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