
1235 :     rawrecord NONE [INT n] allocates a tagless record of length n
1236 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1237 :     and initializes the tag.
1238 :    
1239 :     2. rawupdate of cty
1240 :     rawupdate cty (v,i,x)
1241 :     Assigns to x to the ith component of record v.
1242 :     The storelist is not updated.
1243 :    
1244 :     We use these new primops for both spilling and increment record construction.
1245 :    
1246 :     1. Spilling.
1247 :    
1248 :     This is implemented with a linear scan algorithm (but generalized
1249 :     to trees). The algorithm will create a single spill record at the
1250 :     beginning of the cps function and use rawupdate to spill to it,
1251 :     and SELECT or SELp to reload from it. So both spills and reloads
1252 :     are fine-grain operations. In contrast, in the old algorithm
1253 :     "spills" have to be bundled together in records.
1254 :    
1255 :     Ideally, we should sink the spill record construction to where
1256 :     it is needed. We can even split the spill record into multiple ones
1257 :     at the places where they are needed. But CPS is not a good
1258 :     representation for global code motion, so I'll keep it simple and
1259 :     am not attempting this.
1260 :    
1261 :     2. Incremental record construction (aka record splitting).
1262 :    
1263 :     Long records with many component values which are simulatenously live
1264 :     (recall that single use record addresses are not considered to
1265 :     be live) are constructed with rawrecord and rawupdate.
1266 :     We allocate space on the heap with rawrecord first, then gradually
1267 :     fill it in with rawupdate. This is the technique suggested to me
1268 :     by Matthias.
1269 :    
1270 :     Some restrictions on when this is applicable:
1271 :     1. It is not a VECTOR record. The code generator currently does not handle
1272 :     this case. VECTOR record uses double indirection like arrays.
1273 :     2. All the record component values are defined in the same "basic block"
1274 :     as the record constructor. This is to prevent speculative
1275 :     record construction.
1276 :    
1277 :     ----------------------------------------------------------------------
1278 :     Name: Allen Leung
1279 : leunga 1086 Date: 2002/02/22 01:02:00 EST
1280 :     Tag: leunga-20020222-mlrisc-tools
1281 :    
1282 :     Minor bug fixes in the parser and rewriter
1283 :    
1284 :     ----------------------------------------------------------------------
1285 :     Name: Allen Leung
1286 : leunga 1085 Date: 2002/02/21 20:20:00 EST
1287 :     Tag: leunga-20020221-peephole
1288 :    
1289 :     Regenerated the peephole files. Some contained typos in the specification
1290 :     and some didn't compile because of pretty printing bugs in the old version
1291 :     of 'nowhere'.
1292 :    
1293 :     ----------------------------------------------------------------------
1294 :     Name: Allen Leung
1295 : leunga 1079 Date: 2002/02/19 20:20:00 EST
1296 :     Tag: leunga-20020219-mlrisc-tools
1297 :     Description:
1298 :    
1299 :     Minor bug fixes to the mlrisc-tools library:
1300 :    
1301 :     1. Fixed up parsing colon suffixed keywords
1302 :     2. Added the ability to shut the error messages up
1303 :     3. Reimplemented the pretty printer and fixed up/improved
1304 :     the pretty printing of handle and -> types.
1305 :     4. Fixed up generation of literal symbols in the nowhere tool.
1306 :     5. Added some SML keywords to to sml.sty
1307 :    
1308 :     ----------------------------------------------------------------------
1309 :     Name: Matthias Blume
1310 : blume 1078 Date: 2002/02/19 16:20:00 EST
1311 :     Tag: blume-20020219-cmffi
1312 :     Description:
1313 :    
1314 :     A wild mix of changes, some minor, some major:
1315 :    
1316 :     * All C FFI-related libraries are now anchored under $c:
1317 :     $/c.cm --> $c/c.cm
1318 :     $/c-int.cm --> $c/internals/c-int.cm
1319 :     $/memory.cm --> $c/memory/memory.cm
1320 :    
1321 :     * "make" tool (in CM) now treats its argument pathname slightly
1322 :     differently:
1323 :     1. If the native expansion is an absolute name, then before invoking
1324 :     the "make" command on it, CM will apply OS.Path.mkRelative
1325 :     (with relativeTo = OS.FileSys.getDir()) to it.
1326 :     2. The argument will be passed through to subsequent phases of CM
1327 :     processing without "going native". In particular, if the argument
1328 :     was an anchored path, then "make" will not lose track of that anchor.
1329 :    
1330 :     * Compiler backends now "know" their respective C calling conventions
1331 :     instead of having to be told about it by ml-nlffigen. This relieves
1332 :     ml-nlffigen from one of its burdens.
1333 :    
1334 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1335 :    
1336 :     * Export C_DEBUG and C_Debug from $c/c.cm.
1337 :    
1338 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1339 :     subtyping relationship between incomplete pointers and their complete
1340 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
1341 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
1342 :     the pointer target type is complete. In the incomplete case, 'o
1343 :     will be instantiated with some "'c iobj" -- a type obtained by
1344 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
1345 :    
1346 :     Operations that work on both incomplete and complete pointer types are
1347 :     typed as taking an 'o ptr while operations that require the target to
1348 :     be known are typed as taking some ('t, 'c) obj ptr.
1349 :    
1350 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1351 :     where void is an eqtype without any values. This makes it possible
1352 :     to work on voidptr values using functions meant to operate on light
1353 :     incomplete pointers.
1354 :    
1355 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1356 :     been vastly simplified.
1357 :    
1358 :     ----------------------------------------------------------------------
1359 :     Name: Matthias Blume
1360 : blume 1076 Date: 2002/02/19 10:48:00 EST
1361 :     Tag: blume-20020219-pqfix
1362 :     Description:
1363 :    
1364 :     Applied Chris Okasaki's bug fix for priority queues.
1365 :    
1366 :     ----------------------------------------------------------------------
1367 :     Name: Matthias Blume
1368 : blume 1073 Date: 2002/02/15 17:05:00
1369 : blume 1069 Tag: Release_110_39
1370 :     Description:
1371 :    
1372 : blume 1073 Last-minute retagging is becoming a tradition... :-(
1373 :    
1374 :     This is the working release 110.39.
1375 :    
1376 :     ----------------------------------------------------------------------
1377 :     Name: Matthias Blume
1378 :     Date: 2002/02/15 16:00:00 EST
1379 :     Tag: Release_110_39-orig
1380 :     Description:
1381 :    
1382 : blume 1069 Working release 110.39. New bootfiles.
1383 :    
1384 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
1385 :     with all shells. So I retagged. -Matthias)
1386 :    
1387 : blume 1069 ----------------------------------------------------------------------
1388 :     Name: Matthias Blume
1389 : blume 1068 Date: 2002/02/15 14:17:00 EST
1390 :     Tag: blume-20020215-showbindings
1391 :     Description:
1392 :    
1393 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
1394 :     the latter can be useful for exploring what bindings are available at
1395 :     the interactive prompt. (The first function returns only the list
1396 :     of symbols that are really bound, the second prints those but also the
1397 :     ones that CM's autoloading mechanism knows about.)
1398 :    
1399 :     ----------------------------------------------------------------------
1400 :     Name: Matthias Blume
1401 : blume 1067 Date: 2002/02/15 12:08:00 EST
1402 :     Tag: blume-20020215-iptrs
1403 :     Description:
1404 :    
1405 :     Two improvements to ml-nlffigen:
1406 :    
1407 :     1. Write files only if they do not exist or if their current contents
1408 :     do not coincide with what's being written. (That is, avoid messing
1409 :     with the time stamps unless absolutely necessary.)
1410 :    
1411 :     2. Implement a "repository" mechanism for generated files related
1412 :     to "incomplete pointer types". See the README file for details.
1413 :    
1414 :     ----------------------------------------------------------------------
1415 :     Name: Matthias Blume
1416 : blume 1066 Date: 2002/02/14 11:50:00 EST
1417 :     Tag: blume-20020214-quote
1418 :     Description:
1419 :    
1420 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
1421 :     because of the new and improved tag generation scheme. (Thanks to Allen
1422 :     Leung for pointing it out.)
1423 :    
1424 :     ----------------------------------------------------------------------
1425 : george 1065 Name: Lal George
1426 :     Date: 2002/02/14 09:55:27 EST 2002
1427 :     Tag: george-20020214-isabelle-bug
1428 :     Description:
1429 :    
1430 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1431 :     of Isabelle on the x86.
1432 :    
1433 :     From Allen:
1434 :     -----------
1435 :     I've found the problem:
1436 :    
1437 :     in ra-core.sml, I use the counter "blocked" to keep track of the
1438 :     true number of elements in the freeze queue. When the counter goes
1439 :     to zero, I skip examining the queue. But I've messed up the
1440 :     bookkeeping in combine():
1441 :    
1442 :     else ();
1443 :     case !ucol of
1444 :     PSEUDO => (if !cntv > 0 then
1445 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
1446 :     ^^^^^^^^^^^^^^^^^^^^^^^
1447 :     moveu := mergeMoveList(!movev, !moveu)
1448 :     )
1449 :     else ();
1450 :    
1451 :     combine() is called to coalesce two nodes u and v.
1452 :     I think I was thinking that if the move counts of u and v are both
1453 :     greater than zero then after they are coalesced then one node is
1454 :     removed from the freeze queue. Apparently I was thinking that
1455 :     both u and v are of low degree, but that's clearly not necessarily true.
1456 :    
1457 :    
1458 :     02/12/2002:
1459 :     Here's the patch. HOL now compiles.
1460 :    
1461 :     I don't know how this impact on performance (compile
1462 :     time or runtime). This bug caused the RA (especially on the x86)
1463 :     to go thru the potential spill phase when there are still nodes on the
1464 :     freeze queue.
1465 :    
1466 :    
1467 :    
1468 :    
1469 :     ----------------------------------------------------------------------
1470 : blume 1058 Name: Matthias Blume
1471 : blume 1064 Date: 2002/02/13 22:40:00 EST
1472 :     Tag: blume-20020213-fptr-rtti
1473 :     Description:
1474 :    
1475 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
1476 :     updates.
1477 :    
1478 :     ----------------------------------------------------------------------
1479 :     Name: Matthias Blume
1480 : blume 1063 Date: 2002/02/13 16:41:00 EST
1481 :     Tag: blume-20020213-cmlpq
1482 :     Description:
1483 :    
1484 :     Added new priority queue export symbols (which have just been added to
1485 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
1486 :     not compile and the installer would choke.)
1487 :    
1488 :     ----------------------------------------------------------------------
1489 :     Name: Matthias Blume
1490 : blume 1062 Date: 2002/02/13 16:15:00 EST
1491 :     Tag: blume-20020213-various
1492 :     Description:
1493 :    
1494 :     1. More tweaks to ml-nlffigen:
1495 :    
1496 :     - better internal datastructures (resulting in slight speedup)
1497 :     - "-match" option requires exact match
1498 :     - "localized" gensym counters (untagged structs/unions nested within
1499 :     other structs/unions or within typedefs get a fresh counter; their
1500 :     tag will be prefixed by a concatenation of their parents' tags)
1501 :     - bug fixes (related to calculation of transitive closure of types
1502 :     to be included in the output)
1503 :    
1504 :     2. Minor Basis updates:
1505 :    
1506 :     - added implementations for List.collate and Option.app
1507 :    
1508 :     ----------------------------------------------------------------------
1509 :     Name: Matthias Blume
1510 : blume 1060 Date: 2002/02/11 15:55:00 EST
1511 :     Tag: blume-20020211-gensym
1512 :     Description:
1513 :    
1514 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
1515 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
1516 :     names (ML structure names that correspond to unnamed C structs, unions,
1517 :     and enums), so that separate runs of ml-nlffigen do not clash.
1518 :    
1519 :     ----------------------------------------------------------------------
1520 :     Name: Matthias Blume
1521 : blume 1059 Date: 2002/02/11 12:05:00 EST
1522 :     Tag: blume-20020211-gensml
1523 :     Description:
1524 :    
1525 :     A quick fix for a problem with GenSML (in the pgraph-util library):
1526 :     Make generation of toplevel "local" optional. (Strictly speaking,
1527 :     signature definitions within "local" are not legal SML.)
1528 :    
1529 :     Other than that: updates to INSTALL and cm/TODO.
1530 :    
1531 :     ----------------------------------------------------------------------
1532 :     Name: Matthias Blume
1533 : blume 1058 Date: 2002/02/08 15:00:00 EST
1534 :     Tag: blume-20020208-uniquepid
1535 :     Description:
1536 :    
1537 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
1538 :    
1539 :     1. The installer (config/install.sh) has gotten smarter:
1540 :    
1541 :     - Configuration options are a bit easier to specify now
1542 :     (in config/targets).
1543 :     - Bug in recognizing .tar.bz2 files fixed.
1544 :     - Installer automatically resolves dependencies between
1545 :     configuration options (e.g., if you ask for eXene, you will
1546 :     also get cml -- regardless whether you asked for it or not).
1547 :     - Installer can run in "quieter mode" by setting the environment
1548 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
1549 :     "completely silent", though.
1550 :     - Build HashCons library as part of smlnj-lib.
1551 :    
1552 :     2. A new scheme for assigning persistent identifiers to compilation
1553 :     units (and, by extension, to types etc.) has been put into place.
1554 :     This fixes a long-standing bug where types and even dynamic values
1555 :     can get internally confused, thereby compromising type safety
1556 :     (abstraction) and dynamic correctness. See
1557 :    
1558 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1559 :    
1560 :     for an example of how things could go wrong until now.
1561 :    
1562 :     The downside of the new scheme is that pids are not quite as
1563 :     persistent as they used to be: CM will generate a fresh pid
1564 :     for every compilation unit that it thinks it sees for the first
1565 :     time. That means that if you compile starting from a clean, fresh
1566 :     source tree at two different times, you end up with different
1567 :     binaries.
1568 :    
1569 :     Cutoff recompilation, however, has not been compromised because
1570 :     CM keeps pid information in special caches between runs.
1571 :    
1572 :     ----------------------------------------------------------------------
1573 : george 1053 Name: Lal George
1574 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
1575 :     Tag: <none>
1576 :     Description:
1577 :    
1578 :     Compilers that generate assembly code may produce global labels
1579 :     whose value is resolved at link time. The various peephole optimization
1580 :     modules did not take this in account.
1581 :    
1582 :     TODO. The Labels.addrOf function should really return an option
1583 :     type so that clients are forced to deal with this issue, rather
1584 :     than an exception being raised.
1585 :    
1586 :     ----------------------------------------------------------------------
1587 :     Name: Lal George
1588 : george 1053 Date: 2002/02/06 13:55:02 EST
1589 :     Tag: george-20020206-ra-breakup
1590 :     Description:
1591 :    
1592 :     1. A bug fix from Allen.
1593 :    
1594 :     A typo causes extra fstp %st(0)'s to be generated at compensation
1595 :     edges, which might cause stack underflow traps at runtime. This
1596 :     occurs in fft where there are extraneous fstps right before the 'into'
1597 :     trap instruction (in this case they are harmless since none of the
1598 :     integers overflow.)
1599 :    
1600 : george 1054 2. Pulled out various utility modules that were embedded in the modules
1601 : george 1053 of the register allocator. I need these modules for other purposes, but
1602 : george 1054 they are not complete enough to put into a library (just yet).
1603 : george 1053 ----------------------------------------------------------------------
1604 : blume 1049 Name: Matthias Blume
1605 :     Date: 2002/01/31 16:05:00 EST
1606 :     Tag: blume-20020131-sparc-ccalls
1607 :     Description:
1608 :    
1609 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1610 :     of extra stack space by mistake. Fixed.
1611 :    
1612 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1613 :    
1614 :     ----------------------------------------------------------------------
1615 : leunga 1048 Name: Allen Leung
1616 :     Date: 2002/01/30
1617 :     Tag: leunga-20020130-nowhere-bug-fix
1618 :     Description:
1619 :    
1620 :     MLRISC bug fixes:
1621 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1622 :     2. MachineInt.fromString was negating its value.
1623 :    
1624 :     ----------------------------------------------------------------------
1625 : blume 1039 Name: Matthias Blume
1626 : blume 1047 Date: 2002/01/29
1627 :     Tag: blume-20020129-INSTALL
1628 :     Description:
1629 :    
1630 :     - Added somewhat detailed installation instructions (file INSTALL).
1631 :     - Fixed curl-detection bug in config/install.sh.
1632 :     - It is now possible to select the URL getter using the URLGETTER
1633 :     environment variable:
1634 :    
1635 :     not set / "unknown" --> automatic detection (script tries wget,
1636 :     curl, and lynx)
1637 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1638 :     how to properly invoke them)
1639 :     other --> use $URLGETTER directly, it must take
1640 :     precisely two command-line arguments
1641 :     (source URL and destination file name)
1642 :    
1643 :     ----------------------------------------------------------------------
1644 :     Name: Matthias Blume
1645 : blume 1044 Date: 2002/01/28
1646 : blume 1045 Tag: blume-20020128-sparc-ccalls
1647 :     Description:
1648 :    
1649 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
1650 :     - Make use of the allocParam argument in sparc-c-calls.
1651 :    
1652 :     ----------------------------------------------------------------------
1653 :     Name: Matthias Blume
1654 :     Date: 2002/01/28
1655 : blume 1044 Tag: blume-20020128-allocParam
1656 :     Description:
1657 :    
1658 :     John Reppy: Changes c-calls API to accept client-callback for
1659 :     allocating extra stack space.
1660 :     me: Corresponding changes to mlriscGen (using a dummy argument that
1661 :     does not change the current behavior).
1662 :    
1663 :     ----------------------------------------------------------------------
1664 :     Name: Matthias Blume
1665 : blume 1040 Date: 2002/01/28 12:00:00
1666 :     Tag: Release_110_38
1667 :     Description:
1668 :    
1669 :     This time for real!!!
1670 :    
1671 :     ----------------------------------------------------------------------
1672 :     Name: Matthias Blume
1673 : blume 1039 Date: 2002/01/28 10:56:00 EST
1674 :     Tag: blume-20020128-retraction
1675 :     Description:
1676 :    
1677 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
1678 :     with blume-Release_110_38-retracted.)
1679 :    
1680 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
1681 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1682 :    
1683 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1684 :     and enums fixed. The naming of corresponding ML identifiers should
1685 :     now be consistent again.
1686 :    
1687 :     ----------------------------------------------------------------------
1688 : leunga 1038 Name: Allen Leung
1689 :     Date: 2002/01/27
1690 :     Tag: leunga-20020127-nowhere
1691 :     Description:
1692 :    
1693 :     Added a target called nowhere in the configuration scripts.
1694 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
1695 :     programs with where-clauses into legal SML code) during installation.
1696 :    
1697 :     ----------------------------------------------------------------------
1698 : blume 1036 Name: Matthias Blume
1699 : leunga 1038 Date: 2002/01/25 21:27:00 EST
1700 : blume 1039 Tag: blume-Release_110_38-retracted
1701 : blume 1037 Description:
1702 :    
1703 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
1704 :    
1705 :     README will be added later.
1706 :    
1707 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
1708 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
1709 :    
1710 : blume 1037 ----------------------------------------------------------------------
1711 :     Name: Matthias Blume
1712 : blume 1036 Date: 2002/01/25
1713 :     Tag: blume-20020125-ffi
1714 :     Description:
1715 :    
1716 :     A large number of tweaks and improvements to ml-nlffi-lib and
1717 :     ml-nlffigen:
1718 :    
1719 :     - ML represenation types have been streamlined
1720 :     - getter and setter functions work with concrete values, not abstract
1721 :     ones where possible
1722 :     - ml-nlffigen command line more flexible (see README file there)
1723 :     - some bugs have been fixed (hopefully)
1724 :    
1725 :     ----------------------------------------------------------------------
1726 : george 1034 Name: Lal George
1727 :     Date: 2002/01/24
1728 :     Tag: george-20020124-risc-ra-interface
1729 :     Description:
1730 :    
1731 :     There is a dramatic simplification in the interface to the
1732 :     register allocator for RISC architectures as a result of making
1733 :     parallel copy instructions explicit.
1734 :    
1735 :     ----------------------------------------------------------------------
1736 : blume 1028 Name: Matthias Blume
1737 : blume 1032 Date: 2002/01/22
1738 :     Tag: blume-20020122-x86-ccalls
1739 :     Description:
1740 :    
1741 :     Bug fix for c-calls on x86 (having to do with how char- and
1742 :     short-arguments are being handled).
1743 :    
1744 :     ----------------------------------------------------------------------
1745 :     Name: Matthias Blume
1746 : blume 1031 Date: 2002/01/21
1747 :     Tag: blume-20020121-ff
1748 :     Description:
1749 :    
1750 :     Another day of fiddling with the FFI...
1751 :    
1752 :     1. Bug fix/workaround: CKIT does not complain about negative array
1753 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1754 :     (Otherwise a negative dimension would send it into an infinite loop.)
1755 :    
1756 :     2. Some of the abstract types (light objects, light pointers, most "base"
1757 :     types) in structure C are now eqtypes.
1758 :    
1759 :     3. Added constructors and test functions for NULL function pointers.
1760 :    
1761 :     ----------------------------------------------------------------------
1762 :     Name: Matthias Blume
1763 : blume 1028 Date: 2002/01/18
1764 : blume 1030 Tag: blume-20020118-ready-for-new-release
1765 :     Description:
1766 :    
1767 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1768 :     files shortly.)
1769 :    
1770 :     Maybe we christen this to be 110.38?
1771 :    
1772 :     ----------------------------------------------------------------------
1773 :     Name: Matthias Blume
1774 :     Date: 2002/01/18
1775 : blume 1029 Tag: blume-20020118-more-ffifiddle
1776 :     Description:
1777 :    
1778 :     Today's FFI fiddling:
1779 :    
1780 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1781 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1782 :     of abstract ones.
1783 :    
1784 :     - Provided word-style bit operations etc. for "int" representation
1785 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1786 :     and Long.
1787 :    
1788 :     ----------------------------------------------------------------------
1789 :     Name: Matthias Blume
1790 :     Date: 2002/01/18
1791 : blume 1028 Tag: blume-20020118-use-x86-fp
1792 :     Description:
1793 :    
1794 :     Now that x86-fast-fp seems to be working, I turned it back on again
1795 :     by default. (Seems to work fine now, even with the FFI.)
1796 :    
1797 :     Other than that, I added some documentation about the FFI to
1798 :     src/ml-nlffigen/README and updated the FFI test examples in
1799 :     src/ml-nlffi-lib/Tests/*.
1800 :    
1801 :     ----------------------------------------------------------------------
1802 : leunga 1025 Name: Allen Leung
1803 :     Date: 2002/01/17
1804 :     Tag: leunga-20020117-x86-fast-fp-call
1805 :     Description:
1806 :    
1807 :     1. Fixed a problem with handling return fp values when x86's fast fp
1808 :     mode is turned on.
1809 :    
1810 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1811 :     of %f32.
1812 :    
1813 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1814 :    
1815 :     ----------------------------------------------------------------------
1816 : blume 1021 Name: Matthias Blume
1817 :     Date: 2002/01/16
1818 :     Tag: blume-20020116-ffifiddle
1819 :     Description:
1820 :    
1821 :     More fiddling with the FFI interface:
1822 :    
1823 :     - Make constness 'c instead of rw wherever possible. This eliminates
1824 :     the need for certain explicit coercions. (However, due to ML's
1825 :     value polymorphism, there will still be many cases where explicit
1826 :     coercions are necessary. Phantom types are not the whole answer
1827 :     to modeling a subtyping relationship in ML.)
1828 :    
1829 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1830 :    
1831 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1832 :     clumsy workaround for varargs problem.)
1833 :    
1834 :     ----------------------------------------------------------------------
1835 : george 1016 Name: Lal George
1836 :     Date: 2002/01/15
1837 :     Tag: <none>
1838 :     Description:
1839 :    
1840 :     1. Since COPY instructions are no longer native to the architecture,
1841 :     a generic functor can be used to implement the expandCopies function.
1842 :    
1843 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1844 :     TEXT segment.
1845 :    
1846 :     ----------------------------------------------------------------------
1847 : blume 1011 Name: Matthias Blume
1848 : blume 1015 Date: 2002/01/15
1849 :     Tag: blume-20020115-ffiupdates
1850 :     Description:
1851 :    
1852 :     1. Fix for bug resulting in single-precision float values being returned
1853 :     incorrectly from FFI calls.
1854 :    
1855 :     2. Small modifications to C FFI API:
1856 :    
1857 :     - memory-allocation routines return straight objects (no options)
1858 :     and raise an exception in out-of-memory situations
1859 :     - unsafe extensions to cast between function pointers and pointers
1860 :     from/to ints
1861 :     - added structure C_Debug as an alternative to structure C where
1862 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1863 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1864 :     but also takes a (possibly empty) list of existing library handles
1865 :     that the current library depends on
1866 :    
1867 :     ----------------------------------------------------------------------
1868 :     Name: Matthias Blume
1869 : blume 1011 Date: 2002/01/10
1870 :     Tag: blume-20020110-newffigen
1871 :     Description:
1872 :    
1873 :     1. Updates to portable graph code.
1874 :    
1875 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1876 :     more scalable now so that even huge interfaces such as the one
1877 :     for GTK compile in finite time and space. :-)
1878 :     See src/ml-nlffigen/README for details on what's new.
1879 :    
1880 :     ----------------------------------------------------------------------
1881 : george 1003 Name: Lal George
1882 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1883 :     Tag: george-20011206-rm-native-copy
1884 :     Description:
1885 :    
1886 :     Removed the native COPY and FCOPY instructions
1887 :     from all the architectures and replaced it with the
1888 :     explicit COPY instruction from the previous commit.
1889 :    
1890 :     It is now possible to simplify many of the optimizations
1891 :     modules that manipulate copies. This has not been
1892 :     done in this change.
1893 :    
1894 :     ----------------------------------------------------------------------
1895 :     Name: Lal George
1896 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1897 :     Tag: george-20011206-mlrisc-instruction
1898 :     Description:
1899 : george 824
1900 : george 1003 Changed the representation of instructions from being fully abstract
1901 :     to being partially concrete. That is to say:
1902 :    
1903 :     from
1904 :     type instruction
1905 :    
1906 :     to
1907 :     type instr (* machine instruction *)
1908 :    
1909 :     datatype instruction =
1910 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1911 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1912 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1913 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1914 :     | INSTR of instr
1915 :    
1916 :     This makes the handling of certain special instructions that appear on
1917 :     all architectures easier and uniform.
1918 :    
1919 :     LIVE and KILL say that a list of registers are live or killed at the
1920 :     program point where they appear. No spill code is generated when an
1921 :     element of the 'regs' field is spilled, but the register is moved to
1922 :     the 'spilled' (which is present, more for debugging than anything else).
1923 :    
1924 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1925 :     We used to generate:
1926 :    
1927 :     DEFFREG f1
1928 :     f1 := f2 + f3
1929 :     trapb
1930 :    
1931 :     but now generate:
1932 :    
1933 :     f1 := f2 + f3
1934 :     trapb
1935 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1936 :    
1937 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1938 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1939 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1940 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1941 :     this architecture).
1942 :    
1943 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1944 :     all the architectures. This will result in further simplification of the
1945 :     register allocator that must be aware of them for coalescing purposes, and
1946 :     will also simplify certain aspects of the machine description that provides
1947 :     callbacks related to parallel copies.
1948 :    
1949 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1950 :     machine instruction set!
1951 :    
1952 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1953 :     functions for making porting easier -- essentially converting upper case
1954 :     to lower case. All machine instructions (of type instr) are in upper case,
1955 :     and the lower case form generates an MLRISC instruction. For example on
1956 :     the alpha we have:
1957 :    
1958 :     datatype instr =
1959 :     LDA of {r:cell, b:cell, d:operand}
1960 :     | ...
1961 :    
1962 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1963 :     ...
1964 :    
1965 :     where lda is just (INSTR o LDA), etc.
1966 :    
1967 : leunga 815 ----------------------------------------------------------------------
1968 : blume 986 Name: Matthias Blume
1969 : blume 993 Date: 2001/11/22 21:40:00 EST
1970 :     Tag: Release_110_37
1971 :     Description:
1972 :    
1973 :     Release 110.37. This time for real.
1974 :    
1975 :     ----------------------------------------------------------------------
1976 :     Name: Matthias Blume
1977 : blume 988 Date: 2001/11/21 16:35:00 EST
1978 :     Tag: blume-20011121-foot-in-mouth
1979 :     Description:
1980 :    
1981 :     Removed the "Release_110_37" tag because of a serious bug.
1982 :     This will be re-tagged once the bug is fixed.
1983 :    
1984 :     ----------------------------------------------------------------------
1985 :     Name: Matthias Blume
1986 : blume 987 Date: 2001/11/21 16:14:00 EST
1987 :     Tag: blume-20011121-forgottenfile
1988 :     Description:
1989 :    
1990 :     Forgot to add a file. (Just a .tex-file -- part of
1991 :     the CM manual source.)
1992 :    
1993 :     ----------------------------------------------------------------------
1994 :     Name: Matthias Blume
1995 : blume 986 Date: 2001/11/21 16:10:00 EST
1996 : blume 988 Tag: blume-20011121-invalid_110_37
1997 : blume 986 Description:
1998 :    
1999 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
2000 :     because we found a serious bug in all non-x86 backends.
2001 :     - Matthias
2002 :    
2003 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
2004 :     so that code object name strings are directly inserted into code
2005 :     objects at code generation time. The only business the runtime system
2006 :     has with this is now to read the name strings on occasions.
2007 :     (The encoding of the name string has also changed somewhat.)
2008 :    
2009 :     2. CM now implements a simple "set calculus" for specifying export lists.
2010 :     In particular, it is now possible to refer to the export lists of
2011 :     other libraries/groups/sources and form unions as well as differences.
2012 :     See the latest CM manual for details.
2013 :    
2014 :     3. An separate notion of "proxy" libraries has again be eliminated from
2015 :     CM's model. (Proxy libraries are now simply a special case of using
2016 :     the export list calculus.)
2017 :    
2018 :     4. Some of the existing libraries now take advantage of the new set
2019 :     calculus.
2020 :     (Notice that not all libraries have been converted because some
2021 :     of the existing .cm-files are supposed to be backward compatible
2022 :     with 110.0.x.)
2023 :    
2024 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
2025 :     "General.exnMessage"! The former relies on a certain hook to be
2026 :     initialized, and that often does not happen in the stand-alone case.)
2027 :    
2028 :     ----------------------------------------------------------------------
2029 : george 984 Name: Lal George
2030 :     Date: 2001/11/21 13:56:18 EST
2031 :     Tag: george-2001121-pseudo-ops
2032 :     Description:
2033 :    
2034 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
2035 :     ought to never be any question of incompatabilities with
2036 :     pseudo-op syntax expected by host assemblers.
2037 :    
2038 :     For now, only modules supporting GAS syntax are implemented
2039 :     but more should follow, such as MASM, and vendor assembler
2040 :     syntax, e.g. IBM as, Sun as, etc.
2041 :    
2042 :     ----------------------------------------------------------------------
2043 : blume 977 Name: Matthias Blume
2044 :     Date: 2001/11/14 11:52:00 EST
2045 :     Tag: blume-20011114-srcname
2046 :     Description:
2047 :    
2048 :     1. Routed the name of the current source file to mlriscgen where it
2049 :     should be directly emitted into the code object. (This last part
2050 :     is yet to be done.)
2051 :    
2052 :     2. Some cleanup of the pgraph code to make it match the proposal that
2053 :     I put out the other day. (The proposal notwithstanding, things are
2054 :     still in flux here.)
2055 :    
2056 :     ----------------------------------------------------------------------
2057 : george 976 Name: Lal George
2058 :     Date: 2001/11/14 09:44:04 EST
2059 :     Tag:
2060 :     Description:
2061 :    
2062 :     Fix for a backpatching bug reported by Allen.
2063 :    
2064 :     Because the boundary between short and long span-dependent
2065 :     instructions is +/- 128, there are an astounding number of
2066 :     span-dependent instructions whose size is over estimated.
2067 :    
2068 :     Allen came up with the idea of letting the size of span
2069 :     dependent instructions be non-monotonic, for a maxIter
2070 :     number of times, after which the size must be monotonically
2071 :     increasing.
2072 :    
2073 :     This table shows the number of span-dependent instructions
2074 :     whose size was over-estimated as a function of maxIter, for the
2075 :     file Parse/parse/ml.grm.sml:
2076 :    
2077 :     maxIter # of instructions:
2078 :     10 687
2079 :     20 438
2080 :     30 198
2081 :     40 0
2082 :    
2083 :     In compiling the compiler, there is no significant difference in
2084 :     compilation speed between maxIter=10 and maxIter=40. Actually,
2085 :     my measurements showed that maxIter=40 was a tad faster than
2086 :     maxIter=10! Also 96% of the files in the compiler reach a fix
2087 :     point within 13 iterations, so fixing maxIter at 40, while high,
2088 :     is okay.
2089 :    
2090 :     ----------------------------------------------------------------------
2091 : blume 975 Name: Matthias Blume
2092 :     Date: 2001/10/31 15:25:00 EST
2093 :     Tag: blume-20011031-pgraph
2094 :     Description:
2095 :    
2096 :     CKIT:
2097 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
2098 :     argument identifiers.
2099 :     * Changed the return type of TypeUtil.getFunction accordingly.
2100 :     * Type equality ignores the argument names.
2101 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
2102 :     if there is a mismatch.
2103 :    
2104 :     installation script:
2105 :     * attempts to use "curl" if available (unless "wget" is available as well)
2106 :    
2107 :     CM:
2108 :     * has an experimental implementation of "portable graphs" which I will
2109 :     soon propose as an implementation-independent library format
2110 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2111 :    
2112 :     NLFFI-LIB:
2113 :     * some cleanup (all cosmetic)
2114 :    
2115 :     NLFFIGEN:
2116 :     * temporarily disabled the mechanism that suppresses ML output for
2117 :     C definitions whose identifiers start with an underscore character
2118 :     * generate val bindings for enum constants
2119 :     * user can request that only one style (light or heavy) is being used;
2120 :     default is to use both (command-line arguments: -heavy and -light)
2121 :     * fixed bug in handling of function types involving incomplete pointers
2122 :     * generate ML entry points that take record arguments (i.e., using
2123 :     named arguments) for C functions that have a prototype with named
2124 :     arguments
2125 :     (see changes to CKIT)
2126 :    
2127 :     ----------------------------------------------------------------------
2128 : leunga 972 Name: Allen Leung
2129 :     Date: 2001/10/27 20:34:00 EDT
2130 :     Tag: leunga-20011027-x86-fast-fp-call
2131 :     Description:
2132 :    
2133 :     Fixed the bug described in blume-20010920-slowfp.
2134 :    
2135 :     The fix involves
2136 :     1. generating FCOPYs in FSTP in ia32-svid
2137 :     2. marking a CALL with the appropriate annotation
2138 :    
2139 :     ----------------------------------------------------------------------
2140 : blume 961 Name: Matthias Blume
2141 :     Date: 2001/10/16 11:32:00 EDT
2142 :     Tag: blume-20011016-netbsd
2143 :     Description:
2144 :    
2145 :     Underscore patch from Chris Richards (fixing problem with compiling
2146 :     runtime system under recent NetBSD).
2147 :    
2148 :     ----------------------------------------------------------------------
2149 : george 959 Name: Allen Leung
2150 :     Date: 2001/10/12 17:18:32 EDT 2001
2151 :     Tag: leung-20011012-x86-printflowgraph
2152 :     Description:
2153 :    
2154 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2155 :    
2156 :     ----------------------------------------------------------------------
2157 : george 952 Name: Lal George
2158 : george 958 Date: 2001/10/11 23:51:34 EDT
2159 :     Tag: george-20011011-too-many-instrs
2160 :     Description:
2161 :    
2162 :     The representation of a program point never expected to see more
2163 :     than 65536 instructions in a basic block!
2164 :    
2165 :     ----------------------------------------------------------------------
2166 :     Name: Lal George
2167 : george 952 Date: 2001/10/09 09:41:37 EDT
2168 :     Tag: george-20011008-mlrisc-labels
2169 :     Description:
2170 :    
2171 :     Changed the machine description files to support printing of
2172 :     local and global labels in assembly code, based on host assembler
2173 :     conventions.
2174 :    
2175 :     ----------------------------------------------------------------------
2176 : blume 916 Name: Matthias Blume
2177 : blume 939 Date: 2001/09/25 15:25:00 EDT
2178 :     Tag: blume-20010925-exninfo
2179 :     Description:
2180 :    
2181 :     I provided a non-hook implementation of exnName (at the toplevel) and
2182 :     made the "dummy" implementation of exnMessage (at the toplevel) more
2183 :     useful: if nothing gets "hooked in", then at least you are going to
2184 :     see the exception name and a message indicating why you don't see more.
2185 :    
2186 :     [For the time being, programs that need exnMessage and want to use
2187 :     ml-build should either use General.exnMessage (strongly recommended) or
2188 :     refer to structure General at some other point so that CM sees a
2189 :     static dependency.]
2190 :    
2191 :     [Similar remarks go for "print" and "use": If you want to use their
2192 :     functionality in stand-alone programs generated by ml-build, then use
2193 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2194 :    
2195 :     ----------------------------------------------------------------------
2196 :     Name: Matthias Blume
2197 : blume 938 Date: 2001/09/20 17:28:00 EDT
2198 :     Tag: blume-20010920-slowfp
2199 :     Description:
2200 :    
2201 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2202 :    
2203 :     ----------------------------------------------------------------------
2204 :     Name: Matthias Blume
2205 : blume 937 Date: 2001/09/20 17:20:00 EDT
2206 :     Tag: blume-20010920-canonicalpaths
2207 :     Description:
2208 :    
2209 :     0. Updated the BOOT file (something that I forgot to do earlier).
2210 :    
2211 :     1. Small internal change to CM so that it avoids "/../" in filenames
2212 :     as much as possible (but only where it is safe).
2213 :    
2214 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2215 :     that arguments that contain delimiters are passed through correctly.
2216 :     This change also means that all "special" arguments of the form
2217 :     @SMLxxx... must come first.
2218 :    
2219 :     3. Changed install script to put relative anchor names for tool commands
2220 :     into pathconfig.
2221 :    
2222 :     ----------------------------------------------------------------------
2223 :     Name: Matthias Blume
2224 : blume 932 Date: 2001/09/18 15:35:00 EDT
2225 :     Tag: blume-20010918-readme11036
2226 :     Description:
2227 :    
2228 :     Added README files.
2229 :    
2230 :     ----------------------------------------------------------------------
2231 :     Name: Matthias Blume
2232 : blume 929 Date: 2001/09/18 11:45:00 EDT
2233 :     Tag: Release_110_36 (retag)
2234 :     Description:
2235 :    
2236 :     Fixed mistake in config/preloads. Retagged as 110.36.
2237 :    
2238 :     ----------------------------------------------------------------------
2239 :     Name: Matthias Blume
2240 : blume 928 Date: 2001/09/18 09:40:00 EDT
2241 : blume 929 Tag: Release_110_36_orig (tag changed)
2242 : blume 928 Description:
2243 :    
2244 :     New version (110.36). New bootfiles.
2245 :    
2246 :     ----------------------------------------------------------------------
2247 :     Name: Matthias Blume
2248 : blume 927 Date: 2001/09/14 16:15:00 EDT
2249 :     Tag: blume-20010914-x86fastfp
2250 :     Description:
2251 :    
2252 :     John committed some changes that Allen made, in particular a (hopefully)
2253 :     correctly working version of the x86-fp module.
2254 :    
2255 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2256 :     flag to "true". Everything seems to compile to a fixpoint ok, and
2257 :     "mandelbrot" speeds up by about 15%.
2258 :    
2259 :     ----------------------------------------------------------------------
2260 :     Name: Matthias Blume
2261 : blume 923 Date: 2001/09/13 11:20:00 EDT
2262 :     Tag: blume-20010913-minimal
2263 :     Description:
2264 :    
2265 :     1. Stefan Monnier's patch to fix a miscompilation problem that
2266 :     was brought to light by John Reppy's work on Moby.
2267 :    
2268 :     2. Implemented a minimal "structure Compiler" that contains just
2269 :     "version" and "architecture". The minimal version will be
2270 :     available when the full version is not. This is for backward-
2271 :     compatibility with code that wants to test Compiler.version.
2272 :    
2273 :     ----------------------------------------------------------------------
2274 :     Name: Matthias Blume
2275 : blume 920 Date: 2001/08/28 14:03:00 EDT
2276 :     Tag: blume-20010828-ml-lex
2277 :     Description:
2278 :    
2279 :     Fix for bug 1581, received from Neophytos Michael.
2280 :    
2281 :     ----------------------------------------------------------------------
2282 :     Name: Matthias Blume
2283 : blume 918 Date: 2001/08/27 11:20:00 EDT
2284 :     Tag: blume-20010827-readme11035
2285 :     Description:
2286 :    
2287 :     Fleshed out the README file for 110.35.
2288 :    
2289 :     ----------------------------------------------------------------------
2290 :     Name: Matthias Blume
2291 : blume 916 Date: 2001/08/24 17:10:00 EDT
2292 :     Tag: Release_110_35
2293 :     Description:
2294 :    
2295 :     New version number (110.35). New bootfiles.
2296 :    
2297 :     ----------------------------------------------------------------------
2298 : george 915 Name: Lal George
2299 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
2300 :     Tag: george-20010824-MLRISC-graphs
2301 :     Description:
2302 :    
2303 :     removed clusters from MLRISC completely and replaced with graphs.
2304 :    
2305 :     ----------------------------------------------------------------------
2306 :     Name: Matthias Blume
2307 : blume 905 Date: 2001/08/23 17:50:00 EDT
2308 :     Tag: blume-20010823-toplevel
2309 :     Description:
2310 :    
2311 :     - some reorganization of the code that implements various kinds of
2312 :     environments in the compiler (static, dynamic, symbolic, combined)
2313 :     - re-implemented the EnvRef module so that evalStream works properly
2314 :     (if the stream contains references to "use", "CM.make", etc.)
2315 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2316 :    
2317 :     ----------------------------------------------------------------------
2318 :     Name: Matthias Blume
2319 : blume 904 Date: 2001/08/20 15:50 EDT
2320 :     Tag: blume20010820-slipup
2321 :     Description:
2322 :    
2323 :     I forgot to commit a few files. Here they are...
2324 :    
2325 :     ----------------------------------------------------------------------
2326 :     Name: Matthias Blume
2327 : blume 903 Date: 2001/08/20 15:35:00 EDT
2328 :     Tag: blume-20010820-debugprof
2329 :     Description:
2330 :    
2331 :     !!!! NEW BOOTFILES !!!!
2332 :    
2333 :     This is another round of reorganizing the compiler sources. This
2334 :     time the main goal was to factor out all the "instrumentation"
2335 :     passes (for profiling and backtracing) into their own library.
2336 :     The difficulty was to do it in such a way that it does not depend
2337 :     on elaborate.cm but only on elabdata.cm.
2338 :    
2339 :     Therefore there have been further changes to both elaborate.cm and
2340 :     elabdata.cm -- more "generic" things have been moved from the former
2341 :     to the latter. As a result, I was forced to split the assignment
2342 :     of numbers indicating "primtyc"s into two portions: SML-generic and
2343 :     SML/NJ-specific. Since it would have been awkward to maintain,
2344 :     I bit the bullet and actually _changed_ the mapping between these
2345 :     numbers and primtycs. The bottom line of this is that you need
2346 :     a new set of bin- and bootfiles.
2347 :    
2348 :     I have built new bootfiles for all architectures, so doing a fresh
2349 :     checkout and config/install.sh should be all you need.
2350 :    
2351 :     The newly created library's name is
2352 :    
2353 :     $smlnj/viscomp/debugprof.cm
2354 :    
2355 :     and its sources live under
2356 :    
2357 :     src/compiler/DebugProf
2358 :    
2359 :     ----------------------------------------------------------------------
2360 :     Name: Matthias Blume
2361 : blume 902 Date: 2001/08/15 17:15:00 EDT
2362 :     Tag: blume-20010815-compreorg
2363 :     Description:
2364 :    
2365 :     This is a first cut at reorganizing the CM libraries that make up the
2366 :     core of the compiler. The idea is to separate out pieces that could
2367 :     be used independently by tools, e.g., the parser, the typechecker, etc.
2368 :    
2369 :     The current status is a step in this direction, but it is not quite
2370 :     satisfactory yet. Expect more changes in the future.
2371 :    
2372 :     Here is the current (new) organization...
2373 :    
2374 :     What used to be $smlnj/viscomp/core.cm is now divided into
2375 :     six CM libraries:
2376 :    
2377 :     $smlnj/viscomp/basics.cm
2378 :     /parser.cm
2379 :     /elabdata.cm
2380 :     /elaborate.cm
2381 :     /execute.cm
2382 :     /core.cm
2383 :    
2384 :     The CM files for these libraries live under src/system/smlnj/viscomp.
2385 :     All these libraries are proxy libraries that contain precisely
2386 :     one CM library component. Here are the locations of the components
2387 :     (all within the src/compiler tree):
2388 :    
2389 :     Basics/basics.cm
2390 :     Parse/parser.cm
2391 :     ElabData/elabdata.cm
2392 :     Elaborator/elaborate.cm
2393 :     Execution/execute.cm
2394 :     core.cm
2395 :    
2396 :     [This organization is the same that has been used already
2397 :     for a while for the architecture-specific parts of the visible
2398 :     compiler and for the old version of core.cm.]
2399 :    
2400 :     As you will notice, many source files have been moved from their
2401 :     respective original locations to a new home in one of the above
2402 :     subtrees.
2403 :    
2404 :     The division of labor between the new libraries is the following:
2405 :    
2406 :     basics.cm:
2407 :     - Simple, basic definitions that pertain to many (or all) of
2408 :     the other libraries.
2409 :     parser.cm:
2410 :     - The SML parser, producing output of type Ast.dec.
2411 :     - The type family for Ast is also defined and exported here.
2412 :     elabdata.cm:
2413 :     - The datatypes that describe input and output of the elaborator.
2414 :     This includes types, absyn, and static environments.
2415 :     elaborator.cm:
2416 :     - The SML/NJ type checker and elaborator.
2417 :     This maps an Ast.dec (with a given static environment) to
2418 :     an Absyn.dec (with a new static environment).
2419 :     - This libraries implements certain modules that used to be
2420 :     structures as functors (to remove dependencies on FLINT).
2421 :     execute.cm:
2422 :     - Everything having to do with executing binary code objects.
2423 :     - Dynamic environments.
2424 :     core.cm:
2425 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
2426 :     - Top-level modules.
2427 :     - FLINT (this should eventually become its own library)
2428 :    
2429 :     Notes:
2430 :    
2431 :     I am not 100% happy with the way I separated the elaborator (and its
2432 :     data structures) from FLINT. Two instances of the same problem:
2433 :    
2434 :     1. Data structures contain certain fields that carry FLINT-specific
2435 :     information. I hacked around this using exn and the property list
2436 :     module from smlnj-lib. But the fact that there are middle-end
2437 :     specific fields around at all is a bit annoying.
2438 :    
2439 :     2. The elaborator calculates certain FLINT-related information. I tried
2440 :     to make this as abstract as I could using functorization, but, again,
2441 :     the fact that the elaborator has to perform calculations on behalf
2442 :     of the middle-end at all is not nice.
2443 :    
2444 :     3. Having to used exn and property lists is unfortunate because it
2445 :     weakens type checking. The other alternative (parameterizing
2446 :     nearly *everything*) is not appealing, though.
2447 :    
2448 :     I removed the "rebinding =" warning hack because due to the new organization
2449 :     it was awkward to maintain it. As a result, the compiler now issues some of
2450 :     these warnings when compiling init.cmi during bootstrap compilation. On
2451 :     the plus side, you also get a warning when you do, for example:
2452 :     val op = = Int32.+
2453 :     which was not the case up to now.
2454 :    
2455 :     I placed "assign" and "deref" into the _Core structure so that the
2456 :     code that deals with the "lazy" keyword can find them there. This
2457 :     removes the need for having access to the primitive environment
2458 :     during elaboration.
2459 :    
2460 :     ----------------------------------------------------------------------
2461 :     Name: Matthias Blume
2462 : blume 898 Date: 2001/08/13
2463 :     Tag: blume-20010813-closures
2464 :     Description:
2465 :    
2466 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
2467 :     performance of certain loops by avoiding needless closure allocation.
2468 :    
2469 :     ----------------------------------------------------------------------
2470 : george 890 Name: Lal George
2471 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
2472 :     Tag: george-20010731-x86-fmalloc
2473 :     Description: Fixed bug in x86 calls
2474 :    
2475 :     There was a bug where call instructions would mysteriously
2476 :     vanish. The call instruction had to be one that returned
2477 :     a floating point value.
2478 :    
2479 :     ----------------------------------------------------------------------
2480 :     Name: Lal George
2481 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
2482 :     Tag: george-20010719-simple-cells
2483 :     Description:
2484 :    
2485 :     I have dramatically simplified the interface for CELLS in MLRISC.
2486 :    
2487 :     In summary, the cells interface is broken up into three parts:
2488 :    
2489 :     1. CellsBasis : CELLS_BASIS
2490 :    
2491 :     CellsBasis is a top level structure and common for all
2492 :     architectures. it contains the definitions of basic datatypes
2493 :     and utility functions over these types.
2494 :    
2495 :     2. functor Cells() : CELLS
2496 :    
2497 :     Cells generates an interface for CELLS that incorporates the
2498 :     specific resources on the target architecture, such as the
2499 :     presence of special register classes, their number and size,
2500 :     and various useful substructures.
2501 :    
2502 :     3. <ARCH>CELLS
2503 :    
2504 :     e.g. SparcCells: SPARCCELLS
2505 :    
2506 :     <ARCH>CELLS usually contains additional bindings for special
2507 :     registers on the architecture, such as:
2508 :    
2509 :     val r0 : cell (* register zero *)
2510 :     val y : cell (* Y register *)
2511 :     val psr : cell (* processor status register *)
2512 :     ...
2513 :    
2514 :     The structure returned by applying the Cells functor is opened
2515 :     in this interface.
2516 :    
2517 :     The main implication of all this is that the datatypes for cells is
2518 :     split between CellsBasis and CELLS -- a fairly simple change for user
2519 :     code.
2520 :    
2521 :     In the old scheme the CELLS interface had a definitional binding of
2522 :     the form:
2523 :    
2524 :     signature CELLS = sig
2525 :    
2526 :     structure CellsBasis = CellsBasis
2527 :    
2528 :     ...
2529 :    
2530 :     end
2531 :    
2532 :     With all the sharing constraints that goes on in MLRISC, this old
2533 :     design quickly leads to errors such as:
2534 :    
2535 :     "structure definition spec inside of sharing ... "
2536 :    
2537 :    
2538 :     and appears to require an unacceptable amount of sharing and where
2539 :     constraint hackery.
2540 :    
2541 :     I think this error message (the interaction of definitional specs and
2542 :     sharing) requires more explanation on our web page.
2543 :    
2544 :     ----------------------------------------------------------------------
2545 : blume 826 Name: Matthias Blume
2546 : blume 879 Date: 2001/07/19 15:00:00 EDT
2547 :     Tag: blume-20010719-libreorg
2548 :     Description:
2549 :    
2550 :     This update puts together a fairly extensive but straightforward change
2551 :     to the way the libraries that implement the interactive system are
2552 :     organized:
2553 :    
2554 :     The biggest change is the elimination of structure Compiler. As a
2555 :     replacement for this structure, there is now a CM library
2556 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2557 :     that exports all the substructures of the original structure Compiler
2558 :     directly. So instead of saying Compiler.Foo.bar one now simply
2559 :     says Foo.bar. (The CM libraries actually export a collection of
2560 :     structures that is richer than the collection of substructures of
2561 :     structure Compiler.)
2562 :    
2563 :     To make the transition smooth, there is a separate library called
2564 :     $smlnj/compiler/compiler.cm which puts together and exports the
2565 :     original structure Compiler (or at least something very close to it).
2566 :    
2567 :     There are five members of the original structure Compiler
2568 :     that are not exported directly but which instead became members
2569 :     of a new structure Backend (described by signature BACKEND). These are:
2570 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2571 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
2572 :     architecture (: string).
2573 :    
2574 :     Structure Compiler.Version has become structure CompilerVersion.
2575 :    
2576 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2577 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2578 :     or x86, respectively.
2579 :     Each of these exports the same frontend structures that
2580 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
2581 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2582 :     Hppa, PPC, Sparc, or X86, respectively.
2583 :    
2584 :     Library $smlnj/compiler/all.cm exports the union of the exports of
2585 :     $smlnj/compiler/<arch>.cm
2586 :    
2587 :     There are no structures <Arch>Compiler anymore, use
2588 :     $smlnj/compiler/<arch>.cm instead.
2589 :    
2590 :     Library host-compiler-0.cm is gone. Instead, the internal library
2591 :     that instantiates CM is now called cm0.cm. Selection of the host
2592 :     compiler (backend) is no longer done here but. (Responsibility for it
2593 :     now lies with $smlnj/compiler/current.cm. This seems to be more
2594 :     logical.)
2595 :    
2596 :     Many individual files have been moved or renamed. Some files have
2597 :     been split into multiple files, and some "dead" files have been deleted.
2598 :    
2599 :     Aside from these changes to library organization, there are also changes
2600 :     to the way the code itself is organized:
2601 :    
2602 :     Structure Binfile has been re-implemented in such a way that it no
2603 :     longer needs any knowledge of the compiler. It exclusively deals
2604 :     with the details of binfile layout. It no longer invokes the
2605 :     compiler (for the purpose of creating new prospective binfile
2606 :     content), and it no longer has any knowledge of how to interpret
2607 :     pickles.
2608 :    
2609 :     Structure Compile (: COMPILE) has been stripped down to the bare
2610 :     essentials of compilation. It no longer deals with linking/execution.
2611 :     The interface has been cleaned up considerably.
2612 :    
2613 :     Utility routines for dealing with linking and execution have been
2614 :     moved into their own substructures.
2615 :    
2616 :     (The ultimate goal of these changes is to provide a light-weight
2617 :     binfile loader/linker (at least for, e.g., stable libraries) that
2618 :     does not require CM or the compiler to be present.)
2619 :    
2620 :     CM documentation has been updated to reflect the changes to library
2621 :     organization.
2622 :    
2623 :     ----------------------------------------------------------------------
2624 :     Name: Matthias Blume
2625 : blume 875 Date: 2001/07/10 17:30:00 EDT
2626 :     Tag: Release_110_34
2627 :     Description:
2628 :    
2629 : blume 876 Minor tweak to 110.34 (re-tagged):
2630 :    
2631 :     - README.html file added to CVS repository
2632 :     - runtime compiles properly under FreeBSD 3.X and 4.X
2633 :    
2634 :     ----------------------------------------------------------------------
2635 :     Name: Matthias Blume
2636 :     Date: 2001/07/10 17:30:00 EDT
2637 :     Tag: Release_110_34
2638 :     Description:
2639 :    
2640 : blume 875 New version number (110.34). New bootfiles.
2641 :    
2642 :     ----------------------------------------------------------------------
2643 :     Name: Matthias Blume
2644 : blume 874 Date: 2001/07/09 16:00:00 EDT
2645 :     Tag: blume-20010709-more-varargs
2646 :     Description:
2647 :    
2648 :     I changed the handling of varargs in ml-nlffigen again:
2649 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
2650 :    
2651 :     The immediate effect is that you can actually call a varargs function
2652 :     from ML -- but you can't actually supply any arguments beyond the ones
2653 :     specified explicitly. (For example, you can call printf with its format
2654 :     string, but you cannot pass additional arguments.)
2655 :    
2656 :     This behavior is only marginally more useful than the one before, but
2657 :     it has the advantage that a function or, more importantly, a function
2658 :     type never gets dropped on the floor, thus avoiding follow-up problems with
2659 :     other types that refer to the offending one.
2660 :    
2661 :     ----------------------------------------------------------------------
2662 :     Name: Matthias Blume
2663 : blume 873 Date: 2001/07/09 11:25:00 EDT
2664 :     Tag: blume-20010709-varargs
2665 :     Description:
2666 :    
2667 :     1. ckit-lib.cm now exports structure Error
2668 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2669 :     with a warning accompanied by a source location. Moreover, it
2670 :     merely skips the offending function or type and proceeds with the
2671 :     rest of its work.u As a result, one can safely feed C code containing
2672 :     "..." to ml-nlffigen.
2673 :     3. There are some internal improvements to CM, providing slightly
2674 :     more general string substitutions in the tools subsystem.
2675 :    
2676 :     ----------------------------------------------------------------------
2677 :     Name: Matthias Blume
2678 : blume 854 Date: 2001/06/27 15:10:00 EDT
2679 :     Tag: blume-20010627-concur
2680 :     Description:
2681 :    
2682 :     Fixed a small bug in CM's handling of parallel compilation.
2683 :     (You could observe the bug by Control-C-interrupting an ordinary
2684 :     CMB.make or CM.stabilize and then attaching some compile servers.
2685 :     The result was that all of a sudden the previously interrupted
2686 :     compilation would continue on its own. This was because of
2687 :     an over-optimization: CM did not bother to clean out certain queues
2688 :     when no servers were attached "anyway", resulting in the contents
2689 :     of these queues to grab control when new servers did get attached.)
2690 :    
2691 :     There is also another minor update to the CM manual.
2692 :    
2693 :     ----------------------------------------------------------------------
2694 :     Name: Matthias Blume
2695 : blume 853 Date: 2001/06/26 16:15:00 EDT
2696 :     Tag: blume-20010626-cmdoc
2697 :     Description:
2698 :    
2699 :     Minor typo fixed in CM manual (syntax diagram for libraries).
2700 :    
2701 :     ----------------------------------------------------------------------
2702 :     Name: Matthias Blume
2703 : blume 852 Date: 2001/06/25 22:55:00 EDT
2704 :     Tag: blume-20010625-x86pc
2705 :     Description:
2706 :    
2707 :     Fixed a nasty bug in the X86 assembly code that caused signal
2708 :     handlers to fail (crash) randomly.
2709 :    
2710 :     ----------------------------------------------------------------------
2711 :     Name: Matthias Blume
2712 : blume 846 Date: 2001/06/25 12:05:00 EDT
2713 :     Tag: blume-20010625-nlffigen
2714 :     Description:
2715 :    
2716 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
2717 :     Nick Carter <nbc@andrew.cmu.edu>.
2718 :    
2719 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2720 :     2. Default names for generated files are now derived from the name of
2721 :     the C file *without its directory*. In particular, this causes generated
2722 :     files to be placed locally even if the C file is in some system directory.
2723 :     3. Default names for generated signatures and structures are also derived
2724 :     from the C file name without its directory. This avoids silly things
2725 :     like "structure GL/GL".
2726 :     (Other silly names are still possible because ml-nlffigen does not do
2727 : blume 847 a thorough check of whether generated names are legal ML identifiers.
2728 :     When in doubt, use command line arguments to force particular names.)
2729 : blume 846
2730 :     ----------------------------------------------------------------------
2731 :     Name: Matthias Blume
2732 : blume 845 Date: 2001/06/21 12:25:00 EDT
2733 :     Tag: blume-20010621-eXene
2734 :     Description:
2735 :    
2736 :     eXene now compiles and (sort of) works again.
2737 :    
2738 :     The library name (for version > 110.33) is $/eXene.cm.
2739 :    
2740 :     I also added an new example in src/eXene/examples/nbody. See the
2741 :     README file there for details.
2742 :    
2743 :     ----------------------------------------------------------------------
2744 :     Name: Matthias Blume
2745 : blume 844 Date: 2001/06/20 16:40:00 EDT
2746 :     Tag: blume-20010620-cml
2747 :     Description:
2748 :    
2749 :     CML now compiles and works again.
2750 :    
2751 :     Libraries (for version > 110.33):
2752 :    
2753 :     $cml/cml.cm Main CML library.
2754 :     $cml/basis.cm CML's version of $/basis.cm.
2755 :     $cml/cml-internal.cm Internal helper library.
2756 :     $cml/core-cml.cm Internal helper library.
2757 :     $cml-lib/trace-cml.cm Tracing facility.
2758 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2759 :    
2760 :     The installer (config/install.sh) has been taught how to properly
2761 :     install this stuff.
2762 :    
2763 :     ----------------------------------------------------------------------
2764 :     Name: Matthias Blume
2765 : blume 843 Date: 2001/06/19 17:55:00 EDT
2766 :     Tag: blume-20010619-instantiate
2767 :     Description:
2768 :    
2769 :     This un-breaks the fix for bug 1432.
2770 :     (The bug was originally fixed in 110.9 but I broke it again some
2771 :     time after that.)
2772 :    
2773 :     ----------------------------------------------------------------------
2774 :     Name: Matthias Blume
2775 : blume 842 Date: 2001/06/19 17:25:00 EDT
2776 :     Tag: blume-20010619-signals
2777 :     Description:
2778 :    
2779 :     This should (hopefully) fix the long-standing signal handling bug.
2780 :     (The runtime system was constructing a continuation record with an
2781 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2782 :    
2783 :     ----------------------------------------------------------------------
2784 :     Name: Matthias Blume
2785 : blume 840 Date: 2001/06/15 15:05:00 EDT
2786 : blume 841 Tag: blume-20010615-moresparc
2787 :     Description:
2788 :    
2789 :     Here is a short late-hour update related to Sparc c-calls:
2790 :    
2791 :     -- made handling of double-word arguments a bit smarter
2792 :    
2793 :     -- instruction selection phase tries to collapse certain clumsily
2794 :     constructed ML-Trees; typical example:
2795 :    
2796 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2797 :    
2798 :     This currently has no further impact on SML/NJ since mlriscGen does
2799 :     not seem to generate such patterns in the first place, and c-calls
2800 :     (which did generate them in the beginning) has meanwhile been fixed
2801 :     so as to avoid them as well.
2802 :    
2803 :     ----------------------------------------------------------------------
2804 :     Name: Matthias Blume
2805 :     Date: 2001/06/15 15:05:00 EDT
2806 : blume 840 Tag: blume-20010615-sparc
2807 :     Description:
2808 :    
2809 :     The purpose of this update is to provide an implementation of NLFFI
2810 :     on Sparc machines.
2811 :    
2812 :     Here are the changes in detail:
2813 :    
2814 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2815 :     the Sparc implementation of the c-calls API.
2816 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2817 :     for accessing the ML frame. Thus, we have a real frame pointer and
2818 :     can freely modify %sp without need for an omit-frame-ptr phase.
2819 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2820 :     this case.
2821 :     * ml-nlffigen has been taught to produce code for different architectures
2822 :     and calling conventions.
2823 :     * In a way similar to what was done in the x86 case, the Sparc
2824 :     backend uses its own specific extension to mltree. (For example,
2825 :     it needs to be able to generate UNIMP instructions which are part
2826 :     of the calling convention.)
2827 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2828 :     to make it easier to plug in new machine- and os-dependent parts).
2829 :    
2830 :     There are some other fairly unrelated bug fixes and cleanups as well:
2831 :    
2832 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2833 :     that they properly share their libraries with existing SML/NJ libraries.
2834 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2835 :     follow-up error messages.
2836 :     * Updates to CM/CMB documentation.
2837 :    
2838 :     TODO items:
2839 :    
2840 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2841 :     (The current %o2 is a really bad choice because it is part of the
2842 :     calling conventions, so things might interfere in unexpected ways.)
2843 :    
2844 :     ----------------------------------------------------------------------
2845 :     Name: Matthias Blume
2846 : blume 839 Date: 2001/06/07
2847 :     Tag: blume-20010607-calls
2848 :     Description:
2849 :    
2850 :     A number of internal changes related to C calls and calling conventions:
2851 :    
2852 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2853 :     number of bytes popped implicitly (by the callee). In most cases
2854 :     this field is 0 but on x86/win32 it is some non-zero value. This
2855 :     is information provided for the benefit of the "omit-frameptr" pass.
2856 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2857 :     The instruction selection phase copies its value from the ML-Tree
2858 :     CALL statement.
2859 :     3. On all other architectures, the instruction selection phase checks
2860 :     whether "pops=0" and complains if not.
2861 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2862 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2863 :     up after the call and pops is set to 0. For "stdcall", the caller
2864 :     does nothing, leaving the cleanup to the callee; pops is set to
2865 :     the number of bytes that were pushed onto the stack.
2866 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2867 :     distinguish between "ccall" and "stdcall".
2868 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2869 :     set. (This is needed for implementing the official C calling convention
2870 :     on this architecture.)
2871 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2872 :     work with the latest CM.
2873 :    
2874 :     ----------------------------------------------------------------------
2875 :     Name: Matthias Blume
2876 : blume 838 Date: 2001/06/05 15:10:00 EDT
2877 :     Tag: blume-20010605-cm-index
2878 :     Description:
2879 :    
2880 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2881 :    
2882 :     1. CM can now generate "index files". These are human-readable files
2883 :     that list on a per-.cm-file basis each toplevel symbol defined or
2884 :     imported. The location of the index file for
2885 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2886 :     To enable index-file generation, set CM.Control.generate_index to true
2887 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2888 :    
2889 :     The CM manual has been updated accordingly.
2890 :    
2891 :     2. I made some slight modifications to the c-calls API in MLRISC.
2892 :    
2893 :     a) There is now a callback to support saving/restoring of
2894 :     dedicated but caller-save registers around the actual call
2895 :     instruction.
2896 :     b) One can optionally specify a comment-annotation for the
2897 :     call instruction.
2898 :    
2899 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2900 :     (For example, the comment annotation shows the C prototype of
2901 :     the function being called.)
2902 :    
2903 :     ----------------------------------------------------------------------
2904 :     Name: Matthias Blume
2905 : blume 837 Date: 2001/06/01 13:30:00 EDT
2906 :     Tag: blume-20010601-nlffi-cleanup
2907 :     Description:
2908 :    
2909 :     This is mostly a cleanup of MLFFI stuff:
2910 :    
2911 :     - some signature files have been put into a more exposed place
2912 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2913 :     - ml-nlffigen changed accordingly
2914 :     - tutorial updated
2915 :    
2916 :     Other changes:
2917 :    
2918 :     - author's affiliation in CM manual(s) updated
2919 :     - some more recognized keywords added to Allen's sml.sty
2920 :    
2921 :     ----------------------------------------------------------------------
2922 :     Name: Matthias Blume
2923 : blume 836 Date: 2001/05/25 15:30:00 EDT
2924 :     Tag: blume-20010525-iptr
2925 :     Description:
2926 :    
2927 :     - put the official 110.33-README (as it appears on the ftp server) under
2928 :     CVS
2929 :     - fixed a small bug related to incomplete pointer types in
2930 :     ml-nlffigen
2931 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2932 :     (it does not need the 'f type parameter)
2933 :    
2934 :     ----------------------------------------------------------------------
2935 :     Name: Matthias Blume
2936 : blume 834 Date: 2001/05/23 14:30:00 EDT
2937 :     Tag: Release_110_33
2938 :     Description:
2939 :    
2940 :     New version number (110.33). New bootfiles.
2941 :    
2942 :     ----------------------------------------------------------------------
2943 :     Name: Matthias Blume
2944 : blume 833 Date: 2001/05/22 18:06:00 EDT
2945 :     Tag: blume-20010522-targets
2946 :     Description:
2947 :    
2948 :     Made install.sh use file config/targets.customized if it exists, falling
2949 :     back to config/targets if it doesn't. This way one can have a customized
2950 :     version of the targets file without touching the "real thing", thus
2951 :     eliminating the constant fear of accidentally checking something bogus
2952 :     back into the CVS repository... (File config/targets.customized must
2953 :     not be added to the repository!)
2954 :    
2955 :     ----------------------------------------------------------------------
2956 :     Name: Matthias Blume
2957 : blume 832 Date: 2001/05/22 16:30:00 EDT
2958 :     Tag: blume-20010522-minitut
2959 :     Description:
2960 :    
2961 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2962 :     struct returns.
2963 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2964 :     incomplete, preliminary documentation for NLFFI.
2965 :    
2966 :     ----------------------------------------------------------------------
2967 :     Name: Matthias Blume
2968 : blume 829 Date: 2001/05/14 11:30:00 EDT
2969 :     Tag: blume-20010514-script
2970 :     Description:
2971 :    
2972 :     Some bugs in install script fixed.
2973 :    
2974 :     In addition to that I also made a slight change to the NLFFI API:
2975 :     Functors generated by ml-nlffigen now take the dynamic library as a
2976 :     straight functor argument, not as a suspended one. (The original
2977 :     functor code used to force the suspension right away anyway, so there
2978 :     was nothing gained by this complication of the interface.)
2979 :    
2980 :     ----------------------------------------------------------------------
2981 :     Name: Matthias Blume
2982 : blume 828 Date: 2001/05/11 14:35:00 EDT
2983 :     Tag: blume-20010511-ml-nlffi
2984 :     Description:
2985 :    
2986 :     I finally took the plunge and added my new FFI code to the main
2987 :     repository. For x86-linux it is now ready for prime-time.
2988 :    
2989 :     There are two new subdirectories of "src":
2990 :    
2991 :     - ml-nlffi-lib:
2992 :     The utility library for programs using the FFI interface.
2993 :     Here is the implementation of $/c.cm and its associated low-level
2994 :     partners $/c-int.cm and $/memory.cm.
2995 :     - ml-nlffigen:
2996 :     A stand-alone program for generating ML glue code from C source
2997 :     code.
2998 :    
2999 :     Building ml-nlffigen requires $/ckit-lib.cm.
3000 :    
3001 :     The config/install.sh script has been updates to do the Right Thing
3002 :     (hopefully).
3003 :    
3004 :     Notice that the source tree for the C-Kit will not be put under "src"
3005 :     but directly under the installation root directory. (This is the
3006 :     structure that currently exists on the CVS server when you check out
3007 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
3008 :    
3009 :     Bugs: No documentation yet.
3010 :    
3011 :     ----------------------------------------------------------------------
3012 :     Name: Matthias Blume
3013 : blume 826 Date: 2001/05/09 16:35:00 EDT
3014 :     Tag: blume-20010509-cpscontract
3015 :     Description:
3016 :    
3017 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
3018 :     wrapper/unwrapper elimination did not decrement usage counts and some
3019 :     dead variables got overlooked by the dead-up logic.)
3020 :    
3021 :     ----------------------------------------------------------------------
3022 : george 824 Name: Lal George
3023 :     Date: 2001/05/08 17:26:09 EDT
3024 :     Tag: george-20010508-omit-frameptr
3025 :     Description:
3026 :    
3027 :     Changes to implement the omit-frame-pointer optimization to support
3028 :     raw C calls. For now, there is only support on the Intel x86, but
3029 :     other architectures will follow as more experience is gained with this.
3030 :    
3031 :    
3032 :     ----------------------------------------------------------------------
3033 : blume 818 Name: Matthias Blume
3034 : blume 819 Date: 2001/05/07 14:40:00 EDT
3035 :     Tag: blume-20010507-proxies
3036 :     Description:
3037 :    
3038 :     I made into "proxy libraries" all libraries that qualify for such a
3039 :     change. (A qualifying library is a library that has another library or
3040 :     groups as its sole member and repeats that member's export list
3041 :     verbatim. A proxy library avoids this repetition by omitting its export
3042 :     list, effectively inheriting the list that its (only) member exports.
3043 :     See the CM manual for more explanation.)
3044 :     The main effect is that explicit export lists for these libraries
3045 :     do not have to be kepts in sync, making maintenance a bit easier.
3046 :    
3047 :     I also added copyright notices to many .cm-files.
3048 :    
3049 :     Last but not least, I made a new set of bootfiles.
3050 :    
3051 :     ----------------------------------------------------------------------
3052 :     Name: Matthias Blume
3053 : blume 818 Date: 2001/05/04 17:00:00 EDT
3054 :     Tag: blume-20010504-cm-lsplit
3055 :     Description:
3056 :    
3057 :     0. John merged pending changes to $/smlnj-lib.cm
3058 :    
3059 :     1. Allen's previous change accidentally backed out of one of Lal's
3060 :     earlier changes. I undid this mistake (re-introducing Lal's change).
3061 :    
3062 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3063 :     within the compiler where applicable. There is some code simplification
3064 :     because of that.
3065 :    
3066 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
3067 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
3068 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
3069 :     it can now also be controlled on a per-source basis: CM has been taught
3070 :     a new tool parameter applicable to ML source files.
3071 :    
3072 :     - To turn lambda-splitting off completely:
3073 :     local open Compiler.Control.LambdaSplitting in
3074 :     val _ = set Off
3075 :     end
3076 :     - To make "no lambda-splitting" the global default (but allow per-source
3077 :     overriding); this is the initial setting:
3078 :     local open Compiler.Control.LambdaSplitting in
3079 :     val _ = set (Default NONE)
3080 :     end
3081 :     - To make "lambda-splitting with aggressiveness a" the global default
3082 :     (and allow per-source overriding):
3083 :     local open Compiler.Control.LambdaSplitting in
3084 :     val _ = set (Default (SOME a))
3085 :     end
3086 :    
3087 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
3088 :     write (in the respective .cm-file):
3089 :     a.sml (lambdasplitting:off)
3090 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
3091 :     a.sml (lambdasplitting:on)
3092 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3093 :     <a> is a decimal non-negative integer):
3094 :     a.sml (lambdasplitting:<a>)
3095 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
3096 :     a.sml (lambdasplitting:infinity)
3097 :     - To use the global default for a.sml:
3098 :     a.sml (lambdasplitting:default)
3099 :     or simply
3100 :     a.sml
3101 :    
3102 :     ----------------------------------------------------------------------
3103 : leunga 815 Name: Allen Leung
3104 :     Date: 2001/05/04 01:57:00 EDT
3105 :     Tag: leunga-20010504-sync
3106 :     Description:
3107 : leunga 788
3108 : leunga 815 MLRISC features.
3109 :    
3110 :     1. Fix to CMPXCHG instructions.
3111 :     2. Changed RA interface to allow annotations in callbacks.
3112 :     3. Added a new method to the stream interface to allow annotations updates.
3113 :    
3114 : george 749 ----------------------------------------------------------------------
3115 : blume 812 Name: Matthias Blume
3116 : blume 814 Date: 2001/05/01 11:45:00 EDT
3117 :     Tag: blume-20010501-pcedittmp
3118 :     Description:
3119 :    
3120 :     Changed install.sh to use the current working directory instead of
3121 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
3122 :     of /usr/tmp caused trouble with MacOS X because of file premission
3123 :     problems.
3124 :    
3125 :     ----------------------------------------------------------------------
3126 :     Name: Matthias Blume
3127 : blume 812 Date: 2001/04/20 11:10:00 EDT
3128 :     Tag: blume-20010420-inMLflag
3129 :     Description:
3130 :    
3131 :     - added vp_limitPtrMask to vproc-state.h
3132 :     (for use by the raw-C-calls mechanism to implement proper interrupt
3133 :     handling)
3134 :     - made the ML compiler aware of various data-structure offsets so it
3135 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
3136 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
3137 :     raw C-calls
3138 :    
3139 :     ----------------------------------------------------------------------
3140 : george 810 Name: Lal George
3141 :     Date: 2001/04/20 09:15:28 EDT
3142 :     Tag: george-20010420-macosX
3143 :     Description:
3144 :    
3145 :     - Changes to port to Mac OS X; Darwin.
3146 :    
3147 :     - In the process I found that sqrt was broken on the PPC, because the
3148 :     fsqrt instruction is not implemented.
3149 :    
3150 :     ----------------------------------------------------------------------
3151 : blume 799 Name: Matthias Blume
3152 : blume 808 Date: 2001/04/18 12:45:00 EDT
3153 :     Tag: blume-20010418-ccalls
3154 :     Description:
3155 :    
3156 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
3157 :     (this bug prevented structure arguments containing pointers from being
3158 :     passed correctly)
3159 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
3160 :     structure arguments are represented as a pointer to the beginning
3161 :     of the structure (instead of having a series of synthesized arguments,
3162 :     one for each structure member)
3163 :    
3164 :     - made makeml script's verbosity level configurable via environment
3165 :     variable (MAKEML_VERBOSITY)
3166 :    
3167 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3168 : blume 808 in rawmem-x86.sml; we are now using the real thing
3169 :    
3170 :     ----------------------------------------------------------------------
3171 :     Name: Matthias Blume
3172 : blume 806 Date: 2001/03/22 16:25:00 EST
3173 :     Tag: blume-20010322-bootfiles
3174 :     Description:
3175 :    
3176 :     Created a new set of bootfiles (for your automatic installation convenience).
3177 :    
3178 :     ----------------------------------------------------------------------
3179 :     Name: Matthias Blume
3180 : blume 805 Date: 2001/03/22 15:10:00 EST
3181 :     Tag: blume-20010322-rawmem-parcm
3182 :     Description:
3183 :    
3184 :     1. All "raw memory access" primitives for the new FFI are implemented now
3185 :     (at least on the x86).
3186 :     2. Some further cleanup of CM's parallel make mechanism.
3187 :    
3188 :     ----------------------------------------------------------------------
3189 :     Name: Matthias Blume
3190 : blume 801 Date: 2001/03/19 17:53:00 EST
3191 :     Tag: blume-20010319-parallel
3192 :     Description:
3193 :    
3194 :     Parallel make (using compile servers) now works again.
3195 :    
3196 :     To this end, CM.stabilize and CMB.make have been modified to work in
3197 :     two passes when compile servers are attached:
3198 :     1. Compile everything, do not perform stabilization; this pass
3199 :     uses compile servers
3200 :     2. Stabilize everything; this pass does not use compile servers
3201 :     If there are no compile servers, the two passes are combined into one
3202 :     (as before). Splitting the passes increases the inherent parallelism
3203 :     in the dependency graph because the entire graph including all
3204 :     libraries is available at the same time. This, in turn, improves
3205 :     server utilization. The downside is that the master process will
3206 :     have to do some extra work after compilation is done (because for
3207 :     technical reasons it must re-read all the binfiles during stabilization).
3208 :    
3209 :     ----------------------------------------------------------------------
3210 :     Name: Matthias Blume
3211 : blume 800 Date: 2001/03/16 12:22:00 EST
3212 :     Tag: blume-20010316-bootfiles
3213 :     Description:
3214 :    
3215 :     Created a new set of bootfiles (for your automatic installation convenience).
3216 :    
3217 :     ----------------------------------------------------------------------
3218 :     Name: Matthias Blume
3219 : blume 799 Date: 2001/03/16 11:00:00 EST
3220 :     Tag: blume-20010316-MLTREE-fixup
3221 :     Description:
3222 :    
3223 :     This is a minor fixup for an (untagged) earlier commit by Allen.
3224 :     (A file was missing).
3225 :    
3226 :     ----------------------------------------------------------------------
3227 : leunga 788 Name: Allen Leung
3228 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
3229 :     Tag: leunga-20010305-cut-support
3230 :    
3231 :     1. New support for alternative control-flow in MLTREE.
3232 :     Currently we support
3233 :    
3234 :     FLOW_TO(CALL ...., [k1,...,kn])
3235 :    
3236 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
3237 :     in Moby
3238 :    
3239 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3240 :    
3241 :     2. Register Allocator
3242 :    
3243 :     Changes in interface [from Fermin, John]
3244 :    
3245 :     3. Alpha 8-bit SLL support [Fermin]
3246 :    
3247 :     4. All architectures
3248 :    
3249 :     A new module (ClusterExpandCopies) for expanding parallel copies.
3250 :    
3251 :     ----------------------------------------------------------------------
3252 :     Name: Allen Leung
3253 : leunga 788 Date: 2001/02/27 23:07:00 EST
3254 :     Tag: leunga-20010227-minor-stuff
3255 :    
3256 :     1. Alpha bug fix for CMOVNE
3257 :     2. Handle mltree COND(..,FCMP ...,...)
3258 :     3. Bug fix in simplifier
3259 :    
3260 :     ----------------------------------------------------------------------
3261 : blume 777 Name: Matthias Blume
3262 : blume 781 Date: 2001/01/30 17:50:00 EST
3263 :     Tag: blume-20010130-sync
3264 :     Description:
3265 :    
3266 :     This is just a minor update to sync my devel branch with the main brach.
3267 :     The only visible change is the addition of some README files.
3268 :    
3269 :     ----------------------------------------------------------------------
3270 :     Name: Matthias Blume
3271 : blume 778 Date: 2001/01/12 23:30:00 JST
3272 :     Tag: blume-20010112-bootfiles
3273 :     Description:
3274 :    
3275 :     Made a new set of bootfiles that goes with the current state of the
3276 :     repository.
3277 :    
3278 :     ----------------------------------------------------------------------
3279 :     Name: Matthias Blume
3280 : blume 777 Date: 2001/01/12 21:20:00 JST
3281 :     Tag: blume-20010112-sync
3282 :     Description:
3283 :    
3284 :     I am just flushing out some minor changes that had accumulated in
3285 :     my private branch in order to sync with the main tree. (This is
3286 :     mainly because I had CVS trouble when trying to merge _into_ my
3287 :     private branch.)
3288 :    
3289 :     Most people should be completely unaffected by this.
3290 :    
3291 :     ----------------------------------------------------------------------
3292 : leunga 775 Name: Allen Leung
3293 :     Date: Thu Jan 11 21:03:00 EST 2001
3294 :     Tag: leunga-20010111-labexp=mltree
3295 :     Description:
3296 :    
3297 :     1. Removed the type LabelExp and replace it by MLTree.
3298 :     2. Rewritten mltree-simplify with the pattern matcher tool.
3299 :     3. There were some bugs in alpha code generator which would break
3300 :     64-bit code generation.
3301 :     4. Redo the tools to generate code with the
3302 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3303 :     are now generated by perl scripts.
3304 :    
3305 :     ----------------------------------------------------------------------
3306 : blume 771 Name: Matthias Blume
3307 : blume 774 Date: 2001/01/10 21:55:00 JST
3308 :     Tag: blume-20010110-rcc
3309 :     Description:
3310 :    
3311 :     The RCC stuff now seems to work (but only on the x86).
3312 :     This required hacking of the c-calls interface (and -implementation) in
3313 :     MLRISC.
3314 :    
3315 :     Normal compiler users should be unaffected.
3316 :    
3317 :     ----------------------------------------------------------------------
3318 :     Name: Matthias Blume
3319 : blume 773 Date: 2001/01/09 01:20:00 JST
3320 :     Tag: blume-20010109-rcc
3321 :     Description:
3322 :    
3323 :     This is a fairly big patch, flushing out a large number of pending
3324 :     changes that I made to my development copy over the last couple of days.
3325 :    
3326 :     Of practical relevance at this moment is a workaround for a pickling
3327 :     bug that Allen ran into the other day. The cause of the bug itself is
3328 :     still unknown and it might be hard to fix it properly, but the
3329 :     workaround has some merits of its own (namely somewhat reducing pickling
3330 :     overhead for certain libraries). Therefore, I think this solution should
3331 :     be satisfactory at this time.
3332 :    
3333 :     The rest of the changes (i.e., the vast majority) has to do with my
3334 :     ongoing efforts of providing direct support for C function calls from
3335 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
3336 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3337 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3338 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3339 :     handle RCC.
3340 :    
3341 :     The last part (the changes to mlriscGen.sml) are still known to be
3342 :     wrong on the x86 and not implemented on all other architectures. But
3343 :     the infrastructure is in place. I had to change a few functor
3344 :     signatures in the backend to be able to route the CCalls interface
3345 :     from MLRISC there, and I had to specialize the mltree type (on the
3346 :     x86) to include the necessary extensions. (The extensions themselves
3347 :     were already there and redy to go in MLRISC/x86).
3348 :    
3349 :     Everything should be very happy as soon as someone helps me with
3350 :     mlriscGen.sml...
3351 :    
3352 :     In any case, nothing of this should matter to anyone as long as the
3353 :     new primop is not being used (which is going to be the case unless you
3354 :     find it where I hid it :). The rest of the compiler is completely
3355 :     unaffected.
3356 :    
3357 :     ----------------------------------------------------------------------
3358 :     Name: Matthias Blume
3359 : blume 772 Date: 2001/01/05 00:30:00 JST
3360 :     Tag: blume-20010105-primops
3361 :     Description:
3362 :    
3363 :     Added some experimental support for work that I am doing right now.
3364 :     These changes mostly concern added primops, but there is also a new
3365 :     experimental C library in the runtime system (but currently not enabled
3366 :     anywhere except on Linux/X86).
3367 :    
3368 :     In the course of adding primops (and playing with them), I discovered that
3369 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3370 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
3371 :     To recover, I made type information in INL_PRIM mandatory and changed
3372 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
3373 :     has complete, correct type information (i.e., no bottom types).
3374 :    
3375 :     Since all these changes mean that we need new binfiles, I also bumped the
3376 :     version number to 110.32.1.
3377 :    
3378 :     ----------------------------------------------------------------------
3379 :     Name: Matthias Blume
3380 : blume 771 Date: 2000/12/30 22:10:00 JST
3381 :     Tag: blume-20001230-various
3382 :     Description:
3383 :    
3384 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
3385 :     to each other using path anchors. (See CM manual for explanation.)
3386 :    
3387 :     Updated CM documentation.
3388 :    
3389 :     Fixed some bugs in CM.
3390 :    
3391 :     Implemented "proxy" libraries (= syntactic sugar for CM).
3392 :    
3393 :     Added "-quiet" option to makeml and changed runtime system accordingly.
3394 :    
3395 :     Added cleanup handler for exportML to reset timers and compiler stats.
3396 :    
3397 :     ----------------------------------------------------------------------
3398 : george 761 Name: Lal George
3399 :     Date: 2000/12/22 22:22:58 EST 2000
3400 :     Tag: Release_110_32
3401 :     Description:
3402 :    
3403 :     Infinite precision used throughout MLRISC.
3404 :     see MLRISC/mltree/machine-int.sig
3405 :    
3406 :     ----------------------------------------------------------------------
3407 : blume 756 Name: Matthias Blume
3408 : blume 760 Date: 2000/12/22 23:16:00 JST
3409 :     Tag: blume-20001222-warn
3410 :     Description:
3411 : blume 759
3412 : blume 760 Corrected wording and formatting of some CM warning message which I
3413 :     broke in my previous patch.
3414 : blume 759
3415 :     ----------------------------------------------------------------------
3416 :     Name: Matthias Blume
3417 : blume 763 Date: 2000/12/22 21:20:00 JST
3418 :     Tag: blume-20001222-anchorenv
3419 :     Description:
3420 :    
3421 :     Fixed CM's handling of anchor environments in connection with CMB.make.
3422 :    
3423 :     ----------------------------------------------------------------------
3424 :     Name: Matthias Blume
3425 : blume 758 Date: 2000/12/22 13:15:00 JST
3426 :     Tag: blume-20001222-cleanup
3427 :     Description:
3428 :    
3429 :     Removed src/cm/ffi which does not (and did not) belong here.
3430 :    
3431 :     ----------------------------------------------------------------------
3432 :     Name: Matthias Blume
3433 : blume 757 Date: 2000/12/21 23:55:00 JST
3434 :     Tag: blume-20001221-exn
3435 :     Description:
3436 :    
3437 :     Probably most important: CM no longer silently swallows all exceptions
3438 :     in the compiler.
3439 :     Plus: some other minor CM changes. For example, CM now reports some
3440 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
3441 :    
3442 :     ----------------------------------------------------------------------
3443 :     Name: Matthias Blume
3444 : blume 756 Date: 2000/12/15 00:01:05 JST
3445 :     Tag: blume-20001215-dirtool
3446 :     Description:
3447 :    
3448 :     - "dir" tool added.
3449 :     - improvements and cleanup to Tools structure
3450 :     - documentation updates
3451 :    
3452 :     ----------------------------------------------------------------------
3453 : leunga 752 Name: Allen Leung
3454 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
3455 :     Description:
3456 :     Tag: leunga-20001214-int-inf
3457 :     Description:
3458 :    
3459 :     In IntInf, added these standard functions, which are missing from our
3460 :     implementation:
3461 :    
3462 :     andb : int * int -> int
3463 :     xorb : int * int -> int
3464 :     orb : int * int -> int
3465 :     notb : int -> int
3466 :     << : int * word -> int
3467 :     ~>> : int * word -> int
3468 :    
3469 :     Not tested, I hope they are correct.
3470 :    
3471 :     ----------------------------------------------------------------------
3472 :     Name: Allen Leung
3473 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
3474 :     Description:
3475 :     Tag: leunga-20001208-nowhere
3476 :     Description:
3477 :    
3478 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
3479 :     to generate better error messages etc. Plus a brief manual.
3480 :    
3481 :     ----------------------------------------------------------------------
3482 : george 749 Name: Lal George
3483 :     Date: 2000/12/08 09:54:02 EST 2000
3484 :     Tag: Release_110_31
3485 :     Description:
3486 : george 705
3487 : george 749 - Version 110.31
3488 : leunga 591 ----------------------------------------------------------------------
3489 : leunga 744 Name: Allen Leung
3490 :     Date: Thu Dec 7 22:01:04 EST 2000
3491 :     Tag: leunga-20001207-cell-monster-hack
3492 :     Description:
3493 :    
3494 :     Major MLRISC internal changes. Affect all clients.
3495 :     Summary:
3496 :    
3497 :     1. Type CELLS.cell = int is now replaced by a datatype.
3498 :     As a result, the old regmap is now gone. Almost all interfaces
3499 :     in MLRISC change as a consequence.
3500 :    
3501 :     2. A new brand version of machine description tool (v3.0) that generates
3502 :     modules expecting the new interface. The old version is removed.
3503 :    
3504 :     3. The RA interface has been further abstracted into two new functors.
3505 :     RISC_RA and X86RA. These functors have much simpler interfaces.
3506 :     [See also directory MLRISC/demo.]
3507 :    
3508 :     4. Some other new source->source code generation tools are available:
3509 :    
3510 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
3511 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
3512 :     I use this tool to generate the peephole optimizers---with the new
3513 :     cell type changes, peephole rules are becoming difficult to write
3514 :     without conditional pattern matching.
3515 :    
3516 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
3517 :     cover the entire MLRISC source tree so many things broke.
3518 :    
3519 :     6. CM files have been moved to the subdirectory MLRISC/cm.
3520 :     They are moved because there are a lot of them and they clutter up the
3521 :     root dir.
3522 :    
3523 :     7. More detailed documentation to come...
3524 :    
3525 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
3526 :     a makeml -rebuild first. This is because of other other
3527 :     changes that Matthias has made (see below).
3528 :    
3529 :    
3530 :     ----------------------------------------------------------------------
3531 : blume 733 Name: Matthias Blume
3532 : blume 742 Date: 2000/11/30 23:12:00 JST
3533 :     Tag: blume-20001130-filereorg
3534 :     Description:
3535 :    
3536 :     Some manual updates and some file reorganizations in CM.
3537 :    
3538 :     ----------------------------------------------------------------------
3539 :     Name: Matthias Blume
3540 : blume 737 Date: 2000/11/24 17:45:00 JST
3541 :     Tag: blume-20001124-link
3542 :     Description:
3543 :    
3544 :     Drastically improved link traversal code for the case that the dynamic
3545 :     value was already loaded at bootstrap time. As a result, CM and CMB
3546 :     now both load blazingly fast -- even on a very slow machine. Also,
3547 :     memory consumption has been further reduced by this.
3548 :    
3549 :     Warning: The format of the PIDMAP file has changed. THerefore, to
3550 :     bootstrap you have to do this:
3551 :    
3552 :     1. Run CMB.make
3553 :     2. Make a symbolic link for the boot directory:
3554 :     ln -s sml.boot.ARCH-OS xxx
3555 :     3. "Rebuild" the boot directory:
3556 :     ./makeml -boot xxx -rebuild sml ; rm xxx
3557 :     4. Boot normally:
3558 :     ./makeml
3559 :    
3560 :     ----------------------------------------------------------------------
3561 :     Name: Matthias Blume
3562 : blume 735 Date: 2000/11/21 21:20:00 JST
3563 :     Tag: blume-20001121-tools
3564 :     Description:
3565 :    
3566 :     Continued hacking on autoloading problem -- with success this time.
3567 :     Also changed tool-plugin mechanism. See new CM manual.
3568 :    
3569 :     ----------------------------------------------------------------------
3570 :     Name: Matthias Blume
3571 : blume 734 Date: 2000/11/19 14:30:00 JST
3572 :     Tag: blume-20001119-autoload
3573 :     Description:
3574 :    
3575 :     Some hacking to make autoloading faster. Success for CMB, no success
3576 :     so far for CM. There is a reduced structure CM' that autoloads faster.
3577 :     (This is a temporary, non-documented hack to be eliminated again when
3578 :     the general problem is solved.)
3579 :    
3580 :     ----------------------------------------------------------------------
3581 :     Name: Matthias Blume
3582 : blume 733 Date: 2000/11/17 14:10:00 JST
3583 :     Tag: blume-20001117-pickle-lib
3584 :     Description:
3585 :    
3586 :     1. Eliminated comp-lib.cm
3587 :     2. Made pickle-lib.cm
3588 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
3589 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
3590 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
3591 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
3592 :     Duplicated intmap modules thrown out.
3593 :     6. Hunted down all duplicated SCC code and replaced it with
3594 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
3595 :     7. Rewrote Feedback module.
3596 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
3597 :     should be thrown out and equivalent modules from smlnj-lib.cm
3598 :     should be used (IntRedBlackSet, IntListSet, ...).
3599 :    
3600 :     Confirmed that compiler compiles to fixpoint.
3601 :    
3602 :     ----------------------------------------------------------------------
3603 : leunga 731 Name: Allen Leung
3604 :     Date: 2000/11/10 18:00:00
3605 :     Tag: leunga-20001110-new-x86-fp
3606 :    
3607 :     A new x86 floating point code generator has been added.
3608 :     By default this is turned off. To turn this on, do:
3609 :    
3610 :     CM.autoload "$smlnj/compiler.cm";
3611 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3612 :    
3613 :     Changes:
3614 :    
3615 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
3616 :     2. Changed the extension callback for FTANGENT to generate:
3617 :    
3618 :     fptan
3619 :     fstp %st(0)
3620 :     instead of
3621 :     fptan
3622 :     fstpl ftempmem
3623 :    
3624 :     3. Numerous assembly fixes for x86.
3625 :    
3626 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
3627 :     support for a whole bunch of instructions and addressing modes:
3628 :    
3629 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
3630 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
3631 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
3632 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3633 :     fxch %st(n)
3634 :     fld %st(n)
3635 :     fst %st(n)
3636 :     fst mem
3637 :     fstp %st(n)
3638 :     fucom %st(n)
3639 :     fucomp %st(n)
3640 :    
3641 :     All these are now generated when the fast fp mode is turned on.
3642 :    
3643 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3644 :    
3645 :     ----------------------------------------------------------------------
3646 : blume 729 Name: Matthias Blume
3647 :     Date: 2000/11/09 11:20:00 JST
3648 :     Tag: blume-20001109-scc
3649 :     Description:
3650 :    
3651 :     Eliminated some code duplication:
3652 :    
3653 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
3654 :     (Otherwise the functor is useless.)
3655 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
3656 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
3657 :    
3658 :     ----------------------------------------------------------------------
3659 : george 721 Name: Lal George
3660 :     Date: 2000/11/06 09:02:21 EST 2000
3661 :     Tag: Release_110_30
3662 :     Description:
3663 :    
3664 :     - Version 110.30
3665 :     ----------------------------------------------------------------------
3666 : blume 715 Name: Matthias Blume
3667 : blume 716 Date: 2000/11/04 14:45:00
3668 :     Tag: blume-20001104-mlbuild
3669 :     Description:
3670 :    
3671 :     - Made ml-build faster on startup.
3672 :     - Documentation fixes.
3673 :    
3674 :     ----------------------------------------------------------------------
3675 :     Name: Matthias Blume
3676 : blume 715 Date: 2000/11/02 17:00:00 JST
3677 :     Tag: blume-20001102-condcomp
3678 :     Description:
3679 :    
3680 :     - Small tweaks to pickler -- new BOOTFILES!
3681 :     - Version bumped to 110.29.2.
3682 :     - Added conditional compilation facility to init.cmi (see comment there).
3683 :     ----------------------------------------------------------------------
3684 : leunga 713 Name: Allen Leung
3685 :     Date: 2000/10/23 19:31:00
3686 :     Tag: leunga-20001023-demo-ra
3687 :    
3688 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
3689 :     2. Test programs for the graph library updated
3690 :     3. Some new MLRISC demo programs added
3691 :    
3692 :     ----------------------------------------------------------------------
3693 : blume 710 Name: Matthias Blume
3694 : blume 711 Date: 2000/08/31 22:15:00 JST
3695 :     Tag: blume-20001017-errmsg
3696 :     Description:
3697 :    
3698 :     More error message grief: Where there used to be no messages, there
3699 :     now were some that had bogus error regions. Fixed.
3700 :    
3701 :     ----------------------------------------------------------------------
3702 :     Name: Matthias Blume
3703 : blume 710 Date: 2000/08/31 17:30:00 JST
3704 :     Tag: blume-20001017-v110p29p1
3705 :     Description:
3706 :    
3707 :     I made a version 110.29.1 with new bootfiles.
3708 :    
3709 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
3710 :     CM documentation changes and a small bugfix in CM's error reporting.
3711 :    
3712 :     ----------------------------------------------------------------------
3713 : george 705 Name: Lal George
3714 :     Date: 2000/09/27 14:42:35 EDT
3715 :     Tag: george-20000927-nodestatus
3716 :     Description:
3717 :    
3718 :     Changed the type of the nodestatus, so that:
3719 :    
3720 :     SPILLED(~1) is now SPILLED
3721 :     SPILLED(m) where m>=0 is now MEMREG(m)
3722 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
3723 :    
3724 :     ----------------------------------------------------------------------
3725 : blume 697 Name: Matthias Blume
3726 : blume 703 Date: 2000/09/07 14:45:00 JST
3727 :     Tag: blume-20000907-cmerrmsg
3728 :     Description:
3729 :    
3730 :     Small tweak to CM to avoid getting ML syntax error messages twice.
3731 :    
3732 :     ----------------------------------------------------------------------
3733 :     Name: Matthias Blume
3734 : blume 701 Date: 2000/08/31 18:00:00 JST
3735 :     Tag: blume-20000831-cvsbootfiles
3736 :     Description:
3737 :    
3738 :     New URL for boot files (because the 110.29 files on the BL server do
3739 :     now work correctly with my updated install scripts for yacc and lex).
3740 :    
3741 :     ----------------------------------------------------------------------
3742 :     Name: Matthias Blume
3743 : blume 697 Date: 2000/08/08 12:33:00 JST
3744 :     Tag: blume-20000808-manual
3745 :     Description:
3746 :    
3747 :     Tiny update to CM manual.
3748 :    
3749 :     ----------------------------------------------------------------------
3750 : leunga 695 Name: Allen Leung
3751 :     Date: 2000/08/7 19:31:00
3752 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3753 :    
3754 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3755 :     and Mobdy.
3756 :    
3757 :     1. x86
3758 :    
3759 :     a. Fixes to peephole module by John and Dan.
3760 :     b. Assembly fix to SETcc by Allen.
3761 :     c. Fix to c-call by John.
3762 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3763 :     e. Instruction selection optimization to SETcc as suggested by John.
3764 :    
3765 :     For example,
3766 :    
3767 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3768 :    
3769 :     should generate:
3770 :    
3771 :     MOVL a, x
3772 :     SUBL b, x
3773 :     SHRL 31, x
3774 :    
3775 :     2. IR stuff
3776 :    
3777 :     A bunch of new DJ-graph related algorithms added. These
3778 :     speed up SSA construction.
3779 :    
3780 :     3. SSA + Scheduling
3781 :    
3782 :     Added code for SSA and scheduling to the repository
3783 :    
3784 :     ----------------------------------------------------------------------
3785 : george 694 Name: Lal George
3786 :     Date: 2000/07/27 11:53:14 EDT
3787 :    
3788 :     Tag: lal-20000727-linux-ppc
3789 :     Description:
3790 :    
3791 :     Made changes to support Linux PPC.
3792 :     p.s. I have confirmation that the 110.29 boot files work fine.
3793 :    
3794 :     ----------------------------------------------------------------------
3795 : blume 684 Name: Matthias Blume
3796 : blume 692 Date: 2000/07/27 17:40:00 JST
3797 :     Tag: blume-20000727-scripts
3798 :     Description:
3799 :    
3800 :     !!!! WARNING !!!!
3801 :     You must recompile the runtime system!
3802 :     !!!! WARNING !!!!
3803 :    
3804 :     This is basically another round of script-enhancements:
3805 :    
3806 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3807 :     and undefine CM preprocessor symbols.
3808 :    
3809 :     2. ml-build avoids generating a new heap image if it finds that the
3810 :     existing one is still ok. (The condition is that no ML file had to
3811 :     be recompiled and all ML files are found to be older that the heap
3812 :     file.)
3813 :    
3814 :     To make this work smoothly, I also hacked the runtime system as
3815 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3816 :     (.sparc-solaris, ...) that is currently being used.
3817 :    
3818 :     Moreover, the signature of CM.mk_standalone has changed. See the
3819 :     CM manual.
3820 :    
3821 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3822 :     CM manual for details.)
3823 :    
3824 :     4. More CM manual updates:
3825 :     - all of the above has been documented.
3826 :     - there is now a section describing the (CM-related) command line
3827 :     arguments that are accepted by the "sml" command
3828 :    
3829 :     ----------------------------------------------------------------------
3830 :     Name: Matthias Blume
3831 : blume 691 Date: 2000/07/25 16:20:00 JST
3832 :     Tag: blume-20000725-makedepend
3833 :     Description:
3834 :    
3835 :     Added a script called ml-makedepend. This can be used in makefiles
3836 :     for Unix' make in a way very similar to the "makedepend" command for
3837 :     C.
3838 :    
3839 :     The script internally uses function CM.sources.
3840 :    
3841 :     Synopsis:
3842 :    
3843 :     ml-makedepend [-f makefile] cmfile targetname
3844 :    
3845 :     The default for the makefile is "makefile" (or "Makefile" should
3846 :     "makefile" not exist).
3847 :    
3848 :     ml-makedepend adds a cmfile/targetname-specific section to this
3849 :     makefile (after removing the previous version of this section). The
3850 :     section contains a single dependency specification with targetname on
3851 :     the LHS (targetname is an arbitrary name), and a list of files derived
3852 :     from the cmfile on the RHS. Some of the files on the RHS are
3853 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3854 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3855 :     path names. The makefile writer is responsible for making sure that
3856 :     these variables have correct at the time "make" is invoked.
3857 :    
3858 :     ----------------------------------------------------------------------
3859 :     Name: Matthias Blume
3860 : blume 690 Date: 2000/07/22 23:30:00 JST
3861 :     Tag: blume-20000722-urlupdate
3862 :     Description:
3863 :    
3864 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3865 :    
3866 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3867 :    
3868 :     ----------------------------------------------------------------------
3869 :     Name: Matthias Blume
3870 : blume 686 Date: 2000/07/18 18:00:00 JST
3871 :     Tag: blume-20000718-Version_110_29
3872 :     Description:
3873 :    
3874 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3875 :    
3876 :     2. Updated config/version to 110.29
3877 :    
3878 :     3. Updated config/srcarchiveurl
3879 :    
3880 :     3. New boot files!
3881 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3882 :    
3883 :     ----------------------------------------------------------------------
3884 :     Name: Matthias Blume
3885 : blume 684 Date: 2000/07/11 13:58:00 JST
3886 :     Tag: blume-20000711-doctypo
3887 :     Description:
3888 :    
3889 :     Fixed a few typos in CM manual.
3890 :    
3891 :     ----------------------------------------------------------------------
3892 : leunga 683 Name: Allen Leung
3893 :     Date: 2000/06/15 00:38:00
3894 :     Tag: leunga-20000704-sparc-x86
3895 :    
3896 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3897 :     2. fix to x86 RET bug [found by Dan Grossman]
3898 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3899 :    
3900 :     Affects c-- and moby only
3901 :    
3902 :     ----------------------------------------------------------------------
3903 : blume 680 Name: Matthias Blume
3904 : blume 682 Date: 2000/07/04 15:26:00
3905 :     Tag: blume-20000704-trigger
3906 :     Description:
3907 :    
3908 :     1. Improvements to CM manual.
3909 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3910 :     of getting a back-trace. The function, when called, raises an
3911 :     internal exception which explicitly carries the full back-trace history,
3912 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3913 :     or not). The interactive loop will print that history once it arrives
3914 :     at top level.
3915 :     Short of having all exceptions implicitly carry the full history, the
3916 :     recommended way of using this facility is:
3917 :     - compile your program with instrumentation "on"
3918 :     - run it, when it raises an exception, look at the history
3919 :     - if the history is "cut off" because of some handler, go and modify
3920 :     your program so that it explicitly calls BTrace.trigger
3921 :     - recompile (still instrumented), and rerun; look at the full history
3922 :    
3923 :     ----------------------------------------------------------------------
3924 :     Name: Matthias Blume
3925 :     Date: 2000/07/03 15:36:00 JST
3926 : blume 680 Tag: blume-20000702-manual
3927 :     Description:
3928 :    
3929 :     Small corrections and updates to CM manual.
3930 :    
3931 :     ----------------------------------------------------------------------
3932 : blume 682 Name: Matthias Blume
3933 : blume 679 Date: 2000/06/29 16:04:00 JST
3934 :     Tag: blume-20000629-yacctool
3935 :     Description:
3936 :    
3937 :     Changes:
3938 :    
3939 :     1. Class "mlyacc" now takes separate arguments to pass options to
3940 :     generated .sml- and .sig-files independently.
3941 :     2. Corresponding CM manual updates.
3942 :     3. BTrace module now also reports call sites. (However, for loop clusters
3943 :     it only shows from where the cluster was entered.) There are associated
3944 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3945 :    
3946 :     ----------------------------------------------------------------------
3947 : blume 675 Name: Matthias Blume
3948 : blume 678 Date: 2000/06/27 16:51:00 JST
3949 :     Tag: blume-20000627-noweb
3950 :     Description:
3951 :    
3952 :     Changes:
3953 :    
3954 :     1. Implemented "subdir" and "witness" options for noweb tool.
3955 :     This caused some slight internal changes in CM's tool implementation.
3956 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3957 :     some remaining issues from earlier path anchor changes.
3958 :     3. Updated CM manual accordingly.
3959 :    
3960 :     4. Changed implementation of back-tracing so that I now consider it
3961 :     ready for prime-time.
3962 :    
3963 :     In particular, you don't have to explicitly trigger the back-trace
3964 :     anymore. Instead, if you are running BTrace-instrumented code and
3965 :     there is an uncaught exception (regardless of whether or not it was
3966 :     raised in instrumented code), the top-level evalloop will print
3967 :     the back-trace.
3968 :    
3969 :     Features:
3970 :    
3971 :     - Instrumented and uninstrumented code work together seemlessly.
3972 :     (Of course, uninstrumented code is never mentioned in actual
3973 :     back-traces.)
3974 :    
3975 :     - Asymptotic time- and space-complexity of instrumented code is
3976 :     equal to that of uninstrumented code. (This means that
3977 :     tail-recursion is preserved by the instrumentation phase.)
3978 :    
3979 :     - Modules whose code has been instrumented in different sessions
3980 :     work together without problem.
3981 :    
3982 :     - There is no penalty whatsoever on uninstrumented code.
3983 :    
3984 :     - There is no penalty on "raise" expressions, even in
3985 :     instrumented code.
3986 :    
3987 :     A potential bug (or perhaps it is a feature, too):
3988 :    
3989 :     A back-trace reaches no further than the outermost instrumented
3990 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3991 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3992 :    
3993 :     <exp> handle e => raise e
3994 :    
3995 :     and even
3996 :    
3997 :     <exp> handle Foo => raise Bar
3998 :    
3999 :     and, of course, through
4000 :    
4001 :     <exp> handle Foo => ...
4002 :    
4003 :     if the exception was not Foo.
4004 :    
4005 :     Back-traces always reach right through any un-instrumented code
4006 :     including any of its "handle" expressions, trivial or not.
4007 :    
4008 :     To try this out, do the following:
4009 :    
4010 :     - Erase all existing binfiles for your program.
4011 :     (You may keep binfiles for those modules where you think you
4012 :     definitely don't need back-tracing.)
4013 :     - Turn on back-trace instrumentation:
4014 :     SMLofNJ.Internals.BTrace.mode (SOME true);
4015 :     - Recompile your program. (I.e., run "CM.make" or "use".)
4016 :     - You may now turn instrumentation off again (if you want):
4017 :     SMLofNJ.Internals.BTrace.mode (SOME false);
4018 :     - Run your program as usual. If it raises an exception that
4019 :     reaches the interactive toplevel, then a back-trace will
4020 :     automatically be printed. After that, the toplevel loop
4021 :     will print the exception history as usual.
4022 :    
4023 :     ----------------------------------------------------------------------
4024 :     Name: Matthias Blume
4025 : blume 677 Date: 2000/06/26 09:56:46 JST
4026 :     Tag: blume-20000626-setup
4027 :     Description:
4028 :    
4029 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
4030 :     ML code before and after compiling a file (e.g., to set compiler
4031 :     flags)
4032 :    
4033 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
4034 :     - associated changes to btrace.sml (BTrace instrumentation pass)
4035 :     - cleaner implementation of btimp.sml (BTrace tracing and report
4036 :     module)
4037 :    
4038 :     CM manual: * new path encoding documented
4039 :     * description of setup-parameter to "sml" added
4040 :    
4041 :     The biggest user-visible change to back-tracing is that it is no
4042 :     longer necessary to compile all traced modules within the same
4043 :     session. (This was a real limitation.)
4044 :    
4045 :     ----------------------------------------------------------------------
4046 :     Name: Matthias Blume
4047 : blume 676 Date: 2000/06/24 12:40:00 JST
4048 :     Tag: blume-20000624-startup
4049 :     Description:
4050 :    
4051 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
4052 :     bit too often -- to put it mildly. :)
4053 :    
4054 :     ----------------------------------------------------------------------
4055 :     Name: Matthias Blume
4056 : blume 675 Date: 2000/06/23 18:20:00 JST
4057 :     Tag: blume-20000623-btrace
4058 :     Description:
4059 :    
4060 :     This updates adds a backtrace facility to aid programmers in debugging
4061 :     their programs. This involves the following changes:
4062 :    
4063 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
4064 :     keeping track of the current call stack. When programs are compiled
4065 :     in a special mode, the compiler will insert calls to these hooks
4066 :     into the user program.
4067 :     "Hook" means that it is possible for different implementations of
4068 :     back-tracing to register themselves (at different times).
4069 :    
4070 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
4071 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
4072 :     It can be turned on using this call:
4073 :     SMLofNJ.Internals.BTrace.mode (SOME true);
4074 :     Turning it off again:
4075 :     SMLofNJ.Internals.BTrace.mode (SOME false);
4076 :     Querying the current status:
4077 :     SMLofNJ.Internals.BTrace.mode NONE;
4078 :     Annotated programs are about twice as big as normal ones, and they
4079 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
4080 :     where all hooks do nothing). The slowdown with a plugin that is
4081 :     actually useful (such as the one supplied by default) is even greater,
4082 :     but in the case of the default plugin it is still only an constant
4083 :     factor (amortized).
4084 :    
4085 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
4086 :     with a sub-structure BTrace for controlling back-tracing. In particular,
4087 :     the above-mentioned function "mode" controls whether the annotation
4088 :     phase is invoked by the compiler. Another important function is
4089 :     "trigger": when called it aborts the current execution and causes
4090 :     the top-level loop to print a full back-trace.
4091 :    
4092 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
4093 :     for back-tracing. It keeps track of the dynamic call stack and in
4094 :     addition to that it keeps a partial history at each "level" of that
4095 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
4096 :     d and b (at separate times, dynamically), then the report will show:
4097 :    
4098 :     GOTO d
4099 :     /c
4100 :     GOTO \b
4101 :     CALL a
4102 :    
4103 :     This shows that there was an initial non-tail call of a, then a
4104 :     tail-call to b or c, looping behavior in a cluster of functions that
4105 :     consist of b and c, and then a goto from that cluster (i.e., either from
4106 :     b or from c) to d.
4107 :    
4108 :     Note that (depending on the user program) the amount of information
4109 :     that the back-trace module has to keep track of at each level is bounded
4110 :     by a constant. Thus, the whole implementation has the same asymptotical
4111 :     complexity as the original program (both in space and in time).
4112 :    
4113 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
4114 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
4115 :     which is raised by the "trigger" function mentioned above.
4116 :    
4117 :     Notes on usage:
4118 :    
4119 :     - Annotated code works well together with unannotated code:
4120 :     Unannotated calls simply do not show up at all in the backtrace.
4121 :    
4122 :     - It is not a good idea to let modules that were annotated during
4123 :     different sessions run at the same time. This is because the compiler
4124 :     chooses small integers to identify individual functions, and there
4125 :     will be clashes if different modules were compiled in separate sessions.
4126 :     (Nothing will crash, and you will even be told about the clashes, but
4127 :     back-trace information will in general not be useful.)
4128 :    
4129 :     - Back-tracing can be confused by callcc and capture.
4130 :    
4131 :     - The only way of getting a back-trace right now is to explicitly
4132 :     invoke the "trigger" function from your user program. Eventually, we
4133 :     should make every exception carry back-trace information (if
4134 :     available). But since this creates more overhead at "raise"-time
4135 :     (similar to the current exnHistory overhead), I have not yet
4136 :     implemented this. (The implementation will be rather easy.) With
4137 :     exceptions carrying back-trace information, this facility will be even
4138 :     more useful because users don't need to modify their programs...
4139 :    
4140 :     - While it is possible to compile the compiler with back-trace
4141 :     annotations turned on (I did it to get some confidence in
4142 :     correctness), you must make absolutely sure that core.sml and
4143 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
4144 :     be compiled with annotation because there is no core access yet, but
4145 :     if you compile btimp.sml with annotation, then the system will go into
4146 :     an infinite recursion and crash.)
4147 :     Since CM currently does not know about BTrace, the only way to turn
4148 :     annotations on and off for different modules of the compiler is to
4149 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
4150 :     this is awkward and clumsy.
4151 :    
4152 :     Sample sessions:
4153 :    
4154 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4155 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4156 :     [autoloading]
4157 :     [autoloading done]
4158 :     val it = false : bool
4159 :     - structure X = struct
4160 :     - fun main n = let
4161 :     - fun a (x, 0) = d x
4162 :     - | a (x, n) = b (x, n - 1)
4163 :     - and b (x, n) = c (x, n)
4164 :     - and c (x, n) = a (x, n)
4165 :     - and d x = e (x, 3)
4166 :     - and e (x, 0) = f x
4167 :     - | e (x, n) = e (x, n - 1)
4168 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
4169 :     - | f n = n * g (n - 1)
4170 :     - and g n = a (n, 3)
4171 :     - in
4172 :     - f n
4173 :     - end
4174 :     - end;
4175 :     structure X : sig val main : int -> int end
4176 :     - X.main 3;
4177 :     *** BACK-TRACE ***
4178 :     GOTO stdIn:4.2-13.20: X.main[2].f
4179 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4180 :     GOTO stdIn:4.2-13.20: X.main[2].d
4181 :     / stdIn:4.2-13.20: X.main[2].a
4182 :     | stdIn:4.2-13.20: X.main[2].b
4183 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4184 :     CALL stdIn:4.2-13.20: X.main[2].g
4185 :     GOTO stdIn:4.2-13.20: X.main[2].f
4186 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4187 :     GOTO stdIn:4.2-13.20: X.main[2].d
4188 :     / stdIn:4.2-13.20: X.main[2].a
4189 :     | stdIn:4.2-13.20: X.main[2].b
4190 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4191 :     CALL stdIn:4.2-13.20: X.main[2].g
4192 :     GOTO stdIn:4.2-13.20: X.main[2].f
4193 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4194 :     GOTO stdIn:4.2-13.20: X.main[2].d
4195 :     / stdIn:4.2-13.20: X.main[2].a
4196 :     | stdIn:4.2-13.20: X.main[2].b
4197 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4198 :     CALL stdIn:4.2-13.20: X.main[2].g
4199 :     GOTO stdIn:4.2-13.20: X.main[2].f
4200 :     CALL stdIn:2.15-17.4: X.main[2]
4201 :     -
4202 :    
4203 :     (Note that because of a FLINt bug the above code currently does not
4204 :     compile without BTrace turned on.)
4205 :    
4206 :     Here is another example, using my modified Tiger compiler:
4207 :    
4208 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4209 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4210 :     [autoloading]
4211 :     [autoloading done]
4212 :     val it = false : bool
4213 :     - CM.make "sources.cm";
4214 :     [autoloading]
4215 :     ...
4216 :     [autoloading done]
4217 :     [scanning sources.cm]
4218 :     [parsing (sources.cm):parse.sml]
4219 :     [creating directory CM/SKEL ...]
4220 :     [parsing (sources.cm):tiger.lex.sml]
4221 :     ...
4222 :     [wrote CM/sparc-unix/semant.sml]
4223 :     [compiling (sources.cm):main.sml]
4224 :     [wrote CM/sparc-unix/main.sml]
4225 :     [New bindings added.]
4226 :     val it = true : bool
4227 :     - Main.compile ("../testcases/merge.tig", "foo.out");
4228 :     *** BACK-TRACE ***
4229 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
4230 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4231 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4232 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4233 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4234 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4235 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
4236 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
4237 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
4238 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4239 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
4240 :     CALL main.sml:1.18-118.4: Main.compile[2]
4241 :     -
4242 :    
4243 :     ----------------------------------------------------------------------
4244 : blume 673 Name: Matthias Blumen
4245 :     Date: 2000/06/21 18:00:00 JST
4246 :     Tag: blume-20000621-manual
4247 :     Description:
4248 :    
4249 :     CM manual update: Path environments documented.
4250 :    
4251 :     ----------------------------------------------------------------------
4252 : blume 666 Name: Matthias Blume
4253 : blume 672 Date: 2000/06/19 13:40:00
4254 :     Tag: blume-20000619-manual
4255 :     Description:
4256 :    
4257 :     CM manual and system/README update. This only covers the fact that
4258 :     there are no more implicit anchors. (Path environments and the "bind"
4259 :     option to "cm" have yet to be documented.)
4260 :    
4261 :     ----------------------------------------------------------------------
4262 :     Name: Matthias Blume
4263 : blume 671 Date: 2000/06/19 11:05:00 JST
4264 :     Tag: blume-20000619-chdir-bugfix
4265 :     Description:
4266 :    
4267 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
4268 :    
4269 :     ----------------------------------------------------------------------
4270 :     Name: Matthias Blume
4271 : blume 670 Date: 2000/06/18 22:00:10 JST
4272 :     Tag: blume-20000618-implicit-anchors-really-gone
4273 :     Description:
4274 :    
4275 :     I updates the previous HISTORY entry where I forgot to mention that
4276 :     implicit anchors are no longer with us.
4277 :    
4278 :     The current update also gets rid of the (now useless) controller
4279 :     CM.Control.implicit_anchors.
4280 :    
4281 :     ----------------------------------------------------------------------
4282 :     Name: Matthias Blume
4283 : blume 666 Date: 2000/06/16 17:30:00 JST
4284 :     Tag: blume-20000616-anchorenv
4285 :     Description:
4286 :    
4287 :     This patch implements the long anticipated (just kidding :) "anchor
4288 :     environment" mechanism. In the course of doing this, I also
4289 :     re-implemented CM's internal "SrcPath" module from scratch. The new
4290 :     one should be more robust in certain boundary cases. In any case, it
4291 :     is a lot cleaner than its predecessor (IMHO).
4292 :    
4293 :     This time, although there is yet another boot file format change, I
4294 :     kept the unpickler backward-compatible. As a result, no new bootfiles
4295 :     are necessary and bootstrapping is straightforward. (You cannot read
4296 :     new bootfiles into an old system, but the other way around is no
4297 :     problem.)
4298 :    
4299 :     Visible changes:
4300 :    
4301 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
4302 :     longer recognized at all. This means that such path names are not
4303 :     illegal either. For example, the name basis.cm simply refers to a
4304 :     local file called "basis.cm" (i.e, the name is an ordinary path
4305 :     relative to .cm-files directory). Or, to put it differently, only
4306 :     names that start with $ are anchored paths.
4307 : blume 666
4308 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
4309 :     vanished.
4310 :    
4311 :     John (Reppy) had critizised this as soon as I originally proposed and
4312 : blume 666 implemented it, but at that time I did not really deeply believe
4313 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
4314 :     in another place where it cannot be seen as an abbreviation for
4315 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
4316 :     seems to mean (i.e., it "expands" into the corresponding anchor
4317 :     value).
4318 :    
4319 :     However, when paths are used as members in CM description files, it
4320 :     continues to be true that there must be at least another arc after the
4321 :     anchor. This is now enforced separately during semantic analysis
4322 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
4323 : blume 666
4324 :     ** 2. The "cm" class now accepts an option "bind". The option's value
4325 :     is a sub-option list of precisely two items -- one labeled "anchor"