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 1279 - (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 1279 Date: 2002/10/16 12:00:00 EDT
18 :     Tag: Release_110_42
19 :     Description:
20 :    
21 :     New working release. New bootfiles.
22 :    
23 :     ----------------------------------------------------------------------
24 :     Name: Matthias Blume
25 : blume 1273 Date: 2002/10/10 13:10:00 EDT
26 :     Tag: blume-20021010-ppc-divs
27 :     Description:
28 :    
29 :     The mltree operator DIVS must be implemented with an overflow check on
30 :     the PPC because the hardware indicates divide-by-zero using "overflow" as
31 :     well.
32 :    
33 :     ----------------------------------------------------------------------
34 :     Name: Matthias Blume
35 : blume 1261 Date: 2002/07/23 11:45:00
36 :     Tag: blume-20020723-smlnj-home
37 :     Description:
38 :    
39 :     Sml now senses the SMLNJ_HOME environment variable. If this is set,
40 :     then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
41 :     CM_PATHCONFIG is also set), the path configuration file is assumed
42 :     to be in $SMLNJ_HOME/lib/pathconfig. This way one can easily move
43 :     the entire tree to some other place and everything will "just work".
44 :    
45 :     (Companion commands such as ml-build and ml-makedepend also sense this
46 :     variable.)
47 :    
48 :     ----------------------------------------------------------------------
49 :     Name: Matthias Blume
50 : blume 1259 Date: 2002/07/12 21:19:00 EDT
51 :     Tag: blume-20020712-liveness
52 :     Description:
53 :    
54 :     Exported two useful "step" functions from liveness module (MLRISC).
55 :    
56 :     ----------------------------------------------------------------------
57 :     Name: Matthias Blume
58 : blume 1253 Date: 2002/07/05 16:00 EDT
59 :     Tag: Release_110_41
60 :     Description:
61 :    
62 :     New working release. New bootfiles.
63 :    
64 :     ----------------------------------------------------------------------
65 :     Name: Matthias Blume
66 : blume 1250 Date: 2002/07/05 10:25:00 EDT
67 :     Tag: blume-20020705-btimp
68 :     Description:
69 :    
70 :     Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
71 :     other clients can set up backtracing support.
72 :    
73 :     ----------------------------------------------------------------------
74 :     Name: Matthias Blume
75 : blume 1249 Date: 2002/06/25 17:23:00 EDT
76 :     Tag: blume-20020625-fpmax
77 :     Description:
78 :    
79 :     Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
80 :     case. (The sense of the isNaN test was reversed -- which made min and
81 :     max always return their first argument.)
82 :    
83 :     ----------------------------------------------------------------------
84 :     Name: Matthias Blume
85 : blume 1240 Date: 2002/06/11
86 :     Tag: blume-20020611-unixpath
87 :     Description:
88 :    
89 :     Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
90 :    
91 :     ----------------------------------------------------------------------
92 :     Name: Matthias Blume
93 : blume 1238 Date: 2002/06/10 16:35:00 EDT
94 :     Tag: blume-20020610-ieeereal
95 :     Description:
96 :    
97 :     I back-ported my implementation of IEEEReal.fromString from the
98 :     idlbasis-devel branch so that we can test it.
99 :    
100 :     Another small change is that ppDec tries to give more information
101 :     than just "<sig>" in the case of functors. However, this code is
102 :     broken in some mysterious way if the functor's body's signature
103 :     has not been declared by ascription but gets inferred from the
104 :     implementation. This needs fixing...
105 :    
106 :     ----------------------------------------------------------------------
107 :     Name: Matthias Blume
108 : blume 1229 Date: 2002/05/31
109 :     Tag: blume-20020531-btrace-mode
110 :     Description:
111 :    
112 :     Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by
113 :     the wayside when I switched over to using Controls everywhere.)
114 :    
115 :     ----------------------------------------------------------------------
116 : george 1222 Name: Lal George
117 :     Date: 2002/05/23 12:21:40 EDT
118 :     Tag: george-20020523-visual-labels
119 :     Description:
120 :    
121 :     Labels are now displayed in the graphical output to make
122 :     the fall-through and target blocks obvious.
123 :    
124 :     ----------------------------------------------------------------------
125 : blume 1204 Name: Matthias Blume
126 : blume 1220 Date: 2002/05/22 11:03:00 EDT
127 :     Tag: blume-20020522-shrink
128 :     Description:
129 :    
130 :     John tweaked yesterday's fix for 1131 to handle an out-of-memory
131 :     situation that comes up when allocating huge arrays.
132 :    
133 :     ----------------------------------------------------------------------
134 :     Name: Matthias Blume
135 : blume 1215 Date: 2002/05/21 16:00:00 EDT
136 :     Tag: Release_110_40
137 :     Description:
138 :    
139 :     New working release (110.40). New bootfiles.
140 :    
141 :     [Also: John Reppy fixed GC bug 1131.]
142 :    
143 :     ----------------------------------------------------------------------
144 :     Name: Matthias Blume
145 : blume 1212 Date: 2002/05/21 12:35:00 EDT
146 :     Tag: blume-20020521-cmdoc
147 :     Description:
148 :    
149 :     CM documentation update.
150 :    
151 :     ----------------------------------------------------------------------
152 :     Name: Matthias Blume
153 : blume 1211 Date: 2002/05/21 10:55:00 EDT
154 :     Tag: blume-20020521-misc
155 :     Description:
156 :    
157 :     - John tweaked runtime to be silent on heap export (except when
158 :     GC messages are on).
159 :     - I added a few more things (cross-compiling versions of CMB) to
160 :     config/preloads (as suggestions).
161 :    
162 :     ----------------------------------------------------------------------
163 :     Name: Matthias Blume
164 : blume 1208 Date: 2002/05/20 22:25:00 EDT
165 :     Tag: blume-20020520-controls
166 :     Description:
167 :    
168 :     - Added ControlUtil structure to control-lib.cm.
169 :     - Use it throughout.
170 :     - Used Controls facility to define MLRISC controls (as opposed to
171 :     registering MLRISC control ref cells with Controls after the fact)
172 :     - Fixed messed-up controls priorities.
173 :    
174 :     * Removed again all the stuff from config/preloads that one wouldn't
175 :     be able to preload at the time the initial heap image is built.
176 :     (Many libraries, e.g., CML, do not exist yet at this time. The
177 :     only libraries that can be preloaded via config/preloads are those
178 :     that come bundled with the bootfiles.)
179 :    
180 :     ----------------------------------------------------------------------
181 :     Name: Matthias Blume
182 : blume 1204 Date: 2002/05/20 10:59:00 EDT
183 :     Tag: blume-20020520-preloads
184 :     Description:
185 :    
186 :     Added a lot of commented-out suggestions for things to be included
187 :     in config/preloads.
188 :    
189 :     ----------------------------------------------------------------------
190 : leunga 1203 Name: Allen Leung
191 :     Date: 2002/05/18 14:20:00 EDT
192 :     Tag: leunga-20020518-mdl
193 :     Description:
194 :    
195 :     o Made the mdl tool stuff compile and run again.
196 :     o I've disabled all the stuff that depends on RTL specifications; they
197 :     are all badly broken anyway.
198 :    
199 :     ----------------------------------------------------------------------
200 : blume 1201 Name: Matthias Blume
201 :     Date: 2002/05/17 16:49:00 EDT
202 :     Tag: blume-20020517-controls
203 :     Description:
204 :    
205 :     0. John Reppy made several modifications to the SML/NJ library.
206 :     In particular, there is a shiny new controls-lib.cm.
207 :    
208 :     1. Pushed new controls interface through compiler so that everything
209 :     compiles again.
210 :    
211 :     2. Added FormatComb and FORMAT_COMB to the CML version of the
212 :     SML/NJ library (so that CML compiles again).
213 :    
214 :     3. Modified init scripts because XXX_DEFAULT environment variables
215 :     are no longer with us. (Boot-time initialization is now done
216 :     using the same environment variables that are also used for
217 :     startup-time initialization of controls.)
218 :    
219 :     ----------------------------------------------------------------------
220 : george 1192 Name: Lal George
221 :     Date: 2002/05/15 09:20:10 EDT
222 :     Tag: george-20020515-pseudo-op-decls
223 :     Description:
224 :    
225 :     All pseudo-ops emitted before the first segment declaration
226 :     such as TEXT, DATA, and BSS directives are assumed to be global
227 :     declarations and are emitted first in the assembly file. This is
228 :     useful in a number of situations where one has pseudo-ops that are not
229 :     specific to any segment, and also works around the constraint that one
230 :     cannot have client pseudo-ops in the TEXT segment.
231 :    
232 :     Because no segment is associated with these declarations it is
233 :     an error to allocate any space or objects before the first segment
234 :     directive and an exception will be raised. However, we cannot make
235 :     this check for client pseudo-ops.
236 :    
237 :     These top level declarations are a field in the CFG graph_info.
238 :     In theory you can continue to add to this field after the CFG has been
239 :     built -- provided you know what you are doing;-)
240 :    
241 :     ----------------------------------------------------------------------
242 : blume 1175 Name: Matthias Blume
243 : blume 1190 Date: 2002/05/13 16:40:00 EDT
244 :     Tag: blume-20020513-pp-etc
245 :     Description:
246 :    
247 :     A few minor bugfixes:
248 :    
249 :     - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
250 :     (Bogus printouts for redefined bindings still occur. Compiler
251 :     bug should no longer occur now. We need to redo the prettyprinter
252 :     from scratch.)
253 :    
254 :     - CM pathname printer now also adds escape sequences for ( and )
255 :    
256 :     - commend and docu fixes for ml-nlffi
257 :    
258 :     ----------------------------------------------------------------------
259 :     Name: Matthias Blume
260 : blume 1189 Date: 2002/05/10 16:40:00 EDT
261 :     Tag: blume-20020510-erg-textio
262 :     Description:
263 :    
264 :     Applied the following bugfix provided by Emden Gansner:
265 :    
266 :     Output is corrupted when outputSubstr is used rather than output.
267 :    
268 :     The problem occurs when a substring
269 :    
270 :     ss = (s, dataStart, dataLen)
271 :    
272 :     where dataStart > 0, fills a stream buffer with avail bytes left.
273 :     avail bytes of s, starting at index dataStart, are copied into the
274 :     buffer, the buffer is flushed, and then the remaining dataLen-avail
275 :     bytes of ss are copied into the beginning of the buffer. Instead of
276 :     starting this copy at index dataStart+avail in s, the current code
277 :     starts the copy at index avail.
278 :    
279 :     Fix:
280 :     In text-io-fn.sml, change line 695 from
281 :     val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
282 :     to
283 :     val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
284 :    
285 :     ----------------------------------------------------------------------
286 :     Name: Matthias Blume
287 : blume 1186 Date: 2002/04/12 13:55:00 EDT
288 :     Tag: blume-20020412-assyntax
289 :     Description:
290 :    
291 :     1. Grabbed newer assyntax.h from the XFree86 project.
292 :     2. Fiddled with how to compile X86.prim.asm without warnings.
293 :     3. (Very) Minor cleanup in CM.
294 :    
295 :     ----------------------------------------------------------------------
296 :     Name: Matthias Blume
297 : blume 1185 Date: 2002/04/01 (no joke!) 17:07:00 EST
298 :     Tag: blume-20020401-x86div
299 :     Description:
300 :    
301 :     Added full support for div/mod/rem/quot on the x86, using the machine
302 :     instruction's two results (without clumsily recomputing the remainder)
303 :     directly where appropriate.
304 :    
305 :     Some more extensive power-of-two support was added to the x86 instruction
306 :     selector (avoiding expensive divs, mods, and muls where they can be
307 :     replaced with cheaper shifts and masks). However, this sort of thing
308 :     ought to be done earlier, e.g., within the CPS optimizer so that
309 :     all architectures benefit from it.
310 :    
311 :     The compiler compiles to a fixed point, but changes might be somewhat
312 :     fragile nevertheless. Please, report any strange things that you might
313 :     see wrt. div/mod/quot/rem...
314 :    
315 :     ----------------------------------------------------------------------
316 :     Name: Matthias Blume
317 : blume 1184 Date: 2002/03/29 17:22:00
318 :     Tag: blume-20020329-div
319 :     Description:
320 :    
321 :     Fixed my broken div/mod logic. Unfortunately, this means that the
322 :     inline code for div/mod now has one more comparison than before.
323 :     Fast paths (quotient > 0 or remainder = 0) are not affected, though.
324 :     The problem was with quotient = 0, because that alone does not tell
325 :     us which way the rounding went. One then has to look at whether
326 :     remainder and divisor have the same sign... :(
327 :    
328 :     Anyway, I replaced the bootfiles with fresh ones...
329 :    
330 :     ----------------------------------------------------------------------
331 :     Name: Matthias Blume
332 : blume 1183 Date: 2002/03/29 14:10:00 EST
333 :     Tag: blume-20020329-inlprims
334 :     Description:
335 :    
336 :     NEW BOOTFILES!!! Version number bumped to 110.39.3.
337 :    
338 :     Primops have changed. This means that the bin/boot-file formats have
339 :     changed as well.
340 :    
341 :     To make sure that there is no confusion, I made a new version.
342 :    
343 :    
344 :     CHANGES:
345 :    
346 :     * removed REMT from mltree (remainder should never overflow).
347 :    
348 :     * added primops to deal with divisions of all flavors to the frontend
349 :    
350 :     * handled these primops all the way through so they map to their respective
351 :     MLRISC support
352 :    
353 :     * used these primops in the implementation of Int, Int32, Word, Word32
354 :    
355 :     * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
356 :    
357 :     * parameterized INLMIN, INLMAX, and INLABS by a numkind
358 :    
359 :     * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
360 :     floating point
361 :    
362 :     * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
363 :     and Real (but Real.abs maps to a separate floating-point-only primop)
364 :    
365 :    
366 :     TODO items:
367 :    
368 :     * Hacked Alpha32 instruction selection, disabling the selection of REMx
369 :     instructions because the machine instruction encoder cannot handle
370 :     them. (Hppa, PPC, and Sparc instruction selection did not handle
371 :     REM in the first place, and REM is supported by the x86 machine coder.)
372 :    
373 :     * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
374 :     selection phase. (The two can be streamlined because the hardware
375 :     delivers both quotient and remainder at the same time anyway.)
376 :    
377 :     * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
378 :     (Currently the behavior is inconsistent both across architectures and
379 :     wrt. the draft Basis spec.)
380 :    
381 :     * Word8 should eventually be handled natively, too.
382 :    
383 :     * There seems to be one serious bug in mltree-gen.sml. It appears, though,
384 :     as if there currently is no execution path that could trigger it in
385 :     SML/NJ. (The assumptions underlying functions arith and promotable do not
386 :     hold for things like multiplication and division.)
387 :    
388 :     ----------------------------------------------------------------------
389 :     Name: Matthias Blume
390 : blume 1181 Date: 2002/03/27 16:27:00 EST
391 :     Tag: blume-20020327-mlrisc-divisions
392 :     Description:
393 :    
394 :     Added support for all four division operations (ML's div, mod, quot,
395 :     and rem) to MLRISC. In the course of doing so, I also rationalized
396 :     the naming (no more annoying switch-around of DIV and QUOT), by
397 :     parameterizing the operation by div_rounding_mode (which can be either
398 :     DIV_TO_ZERO or DIV_TO_NEGINF).
399 :    
400 :     The generic MLTreeGen functor takes care of compiling all four
401 :     operations down to only round-to-zero div.
402 :    
403 :     Missing pieces:
404 :    
405 :     * Doing something smarter than relying on MLTreeGen on architectures
406 :     like, e.g., the x86 where hardware division delivers both quotient and
407 :     remainder at the same time. With this, the implementation of the
408 :     round-to-neginf operations could be further streamlined.
409 :    
410 :     * Remove inlining support for div/mod/rem from the frontend and replace it
411 :     with primops that get carried through to the backend. Do this for all
412 :     int and word types.
413 :    
414 :     ----------------------------------------------------------------------
415 :     Name: Matthias Blume
416 : blume 1180 Date: 2002/03/25 17:25:00 EST
417 :     Tag: blume-20020325-divmod
418 :     Description:
419 :    
420 :     I improved (hopefully without breaking them) the implementation of Int.div,
421 :     Int.mod, and Int.rem. For this, the code in translate.sml now takes
422 :     advantage of the following observations:
423 :    
424 :     Let q = x quot y r = x rem y
425 :     d = x div y m = x mod y
426 :    
427 :     where "quot" is the round-to-zero version of integer division that
428 :     hardware usually provides. Then we have:
429 :    
430 :     r = x - q * y where neither the * nor the - will overflow
431 :     d = if q >= 0 orelse x = q * y then q else q - 1
432 :     where neither the * nor the - will overflow
433 :     m = if q >= 0 orelse r = 0 then r else r + y
434 :     where the + will not overflow
435 :    
436 :     This results in substantial simplification of the generated code.
437 :     The following table shows the number of CFG nodes and edges generated
438 :     for
439 :     fun f (x, y) = x OPER y
440 :     (* with OPER \in div, mod, quot, rem *)
441 :    
442 :    
443 :     OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
444 :     --------------------------------------------------------
445 :     div | 24 | 39 | 12 | 16
446 :     mod | 41 | 71 | 12 | 16
447 :     quot | 8 | 10 | 8 | 10
448 :     rem | 10 | 14 | 8 | 10
449 :    
450 :    
451 :     ----------------------------------------------------------------------
452 :     Name: Matthias Blume
453 : blume 1179 Date: 2002/03/25 22:06:00 EST
454 :     Tag: blume-20020325-cprotobug
455 :     Description:
456 :    
457 :     Fixed a bug in cproto (c prototype decoder).
458 :    
459 :     ----------------------------------------------------------------------
460 :     Name: Matthias Blume
461 : blume 1178 Date: 2002/03/25 16:00:00 EST
462 :     Tag: blume-20020325-raw-primops
463 :     Description:
464 :    
465 :     I did some cleanup to Allen's new primop code and
466 :     replaced yesterday's bootfiles with new ones.
467 :     (But they are stored in the same place.)
468 :    
469 :     ----------------------------------------------------------------------
470 :     Name: Matthias Blume
471 : blume 1175 Date: 2002/03/24 22:40:00 EST
472 :     Tag: blume-20020324-bootfiles
473 :     Description:
474 :    
475 :     Made the bootfiles that Allen asked for.
476 :    
477 :     ----------------------------------------------------------------------
478 : leunga 1174 Name: Allen Leung
479 :     Date: 2002/03/23 15:50:00 EST
480 :     Tag: leunga-20020323-flint-cps-rcc-primops
481 :     Description:
482 :    
483 :     1. Changes to FLINT primops:
484 :    
485 :     (* make a call to a C-function;
486 :     * The primop carries C function prototype information and specifies
487 :     * which of its (ML-) arguments are floating point. C prototype
488 :     * information is for use by the backend, ML information is for
489 :     * use by the CPS converter. *)
490 :     | RAW_CCALL of { c_proto: CTypes.c_proto,
491 :     ml_args: ccall_type list,
492 :     ml_res_opt: ccall_type option,
493 :     reentrant : bool
494 :     } option
495 :     (* Allocate uninitialized storage on the heap.
496 :     * The record is meant to hold short-lived C objects, i.e., they
497 :     * are not ML pointers. With the tag, the representation is
498 :     * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
499 :     *)
500 :     | RAW_RECORD of {tag:bool,sz:int}
501 :     and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
502 :    
503 :     2. These CPS primops are now overloaded:
504 :    
505 :     rawload of {kind:numkind}
506 :     rawstore of {kind:numkind}
507 :    
508 :     The one argument form is:
509 :    
510 :     rawload {kind} address
511 :    
512 :     The two argument form is:
513 :    
514 :     rawload {kind} [ml object, byte-offset]
515 :    
516 :     3. RAW_CCALL/RCC now takes two extra arguments:
517 :    
518 :     a. The first is whether the C call is reentrant, i.e., whether
519 :     ML state should be saved and restored.
520 :     b. The second argument is a string argument specifying the name of
521 :     library and the C function.
522 :    
523 :     These things are currently not handled in the code generator, yet.
524 :    
525 :     4. In CProto,
526 :    
527 :     An encoding type of "bool" means "ml object" and is mapped into
528 :     C prototype of PTR. Note that "bool" is different than "string",
529 :     even though "string" is also mapped into PTR, because "bool"
530 :     is assigned an CPS type of BOGt, while "string" is assigned INT32t.
531 :    
532 :     5. Pickler/unpicker
533 :    
534 :     Changed to handle RAW_RECORD and newest RAW_CCALL
535 :    
536 :     6. MLRiscGen,
537 :    
538 :     1. Changed to handle the new rawload/rawstore/rawrecord operators.
539 :     2. Code for handling C Calls has been moved to a new module CPSCCalls,
540 :     in the file CodeGen/cpscompile/cps-c-calls.sml
541 :    
542 :     7. Added the conditional move operator
543 :    
544 :     condmove of branch
545 :    
546 :     to cps. Generation of this is still buggy so it is currently
547 :     disabled.
548 :    
549 :     ----------------------------------------------------------------------
550 : george 1168 Name: Lal George
551 :     Date: 2002/03/22 14:18:25 EST
552 : george 1171 Tag: george-20020322-cps-branch-prob
553 : george 1168 Description:
554 :    
555 : george 1171 Implemented the Ball-Larus branch prediction-heuristics, and
556 : george 1168 incorporated graphical viewers for control flow graphs.
557 :    
558 : george 1171 Ball-Larus Heuristics:
559 : george 1168 ---------------------
560 :     See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
561 :    
562 :     By design it uses the Dempster-Shafer theory for combining
563 :     probabilities. For example, in the function:
564 :    
565 :     fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
566 :    
567 :     the ball-larus heuristics predicts that the n=0 is unlikely
568 :     (OH-heuristic), and the 'then' branch is unlikely because of the
569 :     RH-heuristic -- giving the 'then' branch an even lower combined
570 : george 1171 probability using the Dempster-Shafer theory.
571 : george 1168
572 :     Finally, John Reppy's loop analysis in MLRISC, further lowers the
573 :     probability of the 'then' branch because of the loop in the else
574 :     branch.
575 :    
576 :    
577 :     Graphical Viewing:
578 :     ------------------
579 :     I merely plugged in Allen's graphical viewers into the compiler. The
580 :     additional code is not much. At the top level, saying:
581 :    
582 :     Control.MLRISC.getFlag "cfg-graphical-view" := true;
583 :    
584 :     will display the graphical view of the control flow graph just before
585 :     back-patching. daVinci must be in your path for this to work. If
586 :     daVinci is not available, then the default viewer can be changed
587 :     using:
588 :    
589 :     Control.MLRISC.getString "viewer"
590 :    
591 :     which can be set to "dot" or "vcg" for the corresponding viewers. Of
592 :     course, these viewers must be in your path.
593 :    
594 :     The above will display the compilation unit at the level of clusters,
595 :     many of which are small, boring, and un-interesting. Also setting:
596 :    
597 :     Control.MLRISC.getInt "cfg-graphical-view_size"
598 :    
599 :     will display clusters that are larger than the value set by the above.
600 :    
601 :    
602 :     ----------------------------------------------------------------------
603 : blume 1159 Name: Matthias Blume
604 :     Date: 2002/03/21 22:20:00 EST
605 :     Tag: blume-20020321-kmp-bugfix
606 :     Description:
607 :    
608 :     Changed the interface to the KMP routine in PreString and fixed
609 :     a minor bug in one place where it was used.
610 :    
611 :     ----------------------------------------------------------------------
612 : leunga 1156 Name: Allen Leung
613 : leunga 1158 Date: 2002/03/21 20:30:00 EST
614 :     Tag: leunga-20020321-cfg
615 :     Description:
616 :    
617 :     Fixed a potential problem in cfg edge splitting.
618 :    
619 :     ----------------------------------------------------------------------
620 :     Name: Allen Leung
621 : leunga 1156 Date: 2002/03/21 17:15:00 EST
622 :     Tag: leunga-20020321-x86-fp-cfg
623 :     Description:
624 :    
625 :     1. Recoded the buggy parts of x86-fp.
626 :    
627 :     a. All the block reordering code has been removed.
628 :     We now depend on the block placement phases to do this work.
629 :    
630 :     b. Critical edge splitting code has been simplified and moved into the
631 :     CFG modules, as where they belong.
632 :    
633 :     Both of these were quite buggy and complex. The code is now much, much
634 :     simpler.
635 :    
636 :     2. X86 backend.
637 :    
638 :     a. Added instructions for 64-bit support. Instruction selection for
639 :     64-bit has not been committed, however, since that
640 :     requires changes to MLTREE which haven't been approved by
641 :     Lal and John.
642 :    
643 :     b. Added support for FUCOMI and FUCOMIP when generating code for
644 :     PentiumPro and above. We only generate these instructions in
645 :     the fast-fp mode.
646 :    
647 :     c. Added cases for JP and JNP in X86FreqProps.
648 :    
649 :     3. CFG
650 :    
651 :     CFG now has a bunch of methods for edge splitting and merging.
652 :    
653 :     4. Machine description.
654 :    
655 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
656 :     description things:
657 :    
658 :     rtl-build.{sig,sml} and hppa.mdl fixed.
659 :    
660 :     NOTE: the machine description stuff in the repository is still broken.
661 :     Again, I can't put my fixes in because that involves
662 :     changes to MLTREE.
663 :    
664 :     ----------------------------------------------------------------------
665 : blume 1145 Name: Matthias Blume
666 : blume 1155 Date: 2002/03/20 15:55:00 EST
667 :     Tag: blume-20020320-kmp
668 :     Description:
669 :    
670 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
671 :     it for String.isSubstring, Substring.isSubstring, and
672 :     Substring.position.
673 :    
674 :     (Might need some stress-testing. Simple examples worked fine.)
675 :    
676 :     ----------------------------------------------------------------------
677 :     Name: Matthias Blume
678 : blume 1152 Date: 2002/03/19 16:37:00 EST
679 :     Tag: blume-20020319-witnesses
680 :     Description:
681 :    
682 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
683 :    
684 :     This implements a generic mechanism for changing constness qualifiers
685 :     anywhere within big C types without resorting to outright "casts".
686 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
687 :     modify the constness at the outermost level.)
688 :     The implementation of "convert" is based on the idea of "witness"
689 :     values -- values that are not used by the operation but whose types
690 :     "testify" to their applicability. On the implementation side, "convert"
691 :     is simply a projection (returning its second curried argument). With
692 :     cross-module inlining, it should not result in any machine code being
693 :     generated.
694 :    
695 :     ----------------------------------------------------------------------
696 :     Name: Matthias Blume
697 : blume 1148 Date: 2002/03/15 16:40:00 EST
698 :     Tag: blume-20020315-basis
699 :     Description:
700 :    
701 :     Provided (preliminary?) implementations for
702 :    
703 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
704 :    
705 :     and
706 :    
707 :     Substring.full
708 :    
709 :     Those are in the Basis spec but they were missing in SML/NJ.
710 :    
711 :     ----------------------------------------------------------------------
712 :     Name: Matthias Blume
713 : blume 1145 Date: 2002/03/14 21:30:00 EST
714 :     Tag: blume-20020314-controls
715 :     Description:
716 :    
717 :     Controls:
718 :     ---------
719 :    
720 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
721 :     it into its own library $/controls-lib.cm. The source tree for
722 :     this is under src/smlnj-lib/Controls.
723 :    
724 :     2. Changed the names of types and functions in this interface, so they
725 :     make a bit more "sense":
726 :    
727 :     module -> registry
728 :     'a registry -> 'a group
729 :    
730 :     3. The interface now deals in ref cells only. The getter/setter interface
731 :     is (mostly) gone.
732 :    
733 :     4. Added a function that lets one register an already-existing ref cell.
734 :    
735 :     5. Made the corresponding modifications to the rest of the code so that
736 :     everything compiles again.
737 :    
738 :     6. Changed the implementation of Controls.MLRISC back to something closer
739 :     to the original. In particular, this module (and therefore MLRISC)
740 :     does not depend on Controls. There now is some link-time code in
741 :     int-sys.sml that registers the MLRISC controls with the Controls
742 :     module.
743 :    
744 :     CM:
745 :     ---
746 :    
747 :     * One can now specify the lambda-split aggressiveness in init.cmi.
748 :    
749 :     ----------------------------------------------------------------------
750 : leunga 1142 Name: Allen Leung
751 :     Date: 2002/03/13 17:30:00 EST
752 :     Tag: leunga-20020313-x86-fp-unary
753 :     Description:
754 :    
755 :     Bug fix for:
756 :    
757 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
758 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
759 :     > - fun f(x,(y,z)) = Real.~ y;
760 :     > [autoloading]
761 :     > [autoloading done]
762 :     > fchsl (%eax), 184(%esp)
763 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
764 :     >
765 :     > uncaught exception Error
766 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
767 :    
768 :     The problem was that the code generator did not generate any fp registers
769 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
770 :     translate the pseudo fp instruction. This only happened with unary fp
771 :     operators in certain situations.
772 :    
773 :     ----------------------------------------------------------------------
774 : blume 1137 Name: Matthias Blume
775 : blume 1140 Date: 2002/03/13 14:00:00 EST
776 :     Tag: blume-20020313-overload-etc
777 :     Description:
778 :    
779 :     1. Added _overload as a synonym for overload for backward compatibility.
780 :     (Control.overloadKW must be true for either version to be accepted.)
781 :    
782 :     2. Fixed bug in install script that caused more things to be installed
783 :     than what was requested in config/targets.
784 :    
785 :     3. Made CM aware of the (_)overload construct so that autoloading
786 :     works.
787 :    
788 :     ----------------------------------------------------------------------
789 :     Name: Matthias Blume
790 : blume 1138 Date: 2002/03/12 22:03:00 EST
791 :     Tag: blume-20020312-url
792 :     Description:
793 :    
794 :     Forgot to update BOOT and srcarchiveurl.
795 :    
796 :     ----------------------------------------------------------------------
797 :     Name: Matthias Blume
798 : blume 1137 Date: 2002/03/12 17:30:00 EST
799 :     Tag: blume-20020312-version110392
800 :     Description:
801 :    
802 :     Yet another version number bump (because of small changes to the
803 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
804 :    
805 :     Changes:
806 :    
807 :     The new pid generation scheme described a few weeks ago was overly
808 :     complicated. I implemented a new mechanism that is simpler and
809 :     provides a bit more "stability": Once CM has seen a compilation
810 :     unit, it keeps its identity constant (as long as you do not delete
811 :     those crucial CM/GUID/* files). This means that when you change
812 : blume 1145 an interface, compile, then go back to the old interface, and
813 : blume 1137 compile again, you arrive at the original pid.
814 :    
815 :     There now also is a mechanism that instructs CM to use the plain
816 :     environment hash as a module's pid (effectively making its GUID
817 :     the empty string). For this, "noguid" must be specified as an
818 :     option to the .sml file in question within its .cm file.
819 :     This is most useful for code that is being generated by tools such
820 :     as ml-nlffigen (because during development programmers tend to
821 :     erase the tool's entire output directory tree including CM's cached
822 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
823 :     revert to the old, broken behavior of SML/NJ, but in specific cases
824 :     where there is no danger of interface confusion, its use is ok
825 :     (I think).
826 :    
827 :     ml-nlffigen by default generates "noguid" annotations. They can be
828 :     turned off by specifying -guid in its command line.
829 :    
830 :     ----------------------------------------------------------------------
831 : george 1136 Name: Lal George
832 :     Date: 2002/03/12 12 14:42:36 EST
833 :     Tag: george-20020312-frequency-computation
834 :     Description:
835 :    
836 :     Integrated jump chaining and static block frequency into the
837 :     compiler. More details and numbers later.
838 : leunga 1025
839 : george 1003 ----------------------------------------------------------------------
840 : george 1133 Name: Lal George
841 :     Date: 2002/03/11 11 22:38:53 EST
842 :     Tag: george-20020311-jump-chain-elim
843 :     Description:
844 :    
845 :     Tested the jump chain elimination on all architectures (except the
846 :     hppa). This is on by default right now and is profitable for the
847 :     alpha and x86, however, it may not be profitable for the sparc and ppc
848 :     when compiling the compiler.
849 :    
850 :     The gc test will typically jump to a label at the end of the cluster,
851 :     where there is another jump to an external cluster containing the actual
852 :     code to invoke gc. This is to allow factoring of common gc invocation
853 :     sequences. That is to say, we generate:
854 :    
855 :     f:
856 :     testgc
857 :     ja L1 % jump if above to L1
858 :    
859 :     L1:
860 :     jmp L2
861 :    
862 :    
863 :     After jump chain elimination the 'ja L1' instructions is converted to
864 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
865 :     up being implemented in their long form (if L2 is far away) using:
866 :    
867 :     jbe L3 % jump if below or equal to L3
868 :     jmp L2
869 :     L3:
870 :     ...
871 :    
872 :    
873 :     For large compilation units L2 may be far away.
874 :    
875 :    
876 :     ----------------------------------------------------------------------
877 : blume 1132 Name: Matthias Blume
878 :     Date: 2002/03/11 13:30:00 EST
879 :     Tag: blume-20020311-mltreeeval
880 :     Description:
881 :    
882 :     A functor parameter was missing.
883 :    
884 :     ----------------------------------------------------------------------
885 : leunga 1130 Name: Allen Leung
886 : leunga 1131 Date: 2002/03/11 10:30:00 EST
887 : leunga 1142 Tag: leunga-20020311-runtime-string0
888 : leunga 1131 Description:
889 :    
890 :     The representation of the empty string now points to a
891 :     legal null terminated C string instead of unit. It is now possible
892 :     to convert an ML string into C string with InlineT.CharVector.getData.
893 :     This compiles into one single machine instruction.
894 :    
895 :     ----------------------------------------------------------------------
896 :     Name: Allen Leung
897 : leunga 1130 Date: 2002/03/10 23:55:00 EST
898 :     Tag: leunga-20020310-x86-call
899 :     Description:
900 :    
901 :     Added machine generation for CALL instruction (relative displacement mode)
902 :    
903 :     ----------------------------------------------------------------------
904 : blume 1128 Name: Matthias Blume
905 :     Date: 2002/03/08 16:05:00
906 :     Tag: blume-20020308-entrypoints
907 :     Description:
908 :    
909 :     Version number bumped to 110.39.1. NEW BOOTFILES!
910 :    
911 :     Entrypoints: non-zero offset into a code object where execution should begin.
912 :    
913 :     - Added the notion of an entrypoint to CodeObj.
914 :     - Added reading/writing of entrypoint info to Binfile.
915 :     - Made runtime system bootloader aware of entrypoints.
916 :     - Use the address of the label of the first function given to mlriscGen
917 :     as the entrypoint. This address is currently always 0, but it will
918 :     not be 0 once we turn on block placement.
919 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
920 :     with entry points) from mlriscGen.
921 :    
922 :     ----------------------------------------------------------------------
923 : leunga 1127 Name: Allen Leung
924 :     Date: 2002/03/07 20:45:00 EST
925 :     Tag: leunga-20020307-x86-cmov
926 :     Description:
927 :    
928 :     Bug fixes for CMOVcc on x86.
929 :    
930 :     1. Added machine code generation for CMOVcc
931 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
932 :     3. CMOVcc cannot have an immediate operand as argument.
933 :    
934 :     ----------------------------------------------------------------------
935 : blume 1126 Name: Matthias Blume
936 :     Date: 2002/03/07 16:15:00 EST
937 :     Tag: blume-20020307-controls
938 :     Description:
939 :    
940 :     This is a very large but mostly boring patch which makes (almost)
941 :     every tuneable compiler knob (i.e., pretty much everything under
942 :     Control.* plus a few other things) configurable via both the command
943 :     line and environment variables in the style CM did its configuration
944 :     until now.
945 :    
946 :     Try starting sml with '-h' (or, if you are brave, '-H')
947 :    
948 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
949 :     implements the underlying generic mechanism.
950 :    
951 :     The interface to some of the existing such facilities has changed somewhat.
952 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
953 :     (The getFoo interface is still there for backward-compatibility, but its
954 :     use is deprecated.)
955 :    
956 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
957 :     that one can now twiddle the compiler settings when using this "batch"
958 :     compiler.
959 :    
960 :     TODO items:
961 :    
962 :     We should go through and throw out all controls that are no longer
963 :     connected to anything. Moreover, we should go through and provide
964 :     meaningful (and correct!) documentation strings for those controls
965 :     that still are connected.
966 :    
967 :     Currently, multiple calls to Controls.new are accepted (only the first
968 :     has any effect). Eventually we should make sure that every control
969 :     is being made (via Controls.new) exactly once. Future access can then
970 :     be done using Controls.acc.
971 :    
972 :     Finally, it would probably be a good idea to use the getter-setter
973 :     interface to controls rather than ref cells. For the time being, both
974 :     styles are provided by the Controls module, but getter-setter pairs are
975 :     better if thread-safety is of any concern because they can be wrapped.
976 :    
977 :     *****************************************
978 :    
979 :     One bug fix: The function blockPlacement in three of the MLRISC
980 :     backpatch files used to be hard-wired to one of two possibilities at
981 :     link time (according to the value of the placementFlag). But (I
982 :     think) it should rather sense the flag every time.
983 :    
984 :     *****************************************
985 :    
986 :     Other assorted changes (by other people who did not supply a HISTORY entry):
987 :    
988 :     1. the cross-module inliner now works much better (Monnier)
989 :     2. representation of weights, frequencies, and probabilities in MLRISC
990 :     changed in preparation of using those for weighted block placement
991 :     (Reppy, George)
992 :    
993 :     ----------------------------------------------------------------------
994 : george 1116 Name: Lal George
995 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
996 :     Tag: george-20020307-weighted-block-placement
997 :    
998 :     Tested the weighted block placement optimization on all architectures
999 :     (except the hppa) using AMPL to generate the block and edge frequencies.
1000 :     Changes were required in the machine properties to correctly
1001 :     categorize trap instructions. There is an MLRISC flag
1002 :     "weighted-block-placement" that can be used to enable weighted block
1003 :     placement, but this will be ineffective without block/edge
1004 :     frequencies (coming soon).
1005 :    
1006 :    
1007 :     ----------------------------------------------------------------------
1008 :     Name: Lal George
1009 : george 1116 Date: 2002/03/05 17:24:48 EST
1010 :     Tag: george-20020305-linkage-cluster
1011 :    
1012 : george 1124 In order to support the block placement optimization, a new cluster
1013 :     is generated as the very first cluster (called the linkage cluster).
1014 :     It contains a single jump to the 'real' entry point for the compilation
1015 :     unit. Block placement has no effect on the linkage cluster itself, but
1016 :     all the other clusters have full freedom in the manner in which they
1017 :     reorder blocks or functions.
1018 : george 1116
1019 :     On the x86 the typical linkage code that is generated is:
1020 :     ----------------------
1021 :     .align 2
1022 :     L0:
1023 :     addl $L1-L0, 72(%esp)
1024 : george 1124 jmp L1
1025 : george 1116
1026 :    
1027 :     .align 2
1028 :     L1:
1029 :     ----------------------
1030 :    
1031 :     72(%esp) is the memory location for the stdlink register. This
1032 :     must contain the address of the CPS function being called. In the
1033 : george 1124 above example, it contains the address of L0; before
1034 : george 1116 calling L1 (the real entry point for the compilation unit), it
1035 :     must contain the address for L1, and hence
1036 :    
1037 :     addl $L1-L0, 72(%esp)
1038 :    
1039 : george 1124 I have tested this on all architectures except the hppa.The increase
1040 :     in code size is of course negligible
1041 : george 1116
1042 :     ----------------------------------------------------------------------
1043 : leunga 1115 Name: Allen Leung
1044 :     Date: 2002/03/03 13:20:00 EST
1045 :     Tag: leunga-20020303-mlrisc-tools
1046 :    
1047 :     Added #[ ... ] expressions to mlrisc tools
1048 :    
1049 :     ----------------------------------------------------------------------
1050 : blume 1096 Name: Matthias Blume
1051 : blume 1098 Date: 2002/02/27 12:29:00 EST
1052 :     Tag: blume-20020227-cdebug
1053 :     Description:
1054 :    
1055 :     - made types in structure C and C_Debug to be equal
1056 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1057 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1058 :    
1059 :     ----------------------------------------------------------------------
1060 :     Name: Matthias Blume
1061 : blume 1096 Date: 2002/02/26 12:00:00 EST
1062 :     Tag: blume-20020226-ffi
1063 :     Description:
1064 :    
1065 :     1. Fixed a minor bug in CM's "noweb" tool:
1066 :     If numbering is turned off, then truly don't number (i.e., do not
1067 :     supply the -L option to noweb). The previous behavior was to supply
1068 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
1069 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
1070 :    
1071 :     2. Once again, I reworked some aspects of the FFI:
1072 :    
1073 :     A. The incomplete/complete type business:
1074 :    
1075 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1076 :     gone!
1077 :     - ML types representing an incomplete type are now *equal* to
1078 :     ML types representing their corresponding complete types (just like
1079 :     in C). This is still safe because ml-nlffigen will not generate
1080 :     RTTI for incomplete types, nor will it generate functions that
1081 :     require access to such RTTI. But when ML code generated from both
1082 :     incomplete and complete versions of the C type meet, the ML types
1083 :     are trivially interoperable.
1084 :    
1085 :     NOTE: These changes restore the full generality of the translation
1086 :     (which was previously lost when I eliminated functorization)!
1087 :    
1088 :     B. Enum types:
1089 :    
1090 :     - Structure C now has a type constructor "enum" that is similar to
1091 :     how the "su" constructor works. However, "enum" is not a phantom
1092 :     type because each "T enum" has values (and is isomorphic to
1093 :     MLRep.Signed.int).
1094 :     - There are generic access operations for enum objects (using
1095 :     MLRep.Signed.int).
1096 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
1097 :     * The structure contains the definition of type "mlrep" (the ML-side
1098 :     representation type of the enum). Normally, mlrep is the same
1099 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1100 :     then mlrep will be defined as a datatype -- thus facilitating
1101 :     pattern matching on mlrep values.
1102 :     ("-ec" will be suppressed if there are duplicate values in an
1103 :     enumeration.)
1104 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1105 :     will be generated for each C enum constant xxx.
1106 :     * Conversion functions m2i and i2m convert between mlrep and
1107 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
1108 :     * Coversion functions c and ml convert between mlrep and "tag enum".
1109 :     * Access functions (get/set) fetch and store mlrep values.
1110 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1111 :     enumerations are merged into one single enumeration represented by
1112 :     structure E_'.
1113 :    
1114 :     ----------------------------------------------------------------------
1115 : leunga 1080 Name: Allen Leung
1116 : leunga 1094 Date: 2002/02/25 04:45:00 EST
1117 :     Tag: leunga-20020225-cps-spill
1118 :    
1119 :     This is a new implementation of the CPS spill phase.
1120 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1121 :     In case of problems, replace it with the old file spill.sml
1122 :    
1123 :     The current compiler runs into some serious performance problems when
1124 :     constructing a large record. This can happen when we try to compile a
1125 :     structure with many items. Even a very simple structure like the following
1126 :     makes the compiler slow down.
1127 :    
1128 :     structure Foo = struct
1129 :     val x_1 = 0w1 : Word32.int
1130 :     val x_2 = 0w2 : Word32.int
1131 :     val x_3 = 0w3 : Word32.int
1132 :     ...
1133 :     val x_N = 0wN : Word32.int
1134 :     end
1135 :    
1136 :     The following table shows the compile time, from N=1000 to N=4000,
1137 :     with the old compiler:
1138 :    
1139 :     N
1140 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
1141 :     MLRISC ra 0.06u 0.00s 0.05g
1142 :     (spills = 0 reloads = 0)
1143 :     TOTAL 0.63u 0.07s 0.21g
1144 :    
1145 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
1146 :     MLRISC ra 5.68u 0.59s 3.93g
1147 :     (spills = 0 reloads = 0)
1148 :     TOTAL 14.71u 0.99s 4.81g
1149 :    
1150 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
1151 :     MLRISC ra 5.54u 0.65s 3.91g
1152 :     (spills = 543 reloads = 1082)
1153 :     TOTAL 65.40u 3.13s 6.00g
1154 :    
1155 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
1156 :     MLRISC ra 0.80u 0.10s 0.55g
1157 :     (spills = 42 reloads = 84)
1158 :     TOTAL 129.42u 5.10s 4.13g
1159 :    
1160 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
1161 :     MLRISC ra 2.69u 0.27s 1.38g
1162 :     (spills = 62 reloads = 124)
1163 :     TOTAL 682.48u 19.61s 13.99g
1164 :    
1165 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
1166 :     MLRISC ra 4.96u 0.27s 2.72g
1167 :     (spills = 85 reloads = 170)
1168 :     TOTAL 2375.26u 57.21s 48.00g
1169 :    
1170 :     As you can see the old cps spill module suffers from some serious
1171 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
1172 :     instead of patching the problems up, I'm reimplementing it
1173 : leunga 1094 with a different algorithm. The new code is more modular,
1174 :     smaller when compiled, and substantially faster
1175 :     (O(n log n) time and O(n) space). Timing of the new spill module:
1176 :    
1177 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
1178 :     MLRISC ra 0.25u 0.02s 0.15g
1179 :     (spills=1 reloads=3)
1180 :     TOTAL 7.74u 0.34s 1.62g
1181 :    
1182 :     Implementation details:
1183 :    
1184 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
1185 :     number of live variables at any program point (the bandwidth)
1186 :     does not exceed a certain limit, which is determined by the
1187 :     size of the spill area.
1188 :    
1189 :     When the bandwidth is too large, we decrease the register pressure by
1190 :     packing live variables into spill records. How we achieve this is
1191 :     completely different than what we did in the old code.
1192 :    
1193 :     First, there is something about the MLRiscGen code generator
1194 :     that we should be aware of:
1195 :    
1196 :     o MLRiscGen performs code motion!
1197 :    
1198 :     In particular, it will move floating point computations and
1199 :     address computations involving only the heap pointer to
1200 :     their use sites (if there is only a single use).
1201 :     What this means is that if we have a CPS record construction
1202 :     statement
1203 :    
1204 :     RECORD(k,vl,w,e)
1205 :    
1206 :     we should never count the new record address w as live if w
1207 :     has only one use (which is often the case).
1208 :    
1209 :     We should do something similar to floating point, but the transformation
1210 :     there is much more complex, so I won't deal with that.
1211 :    
1212 :     Secondly, there are now two new cps primops at our disposal:
1213 :    
1214 :     1. rawrecord of record_kind option
1215 :     This pure operator allocates some uninitialized storage from the heap.
1216 :     There are two forms:
1217 :    
1218 :     rawrecord NONE [INT n] allocates a tagless record of length n
1219 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1220 :     and initializes the tag.
1221 :    
1222 :     2. rawupdate of cty
1223 :     rawupdate cty (v,i,x)
1224 :     Assigns to x to the ith component of record v.
1225 :     The storelist is not updated.
1226 :    
1227 :     We use these new primops for both spilling and increment record construction.
1228 :    
1229 :     1. Spilling.
1230 :    
1231 :     This is implemented with a linear scan algorithm (but generalized
1232 :     to trees). The algorithm will create a single spill record at the
1233 :     beginning of the cps function and use rawupdate to spill to it,
1234 :     and SELECT or SELp to reload from it. So both spills and reloads
1235 :     are fine-grain operations. In contrast, in the old algorithm
1236 :     "spills" have to be bundled together in records.
1237 :    
1238 :     Ideally, we should sink the spill record construction to where
1239 :     it is needed. We can even split the spill record into multiple ones
1240 :     at the places where they are needed. But CPS is not a good
1241 :     representation for global code motion, so I'll keep it simple and
1242 :     am not attempting this.
1243 :    
1244 :     2. Incremental record construction (aka record splitting).
1245 :    
1246 :     Long records with many component values which are simulatenously live
1247 :     (recall that single use record addresses are not considered to
1248 :     be live) are constructed with rawrecord and rawupdate.
1249 :     We allocate space on the heap with rawrecord first, then gradually
1250 :     fill it in with rawupdate. This is the technique suggested to me
1251 :     by Matthias.
1252 :    
1253 :     Some restrictions on when this is applicable:
1254 :     1. It is not a VECTOR record. The code generator currently does not handle
1255 :     this case. VECTOR record uses double indirection like arrays.
1256 :     2. All the record component values are defined in the same "basic block"
1257 :     as the record constructor. This is to prevent speculative
1258 :     record construction.
1259 :    
1260 :     ----------------------------------------------------------------------
1261 :     Name: Allen Leung
1262 : leunga 1086 Date: 2002/02/22 01:02:00 EST
1263 :     Tag: leunga-20020222-mlrisc-tools
1264 :    
1265 :     Minor bug fixes in the parser and rewriter
1266 :    
1267 :     ----------------------------------------------------------------------
1268 :     Name: Allen Leung
1269 : leunga 1085 Date: 2002/02/21 20:20:00 EST
1270 :     Tag: leunga-20020221-peephole
1271 :    
1272 :     Regenerated the peephole files. Some contained typos in the specification
1273 :     and some didn't compile because of pretty printing bugs in the old version
1274 :     of 'nowhere'.
1275 :    
1276 :     ----------------------------------------------------------------------
1277 :     Name: Allen Leung
1278 : leunga 1079 Date: 2002/02/19 20:20:00 EST
1279 :     Tag: leunga-20020219-mlrisc-tools
1280 :     Description:
1281 :    
1282 :     Minor bug fixes to the mlrisc-tools library:
1283 :    
1284 :     1. Fixed up parsing colon suffixed keywords
1285 :     2. Added the ability to shut the error messages up
1286 :     3. Reimplemented the pretty printer and fixed up/improved
1287 :     the pretty printing of handle and -> types.
1288 :     4. Fixed up generation of literal symbols in the nowhere tool.
1289 :     5. Added some SML keywords to to sml.sty
1290 :    
1291 :     ----------------------------------------------------------------------
1292 :     Name: Matthias Blume
1293 : blume 1078 Date: 2002/02/19 16:20:00 EST
1294 :     Tag: blume-20020219-cmffi
1295 :     Description:
1296 :    
1297 :     A wild mix of changes, some minor, some major:
1298 :    
1299 :     * All C FFI-related libraries are now anchored under $c:
1300 :     $/c.cm --> $c/c.cm
1301 :     $/c-int.cm --> $c/internals/c-int.cm
1302 :     $/memory.cm --> $c/memory/memory.cm
1303 :    
1304 :     * "make" tool (in CM) now treats its argument pathname slightly
1305 :     differently:
1306 :     1. If the native expansion is an absolute name, then before invoking
1307 :     the "make" command on it, CM will apply OS.Path.mkRelative
1308 :     (with relativeTo = OS.FileSys.getDir()) to it.
1309 :     2. The argument will be passed through to subsequent phases of CM
1310 :     processing without "going native". In particular, if the argument
1311 :     was an anchored path, then "make" will not lose track of that anchor.
1312 :    
1313 :     * Compiler backends now "know" their respective C calling conventions
1314 :     instead of having to be told about it by ml-nlffigen. This relieves
1315 :     ml-nlffigen from one of its burdens.
1316 :    
1317 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1318 :    
1319 :     * Export C_DEBUG and C_Debug from $c/c.cm.
1320 :    
1321 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1322 :     subtyping relationship between incomplete pointers and their complete
1323 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
1324 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
1325 :     the pointer target type is complete. In the incomplete case, 'o
1326 :     will be instantiated with some "'c iobj" -- a type obtained by
1327 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
1328 :    
1329 :     Operations that work on both incomplete and complete pointer types are
1330 :     typed as taking an 'o ptr while operations that require the target to
1331 :     be known are typed as taking some ('t, 'c) obj ptr.
1332 :    
1333 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1334 :     where void is an eqtype without any values. This makes it possible
1335 :     to work on voidptr values using functions meant to operate on light
1336 :     incomplete pointers.
1337 :    
1338 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1339 :     been vastly simplified.
1340 :    
1341 :     ----------------------------------------------------------------------
1342 :     Name: Matthias Blume
1343 : blume 1076 Date: 2002/02/19 10:48:00 EST
1344 :     Tag: blume-20020219-pqfix
1345 :     Description:
1346 :    
1347 :     Applied Chris Okasaki's bug fix for priority queues.
1348 :    
1349 :     ----------------------------------------------------------------------
1350 :     Name: Matthias Blume
1351 : blume 1073 Date: 2002/02/15 17:05:00
1352 : blume 1069 Tag: Release_110_39
1353 :     Description:
1354 :    
1355 : blume 1073 Last-minute retagging is becoming a tradition... :-(
1356 :    
1357 :     This is the working release 110.39.
1358 :    
1359 :     ----------------------------------------------------------------------
1360 :     Name: Matthias Blume
1361 :     Date: 2002/02/15 16:00:00 EST
1362 :     Tag: Release_110_39-orig
1363 :     Description:
1364 :    
1365 : blume 1069 Working release 110.39. New bootfiles.
1366 :    
1367 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
1368 :     with all shells. So I retagged. -Matthias)
1369 :    
1370 : blume 1069 ----------------------------------------------------------------------
1371 :     Name: Matthias Blume
1372 : blume 1068 Date: 2002/02/15 14:17:00 EST
1373 :     Tag: blume-20020215-showbindings
1374 :     Description:
1375 :    
1376 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
1377 :     the latter can be useful for exploring what bindings are available at
1378 :     the interactive prompt. (The first function returns only the list
1379 :     of symbols that are really bound, the second prints those but also the
1380 :     ones that CM's autoloading mechanism knows about.)
1381 :    
1382 :     ----------------------------------------------------------------------
1383 :     Name: Matthias Blume
1384 : blume 1067 Date: 2002/02/15 12:08:00 EST
1385 :     Tag: blume-20020215-iptrs
1386 :     Description:
1387 :    
1388 :     Two improvements to ml-nlffigen:
1389 :    
1390 :     1. Write files only if they do not exist or if their current contents
1391 :     do not coincide with what's being written. (That is, avoid messing
1392 :     with the time stamps unless absolutely necessary.)
1393 :    
1394 :     2. Implement a "repository" mechanism for generated files related
1395 :     to "incomplete pointer types". See the README file for details.
1396 :    
1397 :     ----------------------------------------------------------------------
1398 :     Name: Matthias Blume
1399 : blume 1066 Date: 2002/02/14 11:50:00 EST
1400 :     Tag: blume-20020214-quote
1401 :     Description:
1402 :    
1403 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
1404 :     because of the new and improved tag generation scheme. (Thanks to Allen
1405 :     Leung for pointing it out.)
1406 :    
1407 :     ----------------------------------------------------------------------
1408 : george 1065 Name: Lal George
1409 :     Date: 2002/02/14 09:55:27 EST 2002
1410 :     Tag: george-20020214-isabelle-bug
1411 :     Description:
1412 :    
1413 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1414 :     of Isabelle on the x86.
1415 :    
1416 :     From Allen:
1417 :     -----------
1418 :     I've found the problem:
1419 :    
1420 :     in ra-core.sml, I use the counter "blocked" to keep track of the
1421 :     true number of elements in the freeze queue. When the counter goes
1422 :     to zero, I skip examining the queue. But I've messed up the
1423 :     bookkeeping in combine():
1424 :    
1425 :     else ();
1426 :     case !ucol of
1427 :     PSEUDO => (if !cntv > 0 then
1428 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
1429 :     ^^^^^^^^^^^^^^^^^^^^^^^
1430 :     moveu := mergeMoveList(!movev, !moveu)
1431 :     )
1432 :     else ();
1433 :    
1434 :     combine() is called to coalesce two nodes u and v.
1435 :     I think I was thinking that if the move counts of u and v are both
1436 :     greater than zero then after they are coalesced then one node is
1437 :     removed from the freeze queue. Apparently I was thinking that
1438 :     both u and v are of low degree, but that's clearly not necessarily true.
1439 :    
1440 :    
1441 :     02/12/2002:
1442 :     Here's the patch. HOL now compiles.
1443 :    
1444 :     I don't know how this impact on performance (compile
1445 :     time or runtime). This bug caused the RA (especially on the x86)
1446 :     to go thru the potential spill phase when there are still nodes on the
1447 :     freeze queue.
1448 :    
1449 :    
1450 :    
1451 :    
1452 :     ----------------------------------------------------------------------
1453 : blume 1058 Name: Matthias Blume
1454 : blume 1064 Date: 2002/02/13 22:40:00 EST
1455 :     Tag: blume-20020213-fptr-rtti
1456 :     Description:
1457 :    
1458 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
1459 :     updates.
1460 :    
1461 :     ----------------------------------------------------------------------
1462 :     Name: Matthias Blume
1463 : blume 1063 Date: 2002/02/13 16:41:00 EST
1464 :     Tag: blume-20020213-cmlpq
1465 :     Description:
1466 :    
1467 :     Added new priority queue export symbols (which have just been added to
1468 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
1469 :     not compile and the installer would choke.)
1470 :    
1471 :     ----------------------------------------------------------------------
1472 :     Name: Matthias Blume
1473 : blume 1062 Date: 2002/02/13 16:15:00 EST
1474 :     Tag: blume-20020213-various
1475 :     Description:
1476 :    
1477 :     1. More tweaks to ml-nlffigen:
1478 :    
1479 :     - better internal datastructures (resulting in slight speedup)
1480 :     - "-match" option requires exact match
1481 :     - "localized" gensym counters (untagged structs/unions nested within
1482 :     other structs/unions or within typedefs get a fresh counter; their
1483 :     tag will be prefixed by a concatenation of their parents' tags)
1484 :     - bug fixes (related to calculation of transitive closure of types
1485 :     to be included in the output)
1486 :    
1487 :     2. Minor Basis updates:
1488 :    
1489 :     - added implementations for List.collate and Option.app
1490 :    
1491 :     ----------------------------------------------------------------------
1492 :     Name: Matthias Blume
1493 : blume 1060 Date: 2002/02/11 15:55:00 EST
1494 :     Tag: blume-20020211-gensym
1495 :     Description:
1496 :    
1497 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
1498 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
1499 :     names (ML structure names that correspond to unnamed C structs, unions,
1500 :     and enums), so that separate runs of ml-nlffigen do not clash.
1501 :    
1502 :     ----------------------------------------------------------------------
1503 :     Name: Matthias Blume
1504 : blume 1059 Date: 2002/02/11 12:05:00 EST
1505 :     Tag: blume-20020211-gensml
1506 :     Description:
1507 :    
1508 :     A quick fix for a problem with GenSML (in the pgraph-util library):
1509 :     Make generation of toplevel "local" optional. (Strictly speaking,
1510 :     signature definitions within "local" are not legal SML.)
1511 :    
1512 :     Other than that: updates to INSTALL and cm/TODO.
1513 :    
1514 :     ----------------------------------------------------------------------
1515 :     Name: Matthias Blume
1516 : blume 1058 Date: 2002/02/08 15:00:00 EST
1517 :     Tag: blume-20020208-uniquepid
1518 :     Description:
1519 :    
1520 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
1521 :    
1522 :     1. The installer (config/install.sh) has gotten smarter:
1523 :    
1524 :     - Configuration options are a bit easier to specify now
1525 :     (in config/targets).
1526 :     - Bug in recognizing .tar.bz2 files fixed.
1527 :     - Installer automatically resolves dependencies between
1528 :     configuration options (e.g., if you ask for eXene, you will
1529 :     also get cml -- regardless whether you asked for it or not).
1530 :     - Installer can run in "quieter mode" by setting the environment
1531 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
1532 :     "completely silent", though.
1533 :     - Build HashCons library as part of smlnj-lib.
1534 :    
1535 :     2. A new scheme for assigning persistent identifiers to compilation
1536 :     units (and, by extension, to types etc.) has been put into place.
1537 :     This fixes a long-standing bug where types and even dynamic values
1538 :     can get internally confused, thereby compromising type safety
1539 :     (abstraction) and dynamic correctness. See
1540 :    
1541 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1542 :    
1543 :     for an example of how things could go wrong until now.
1544 :    
1545 :     The downside of the new scheme is that pids are not quite as
1546 :     persistent as they used to be: CM will generate a fresh pid
1547 :     for every compilation unit that it thinks it sees for the first
1548 :     time. That means that if you compile starting from a clean, fresh
1549 :     source tree at two different times, you end up with different
1550 :     binaries.
1551 :    
1552 :     Cutoff recompilation, however, has not been compromised because
1553 :     CM keeps pid information in special caches between runs.
1554 :    
1555 :     ----------------------------------------------------------------------
1556 : george 1053 Name: Lal George
1557 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
1558 :     Tag: <none>
1559 :     Description:
1560 :    
1561 :     Compilers that generate assembly code may produce global labels
1562 :     whose value is resolved at link time. The various peephole optimization
1563 :     modules did not take this in account.
1564 :    
1565 :     TODO. The Labels.addrOf function should really return an option
1566 :     type so that clients are forced to deal with this issue, rather
1567 :     than an exception being raised.
1568 :    
1569 :     ----------------------------------------------------------------------
1570 :     Name: Lal George
1571 : george 1053 Date: 2002/02/06 13:55:02 EST
1572 :     Tag: george-20020206-ra-breakup
1573 :     Description:
1574 :    
1575 :     1. A bug fix from Allen.
1576 :    
1577 :     A typo causes extra fstp %st(0)'s to be generated at compensation
1578 :     edges, which might cause stack underflow traps at runtime. This
1579 :     occurs in fft where there are extraneous fstps right before the 'into'
1580 :     trap instruction (in this case they are harmless since none of the
1581 :     integers overflow.)
1582 :    
1583 : george 1054 2. Pulled out various utility modules that were embedded in the modules
1584 : george 1053 of the register allocator. I need these modules for other purposes, but
1585 : george 1054 they are not complete enough to put into a library (just yet).
1586 : george 1053 ----------------------------------------------------------------------
1587 : blume 1049 Name: Matthias Blume
1588 :     Date: 2002/01/31 16:05:00 EST
1589 :     Tag: blume-20020131-sparc-ccalls
1590 :     Description:
1591 :    
1592 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1593 :     of extra stack space by mistake. Fixed.
1594 :    
1595 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1596 :    
1597 :     ----------------------------------------------------------------------
1598 : leunga 1048 Name: Allen Leung
1599 :     Date: 2002/01/30
1600 :     Tag: leunga-20020130-nowhere-bug-fix
1601 :     Description:
1602 :    
1603 :     MLRISC bug fixes:
1604 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1605 :     2. MachineInt.fromString was negating its value.
1606 :    
1607 :     ----------------------------------------------------------------------
1608 : blume 1039 Name: Matthias Blume
1609 : blume 1047 Date: 2002/01/29
1610 :     Tag: blume-20020129-INSTALL
1611 :     Description:
1612 :    
1613 :     - Added somewhat detailed installation instructions (file INSTALL).
1614 :     - Fixed curl-detection bug in config/install.sh.
1615 :     - It is now possible to select the URL getter using the URLGETTER
1616 :     environment variable:
1617 :    
1618 :     not set / "unknown" --> automatic detection (script tries wget,
1619 :     curl, and lynx)
1620 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1621 :     how to properly invoke them)
1622 :     other --> use $URLGETTER directly, it must take
1623 :     precisely two command-line arguments
1624 :     (source URL and destination file name)
1625 :    
1626 :     ----------------------------------------------------------------------
1627 :     Name: Matthias Blume
1628 : blume 1044 Date: 2002/01/28
1629 : blume 1045 Tag: blume-20020128-sparc-ccalls
1630 :     Description:
1631 :    
1632 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
1633 :     - Make use of the allocParam argument in sparc-c-calls.
1634 :    
1635 :     ----------------------------------------------------------------------
1636 :     Name: Matthias Blume
1637 :     Date: 2002/01/28
1638 : blume 1044 Tag: blume-20020128-allocParam
1639 :     Description:
1640 :    
1641 :     John Reppy: Changes c-calls API to accept client-callback for
1642 :     allocating extra stack space.
1643 :     me: Corresponding changes to mlriscGen (using a dummy argument that
1644 :     does not change the current behavior).
1645 :    
1646 :     ----------------------------------------------------------------------
1647 :     Name: Matthias Blume
1648 : blume 1040 Date: 2002/01/28 12:00:00
1649 :     Tag: Release_110_38
1650 :     Description:
1651 :    
1652 :     This time for real!!!
1653 :    
1654 :     ----------------------------------------------------------------------
1655 :     Name: Matthias Blume
1656 : blume 1039 Date: 2002/01/28 10:56:00 EST
1657 :     Tag: blume-20020128-retraction
1658 :     Description:
1659 :    
1660 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
1661 :     with blume-Release_110_38-retracted.)
1662 :    
1663 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
1664 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1665 :    
1666 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1667 :     and enums fixed. The naming of corresponding ML identifiers should
1668 :     now be consistent again.
1669 :    
1670 :     ----------------------------------------------------------------------
1671 : leunga 1038 Name: Allen Leung
1672 :     Date: 2002/01/27
1673 :     Tag: leunga-20020127-nowhere
1674 :     Description:
1675 :    
1676 :     Added a target called nowhere in the configuration scripts.
1677 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
1678 :     programs with where-clauses into legal SML code) during installation.
1679 :    
1680 :     ----------------------------------------------------------------------
1681 : blume 1036 Name: Matthias Blume
1682 : leunga 1038 Date: 2002/01/25 21:27:00 EST
1683 : blume 1039 Tag: blume-Release_110_38-retracted
1684 : blume 1037 Description:
1685 :    
1686 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
1687 :    
1688 :     README will be added later.
1689 :    
1690 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
1691 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
1692 :    
1693 : blume 1037 ----------------------------------------------------------------------
1694 :     Name: Matthias Blume
1695 : blume 1036 Date: 2002/01/25
1696 :     Tag: blume-20020125-ffi
1697 :     Description:
1698 :    
1699 :     A large number of tweaks and improvements to ml-nlffi-lib and
1700 :     ml-nlffigen:
1701 :    
1702 :     - ML represenation types have been streamlined
1703 :     - getter and setter functions work with concrete values, not abstract
1704 :     ones where possible
1705 :     - ml-nlffigen command line more flexible (see README file there)
1706 :     - some bugs have been fixed (hopefully)
1707 :    
1708 :     ----------------------------------------------------------------------
1709 : george 1034 Name: Lal George
1710 :     Date: 2002/01/24
1711 :     Tag: george-20020124-risc-ra-interface
1712 :     Description:
1713 :    
1714 :     There is a dramatic simplification in the interface to the
1715 :     register allocator for RISC architectures as a result of making
1716 :     parallel copy instructions explicit.
1717 :    
1718 :     ----------------------------------------------------------------------
1719 : blume 1028 Name: Matthias Blume
1720 : blume 1032 Date: 2002/01/22
1721 :     Tag: blume-20020122-x86-ccalls
1722 :     Description:
1723 :    
1724 :     Bug fix for c-calls on x86 (having to do with how char- and
1725 :     short-arguments are being handled).
1726 :    
1727 :     ----------------------------------------------------------------------
1728 :     Name: Matthias Blume
1729 : blume 1031 Date: 2002/01/21
1730 :     Tag: blume-20020121-ff
1731 :     Description:
1732 :    
1733 :     Another day of fiddling with the FFI...
1734 :    
1735 :     1. Bug fix/workaround: CKIT does not complain about negative array
1736 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1737 :     (Otherwise a negative dimension would send it into an infinite loop.)
1738 :    
1739 :     2. Some of the abstract types (light objects, light pointers, most "base"
1740 :     types) in structure C are now eqtypes.
1741 :    
1742 :     3. Added constructors and test functions for NULL function pointers.
1743 :    
1744 :     ----------------------------------------------------------------------
1745 :     Name: Matthias Blume
1746 : blume 1028 Date: 2002/01/18
1747 : blume 1030 Tag: blume-20020118-ready-for-new-release
1748 :     Description:
1749 :    
1750 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1751 :     files shortly.)
1752 :    
1753 :     Maybe we christen this to be 110.38?
1754 :    
1755 :     ----------------------------------------------------------------------
1756 :     Name: Matthias Blume
1757 :     Date: 2002/01/18
1758 : blume 1029 Tag: blume-20020118-more-ffifiddle
1759 :     Description:
1760 :    
1761 :     Today's FFI fiddling:
1762 :    
1763 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1764 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1765 :     of abstract ones.
1766 :    
1767 :     - Provided word-style bit operations etc. for "int" representation
1768 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1769 :     and Long.
1770 :    
1771 :     ----------------------------------------------------------------------
1772 :     Name: Matthias Blume
1773 :     Date: 2002/01/18
1774 : blume 1028 Tag: blume-20020118-use-x86-fp
1775 :     Description:
1776 :    
1777 :     Now that x86-fast-fp seems to be working, I turned it back on again
1778 :     by default. (Seems to work fine now, even with the FFI.)
1779 :    
1780 :     Other than that, I added some documentation about the FFI to
1781 :     src/ml-nlffigen/README and updated the FFI test examples in
1782 :     src/ml-nlffi-lib/Tests/*.
1783 :    
1784 :     ----------------------------------------------------------------------
1785 : leunga 1025 Name: Allen Leung
1786 :     Date: 2002/01/17
1787 :     Tag: leunga-20020117-x86-fast-fp-call
1788 :     Description:
1789 :    
1790 :     1. Fixed a problem with handling return fp values when x86's fast fp
1791 :     mode is turned on.
1792 :    
1793 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1794 :     of %f32.
1795 :    
1796 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1797 :    
1798 :     ----------------------------------------------------------------------
1799 : blume 1021 Name: Matthias Blume
1800 :     Date: 2002/01/16
1801 :     Tag: blume-20020116-ffifiddle
1802 :     Description:
1803 :    
1804 :     More fiddling with the FFI interface:
1805 :    
1806 :     - Make constness 'c instead of rw wherever possible. This eliminates
1807 :     the need for certain explicit coercions. (However, due to ML's
1808 :     value polymorphism, there will still be many cases where explicit
1809 :     coercions are necessary. Phantom types are not the whole answer
1810 :     to modeling a subtyping relationship in ML.)
1811 :    
1812 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1813 :    
1814 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1815 :     clumsy workaround for varargs problem.)
1816 :    
1817 :     ----------------------------------------------------------------------
1818 : george 1016 Name: Lal George
1819 :     Date: 2002/01/15
1820 :     Tag: <none>
1821 :     Description:
1822 :    
1823 :     1. Since COPY instructions are no longer native to the architecture,
1824 :     a generic functor can be used to implement the expandCopies function.
1825 :    
1826 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1827 :     TEXT segment.
1828 :    
1829 :     ----------------------------------------------------------------------
1830 : blume 1011 Name: Matthias Blume
1831 : blume 1015 Date: 2002/01/15
1832 :     Tag: blume-20020115-ffiupdates
1833 :     Description:
1834 :    
1835 :     1. Fix for bug resulting in single-precision float values being returned
1836 :     incorrectly from FFI calls.
1837 :    
1838 :     2. Small modifications to C FFI API:
1839 :    
1840 :     - memory-allocation routines return straight objects (no options)
1841 :     and raise an exception in out-of-memory situations
1842 :     - unsafe extensions to cast between function pointers and pointers
1843 :     from/to ints
1844 :     - added structure C_Debug as an alternative to structure C where
1845 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1846 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1847 :     but also takes a (possibly empty) list of existing library handles
1848 :     that the current library depends on
1849 :    
1850 :     ----------------------------------------------------------------------
1851 :     Name: Matthias Blume
1852 : blume 1011 Date: 2002/01/10
1853 :     Tag: blume-20020110-newffigen
1854 :     Description:
1855 :    
1856 :     1. Updates to portable graph code.
1857 :    
1858 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1859 :     more scalable now so that even huge interfaces such as the one
1860 :     for GTK compile in finite time and space. :-)
1861 :     See src/ml-nlffigen/README for details on what's new.
1862 :    
1863 :     ----------------------------------------------------------------------
1864 : george 1003 Name: Lal George
1865 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1866 :     Tag: george-20011206-rm-native-copy
1867 :     Description:
1868 :    
1869 :     Removed the native COPY and FCOPY instructions
1870 :     from all the architectures and replaced it with the
1871 :     explicit COPY instruction from the previous commit.
1872 :    
1873 :     It is now possible to simplify many of the optimizations
1874 :     modules that manipulate copies. This has not been
1875 :     done in this change.
1876 :    
1877 :     ----------------------------------------------------------------------
1878 :     Name: Lal George
1879 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1880 :     Tag: george-20011206-mlrisc-instruction
1881 :     Description:
1882 : george 824
1883 : george 1003 Changed the representation of instructions from being fully abstract
1884 :     to being partially concrete. That is to say:
1885 :    
1886 :     from
1887 :     type instruction
1888 :    
1889 :     to
1890 :     type instr (* machine instruction *)
1891 :    
1892 :     datatype instruction =
1893 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1894 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1895 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1896 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1897 :     | INSTR of instr
1898 :    
1899 :     This makes the handling of certain special instructions that appear on
1900 :     all architectures easier and uniform.
1901 :    
1902 :     LIVE and KILL say that a list of registers are live or killed at the
1903 :     program point where they appear. No spill code is generated when an
1904 :     element of the 'regs' field is spilled, but the register is moved to
1905 :     the 'spilled' (which is present, more for debugging than anything else).
1906 :    
1907 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1908 :     We used to generate:
1909 :    
1910 :     DEFFREG f1
1911 :     f1 := f2 + f3
1912 :     trapb
1913 :    
1914 :     but now generate:
1915 :    
1916 :     f1 := f2 + f3
1917 :     trapb
1918 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1919 :    
1920 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1921 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1922 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1923 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1924 :     this architecture).
1925 :    
1926 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1927 :     all the architectures. This will result in further simplification of the
1928 :     register allocator that must be aware of them for coalescing purposes, and
1929 :     will also simplify certain aspects of the machine description that provides
1930 :     callbacks related to parallel copies.
1931 :    
1932 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1933 :     machine instruction set!
1934 :    
1935 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1936 :     functions for making porting easier -- essentially converting upper case
1937 :     to lower case. All machine instructions (of type instr) are in upper case,
1938 :     and the lower case form generates an MLRISC instruction. For example on
1939 :     the alpha we have:
1940 :    
1941 :     datatype instr =
1942 :     LDA of {r:cell, b:cell, d:operand}
1943 :     | ...
1944 :    
1945 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1946 :     ...
1947 :    
1948 :     where lda is just (INSTR o LDA), etc.
1949 :    
1950 : leunga 815 ----------------------------------------------------------------------
1951 : blume 986 Name: Matthias Blume
1952 : blume 993 Date: 2001/11/22 21:40:00 EST
1953 :     Tag: Release_110_37
1954 :     Description:
1955 :    
1956 :     Release 110.37. This time for real.
1957 :    
1958 :     ----------------------------------------------------------------------
1959 :     Name: Matthias Blume
1960 : blume 988 Date: 2001/11/21 16:35:00 EST
1961 :     Tag: blume-20011121-foot-in-mouth
1962 :     Description:
1963 :    
1964 :     Removed the "Release_110_37" tag because of a serious bug.
1965 :     This will be re-tagged once the bug is fixed.
1966 :    
1967 :     ----------------------------------------------------------------------
1968 :     Name: Matthias Blume
1969 : blume 987 Date: 2001/11/21 16:14:00 EST
1970 :     Tag: blume-20011121-forgottenfile
1971 :     Description:
1972 :    
1973 :     Forgot to add a file. (Just a .tex-file -- part of
1974 :     the CM manual source.)
1975 :    
1976 :     ----------------------------------------------------------------------
1977 :     Name: Matthias Blume
1978 : blume 986 Date: 2001/11/21 16:10:00 EST
1979 : blume 988 Tag: blume-20011121-invalid_110_37
1980 : blume 986 Description:
1981 :    
1982 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
1983 :     because we found a serious bug in all non-x86 backends.
1984 :     - Matthias
1985 :    
1986 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
1987 :     so that code object name strings are directly inserted into code
1988 :     objects at code generation time. The only business the runtime system
1989 :     has with this is now to read the name strings on occasions.
1990 :     (The encoding of the name string has also changed somewhat.)
1991 :    
1992 :     2. CM now implements a simple "set calculus" for specifying export lists.
1993 :     In particular, it is now possible to refer to the export lists of
1994 :     other libraries/groups/sources and form unions as well as differences.
1995 :     See the latest CM manual for details.
1996 :    
1997 :     3. An separate notion of "proxy" libraries has again be eliminated from
1998 :     CM's model. (Proxy libraries are now simply a special case of using
1999 :     the export list calculus.)
2000 :    
2001 :     4. Some of the existing libraries now take advantage of the new set
2002 :     calculus.
2003 :     (Notice that not all libraries have been converted because some
2004 :     of the existing .cm-files are supposed to be backward compatible
2005 :     with 110.0.x.)
2006 :    
2007 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
2008 :     "General.exnMessage"! The former relies on a certain hook to be
2009 :     initialized, and that often does not happen in the stand-alone case.)
2010 :    
2011 :     ----------------------------------------------------------------------
2012 : george 984 Name: Lal George
2013 :     Date: 2001/11/21 13:56:18 EST
2014 :     Tag: george-2001121-pseudo-ops
2015 :     Description:
2016 :    
2017 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
2018 :     ought to never be any question of incompatabilities with
2019 :     pseudo-op syntax expected by host assemblers.
2020 :    
2021 :     For now, only modules supporting GAS syntax are implemented
2022 :     but more should follow, such as MASM, and vendor assembler
2023 :     syntax, e.g. IBM as, Sun as, etc.
2024 :    
2025 :     ----------------------------------------------------------------------
2026 : blume 977 Name: Matthias Blume
2027 :     Date: 2001/11/14 11:52:00 EST
2028 :     Tag: blume-20011114-srcname
2029 :     Description:
2030 :    
2031 :     1. Routed the name of the current source file to mlriscgen where it
2032 :     should be directly emitted into the code object. (This last part
2033 :     is yet to be done.)
2034 :    
2035 :     2. Some cleanup of the pgraph code to make it match the proposal that
2036 :     I put out the other day. (The proposal notwithstanding, things are
2037 :     still in flux here.)
2038 :    
2039 :     ----------------------------------------------------------------------
2040 : george 976 Name: Lal George
2041 :     Date: 2001/11/14 09:44:04 EST
2042 :     Tag:
2043 :     Description:
2044 :    
2045 :     Fix for a backpatching bug reported by Allen.
2046 :    
2047 :     Because the boundary between short and long span-dependent
2048 :     instructions is +/- 128, there are an astounding number of
2049 :     span-dependent instructions whose size is over estimated.
2050 :    
2051 :     Allen came up with the idea of letting the size of span
2052 :     dependent instructions be non-monotonic, for a maxIter
2053 :     number of times, after which the size must be monotonically
2054 :     increasing.
2055 :    
2056 :     This table shows the number of span-dependent instructions
2057 :     whose size was over-estimated as a function of maxIter, for the
2058 :     file Parse/parse/ml.grm.sml:
2059 :    
2060 :     maxIter # of instructions:
2061 :     10 687
2062 :     20 438
2063 :     30 198
2064 :     40 0
2065 :    
2066 :     In compiling the compiler, there is no significant difference in
2067 :     compilation speed between maxIter=10 and maxIter=40. Actually,
2068 :     my measurements showed that maxIter=40 was a tad faster than
2069 :     maxIter=10! Also 96% of the files in the compiler reach a fix
2070 :     point within 13 iterations, so fixing maxIter at 40, while high,
2071 :     is okay.
2072 :    
2073 :     ----------------------------------------------------------------------
2074 : blume 975 Name: Matthias Blume
2075 :     Date: 2001/10/31 15:25:00 EST
2076 :     Tag: blume-20011031-pgraph
2077 :     Description:
2078 :    
2079 :     CKIT:
2080 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
2081 :     argument identifiers.
2082 :     * Changed the return type of TypeUtil.getFunction accordingly.
2083 :     * Type equality ignores the argument names.
2084 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
2085 :     if there is a mismatch.
2086 :    
2087 :     installation script:
2088 :     * attempts to use "curl" if available (unless "wget" is available as well)
2089 :    
2090 :     CM:
2091 :     * has an experimental implementation of "portable graphs" which I will
2092 :     soon propose as an implementation-independent library format
2093 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2094 :    
2095 :     NLFFI-LIB:
2096 :     * some cleanup (all cosmetic)
2097 :    
2098 :     NLFFIGEN:
2099 :     * temporarily disabled the mechanism that suppresses ML output for
2100 :     C definitions whose identifiers start with an underscore character
2101 :     * generate val bindings for enum constants
2102 :     * user can request that only one style (light or heavy) is being used;
2103 :     default is to use both (command-line arguments: -heavy and -light)
2104 :     * fixed bug in handling of function types involving incomplete pointers
2105 :     * generate ML entry points that take record arguments (i.e., using
2106 :     named arguments) for C functions that have a prototype with named
2107 :     arguments
2108 :     (see changes to CKIT)
2109 :    
2110 :     ----------------------------------------------------------------------
2111 : leunga 972 Name: Allen Leung
2112 :     Date: 2001/10/27 20:34:00 EDT
2113 :     Tag: leunga-20011027-x86-fast-fp-call
2114 :     Description:
2115 :    
2116 :     Fixed the bug described in blume-20010920-slowfp.
2117 :    
2118 :     The fix involves
2119 :     1. generating FCOPYs in FSTP in ia32-svid
2120 :     2. marking a CALL with the appropriate annotation
2121 :    
2122 :     ----------------------------------------------------------------------
2123 : blume 961 Name: Matthias Blume
2124 :     Date: 2001/10/16 11:32:00 EDT
2125 :     Tag: blume-20011016-netbsd
2126 :     Description:
2127 :    
2128 :     Underscore patch from Chris Richards (fixing problem with compiling
2129 :     runtime system under recent NetBSD).
2130 :    
2131 :     ----------------------------------------------------------------------
2132 : george 959 Name: Allen Leung
2133 :     Date: 2001/10/12 17:18:32 EDT 2001
2134 :     Tag: leung-20011012-x86-printflowgraph
2135 :     Description:
2136 :    
2137 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2138 :    
2139 :     ----------------------------------------------------------------------
2140 : george 952 Name: Lal George
2141 : george 958 Date: 2001/10/11 23:51:34 EDT
2142 :     Tag: george-20011011-too-many-instrs
2143 :     Description:
2144 :    
2145 :     The representation of a program point never expected to see more
2146 :     than 65536 instructions in a basic block!
2147 :    
2148 :     ----------------------------------------------------------------------
2149 :     Name: Lal George
2150 : george 952 Date: 2001/10/09 09:41:37 EDT
2151 :     Tag: george-20011008-mlrisc-labels
2152 :     Description:
2153 :    
2154 :     Changed the machine description files to support printing of
2155 :     local and global labels in assembly code, based on host assembler
2156 :     conventions.
2157 :    
2158 :     ----------------------------------------------------------------------
2159 : blume 916 Name: Matthias Blume
2160 : blume 939 Date: 2001/09/25 15:25:00 EDT
2161 :     Tag: blume-20010925-exninfo
2162 :     Description:
2163 :    
2164 :     I provided a non-hook implementation of exnName (at the toplevel) and
2165 :     made the "dummy" implementation of exnMessage (at the toplevel) more
2166 :     useful: if nothing gets "hooked in", then at least you are going to
2167 :     see the exception name and a message indicating why you don't see more.
2168 :    
2169 :     [For the time being, programs that need exnMessage and want to use
2170 :     ml-build should either use General.exnMessage (strongly recommended) or
2171 :     refer to structure General at some other point so that CM sees a
2172 :     static dependency.]
2173 :    
2174 :     [Similar remarks go for "print" and "use": If you want to use their
2175 :     functionality in stand-alone programs generated by ml-build, then use
2176 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2177 :    
2178 :     ----------------------------------------------------------------------
2179 :     Name: Matthias Blume
2180 : blume 938 Date: 2001/09/20 17:28:00 EDT
2181 :     Tag: blume-20010920-slowfp
2182 :     Description:
2183 :    
2184 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2185 :    
2186 :     ----------------------------------------------------------------------
2187 :     Name: Matthias Blume
2188 : blume 937 Date: 2001/09/20 17:20:00 EDT
2189 :     Tag: blume-20010920-canonicalpaths
2190 :     Description:
2191 :    
2192 :     0. Updated the BOOT file (something that I forgot to do earlier).
2193 :    
2194 :     1. Small internal change to CM so that it avoids "/../" in filenames
2195 :     as much as possible (but only where it is safe).
2196 :    
2197 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2198 :     that arguments that contain delimiters are passed through correctly.
2199 :     This change also means that all "special" arguments of the form
2200 :     @SMLxxx... must come first.
2201 :    
2202 :     3. Changed install script to put relative anchor names for tool commands
2203 :     into pathconfig.
2204 :    
2205 :     ----------------------------------------------------------------------
2206 :     Name: Matthias Blume
2207 : blume 932 Date: 2001/09/18 15:35:00 EDT
2208 :     Tag: blume-20010918-readme11036
2209 :     Description:
2210 :    
2211 :     Added README files.
2212 :    
2213 :     ----------------------------------------------------------------------
2214 :     Name: Matthias Blume
2215 : blume 929 Date: 2001/09/18 11:45:00 EDT
2216 :     Tag: Release_110_36 (retag)
2217 :     Description:
2218 :    
2219 :     Fixed mistake in config/preloads. Retagged as 110.36.
2220 :    
2221 :     ----------------------------------------------------------------------
2222 :     Name: Matthias Blume
2223 : blume 928 Date: 2001/09/18 09:40:00 EDT
2224 : blume 929 Tag: Release_110_36_orig (tag changed)
2225 : blume 928 Description:
2226 :    
2227 :     New version (110.36). New bootfiles.
2228 :    
2229 :     ----------------------------------------------------------------------
2230 :     Name: Matthias Blume
2231 : blume 927 Date: 2001/09/14 16:15:00 EDT
2232 :     Tag: blume-20010914-x86fastfp
2233 :     Description:
2234 :    
2235 :     John committed some changes that Allen made, in particular a (hopefully)
2236 :     correctly working version of the x86-fp module.
2237 :    
2238 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2239 :     flag to "true". Everything seems to compile to a fixpoint ok, and
2240 :     "mandelbrot" speeds up by about 15%.
2241 :    
2242 :     ----------------------------------------------------------------------
2243 :     Name: Matthias Blume
2244 : blume 923 Date: 2001/09/13 11:20:00 EDT
2245 :     Tag: blume-20010913-minimal
2246 :     Description:
2247 :    
2248 :     1. Stefan Monnier's patch to fix a miscompilation problem that
2249 :     was brought to light by John Reppy's work on Moby.
2250 :    
2251 :     2. Implemented a minimal "structure Compiler" that contains just
2252 :     "version" and "architecture". The minimal version will be
2253 :     available when the full version is not. This is for backward-
2254 :     compatibility with code that wants to test Compiler.version.
2255 :    
2256 :     ----------------------------------------------------------------------
2257 :     Name: Matthias Blume
2258 : blume 920 Date: 2001/08/28 14:03:00 EDT
2259 :     Tag: blume-20010828-ml-lex
2260 :     Description:
2261 :    
2262 :     Fix for bug 1581, received from Neophytos Michael.
2263 :    
2264 :     ----------------------------------------------------------------------
2265 :     Name: Matthias Blume
2266 : blume 918 Date: 2001/08/27 11:20:00 EDT
2267 :     Tag: blume-20010827-readme11035
2268 :     Description:
2269 :    
2270 :     Fleshed out the README file for 110.35.
2271 :    
2272 :     ----------------------------------------------------------------------
2273 :     Name: Matthias Blume
2274 : blume 916 Date: 2001/08/24 17:10:00 EDT
2275 :     Tag: Release_110_35
2276 :     Description:
2277 :    
2278 :     New version number (110.35). New bootfiles.
2279 :    
2280 :     ----------------------------------------------------------------------
2281 : george 915 Name: Lal George
2282 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
2283 :     Tag: george-20010824-MLRISC-graphs
2284 :     Description:
2285 :    
2286 :     removed clusters from MLRISC completely and replaced with graphs.
2287 :    
2288 :     ----------------------------------------------------------------------
2289 :     Name: Matthias Blume
2290 : blume 905 Date: 2001/08/23 17:50:00 EDT
2291 :     Tag: blume-20010823-toplevel
2292 :     Description:
2293 :    
2294 :     - some reorganization of the code that implements various kinds of
2295 :     environments in the compiler (static, dynamic, symbolic, combined)
2296 :     - re-implemented the EnvRef module so that evalStream works properly
2297 :     (if the stream contains references to "use", "CM.make", etc.)
2298 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2299 :    
2300 :     ----------------------------------------------------------------------
2301 :     Name: Matthias Blume
2302 : blume 904 Date: 2001/08/20 15:50 EDT
2303 :     Tag: blume20010820-slipup
2304 :     Description:
2305 :    
2306 :     I forgot to commit a few files. Here they are...
2307 :    
2308 :     ----------------------------------------------------------------------
2309 :     Name: Matthias Blume
2310 : blume 903 Date: 2001/08/20 15:35:00 EDT
2311 :     Tag: blume-20010820-debugprof
2312 :     Description:
2313 :    
2314 :     !!!! NEW BOOTFILES !!!!
2315 :    
2316 :     This is another round of reorganizing the compiler sources. This
2317 :     time the main goal was to factor out all the "instrumentation"
2318 :     passes (for profiling and backtracing) into their own library.
2319 :     The difficulty was to do it in such a way that it does not depend
2320 :     on elaborate.cm but only on elabdata.cm.
2321 :    
2322 :     Therefore there have been further changes to both elaborate.cm and
2323 :     elabdata.cm -- more "generic" things have been moved from the former
2324 :     to the latter. As a result, I was forced to split the assignment
2325 :     of numbers indicating "primtyc"s into two portions: SML-generic and
2326 :     SML/NJ-specific. Since it would have been awkward to maintain,
2327 :     I bit the bullet and actually _changed_ the mapping between these
2328 :     numbers and primtycs. The bottom line of this is that you need
2329 :     a new set of bin- and bootfiles.
2330 :    
2331 :     I have built new bootfiles for all architectures, so doing a fresh
2332 :     checkout and config/install.sh should be all you need.
2333 :    
2334 :     The newly created library's name is
2335 :    
2336 :     $smlnj/viscomp/debugprof.cm
2337 :    
2338 :     and its sources live under
2339 :    
2340 :     src/compiler/DebugProf
2341 :    
2342 :     ----------------------------------------------------------------------
2343 :     Name: Matthias Blume
2344 : blume 902 Date: 2001/08/15 17:15:00 EDT
2345 :     Tag: blume-20010815-compreorg
2346 :     Description:
2347 :    
2348 :     This is a first cut at reorganizing the CM libraries that make up the
2349 :     core of the compiler. The idea is to separate out pieces that could
2350 :     be used independently by tools, e.g., the parser, the typechecker, etc.
2351 :    
2352 :     The current status is a step in this direction, but it is not quite
2353 :     satisfactory yet. Expect more changes in the future.
2354 :    
2355 :     Here is the current (new) organization...
2356 :    
2357 :     What used to be $smlnj/viscomp/core.cm is now divided into
2358 :     six CM libraries:
2359 :    
2360 :     $smlnj/viscomp/basics.cm
2361 :     /parser.cm
2362 :     /elabdata.cm
2363 :     /elaborate.cm
2364 :     /execute.cm
2365 :     /core.cm
2366 :    
2367 :     The CM files for these libraries live under src/system/smlnj/viscomp.
2368 :     All these libraries are proxy libraries that contain precisely
2369 :     one CM library component. Here are the locations of the components
2370 :     (all within the src/compiler tree):
2371 :    
2372 :     Basics/basics.cm
2373 :     Parse/parser.cm
2374 :     ElabData/elabdata.cm
2375 :     Elaborator/elaborate.cm
2376 :     Execution/execute.cm
2377 :     core.cm
2378 :    
2379 :     [This organization is the same that has been used already
2380 :     for a while for the architecture-specific parts of the visible
2381 :     compiler and for the old version of core.cm.]
2382 :    
2383 :     As you will notice, many source files have been moved from their
2384 :     respective original locations to a new home in one of the above
2385 :     subtrees.
2386 :    
2387 :     The division of labor between the new libraries is the following:
2388 :    
2389 :     basics.cm:
2390 :     - Simple, basic definitions that pertain to many (or all) of
2391 :     the other libraries.
2392 :     parser.cm:
2393 :     - The SML parser, producing output of type Ast.dec.
2394 :     - The type family for Ast is also defined and exported here.
2395 :     elabdata.cm:
2396 :     - The datatypes that describe input and output of the elaborator.
2397 :     This includes types, absyn, and static environments.
2398 :     elaborator.cm:
2399 :     - The SML/NJ type checker and elaborator.
2400 :     This maps an Ast.dec (with a given static environment) to
2401 :     an Absyn.dec (with a new static environment).
2402 :     - This libraries implements certain modules that used to be
2403 :     structures as functors (to remove dependencies on FLINT).
2404 :     execute.cm:
2405 :     - Everything having to do with executing binary code objects.
2406 :     - Dynamic environments.
2407 :     core.cm:
2408 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
2409 :     - Top-level modules.
2410 :     - FLINT (this should eventually become its own library)
2411 :    
2412 :     Notes:
2413 :    
2414 :     I am not 100% happy with the way I separated the elaborator (and its
2415 :     data structures) from FLINT. Two instances of the same problem:
2416 :    
2417 :     1. Data structures contain certain fields that carry FLINT-specific
2418 :     information. I hacked around this using exn and the property list
2419 :     module from smlnj-lib. But the fact that there are middle-end
2420 :     specific fields around at all is a bit annoying.
2421 :    
2422 :     2. The elaborator calculates certain FLINT-related information. I tried
2423 :     to make this as abstract as I could using functorization, but, again,
2424 :     the fact that the elaborator has to perform calculations on behalf
2425 :     of the middle-end at all is not nice.
2426 :    
2427 :     3. Having to used exn and property lists is unfortunate because it
2428 :     weakens type checking. The other alternative (parameterizing
2429 :     nearly *everything*) is not appealing, though.
2430 :    
2431 :     I removed the "rebinding =" warning hack because due to the new organization
2432 :     it was awkward to maintain it. As a result, the compiler now issues some of
2433 :     these warnings when compiling init.cmi during bootstrap compilation. On
2434 :     the plus side, you also get a warning when you do, for example:
2435 :     val op = = Int32.+
2436 :     which was not the case up to now.
2437 :    
2438 :     I placed "assign" and "deref" into the _Core structure so that the
2439 :     code that deals with the "lazy" keyword can find them there. This
2440 :     removes the need for having access to the primitive environment
2441 :     during elaboration.
2442 :    
2443 :     ----------------------------------------------------------------------
2444 :     Name: Matthias Blume
2445 : blume 898 Date: 2001/08/13
2446 :     Tag: blume-20010813-closures
2447 :     Description:
2448 :    
2449 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
2450 :     performance of certain loops by avoiding needless closure allocation.
2451 :    
2452 :     ----------------------------------------------------------------------
2453 : george 890 Name: Lal George
2454 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
2455 :     Tag: george-20010731-x86-fmalloc
2456 :     Description: Fixed bug in x86 calls
2457 :    
2458 :     There was a bug where call instructions would mysteriously
2459 :     vanish. The call instruction had to be one that returned
2460 :     a floating point value.
2461 :    
2462 :     ----------------------------------------------------------------------
2463 :     Name: Lal George
2464 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
2465 :     Tag: george-20010719-simple-cells
2466 :     Description:
2467 :    
2468 :     I have dramatically simplified the interface for CELLS in MLRISC.
2469 :    
2470 :     In summary, the cells interface is broken up into three parts:
2471 :    
2472 :     1. CellsBasis : CELLS_BASIS
2473 :    
2474 :     CellsBasis is a top level structure and common for all
2475 :     architectures. it contains the definitions of basic datatypes
2476 :     and utility functions over these types.
2477 :    
2478 :     2. functor Cells() : CELLS
2479 :    
2480 :     Cells generates an interface for CELLS that incorporates the
2481 :     specific resources on the target architecture, such as the
2482 :     presence of special register classes, their number and size,
2483 :     and various useful substructures.
2484 :    
2485 :     3. <ARCH>CELLS
2486 :    
2487 :     e.g. SparcCells: SPARCCELLS
2488 :    
2489 :     <ARCH>CELLS usually contains additional bindings for special
2490 :     registers on the architecture, such as:
2491 :    
2492 :     val r0 : cell (* register zero *)
2493 :     val y : cell (* Y register *)
2494 :     val psr : cell (* processor status register *)
2495 :     ...
2496 :    
2497 :     The structure returned by applying the Cells functor is opened
2498 :     in this interface.
2499 :    
2500 :     The main implication of all this is that the datatypes for cells is
2501 :     split between CellsBasis and CELLS -- a fairly simple change for user
2502 :     code.
2503 :    
2504 :     In the old scheme the CELLS interface had a definitional binding of
2505 :     the form:
2506 :    
2507 :     signature CELLS = sig
2508 :    
2509 :     structure CellsBasis = CellsBasis
2510 :    
2511 :     ...
2512 :    
2513 :     end
2514 :    
2515 :     With all the sharing constraints that goes on in MLRISC, this old
2516 :     design quickly leads to errors such as:
2517 :    
2518 :     "structure definition spec inside of sharing ... "
2519 :    
2520 :    
2521 :     and appears to require an unacceptable amount of sharing and where
2522 :     constraint hackery.
2523 :    
2524 :     I think this error message (the interaction of definitional specs and
2525 :     sharing) requires more explanation on our web page.
2526 :    
2527 :     ----------------------------------------------------------------------
2528 : blume 826 Name: Matthias Blume
2529 : blume 879 Date: 2001/07/19 15:00:00 EDT
2530 :     Tag: blume-20010719-libreorg
2531 :     Description:
2532 :    
2533 :     This update puts together a fairly extensive but straightforward change
2534 :     to the way the libraries that implement the interactive system are
2535 :     organized:
2536 :    
2537 :     The biggest change is the elimination of structure Compiler. As a
2538 :     replacement for this structure, there is now a CM library
2539 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2540 :     that exports all the substructures of the original structure Compiler
2541 :     directly. So instead of saying Compiler.Foo.bar one now simply
2542 :     says Foo.bar. (The CM libraries actually export a collection of
2543 :     structures that is richer than the collection of substructures of
2544 :     structure Compiler.)
2545 :    
2546 :     To make the transition smooth, there is a separate library called
2547 :     $smlnj/compiler/compiler.cm which puts together and exports the
2548 :     original structure Compiler (or at least something very close to it).
2549 :    
2550 :     There are five members of the original structure Compiler
2551 :     that are not exported directly but which instead became members
2552 :     of a new structure Backend (described by signature BACKEND). These are:
2553 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2554 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
2555 :     architecture (: string).
2556 :    
2557 :     Structure Compiler.Version has become structure CompilerVersion.
2558 :    
2559 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2560 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2561 :     or x86, respectively.
2562 :     Each of these exports the same frontend structures that
2563 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
2564 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2565 :     Hppa, PPC, Sparc, or X86, respectively.
2566 :    
2567 :     Library $smlnj/compiler/all.cm exports the union of the exports of
2568 :     $smlnj/compiler/<arch>.cm
2569 :    
2570 :     There are no structures <Arch>Compiler anymore, use
2571 :     $smlnj/compiler/<arch>.cm instead.
2572 :    
2573 :     Library host-compiler-0.cm is gone. Instead, the internal library
2574 :     that instantiates CM is now called cm0.cm. Selection of the host
2575 :     compiler (backend) is no longer done here but. (Responsibility for it
2576 :     now lies with $smlnj/compiler/current.cm. This seems to be more
2577 :     logical.)
2578 :    
2579 :     Many individual files have been moved or renamed. Some files have
2580 :     been split into multiple files, and some "dead" files have been deleted.
2581 :    
2582 :     Aside from these changes to library organization, there are also changes
2583 :     to the way the code itself is organized:
2584 :    
2585 :     Structure Binfile has been re-implemented in such a way that it no
2586 :     longer needs any knowledge of the compiler. It exclusively deals
2587 :     with the details of binfile layout. It no longer invokes the
2588 :     compiler (for the purpose of creating new prospective binfile
2589 :     content), and it no longer has any knowledge of how to interpret
2590 :     pickles.
2591 :    
2592 :     Structure Compile (: COMPILE) has been stripped down to the bare
2593 :     essentials of compilation. It no longer deals with linking/execution.
2594 :     The interface has been cleaned up considerably.
2595 :    
2596 :     Utility routines for dealing with linking and execution have been
2597 :     moved into their own substructures.
2598 :    
2599 :     (The ultimate goal of these changes is to provide a light-weight
2600 :     binfile loader/linker (at least for, e.g., stable libraries) that
2601 :     does not require CM or the compiler to be present.)
2602 :    
2603 :     CM documentation has been updated to reflect the changes to library
2604 :     organization.
2605 :    
2606 :     ----------------------------------------------------------------------
2607 :     Name: Matthias Blume
2608 : blume 875 Date: 2001/07/10 17:30:00 EDT
2609 :     Tag: Release_110_34
2610 :     Description:
2611 :    
2612 : blume 876 Minor tweak to 110.34 (re-tagged):
2613 :    
2614 :     - README.html file added to CVS repository
2615 :     - runtime compiles properly under FreeBSD 3.X and 4.X
2616 :    
2617 :     ----------------------------------------------------------------------
2618 :     Name: Matthias Blume
2619 :     Date: 2001/07/10 17:30:00 EDT
2620 :     Tag: Release_110_34
2621 :     Description:
2622 :    
2623 : blume 875 New version number (110.34). New bootfiles.
2624 :    
2625 :     ----------------------------------------------------------------------
2626 :     Name: Matthias Blume
2627 : blume 874 Date: 2001/07/09 16:00:00 EDT
2628 :     Tag: blume-20010709-more-varargs
2629 :     Description:
2630 :    
2631 :     I changed the handling of varargs in ml-nlffigen again:
2632 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
2633 :    
2634 :     The immediate effect is that you can actually call a varargs function
2635 :     from ML -- but you can't actually supply any arguments beyond the ones
2636 :     specified explicitly. (For example, you can call printf with its format
2637 :     string, but you cannot pass additional arguments.)
2638 :    
2639 :     This behavior is only marginally more useful than the one before, but
2640 :     it has the advantage that a function or, more importantly, a function
2641 :     type never gets dropped on the floor, thus avoiding follow-up problems with
2642 :     other types that refer to the offending one.
2643 :    
2644 :     ----------------------------------------------------------------------
2645 :     Name: Matthias Blume
2646 : blume 873 Date: 2001/07/09 11:25:00 EDT
2647 :     Tag: blume-20010709-varargs
2648 :     Description:
2649 :    
2650 :     1. ckit-lib.cm now exports structure Error
2651 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2652 :     with a warning accompanied by a source location. Moreover, it
2653 :     merely skips the offending function or type and proceeds with the
2654 :     rest of its work.u As a result, one can safely feed C code containing
2655 :     "..." to ml-nlffigen.
2656 :     3. There are some internal improvements to CM, providing slightly
2657 :     more general string substitutions in the tools subsystem.
2658 :    
2659 :     ----------------------------------------------------------------------
2660 :     Name: Matthias Blume
2661 : blume 854 Date: 2001/06/27 15:10:00 EDT
2662 :     Tag: blume-20010627-concur
2663 :     Description:
2664 :    
2665 :     Fixed a small bug in CM's handling of parallel compilation.
2666 :     (You could observe the bug by Control-C-interrupting an ordinary
2667 :     CMB.make or CM.stabilize and then attaching some compile servers.
2668 :     The result was that all of a sudden the previously interrupted
2669 :     compilation would continue on its own. This was because of
2670 :     an over-optimization: CM did not bother to clean out certain queues
2671 :     when no servers were attached "anyway", resulting in the contents
2672 :     of these queues to grab control when new servers did get attached.)
2673 :    
2674 :     There is also another minor update to the CM manual.
2675 :    
2676 :     ----------------------------------------------------------------------
2677 :     Name: Matthias Blume
2678 : blume 853 Date: 2001/06/26 16:15:00 EDT
2679 :     Tag: blume-20010626-cmdoc
2680 :     Description:
2681 :    
2682 :     Minor typo fixed in CM manual (syntax diagram for libraries).
2683 :    
2684 :     ----------------------------------------------------------------------
2685 :     Name: Matthias Blume
2686 : blume 852 Date: 2001/06/25 22:55:00 EDT
2687 :     Tag: blume-20010625-x86pc
2688 :     Description:
2689 :    
2690 :     Fixed a nasty bug in the X86 assembly code that caused signal
2691 :     handlers to fail (crash) randomly.
2692 :    
2693 :     ----------------------------------------------------------------------
2694 :     Name: Matthias Blume
2695 : blume 846 Date: 2001/06/25 12:05:00 EDT
2696 :     Tag: blume-20010625-nlffigen
2697 :     Description:
2698 :    
2699 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
2700 :     Nick Carter <nbc@andrew.cmu.edu>.
2701 :    
2702 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2703 :     2. Default names for generated files are now derived from the name of
2704 :     the C file *without its directory*. In particular, this causes generated
2705 :     files to be placed locally even if the C file is in some system directory.
2706 :     3. Default names for generated signatures and structures are also derived
2707 :     from the C file name without its directory. This avoids silly things
2708 :     like "structure GL/GL".
2709 :     (Other silly names are still possible because ml-nlffigen does not do
2710 : blume 847 a thorough check of whether generated names are legal ML identifiers.
2711 :     When in doubt, use command line arguments to force particular names.)
2712 : blume 846
2713 :     ----------------------------------------------------------------------
2714 :     Name: Matthias Blume
2715 : blume 845 Date: 2001/06/21 12:25:00 EDT
2716 :     Tag: blume-20010621-eXene
2717 :     Description:
2718 :    
2719 :     eXene now compiles and (sort of) works again.
2720 :    
2721 :     The library name (for version > 110.33) is $/eXene.cm.
2722 :    
2723 :     I also added an new example in src/eXene/examples/nbody. See the
2724 :     README file there for details.
2725 :    
2726 :     ----------------------------------------------------------------------
2727 :     Name: Matthias Blume
2728 : blume 844 Date: 2001/06/20 16:40:00 EDT
2729 :     Tag: blume-20010620-cml
2730 :     Description:
2731 :    
2732 :     CML now compiles and works again.
2733 :    
2734 :     Libraries (for version > 110.33):
2735 :    
2736 :     $cml/cml.cm Main CML library.
2737 :     $cml/basis.cm CML's version of $/basis.cm.
2738 :     $cml/cml-internal.cm Internal helper library.
2739 :     $cml/core-cml.cm Internal helper library.
2740 :     $cml-lib/trace-cml.cm Tracing facility.
2741 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2742 :    
2743 :     The installer (config/install.sh) has been taught how to properly
2744 :     install this stuff.
2745 :    
2746 :     ----------------------------------------------------------------------
2747 :     Name: Matthias Blume
2748 : blume 843 Date: 2001/06/19 17:55:00 EDT
2749 :     Tag: blume-20010619-instantiate
2750 :     Description:
2751 :    
2752 :     This un-breaks the fix for bug 1432.
2753 :     (The bug was originally fixed in 110.9 but I broke it again some
2754 :     time after that.)
2755 :    
2756 :     ----------------------------------------------------------------------
2757 :     Name: Matthias Blume
2758 : blume 842 Date: 2001/06/19 17:25:00 EDT
2759 :     Tag: blume-20010619-signals
2760 :     Description:
2761 :    
2762 :     This should (hopefully) fix the long-standing signal handling bug.
2763 :     (The runtime system was constructing a continuation record with an
2764 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2765 :    
2766 :     ----------------------------------------------------------------------
2767 :     Name: Matthias Blume
2768 : blume 840 Date: 2001/06/15 15:05:00 EDT
2769 : blume 841 Tag: blume-20010615-moresparc
2770 :     Description:
2771 :    
2772 :     Here is a short late-hour update related to Sparc c-calls:
2773 :    
2774 :     -- made handling of double-word arguments a bit smarter
2775 :    
2776 :     -- instruction selection phase tries to collapse certain clumsily
2777 :     constructed ML-Trees; typical example:
2778 :    
2779 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2780 :    
2781 :     This currently has no further impact on SML/NJ since mlriscGen does
2782 :     not seem to generate such patterns in the first place, and c-calls
2783 :     (which did generate them in the beginning) has meanwhile been fixed
2784 :     so as to avoid them as well.
2785 :    
2786 :     ----------------------------------------------------------------------
2787 :     Name: Matthias Blume
2788 :     Date: 2001/06/15 15:05:00 EDT
2789 : blume 840 Tag: blume-20010615-sparc
2790 :     Description:
2791 :    
2792 :     The purpose of this update is to provide an implementation of NLFFI
2793 :     on Sparc machines.
2794 :    
2795 :     Here are the changes in detail:
2796 :    
2797 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2798 :     the Sparc implementation of the c-calls API.
2799 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2800 :     for accessing the ML frame. Thus, we have a real frame pointer and
2801 :     can freely modify %sp without need for an omit-frame-ptr phase.
2802 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2803 :     this case.
2804 :     * ml-nlffigen has been taught to produce code for different architectures
2805 :     and calling conventions.
2806 :     * In a way similar to what was done in the x86 case, the Sparc
2807 :     backend uses its own specific extension to mltree. (For example,
2808 :     it needs to be able to generate UNIMP instructions which are part
2809 :     of the calling convention.)
2810 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2811 :     to make it easier to plug in new machine- and os-dependent parts).
2812 :    
2813 :     There are some other fairly unrelated bug fixes and cleanups as well:
2814 :    
2815 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2816 :     that they properly share their libraries with existing SML/NJ libraries.
2817 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2818 :     follow-up error messages.
2819 :     * Updates to CM/CMB documentation.
2820 :    
2821 :     TODO items:
2822 :    
2823 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2824 :     (The current %o2 is a really bad choice because it is part of the
2825 :     calling conventions, so things might interfere in unexpected ways.)
2826 :    
2827 :     ----------------------------------------------------------------------
2828 :     Name: Matthias Blume
2829 : blume 839 Date: 2001/06/07
2830 :     Tag: blume-20010607-calls
2831 :     Description:
2832 :    
2833 :     A number of internal changes related to C calls and calling conventions:
2834 :    
2835 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2836 :     number of bytes popped implicitly (by the callee). In most cases
2837 :     this field is 0 but on x86/win32 it is some non-zero value. This
2838 :     is information provided for the benefit of the "omit-frameptr" pass.
2839 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2840 :     The instruction selection phase copies its value from the ML-Tree
2841 :     CALL statement.
2842 :     3. On all other architectures, the instruction selection phase checks
2843 :     whether "pops=0" and complains if not.
2844 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2845 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2846 :     up after the call and pops is set to 0. For "stdcall", the caller
2847 :     does nothing, leaving the cleanup to the callee; pops is set to
2848 :     the number of bytes that were pushed onto the stack.
2849 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2850 :     distinguish between "ccall" and "stdcall".
2851 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2852 :     set. (This is needed for implementing the official C calling convention
2853 :     on this architecture.)
2854 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2855 :     work with the latest CM.
2856 :    
2857 :     ----------------------------------------------------------------------
2858 :     Name: Matthias Blume
2859 : blume 838 Date: 2001/06/05 15:10:00 EDT
2860 :     Tag: blume-20010605-cm-index
2861 :     Description:
2862 :    
2863 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2864 :    
2865 :     1. CM can now generate "index files". These are human-readable files
2866 :     that list on a per-.cm-file basis each toplevel symbol defined or
2867 :     imported. The location of the index file for
2868 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2869 :     To enable index-file generation, set CM.Control.generate_index to true
2870 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2871 :    
2872 :     The CM manual has been updated accordingly.
2873 :    
2874 :     2. I made some slight modifications to the c-calls API in MLRISC.
2875 :    
2876 :     a) There is now a callback to support saving/restoring of
2877 :     dedicated but caller-save registers around the actual call
2878 :     instruction.
2879 :     b) One can optionally specify a comment-annotation for the
2880 :     call instruction.
2881 :    
2882 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2883 :     (For example, the comment annotation shows the C prototype of
2884 :     the function being called.)
2885 :    
2886 :     ----------------------------------------------------------------------
2887 :     Name: Matthias Blume
2888 : blume 837 Date: 2001/06/01 13:30:00 EDT
2889 :     Tag: blume-20010601-nlffi-cleanup
2890 :     Description:
2891 :    
2892 :     This is mostly a cleanup of MLFFI stuff:
2893 :    
2894 :     - some signature files have been put into a more exposed place
2895 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2896 :     - ml-nlffigen changed accordingly
2897 :     - tutorial updated
2898 :    
2899 :     Other changes:
2900 :    
2901 :     - author's affiliation in CM manual(s) updated
2902 :     - some more recognized keywords added to Allen's sml.sty
2903 :    
2904 :     ----------------------------------------------------------------------
2905 :     Name: Matthias Blume
2906 : blume 836 Date: 2001/05/25 15:30:00 EDT
2907 :     Tag: blume-20010525-iptr
2908 :     Description:
2909 :    
2910 :     - put the official 110.33-README (as it appears on the ftp server) under
2911 :     CVS
2912 :     - fixed a small bug related to incomplete pointer types in
2913 :     ml-nlffigen
2914 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2915 :     (it does not need the 'f type parameter)
2916 :    
2917 :     ----------------------------------------------------------------------
2918 :     Name: Matthias Blume
2919 : blume 834 Date: 2001/05/23 14:30:00 EDT
2920 :     Tag: Release_110_33
2921 :     Description:
2922 :    
2923 :     New version number (110.33). New bootfiles.
2924 :    
2925 :     ----------------------------------------------------------------------
2926 :     Name: Matthias Blume
2927 : blume 833 Date: 2001/05/22 18:06:00 EDT
2928 :     Tag: blume-20010522-targets
2929 :     Description:
2930 :    
2931 :     Made install.sh use file config/targets.customized if it exists, falling
2932 :     back to config/targets if it doesn't. This way one can have a customized
2933 :     version of the targets file without touching the "real thing", thus
2934 :     eliminating the constant fear of accidentally checking something bogus
2935 :     back into the CVS repository... (File config/targets.customized must
2936 :     not be added to the repository!)
2937 :    
2938 :     ----------------------------------------------------------------------
2939 :     Name: Matthias Blume
2940 : blume 832 Date: 2001/05/22 16:30:00 EDT
2941 :     Tag: blume-20010522-minitut
2942 :     Description:
2943 :    
2944 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2945 :     struct returns.
2946 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2947 :     incomplete, preliminary documentation for NLFFI.
2948 :    
2949 :     ----------------------------------------------------------------------
2950 :     Name: Matthias Blume
2951 : blume 829 Date: 2001/05/14 11:30:00 EDT
2952 :     Tag: blume-20010514-script
2953 :     Description:
2954 :    
2955 :     Some bugs in install script fixed.
2956 :    
2957 :     In addition to that I also made a slight change to the NLFFI API:
2958 :     Functors generated by ml-nlffigen now take the dynamic library as a
2959 :     straight functor argument, not as a suspended one. (The original
2960 :     functor code used to force the suspension right away anyway, so there
2961 :     was nothing gained by this complication of the interface.)
2962 :    
2963 :     ----------------------------------------------------------------------
2964 :     Name: Matthias Blume
2965 : blume 828 Date: 2001/05/11 14:35:00 EDT
2966 :     Tag: blume-20010511-ml-nlffi
2967 :     Description:
2968 :    
2969 :     I finally took the plunge and added my new FFI code to the main
2970 :     repository. For x86-linux it is now ready for prime-time.
2971 :    
2972 :     There are two new subdirectories of "src":
2973 :    
2974 :     - ml-nlffi-lib:
2975 :     The utility library for programs using the FFI interface.
2976 :     Here is the implementation of $/c.cm and its associated low-level
2977 :     partners $/c-int.cm and $/memory.cm.
2978 :     - ml-nlffigen:
2979 :     A stand-alone program for generating ML glue code from C source
2980 :     code.
2981 :    
2982 :     Building ml-nlffigen requires $/ckit-lib.cm.
2983 :    
2984 :     The config/install.sh script has been updates to do the Right Thing
2985 :     (hopefully).
2986 :    
2987 :     Notice that the source tree for the C-Kit will not be put under "src"
2988 :     but directly under the installation root directory. (This is the
2989 :     structure that currently exists on the CVS server when you check out
2990 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
2991 :    
2992 :     Bugs: No documentation yet.
2993 :    
2994 :     ----------------------------------------------------------------------
2995 :     Name: Matthias Blume
2996 : blume 826 Date: 2001/05/09 16:35:00 EDT
2997 :     Tag: blume-20010509-cpscontract
2998 :     Description:
2999 :    
3000 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
3001 :     wrapper/unwrapper elimination did not decrement usage counts and some
3002 :     dead variables got overlooked by the dead-up logic.)
3003 :    
3004 :     ----------------------------------------------------------------------
3005 : george 824 Name: Lal George
3006 :     Date: 2001/05/08 17:26:09 EDT
3007 :     Tag: george-20010508-omit-frameptr
3008 :     Description:
3009 :    
3010 :     Changes to implement the omit-frame-pointer optimization to support
3011 :     raw C calls. For now, there is only support on the Intel x86, but
3012 :     other architectures will follow as more experience is gained with this.
3013 :    
3014 :    
3015 :     ----------------------------------------------------------------------
3016 : blume 818 Name: Matthias Blume
3017 : blume 819 Date: 2001/05/07 14:40:00 EDT
3018 :     Tag: blume-20010507-proxies
3019 :     Description:
3020 :    
3021 :     I made into "proxy libraries" all libraries that qualify for such a
3022 :     change. (A qualifying library is a library that has another library or
3023 :     groups as its sole member and repeats that member's export list
3024 :     verbatim. A proxy library avoids this repetition by omitting its export
3025 :     list, effectively inheriting the list that its (only) member exports.
3026 :     See the CM manual for more explanation.)
3027 :     The main effect is that explicit export lists for these libraries
3028 :     do not have to be kepts in sync, making maintenance a bit easier.
3029 :    
3030 :     I also added copyright notices to many .cm-files.
3031 :    
3032 :     Last but not least, I made a new set of bootfiles.
3033 :    
3034 :     ----------------------------------------------------------------------
3035 :     Name: Matthias Blume
3036 : blume 818 Date: 2001/05/04 17:00:00 EDT
3037 :     Tag: blume-20010504-cm-lsplit
3038 :     Description:
3039 :    
3040 :     0. John merged pending changes to $/smlnj-lib.cm
3041 :    
3042 :     1. Allen's previous change accidentally backed out of one of Lal's
3043 :     earlier changes. I undid this mistake (re-introducing Lal's change).
3044 :    
3045 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3046 :     within the compiler where applicable. There is some code simplification
3047 :     because of that.
3048 :    
3049 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
3050 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
3051 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
3052 :     it can now also be controlled on a per-source basis: CM has been taught
3053 :     a new tool parameter applicable to ML source files.
3054 :    
3055 :     - To turn lambda-splitting off completely:
3056 :     local open Compiler.Control.LambdaSplitting in
3057 :     val _ = set Off
3058 :     end
3059 :     - To make "no lambda-splitting" the global default (but allow per-source
3060 :     overriding); this is the initial setting:
3061 :     local open Compiler.Control.LambdaSplitting in
3062 :     val _ = set (Default NONE)
3063 :     end
3064 :     - To make "lambda-splitting with aggressiveness a" the global default
3065 :     (and allow per-source overriding):
3066 :     local open Compiler.Control.LambdaSplitting in
3067 :     val _ = set (Default (SOME a))
3068 :     end
3069 :    
3070 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
3071 :     write (in the respective .cm-file):
3072 :     a.sml (lambdasplitting:off)
3073 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
3074 :     a.sml (lambdasplitting:on)
3075 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3076 :     <a> is a decimal non-negative integer):
3077 :     a.sml (lambdasplitting:<a>)
3078 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
3079 :     a.sml (lambdasplitting:infinity)
3080 :     - To use the global default for a.sml:
3081 :     a.sml (lambdasplitting:default)
3082 :     or simply
3083 :     a.sml
3084 :    
3085 :     ----------------------------------------------------------------------
3086 : leunga 815 Name: Allen Leung
3087 :     Date: 2001/05/04 01:57:00 EDT
3088 :     Tag: leunga-20010504-sync
3089 :     Description:
3090 : leunga 788
3091 : leunga 815 MLRISC features.
3092 :    
3093 :     1. Fix to CMPXCHG instructions.
3094 :     2. Changed RA interface to allow annotations in callbacks.
3095 :     3. Added a new method to the stream interface to allow annotations updates.
3096 :    
3097 : george 749 ----------------------------------------------------------------------
3098 : blume 812 Name: Matthias Blume
3099 : blume 814 Date: 2001/05/01 11:45:00 EDT
3100 :     Tag: blume-20010501-pcedittmp
3101 :     Description:
3102 :    
3103 :     Changed install.sh to use the current working directory instead of
3104 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
3105 :     of /usr/tmp caused trouble with MacOS X because of file premission
3106 :     problems.
3107 :    
3108 :     ----------------------------------------------------------------------
3109 :     Name: Matthias Blume
3110 : blume 812 Date: 2001/04/20 11:10:00 EDT
3111 :     Tag: blume-20010420-inMLflag
3112 :     Description:
3113 :    
3114 :     - added vp_limitPtrMask to vproc-state.h
3115 :     (for use by the raw-C-calls mechanism to implement proper interrupt
3116 :     handling)
3117 :     - made the ML compiler aware of various data-structure offsets so it
3118 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
3119 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
3120 :     raw C-calls
3121 :    
3122 :     ----------------------------------------------------------------------
3123 : george 810 Name: Lal George
3124 :     Date: 2001/04/20 09:15:28 EDT
3125 :     Tag: george-20010420-macosX
3126 :     Description:
3127 :    
3128 :     - Changes to port to Mac OS X; Darwin.
3129 :    
3130 :     - In the process I found that sqrt was broken on the PPC, because the
3131 :     fsqrt instruction is not implemented.
3132 :    
3133 :     ----------------------------------------------------------------------
3134 : blume 799 Name: Matthias Blume
3135 : blume 808 Date: 2001/04/18 12:45:00 EDT
3136 :     Tag: blume-20010418-ccalls
3137 :     Description:
3138 :    
3139 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
3140 :     (this bug prevented structure arguments containing pointers from being
3141 :     passed correctly)
3142 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
3143 :     structure arguments are represented as a pointer to the beginning
3144 :     of the structure (instead of having a series of synthesized arguments,
3145 :     one for each structure member)
3146 :    
3147 :     - made makeml script's verbosity level configurable via environment
3148 :     variable (MAKEML_VERBOSITY)
3149 :    
3150 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3151 : blume 808 in rawmem-x86.sml; we are now using the real thing
3152 :    
3153 :     ----------------------------------------------------------------------
3154 :     Name: Matthias Blume
3155 : blume 806 Date: 2001/03/22 16:25:00 EST
3156 :     Tag: blume-20010322-bootfiles
3157 :     Description:
3158 :    
3159 :     Created a new set of bootfiles (for your automatic installation convenience).
3160 :    
3161 :     ----------------------------------------------------------------------
3162 :     Name: Matthias Blume
3163 : blume 805 Date: 2001/03/22 15:10:00 EST
3164 :     Tag: blume-20010322-rawmem-parcm
3165 :     Description:
3166 :    
3167 :     1. All "raw memory access" primitives for the new FFI are implemented now
3168 :     (at least on the x86).
3169 :     2. Some further cleanup of CM's parallel make mechanism.
3170 :    
3171 :     ----------------------------------------------------------------------
3172 :     Name: Matthias Blume
3173 : blume 801 Date: 2001/03/19 17:53:00 EST
3174 :     Tag: blume-20010319-parallel
3175 :     Description:
3176 :    
3177 :     Parallel make (using compile servers) now works again.
3178 :    
3179 :     To this end, CM.stabilize and CMB.make have been modified to work in
3180 :     two passes when compile servers are attached:
3181 :     1. Compile everything, do not perform stabilization; this pass
3182 :     uses compile servers
3183 :     2. Stabilize everything; this pass does not use compile servers
3184 :     If there are no compile servers, the two passes are combined into one
3185 :     (as before). Splitting the passes increases the inherent parallelism
3186 :     in the dependency graph because the entire graph including all
3187 :     libraries is available at the same time. This, in turn, improves
3188 :     server utilization. The downside is that the master process will
3189 :     have to do some extra work after compilation is done (because for
3190 :     technical reasons it must re-read all the binfiles during stabilization).
3191 :    
3192 :     ----------------------------------------------------------------------
3193 :     Name: Matthias Blume
3194 : blume 800 Date: 2001/03/16 12:22:00 EST
3195 :     Tag: blume-20010316-bootfiles
3196 :     Description:
3197 :    
3198 :     Created a new set of bootfiles (for your automatic installation convenience).
3199 :    
3200 :     ----------------------------------------------------------------------
3201 :     Name: Matthias Blume
3202 : blume 799 Date: 2001/03/16 11:00:00 EST
3203 :     Tag: blume-20010316-MLTREE-fixup
3204 :     Description:
3205 :    
3206 :     This is a minor fixup for an (untagged) earlier commit by Allen.
3207 :     (A file was missing).
3208 :    
3209 :     ----------------------------------------------------------------------
3210 : leunga 788 Name: Allen Leung
3211 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
3212 :     Tag: leunga-20010305-cut-support
3213 :    
3214 :     1. New support for alternative control-flow in MLTREE.
3215 :     Currently we support
3216 :    
3217 :     FLOW_TO(CALL ...., [k1,...,kn])
3218 :    
3219 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
3220 :     in Moby
3221 :    
3222 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3223 :    
3224 :     2. Register Allocator
3225 :    
3226 :     Changes in interface [from Fermin, John]
3227 :    
3228 :     3. Alpha 8-bit SLL support [Fermin]
3229 :    
3230 :     4. All architectures
3231 :    
3232 :     A new module (ClusterExpandCopies) for expanding parallel copies.
3233 :    
3234 :     ----------------------------------------------------------------------
3235 :     Name: Allen Leung
3236 : leunga 788 Date: 2001/02/27 23:07:00 EST
3237 :     Tag: leunga-20010227-minor-stuff
3238 :    
3239 :     1. Alpha bug fix for CMOVNE
3240 :     2. Handle mltree COND(..,FCMP ...,...)
3241 :     3. Bug fix in simplifier
3242 :    
3243 :     ----------------------------------------------------------------------
3244 : blume 777 Name: Matthias Blume
3245 : blume 781 Date: 2001/01/30 17:50:00 EST
3246 :     Tag: blume-20010130-sync
3247 :     Description:
3248 :    
3249 :     This is just a minor update to sync my devel branch with the main brach.
3250 :     The only visible change is the addition of some README files.
3251 :    
3252 :     ----------------------------------------------------------------------
3253 :     Name: Matthias Blume
3254 : blume 778 Date: 2001/01/12 23:30:00 JST
3255 :     Tag: blume-20010112-bootfiles
3256 :     Description:
3257 :    
3258 :     Made a new set of bootfiles that goes with the current state of the
3259 :     repository.
3260 :    
3261 :     ----------------------------------------------------------------------
3262 :     Name: Matthias Blume
3263 : blume 777 Date: 2001/01/12 21:20:00 JST
3264 :     Tag: blume-20010112-sync
3265 :     Description:
3266 :    
3267 :     I am just flushing out some minor changes that had accumulated in
3268 :     my private branch in order to sync with the main tree. (This is
3269 :     mainly because I had CVS trouble when trying to merge _into_ my
3270 :     private branch.)
3271 :    
3272 :     Most people should be completely unaffected by this.
3273 :    
3274 :     ----------------------------------------------------------------------
3275 : leunga 775 Name: Allen Leung
3276 :     Date: Thu Jan 11 21:03:00 EST 2001
3277 :     Tag: leunga-20010111-labexp=mltree
3278 :     Description:
3279 :    
3280 :     1. Removed the type LabelExp and replace it by MLTree.
3281 :     2. Rewritten mltree-simplify with the pattern matcher tool.
3282 :     3. There were some bugs in alpha code generator which would break
3283 :     64-bit code generation.
3284 :     4. Redo the tools to generate code with the
3285 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3286 :     are now generated by perl scripts.
3287 :    
3288 :     ----------------------------------------------------------------------
3289 : blume 771 Name: Matthias Blume
3290 : blume 774 Date: 2001/01/10 21:55:00 JST
3291 :     Tag: blume-20010110-rcc
3292 :     Description:
3293 :    
3294 :     The RCC stuff now seems to work (but only on the x86).
3295 :     This required hacking of the c-calls interface (and -implementation) in
3296 :     MLRISC.
3297 :    
3298 :     Normal compiler users should be unaffected.
3299 :    
3300 :     ----------------------------------------------------------------------
3301 :     Name: Matthias Blume
3302 : blume 773 Date: 2001/01/09 01:20:00 JST
3303 :     Tag: blume-20010109-rcc
3304 :     Description:
3305 :    
3306 :     This is a fairly big patch, flushing out a large number of pending
3307 :     changes that I made to my development copy over the last couple of days.
3308 :    
3309 :     Of practical relevance at this moment is a workaround for a pickling
3310 :     bug that Allen ran into the other day. The cause of the bug itself is
3311 :     still unknown and it might be hard to fix it properly, but the
3312 :     workaround has some merits of its own (namely somewhat reducing pickling
3313 :     overhead for certain libraries). Therefore, I think this solution should
3314 :     be satisfactory at this time.
3315 :    
3316 :     The rest of the changes (i.e., the vast majority) has to do with my
3317 :     ongoing efforts of providing direct support for C function calls from
3318 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
3319 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3320 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3321 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3322 :     handle RCC.
3323 :    
3324 :     The last part (the changes to mlriscGen.sml) are still known to be
3325 :     wrong on the x86 and not implemented on all other architectures. But
3326 :     the infrastructure is in place. I had to change a few functor
3327 :     signatures in the backend to be able to route the CCalls interface
3328 :     from MLRISC there, and I had to specialize the mltree type (on the
3329 :     x86) to include the necessary extensions. (The extensions themselves
3330 :     were already there and redy to go in MLRISC/x86).
3331 :    
3332 :     Everything should be very happy as soon as someone helps me with
3333 :     mlriscGen.sml...
3334 :    
3335 :     In any case, nothing of this should matter to anyone as long as the
3336 :     new primop is not being used (which is going to be the case unless you
3337 :     find it where I hid it :). The rest of the compiler is completely
3338 :     unaffected.
3339 :    
3340 :     ----------------------------------------------------------------------
3341 :     Name: Matthias Blume
3342 : blume 772 Date: 2001/01/05 00:30:00 JST
3343 :     Tag: blume-20010105-primops
3344 :     Description:
3345 :    
3346 :     Added some experimental support for work that I am doing right now.
3347 :     These changes mostly concern added primops, but there is also a new
3348 :     experimental C library in the runtime system (but currently not enabled
3349 :     anywhere except on Linux/X86).
3350 :    
3351 :     In the course of adding primops (and playing with them), I discovered that
3352 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3353 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
3354 :     To recover, I made type information in INL_PRIM mandatory and changed
3355 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
3356 :     has complete, correct type information (i.e., no bottom types).
3357 :    
3358 :     Since all these changes mean that we need new binfiles, I also bumped the
3359 :     version number to 110.32.1.
3360 :    
3361 :     ----------------------------------------------------------------------
3362 :     Name: Matthias Blume
3363 : blume 771 Date: 2000/12/30 22:10:00 JST
3364 :     Tag: blume-20001230-various
3365 :     Description:
3366 :    
3367 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
3368 :     to each other using path anchors. (See CM manual for explanation.)
3369 :    
3370 :     Updated CM documentation.
3371 :    
3372 :     Fixed some bugs in CM.
3373 :    
3374 :     Implemented "proxy" libraries (= syntactic sugar for CM).
3375 :    
3376 :     Added "-quiet" option to makeml and changed runtime system accordingly.
3377 :    
3378 :     Added cleanup handler for exportML to reset timers and compiler stats.
3379 :    
3380 :     ----------------------------------------------------------------------
3381 : george 761 Name: Lal George
3382 :     Date: 2000/12/22 22:22:58 EST 2000
3383 :     Tag: Release_110_32
3384 :     Description:
3385 :    
3386 :     Infinite precision used throughout MLRISC.
3387 :     see MLRISC/mltree/machine-int.sig
3388 :    
3389 :     ----------------------------------------------------------------------
3390 : blume 756 Name: Matthias Blume
3391 : blume 760 Date: 2000/12/22 23:16:00 JST
3392 :     Tag: blume-20001222-warn
3393 :     Description:
3394 : blume 759
3395 : blume 760 Corrected wording and formatting of some CM warning message which I
3396 :     broke in my previous patch.
3397 : blume 759
3398 :     ----------------------------------------------------------------------
3399 :     Name: Matthias Blume
3400 : blume 763 Date: 2000/12/22 21:20:00 JST
3401 :     Tag: blume-20001222-anchorenv
3402 :     Description:
3403 :    
3404 :     Fixed CM's handling of anchor environments in connection with CMB.make.
3405 :    
3406 :     ----------------------------------------------------------------------
3407 :     Name: Matthias Blume
3408 : blume 758 Date: 2000/12/22 13:15:00 JST
3409 :     Tag: blume-20001222-cleanup
3410 :     Description:
3411 :    
3412 :     Removed src/cm/ffi which does not (and did not) belong here.
3413 :    
3414 :     ----------------------------------------------------------------------
3415 :     Name: Matthias Blume
3416 : blume 757 Date: 2000/12/21 23:55:00 JST
3417 :     Tag: blume-20001221-exn
3418 :     Description:
3419 :    
3420 :     Probably most important: CM no longer silently swallows all exceptions
3421 :     in the compiler.
3422 :     Plus: some other minor CM changes. For example, CM now reports some
3423 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
3424 :    
3425 :     ----------------------------------------------------------------------
3426 :     Name: Matthias Blume
3427 : blume 756 Date: 2000/12/15 00:01:05 JST
3428 :     Tag: blume-20001215-dirtool
3429 :     Description:
3430 :    
3431 :     - "dir" tool added.
3432 :     - improvements and cleanup to Tools structure
3433 :     - documentation updates
3434 :    
3435 :     ----------------------------------------------------------------------
3436 : leunga 752 Name: Allen Leung
3437 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
3438 :     Description:
3439 :     Tag: leunga-20001214-int-inf
3440 :     Description:
3441 :    
3442 :     In IntInf, added these standard functions, which are missing from our
3443 :     implementation:
3444 :    
3445 :     andb : int * int -> int
3446 :     xorb : int * int -> int
3447 :     orb : int * int -> int
3448 :     notb : int -> int
3449 :     << : int * word -> int
3450 :     ~>> : int * word -> int
3451 :    
3452 :     Not tested, I hope they are correct.
3453 :    
3454 :     ----------------------------------------------------------------------
3455 :     Name: Allen Leung
3456 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
3457 :     Description:
3458 :     Tag: leunga-20001208-nowhere
3459 :     Description:
3460 :    
3461 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
3462 :     to generate better error messages etc. Plus a brief manual.
3463 :    
3464 :     ----------------------------------------------------------------------
3465 : george 749 Name: Lal George
3466 :     Date: 2000/12/08 09:54:02 EST 2000
3467 :     Tag: Release_110_31
3468 :     Description:
3469 : george 705
3470 : george 749 - Version 110.31
3471 : leunga 591 ----------------------------------------------------------------------
3472 : leunga 744 Name: Allen Leung
3473 :     Date: Thu Dec 7 22:01:04 EST 2000
3474 :     Tag: leunga-20001207-cell-monster-hack
3475 :     Description:
3476 :    
3477 :     Major MLRISC internal changes. Affect all clients.
3478 :     Summary:
3479 :    
3480 :     1. Type CELLS.cell = int is now replaced by a datatype.
3481 :     As a result, the old regmap is now gone. Almost all interfaces
3482 :     in MLRISC change as a consequence.
3483 :    
3484 :     2. A new brand version of machine description tool (v3.0) that generates
3485 :     modules expecting the new interface. The old version is removed.
3486 :    
3487 :     3. The RA interface has been further abstracted into two new functors.
3488 :     RISC_RA and X86RA. These functors have much simpler interfaces.
3489 :     [See also directory MLRISC/demo.]
3490 :    
3491 :     4. Some other new source->source code generation tools are available:
3492 :    
3493 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
3494 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
3495 :     I use this tool to generate the peephole optimizers---with the new
3496 :     cell type changes, peephole rules are becoming difficult to write
3497 :     without conditional pattern matching.
3498 :    
3499 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
3500 :     cover the entire MLRISC source tree so many things broke.
3501 :    
3502 :     6. CM files have been moved to the subdirectory MLRISC/cm.
3503 :     They are moved because there are a lot of them and they clutter up the
3504 :     root dir.
3505 :    
3506 :     7. More detailed documentation to come...
3507 :    
3508 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
3509 :     a makeml -rebuild first. This is because of other other
3510 :     changes that Matthias has made (see below).
3511 :    
3512 :    
3513 :     ----------------------------------------------------------------------
3514 : blume 733 Name: Matthias Blume
3515 : blume 742 Date: 2000/11/30 23:12:00 JST
3516 :     Tag: blume-20001130-filereorg
3517 :     Description:
3518 :    
3519 :     Some manual updates and some file reorganizations in CM.
3520 :    
3521 :     ----------------------------------------------------------------------
3522 :     Name: Matthias Blume
3523 : blume 737 Date: 2000/11/24 17:45:00 JST
3524 :     Tag: blume-20001124-link
3525 :     Description:
3526 :    
3527 :     Drastically improved link traversal code for the case that the dynamic
3528 :     value was already loaded at bootstrap time. As a result, CM and CMB
3529 :     now both load blazingly fast -- even on a very slow machine. Also,
3530 :     memory consumption has been further reduced by this.
3531 :    
3532 :     Warning: The format of the PIDMAP file has changed. THerefore, to
3533 :     bootstrap you have to do this:
3534 :    
3535 :     1. Run CMB.make
3536 :     2. Make a symbolic link for the boot directory:
3537 :     ln -s sml.boot.ARCH-OS xxx
3538 :     3. "Rebuild" the boot directory:
3539 :     ./makeml -boot xxx -rebuild sml ; rm xxx
3540 :     4. Boot normally:
3541 :     ./makeml
3542 :    
3543 :     ----------------------------------------------------------------------
3544 :     Name: Matthias Blume
3545 : blume 735 Date: 2000/11/21 21:20:00 JST
3546 :     Tag: blume-20001121-tools
3547 :     Description:
3548 :    
3549 :     Continued hacking on autoloading problem -- with success this time.
3550 :     Also changed tool-plugin mechanism. See new CM manual.
3551 :    
3552 :     ----------------------------------------------------------------------
3553 :     Name: Matthias Blume
3554 : blume 734 Date: 2000/11/19 14:30:00 JST
3555 :     Tag: blume-20001119-autoload
3556 :     Description:
3557 :    
3558 :     Some hacking to make autoloading faster. Success for CMB, no success
3559 :     so far for CM. There is a reduced structure CM' that autoloads faster.
3560 :     (This is a temporary, non-documented hack to be eliminated again when
3561 :     the general problem is solved.)
3562 :    
3563 :     ----------------------------------------------------------------------
3564 :     Name: Matthias Blume
3565 : blume 733 Date: 2000/11/17 14:10:00 JST
3566 :     Tag: blume-20001117-pickle-lib
3567 :     Description:
3568 :    
3569 :     1. Eliminated comp-lib.cm
3570 :     2. Made pickle-lib.cm
3571 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
3572 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
3573 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
3574 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
3575 :     Duplicated intmap modules thrown out.
3576 :     6. Hunted down all duplicated SCC code and replaced it with
3577 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
3578 :     7. Rewrote Feedback module.
3579 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
3580 :     should be thrown out and equivalent modules from smlnj-lib.cm
3581 :     should be used (IntRedBlackSet, IntListSet, ...).
3582 :    
3583 :     Confirmed that compiler compiles to fixpoint.
3584 :    
3585 :     ----------------------------------------------------------------------
3586 : leunga 731 Name: Allen Leung
3587 :     Date: 2000/11/10 18:00:00
3588 :     Tag: leunga-20001110-new-x86-fp
3589 :    
3590 :     A new x86 floating point code generator has been added.
3591 :     By default this is turned off. To turn this on, do:
3592 :    
3593 :     CM.autoload "$smlnj/compiler.cm";
3594 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3595 :    
3596 :     Changes:
3597 :    
3598 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
3599 :     2. Changed the extension callback for FTANGENT to generate:
3600 :    
3601 :     fptan
3602 :     fstp %st(0)
3603 :     instead of
3604 :     fptan
3605 :     fstpl ftempmem
3606 :    
3607 :     3. Numerous assembly fixes for x86.
3608 :    
3609 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
3610 :     support for a whole bunch of instructions and addressing modes:
3611 :    
3612 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
3613 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
3614 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
3615 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3616 :     fxch %st(n)
3617 :     fld %st(n)
3618 :     fst %st(n)
3619 :     fst mem
3620 :     fstp %st(n)
3621 :     fucom %st(n)
3622 :     fucomp %st(n)
3623 :    
3624 :     All these are now generated when the fast fp mode is turned on.
3625 :    
3626 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3627 :    
3628 :     ----------------------------------------------------------------------
3629 : blume 729 Name: Matthias Blume
3630 :     Date: 2000/11/09 11:20:00 JST
3631 :     Tag: blume-20001109-scc
3632 :     Description:
3633 :    
3634 :     Eliminated some code duplication:
3635 :    
3636 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
3637 :     (Otherwise the functor is useless.)
3638 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
3639 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
3640 :    
3641 :     ----------------------------------------------------------------------
3642 : george 721 Name: Lal George
3643 :     Date: 2000/11/06 09:02:21 EST 2000
3644 :     Tag: Release_110_30
3645 :     Description:
3646 :    
3647 :     - Version 110.30
3648 :     ----------------------------------------------------------------------
3649 : blume 715 Name: Matthias Blume
3650 : blume 716 Date: 2000/11/04 14:45:00
3651 :     Tag: blume-20001104-mlbuild
3652 :     Description:
3653 :    
3654 :     - Made ml-build faster on startup.
3655 :     - Documentation fixes.
3656 :    
3657 :     ----------------------------------------------------------------------
3658 :     Name: Matthias Blume
3659 : blume 715 Date: 2000/11/02 17:00:00 JST
3660 :     Tag: blume-20001102-condcomp
3661 :     Description:
3662 :    
3663 :     - Small tweaks to pickler -- new BOOTFILES!
3664 :     - Version bumped to 110.29.2.
3665 :     - Added conditional compilation facility to init.cmi (see comment there).
3666 :     ----------------------------------------------------------------------
3667 : leunga 713 Name: Allen Leung
3668 :     Date: 2000/10/23 19:31:00
3669 :     Tag: leunga-20001023-demo-ra
3670 :    
3671 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
3672 :     2. Test programs for the graph library updated
3673 :     3. Some new MLRISC demo programs added
3674 :    
3675 :     ----------------------------------------------------------------------
3676 : blume 710 Name: Matthias Blume
3677 : blume 711 Date: 2000/08/31 22:15:00 JST
3678 :     Tag: blume-20001017-errmsg
3679 :     Description:
3680 :    
3681 :     More error message grief: Where there used to be no messages, there
3682 :     now were some that had bogus error regions. Fixed.
3683 :    
3684 :     ----------------------------------------------------------------------
3685 :     Name: Matthias Blume
3686 : blume 710 Date: 2000/08/31 17:30:00 JST
3687 :     Tag: blume-20001017-v110p29p1
3688 :     Description:
3689 :    
3690 :     I made a version 110.29.1 with new bootfiles.
3691 :    
3692 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
3693 :     CM documentation changes and a small bugfix in CM's error reporting.
3694 :    
3695 :     ----------------------------------------------------------------------
3696 : george 705 Name: Lal George
3697 :     Date: 2000/09/27 14:42:35 EDT
3698 :     Tag: george-20000927-nodestatus
3699 :     Description:
3700 :    
3701 :     Changed the type of the nodestatus, so that:
3702 :    
3703 :     SPILLED(~1) is now SPILLED
3704 :     SPILLED(m) where m>=0 is now MEMREG(m)
3705 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
3706 :    
3707 :     ----------------------------------------------------------------------
3708 : blume 697 Name: Matthias Blume
3709 : blume 703 Date: 2000/09/07 14:45:00 JST
3710 :     Tag: blume-20000907-cmerrmsg
3711 :     Description:
3712 :    
3713 :     Small tweak to CM to avoid getting ML syntax error messages twice.
3714 :    
3715 :     ----------------------------------------------------------------------
3716 :     Name: Matthias Blume
3717 : blume 701 Date: 2000/08/31 18:00:00 JST
3718 :     Tag: blume-20000831-cvsbootfiles
3719 :     Description:
3720 :    
3721 :     New URL for boot files (because the 110.29 files on the BL server do
3722 :     now work correctly with my updated install scripts for yacc and lex).
3723 :    
3724 :     ----------------------------------------------------------------------
3725 :     Name: Matthias Blume
3726 : blume 697 Date: 2000/08/08 12:33:00 JST
3727 :     Tag: blume-20000808-manual
3728 :     Description:
3729 :    
3730 :     Tiny update to CM manual.
3731 :    
3732 :     ----------------------------------------------------------------------
3733 : leunga 695 Name: Allen Leung
3734 :     Date: 2000/08/7 19:31:00
3735 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3736 :    
3737 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3738 :     and Mobdy.
3739 :    
3740 :     1. x86
3741 :    
3742 :     a. Fixes to peephole module by John and Dan.
3743 :     b. Assembly fix to SETcc by Allen.
3744 :     c. Fix to c-call by John.
3745 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3746 :     e. Instruction selection optimization to SETcc as suggested by John.
3747 :    
3748 :     For example,
3749 :    
3750 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3751 :    
3752 :     should generate:
3753 :    
3754 :     MOVL a, x
3755 :     SUBL b, x
3756 :     SHRL 31, x
3757 :    
3758 :     2. IR stuff
3759 :    
3760 :     A bunch of new DJ-graph related algorithms added. These
3761 :     speed up SSA construction.
3762 :    
3763 :     3. SSA + Scheduling
3764 :    
3765 :     Added code for SSA and scheduling to the repository
3766 :    
3767 :     ----------------------------------------------------------------------
3768 : george 694 Name: Lal George
3769 :     Date: 2000/07/27 11:53:14 EDT
3770 :    
3771 :     Tag: lal-20000727-linux-ppc
3772 :     Description:
3773 :    
3774 :     Made changes to support Linux PPC.
3775 :     p.s. I have confirmation that the 110.29 boot files work fine.
3776 :    
3777 :     ----------------------------------------------------------------------
3778 : blume 684 Name: Matthias Blume
3779 : blume 692 Date: 2000/07/27 17:40:00 JST
3780 :     Tag: blume-20000727-scripts
3781 :     Description:
3782 :    
3783 :     !!!! WARNING !!!!
3784 :     You must recompile the runtime system!
3785 :     !!!! WARNING !!!!
3786 :    
3787 :     This is basically another round of script-enhancements:
3788 :    
3789 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3790 :     and undefine CM preprocessor symbols.
3791 :    
3792 :     2. ml-build avoids generating a new heap image if it finds that the
3793 :     existing one is still ok. (The condition is that no ML file had to
3794 :     be recompiled and all ML files are found to be older that the heap
3795 :     file.)
3796 :    
3797 :     To make this work smoothly, I also hacked the runtime system as
3798 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3799 :     (.sparc-solaris, ...) that is currently being used.
3800 :    
3801 :     Moreover, the signature of CM.mk_standalone has changed. See the
3802 :     CM manual.
3803 :    
3804 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3805 :     CM manual for details.)
3806 :    
3807 :     4. More CM manual updates:
3808 :     - all of the above has been documented.
3809 :     - there is now a section describing the (CM-related) command line
3810 :     arguments that are accepted by the "sml" command
3811 :    
3812 :     ----------------------------------------------------------------------
3813 :     Name: Matthias Blume
3814 : blume 691 Date: 2000/07/25 16:20:00 JST
3815 :     Tag: blume-20000725-makedepend
3816 :     Description:
3817 :    
3818 :     Added a script called ml-makedepend. This can be used in makefiles
3819 :     for Unix' make in a way very similar to the "makedepend" command for
3820 :     C.
3821 :    
3822 :     The script internally uses function CM.sources.
3823 :    
3824 :     Synopsis:
3825 :    
3826 :     ml-makedepend [-f makefile] cmfile targetname
3827 :    
3828 :     The default for the makefile is "makefile" (or "Makefile" should
3829 :     "makefile" not exist).
3830 :    
3831 :     ml-makedepend adds a cmfile/targetname-specific section to this
3832 :     makefile (after removing the previous version of this section). The
3833 :     section contains a single dependency specification with targetname on
3834 :     the LHS (targetname is an arbitrary name), and a list of files derived
3835 :     from the cmfile on the RHS. Some of the files on the RHS are
3836 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3837 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3838 :     path names. The makefile writer is responsible for making sure that
3839 :     these variables have correct at the time "make" is invoked.
3840 :    
3841 :     ----------------------------------------------------------------------
3842 :     Name: Matthias Blume
3843 : blume 690 Date: 2000/07/22 23:30:00 JST
3844 :     Tag: blume-20000722-urlupdate
3845 :     Description:
3846 :    
3847 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3848 :    
3849 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3850 :    
3851 :     ----------------------------------------------------------------------
3852 :     Name: Matthias Blume
3853 : blume 686 Date: 2000/07/18 18:00:00 JST
3854 :     Tag: blume-20000718-Version_110_29
3855 :     Description:
3856 :    
3857 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3858 :    
3859 :     2. Updated config/version to 110.29
3860 :    
3861 :     3. Updated config/srcarchiveurl
3862 :    
3863 :     3. New boot files!
3864 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3865 :    
3866 :     ----------------------------------------------------------------------
3867 :     Name: Matthias Blume
3868 : blume 684 Date: 2000/07/11 13:58:00 JST
3869 :     Tag: blume-20000711-doctypo
3870 :     Description:
3871 :    
3872 :     Fixed a few typos in CM manual.
3873 :    
3874 :     ----------------------------------------------------------------------
3875 : leunga 683 Name: Allen Leung
3876 :     Date: 2000/06/15 00:38:00
3877 :     Tag: leunga-20000704-sparc-x86
3878 :    
3879 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3880 :     2. fix to x86 RET bug [found by Dan Grossman]
3881 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3882 :    
3883 :     Affects c-- and moby only
3884 :    
3885 :     ----------------------------------------------------------------------
3886 : blume 680 Name: Matthias Blume
3887 : blume 682 Date: 2000/07/04 15:26:00
3888 :     Tag: blume-20000704-trigger
3889 :     Description:
3890 :    
3891 :     1. Improvements to CM manual.
3892 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3893 :     of getting a back-trace. The function, when called, raises an
3894 :     internal exception which explicitly carries the full back-trace history,
3895 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3896 :     or not). The interactive loop will print that history once it arrives
3897 :     at top level.
3898 :     Short of having all exceptions implicitly carry the full history, the
3899 :     recommended way of using this facility is:
3900 :     - compile your program with instrumentation "on"
3901 :     - run it, when it raises an exception, look at the history
3902 :     - if the history is "cut off" because of some handler, go and modify
3903 :     your program so that it explicitly calls BTrace.trigger
3904 :     - recompile (still instrumented), and rerun; look at the full history
3905 :    
3906 :     ----------------------------------------------------------------------
3907 :     Name: Matthias Blume
3908 :     Date: 2000/07/03 15:36:00 JST
3909 : blume 680 Tag: blume-20000702-manual
3910 :     Description:
3911 :    
3912 :     Small corrections and updates to CM manual.
3913 :    
3914 :     ----------------------------------------------------------------------
3915 : blume 682 Name: Matthias Blume
3916 : blume 679 Date: 2000/06/29 16:04:00 JST
3917 :     Tag: blume-20000629-yacctool
3918 :     Description:
3919 :    
3920 :     Changes:
3921 :    
3922 :     1. Class "mlyacc" now takes separate arguments to pass options to
3923 :     generated .sml- and .sig-files independently.
3924 :     2. Corresponding CM manual updates.
3925 :     3. BTrace module now also reports call sites. (However, for loop clusters
3926 :     it only shows from where the cluster was entered.) There are associated
3927 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3928 :    
3929 :     ----------------------------------------------------------------------
3930 : blume 675 Name: Matthias Blume
3931 : blume 678 Date: 2000/06/27 16:51:00 JST
3932 :     Tag: blume-20000627-noweb
3933 :     Description:
3934 :    
3935 :     Changes:
3936 :    
3937 :     1. Implemented "subdir" and "witness" options for noweb tool.
3938 :     This caused some slight internal changes in CM's tool implementation.
3939 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3940 :     some remaining issues from earlier path anchor changes.
3941 :     3. Updated CM manual accordingly.
3942 :    
3943 :     4. Changed implementation of back-tracing so that I now consider it
3944 :     ready for prime-time.
3945 :    
3946 :     In particular, you don't have to explicitly trigger the back-trace
3947 :     anymore. Instead, if you are running BTrace-instrumented code and
3948 :     there is an uncaught exception (regardless of whether or not it was
3949 :     raised in instrumented code), the top-level evalloop will print
3950 :     the back-trace.
3951 :    
3952 :     Features:
3953 :    
3954 :     - Instrumented and uninstrumented code work together seemlessly.
3955 :     (Of course, uninstrumented code is never mentioned in actual
3956 :     back-traces.)
3957 :    
3958 :     - Asymptotic time- and space-complexity of instrumented code is
3959 :     equal to that of uninstrumented code. (This means that
3960 :     tail-recursion is preserved by the instrumentation phase.)
3961 :    
3962 :     - Modules whose code has been instrumented in different sessions
3963 :     work together without problem.
3964 :    
3965 :     - There is no penalty whatsoever on uninstrumented code.
3966 :    
3967 :     - There is no penalty on "raise" expressions, even in
3968 :     instrumented code.
3969 :    
3970 :     A potential bug (or perhaps it is a feature, too):
3971 :    
3972 :     A back-trace reaches no further than the outermost instrumented
3973 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3974 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3975 :    
3976 :     <exp> handle e => raise e
3977 :    
3978 :     and even
3979 :    
3980 :     <exp> handle Foo => raise Bar
3981 :    
3982 :     and, of course, through
3983 :    
3984 :     <exp> handle Foo => ...
3985 :    
3986 :     if the exception was not Foo.
3987 :    
3988 :     Back-traces always reach right through any un-instrumented code
3989 :     including any of its "handle" expressions, trivial or not.
3990 :    
3991 :     To try this out, do the following:
3992 :    
3993 :     - Erase all existing binfiles for your program.
3994 :     (You may keep binfiles for those modules where you think you
3995 :     definitely don't need back-tracing.)
3996 :     - Turn on back-trace instrumentation:
3997 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3998 :     - Recompile your program. (I.e., run "CM.make" or "use".)
3999 :     - You may now turn instrumentation off again (if you want):
4000 :     SMLofNJ.Internals.BTrace.mode (SOME false);
4001 :     - Run your program as usual. If it raises an exception that
4002 :     reaches the interactive toplevel, then a back-trace will
4003 :     automatically be printed. After that, the toplevel loop
4004 :     will print the exception history as usual.
4005 :    
4006 :     ----------------------------------------------------------------------
4007 :     Name: Matthias Blume
4008 : blume 677 Date: 2000/06/26 09:56:46 JST
4009 :     Tag: blume-20000626-setup
4010 :     Description:
4011 :    
4012 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
4013 :     ML code before and after compiling a file (e.g., to set compiler
4014 :     flags)
4015 :    
4016 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
4017 :     - associated changes to btrace.sml (BTrace instrumentation pass)
4018 :     - cleaner implementation of btimp.sml (BTrace tracing and report
4019 :     module)
4020 :    
4021 :     CM manual: * new path encoding documented
4022 :     * description of setup-parameter to "sml" added
4023 :    
4024 :     The biggest user-visible change to back-tracing is that it is no
4025 :     longer necessary to compile all traced modules within the same
4026 :     session. (This was a real limitation.)
4027 :    
4028 :     ----------------------------------------------------------------------
4029 :     Name: Matthias Blume
4030 : blume 676 Date: 2000/06/24 12:40:00 JST
4031 :     Tag: blume-20000624-startup
4032 :     Description:
4033 :    
4034 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
4035 :     bit too often -- to put it mildly. :)
4036 :    
4037 :     ----------------------------------------------------------------------
4038 :     Name: Matthias Blume
4039 : blume 675 Date: 2000/06/23 18:20:00 JST
4040 :     Tag: blume-20000623-btrace
4041 :     Description:
4042 :    
4043 :     This updates adds a backtrace facility to aid programmers in debugging
4044 :     their programs. This involves the following changes:
4045 :    
4046 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
4047 :     keeping track of the current call stack. When programs are compiled
4048 :     in a special mode, the compiler will insert calls to these hooks
4049 :     into the user program.
4050 :     "Hook" means that it is possible for different implementations of
4051 :     back-tracing to register themselves (at different times).
4052 :    
4053 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
4054 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
4055 :     It can be turned on using this call:
4056 :     SMLofNJ.Internals.BTrace.mode (SOME true);
4057 :     Turning it off again:
4058 :     SMLofNJ.Internals.BTrace.mode (SOME false);
4059 :     Querying the current status:
4060 :     SMLofNJ.Internals.BTrace.mode NONE;
4061 :     Annotated programs are about twice as big as normal ones, and they
4062 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
4063 :     where all hooks do nothing). The slowdown with a plugin that is
4064 :     actually useful (such as the one supplied by default) is even greater,
4065 :     but in the case of the default plugin it is still only an constant
4066 :     factor (amortized).
4067 :    
4068 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
4069 :     with a sub-structure BTrace for controlling back-tracing. In particular,
4070 :     the above-mentioned function "mode" controls whether the annotation
4071 :     phase is invoked by the compiler. Another important function is
4072 :     "trigger": when called it aborts the current execution and causes
4073 :     the top-level loop to print a full back-trace.
4074 :    
4075 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
4076 :     for back-tracing. It keeps track of the dynamic call stack and in
4077 :     addition to that it keeps a partial history at each "level" of that
4078 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
4079 :     d and b (at separate times, dynamically), then the report will show:
4080 :    
4081 :     GOTO d
4082 :     /c
4083 :     GOTO \b
4084 :     CALL a
4085 :    
4086 :     This shows that there was an initial non-tail call of a, then a
4087 :     tail-call to b or c, looping behavior in a cluster of functions that
4088 :     consist of b and c, and then a goto from that cluster (i.e., either from
4089 :     b or from c) to d.
4090 :    
4091 :     Note that (depending on the user program) the amount of information
4092 :     that the back-trace module has to keep track of at each level is bounded
4093 :     by a constant. Thus, the whole implementation has the same asymptotical
4094 :     complexity as the original program (both in space and in time).
4095 :    
4096 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
4097 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
4098 :     which is raised by the "trigger" function mentioned above.
4099 :    
4100 :     Notes on usage:
4101 :    
4102 :     - Annotated code works well together with unannotated code:
4103 :     Unannotated calls simply do not show up at all in the backtrace.
4104 :    
4105 :     - It is not a good idea to let modules that were annotated during
4106 :     different sessions run at the same time. This is because the compiler
4107 :     chooses small integers to identify individual functions, and there
4108 :     will be clashes if different modules were compiled in separate sessions.
4109 :     (Nothing will crash, and you will even be told about the clashes, but
4110 :     back-trace information will in general not be useful.)
4111 :    
4112 :     - Back-tracing can be confused by callcc and capture.
4113 :    
4114 :     - The only way of getting a back-trace right now is to explicitly
4115 :     invoke the "trigger" function from your user program. Eventually, we
4116 :     should make every exception carry back-trace information (if
4117 :     available). But since this creates more overhead at "raise"-time
4118 :     (similar to the current exnHistory overhead), I have not yet
4119 :     implemented this. (The implementation will be rather easy.) With
4120 :     exceptions carrying back-trace information, this facility will be even
4121 :     more useful because users don't need to modify their programs...
4122 :    
4123 :     - While it is possible to compile the compiler with back-trace
4124 :     annotations turned on (I did it to get some confidence in
4125 :     correctness), you must make absolutely sure that core.sml and
4126 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
4127 :     be compiled with annotation because there is no core access yet, but
4128 :     if you compile btimp.sml with annotation, then the system will go into
4129 :     an infinite recursion and crash.)
4130 :     Since CM currently does not know about BTrace, the only way to turn
4131 :     annotations on and off for different modules of the compiler is to
4132 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
4133 :     this is awkward and clumsy.
4134 :    
4135 :     Sample sessions:
4136 :    
4137 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4138 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4139 :     [autoloading]
4140 :     [autoloading done]
4141 :     val it = false : bool
4142 :     - structure X = struct
4143 :     - fun main n = let
4144 :     - fun a (x, 0) = d x
4145 :     - | a (x, n) = b (x, n - 1)
4146 :     - and b (x, n) = c (x, n)
4147 :     - and c (x, n) = a (x, n)
4148 :     - and d x = e (x, 3)
4149 :     - and e (x, 0) = f x
4150 :     - | e (x, n) = e (x, n - 1)
4151 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
4152 :     - | f n = n * g (n - 1)
4153 :     - and g n = a (n, 3)
4154 :     - in
4155 :     - f n
4156 :     - end
4157 :     - end;
4158 :     structure X : sig val main : int -> int end
4159 :     - X.main 3;
4160 :     *** BACK-TRACE ***
4161 :     GOTO stdIn:4.2-13.20: X.main[2].f
4162 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4163 :     GOTO stdIn:4.2-13.20: X.main[2].d
4164 :     / stdIn:4.2-13.20: X.main[2].a
4165 :     | stdIn:4.2-13.20: X.main[2].b
4166 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4167 :     CALL stdIn:4.2-13.20: X.main[2].g
4168 :     GOTO stdIn:4.2-13.20: X.main[2].f
4169 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4170 :     GOTO stdIn:4.2-13.20: X.main[2].d
4171 :     / stdIn:4.2-13.20: X.main[2].a
4172 :     | stdIn:4.2-13.20: X.main[2].b
4173 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4174 :     CALL stdIn:4.2-13.20: X.main[2].g
4175 :     GOTO stdIn:4.2-13.20: X.main[2].f
4176 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4177 :     GOTO stdIn:4.2-13.20: X.main[2].d
4178 :     / stdIn:4.2-13.20: X.main[2].a
4179 :     | stdIn:4.2-13.20: X.main[2].b
4180 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4181 :     CALL stdIn:4.2-13.20: X.main[2].g
4182 :     GOTO stdIn:4.2-13.20: X.main[2].f
4183 :     CALL stdIn:2.15-17.4: X.main[2]
4184 :     -
4185 :    
4186 :     (Note that because of a FLINt bug the above code currently does not
4187 :     compile without BTrace turned on.)
4188 :    
4189 :     Here is another example, using my modified Tiger compiler:
4190 :    
4191 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4192 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4193 :     [autoloading]
4194 :     [autoloading done]
4195 :     val it = false : bool
4196 :     - CM.make "sources.cm";
4197 :     [autoloading]
4198 :     ...
4199 :     [autoloading done]
4200 :     [scanning sources.cm]
4201 :     [parsing (sources.cm):parse.sml]
4202 :     [creating directory CM/SKEL ...]
4203 :     [parsing (sources.cm):tiger.lex.sml]
4204 :     ...
4205 :     [wrote CM/sparc-unix/semant.sml]
4206 :     [compiling (sources.cm):main.sml]
4207 :     [wrote CM/sparc-unix/main.sml]
4208 :     [New bindings added.]
4209 :     val it = true : bool
4210 :     - Main.compile ("../testcases/merge.tig", "foo.out");
4211 :     *** BACK-TRACE ***
4212 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
4213 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4214 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4215 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4216 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4217 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4218 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
4219 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
4220 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
4221 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4222 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
4223 :     CALL main.sml:1.18-118.4: Main.compile[2]
4224 :     -
4225 :    
4226 :     ----------------------------------------------------------------------
4227 : blume 673 Name: Matthias Blumen
4228 :     Date: 2000/06/21 18:00:00 JST
4229 :     Tag: blume-20000621-manual
4230 :     Description:
4231 :    
4232 :     CM manual update: Path environments documented.
4233 :    
4234 :     ----------------------------------------------------------------------
4235 : blume 666 Name: Matthias Blume
4236 : blume 672 Date: 2000/06/19 13:40:00
4237 :     Tag: blume-20000619-manual
4238 :     Description:
4239 :    
4240 :     CM manual and system/README update. This only covers the fact that
4241 :     there are no more implicit anchors. (Path environments and the "bind"
4242 :     option to "cm" have yet to be documented.)
4243 :    
4244 :     ----------------------------------------------------------------------
4245 :     Name: Matthias Blume
4246 : blume 671 Date: 2000/06/19 11:05:00 JST
4247 :     Tag: blume-20000619-chdir-bugfix
4248 :     Description:
4249 :    
4250 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
4251 :    
4252 :     ----------------------------------------------------------------------
4253 :     Name: Matthias Blume
4254 : blume 670 Date: 2000/06/18 22:00:10 JST
4255 :     Tag: blume-20000618-implicit-anchors-really-gone
4256 :     Description:
4257 :    
4258 :     I updates the previous HISTORY entry where I forgot to mention that
4259 :     implicit anchors are no longer with us.
4260 :    
4261 :     The current update also gets rid of the (now useless) controller
4262 :     CM.Control.implicit_anchors.
4263 :    
4264 :     ----------------------------------------------------------------------
4265 :     Name: Matthias Blume
4266 : blume 666 Date: 2000/06/16 17:30:00 JST
4267 :     Tag: blume-20000616-anchorenv
4268 :     Description:
4269 :    
4270 :     This patch implements the long anticipated (just kidding :) "anchor
4271 :     environment" mechanism. In the course of doing this, I also
4272 :     re-implemented CM's internal "SrcPath" module from scratch. The new
4273 :     one should be more robust in certain boundary cases. In any case, it
4274 :     is a lot cleaner than its predecessor (IMHO).
4275 :    
4276 :     This time, although there is yet another boot file format change, I
4277 :     kept the unpickler backward-compatible. As a result, no new bootfiles
4278 :     are necessary and bootstrapping is straightforward. (You cannot read
4279 :     new bootfiles into an old system, but the other way around is no
4280 :     problem.)
4281 :    
4282 :     Visible changes:
4283 :    
4284 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
4285 :     longer recognized at all. This means that such path names are not
4286 :     illegal either. For example, the name basis.cm simply refers to a
4287 :     local file called "basis.cm" (i.e, the name is an ordinary path
4288 :     relative to .cm-files directory). Or, to put it differently, only
4289 :     names that start with $ are anchored paths.
4290 : blume 666
4291 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
4292 :     vanished.
4293 :    
4294 :     John (Reppy) had critizised this as soon as I originally proposed and
4295 : blume 666 implemented it, but at that time I did not really deeply believe
4296 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
4297 :     in another place where it cannot be seen as an abbreviation for
4298 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
4299 :     seems to mean (i.e., it "expands" into the corresponding anchor
4300 :     value).
4301 :    
4302 :     However, when paths are used as members in CM description files, it
4303 :     continues to be true that there must be at least another arc after the
4304 :     anchor. This is now enforced separately during semantic analysis
4305 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
4306 : blume 666
4307 :     ** 2. The "cm" class now accepts an option "bind". The option's value
4308 :     is a sub-option list of precisely two items -- one labeled "anchor"
4309 :     and the other one labeled "value". As you might expect, "anchor" is
4310 :     used to specify an anchor name to be bound, and "value" specifies what
4311 :     the anchor is being bound to.
4312 :    
4313 :     The value must be a directory name and can be given in either standard
4314 :     syntax (including the possibility that it is itself an anchored path)
4315 :     or native syntax.
4316 :    
4317 :     Examples:
4318 :    
4319 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
4320 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
4321 :    
4322 :     and so on.
4323 :    
4324 :     The meaning of this is that the .cm-file will be processed with an
4325 :     augmented anchor environment where the given anchor(s) is/are bound to
4326 :     the given values(s).
4327 :    
4328 :     The rationale for having this feature is this: Suppose you are trying
4329 :     to use two different (already stable) libraries a.cm and b.cm (that
4330 :     you perhaps didn't write yourself). Further, suppose each of these
4331 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
4332 :     Normally you would now have a problem because the anchor "lib" can not
4333 :     be bound to more than one value globally. Therefore, the project that
4334 :     uses both a.cm and b.cm must locally redirect the anchor to some other
4335 :     place:
4336 :    
4337 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
4338 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
4339 :    
4340 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
4341 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
4342 :    
4343 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
4344 :     you when you do so at the time of CM.stabilize). Therefore, you can
4345 :     also use an anchored path as the value:
4346 :    
4347 :     a.cm (bind:(anchor:lib value:$a-lib))
4348 :     b.cm (bind:(anchor:lib value:$b-lib))
4349 :    
4350 :     Now you can globally configure (using the usual CM.Anchor.anchor or