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 1211 - (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 1204 Name: Matthias Blume
17 : blume 1211 Date: 2002/05/21 10:55:00 EDT
18 :     Tag: blume-20020521-misc
19 :     Description:
20 :    
21 :     - John tweaked runtime to be silent on heap export (except when
22 :     GC messages are on).
23 :     - I added a few more things (cross-compiling versions of CMB) to
24 :     config/preloads (as suggestions).
25 :    
26 :     ----------------------------------------------------------------------
27 :     Name: Matthias Blume
28 : blume 1208 Date: 2002/05/20 22:25:00 EDT
29 :     Tag: blume-20020520-controls
30 :     Description:
31 :    
32 :     - Added ControlUtil structure to control-lib.cm.
33 :     - Use it throughout.
34 :     - Used Controls facility to define MLRISC controls (as opposed to
35 :     registering MLRISC control ref cells with Controls after the fact)
36 :     - Fixed messed-up controls priorities.
37 :    
38 :     * Removed again all the stuff from config/preloads that one wouldn't
39 :     be able to preload at the time the initial heap image is built.
40 :     (Many libraries, e.g., CML, do not exist yet at this time. The
41 :     only libraries that can be preloaded via config/preloads are those
42 :     that come bundled with the bootfiles.)
43 :    
44 :     ----------------------------------------------------------------------
45 :     Name: Matthias Blume
46 : blume 1204 Date: 2002/05/20 10:59:00 EDT
47 :     Tag: blume-20020520-preloads
48 :     Description:
49 :    
50 :     Added a lot of commented-out suggestions for things to be included
51 :     in config/preloads.
52 :    
53 :     ----------------------------------------------------------------------
54 : leunga 1203 Name: Allen Leung
55 :     Date: 2002/05/18 14:20:00 EDT
56 :     Tag: leunga-20020518-mdl
57 :     Description:
58 :    
59 :     o Made the mdl tool stuff compile and run again.
60 :     o I've disabled all the stuff that depends on RTL specifications; they
61 :     are all badly broken anyway.
62 :    
63 :     ----------------------------------------------------------------------
64 : blume 1201 Name: Matthias Blume
65 :     Date: 2002/05/17 16:49:00 EDT
66 :     Tag: blume-20020517-controls
67 :     Description:
68 :    
69 :     0. John Reppy made several modifications to the SML/NJ library.
70 :     In particular, there is a shiny new controls-lib.cm.
71 :    
72 :     1. Pushed new controls interface through compiler so that everything
73 :     compiles again.
74 :    
75 :     2. Added FormatComb and FORMAT_COMB to the CML version of the
76 :     SML/NJ library (so that CML compiles again).
77 :    
78 :     3. Modified init scripts because XXX_DEFAULT environment variables
79 :     are no longer with us. (Boot-time initialization is now done
80 :     using the same environment variables that are also used for
81 :     startup-time initialization of controls.)
82 :    
83 :     ----------------------------------------------------------------------
84 : george 1192 Name: Lal George
85 :     Date: 2002/05/15 09:20:10 EDT
86 :     Tag: george-20020515-pseudo-op-decls
87 :     Description:
88 :    
89 :     All pseudo-ops emitted before the first segment declaration
90 :     such as TEXT, DATA, and BSS directives are assumed to be global
91 :     declarations and are emitted first in the assembly file. This is
92 :     useful in a number of situations where one has pseudo-ops that are not
93 :     specific to any segment, and also works around the constraint that one
94 :     cannot have client pseudo-ops in the TEXT segment.
95 :    
96 :     Because no segment is associated with these declarations it is
97 :     an error to allocate any space or objects before the first segment
98 :     directive and an exception will be raised. However, we cannot make
99 :     this check for client pseudo-ops.
100 :    
101 :     These top level declarations are a field in the CFG graph_info.
102 :     In theory you can continue to add to this field after the CFG has been
103 :     built -- provided you know what you are doing;-)
104 :    
105 :     ----------------------------------------------------------------------
106 : blume 1175 Name: Matthias Blume
107 : blume 1190 Date: 2002/05/13 16:40:00 EDT
108 :     Tag: blume-20020513-pp-etc
109 :     Description:
110 :    
111 :     A few minor bugfixes:
112 :    
113 :     - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
114 :     (Bogus printouts for redefined bindings still occur. Compiler
115 :     bug should no longer occur now. We need to redo the prettyprinter
116 :     from scratch.)
117 :    
118 :     - CM pathname printer now also adds escape sequences for ( and )
119 :    
120 :     - commend and docu fixes for ml-nlffi
121 :    
122 :     ----------------------------------------------------------------------
123 :     Name: Matthias Blume
124 : blume 1189 Date: 2002/05/10 16:40:00 EDT
125 :     Tag: blume-20020510-erg-textio
126 :     Description:
127 :    
128 :     Applied the following bugfix provided by Emden Gansner:
129 :    
130 :     Output is corrupted when outputSubstr is used rather than output.
131 :    
132 :     The problem occurs when a substring
133 :    
134 :     ss = (s, dataStart, dataLen)
135 :    
136 :     where dataStart > 0, fills a stream buffer with avail bytes left.
137 :     avail bytes of s, starting at index dataStart, are copied into the
138 :     buffer, the buffer is flushed, and then the remaining dataLen-avail
139 :     bytes of ss are copied into the beginning of the buffer. Instead of
140 :     starting this copy at index dataStart+avail in s, the current code
141 :     starts the copy at index avail.
142 :    
143 :     Fix:
144 :     In text-io-fn.sml, change line 695 from
145 :     val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
146 :     to
147 :     val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
148 :    
149 :     ----------------------------------------------------------------------
150 :     Name: Matthias Blume
151 : blume 1186 Date: 2002/04/12 13:55:00 EDT
152 :     Tag: blume-20020412-assyntax
153 :     Description:
154 :    
155 :     1. Grabbed newer assyntax.h from the XFree86 project.
156 :     2. Fiddled with how to compile X86.prim.asm without warnings.
157 :     3. (Very) Minor cleanup in CM.
158 :    
159 :     ----------------------------------------------------------------------
160 :     Name: Matthias Blume
161 : blume 1185 Date: 2002/04/01 (no joke!) 17:07:00 EST
162 :     Tag: blume-20020401-x86div
163 :     Description:
164 :    
165 :     Added full support for div/mod/rem/quot on the x86, using the machine
166 :     instruction's two results (without clumsily recomputing the remainder)
167 :     directly where appropriate.
168 :    
169 :     Some more extensive power-of-two support was added to the x86 instruction
170 :     selector (avoiding expensive divs, mods, and muls where they can be
171 :     replaced with cheaper shifts and masks). However, this sort of thing
172 :     ought to be done earlier, e.g., within the CPS optimizer so that
173 :     all architectures benefit from it.
174 :    
175 :     The compiler compiles to a fixed point, but changes might be somewhat
176 :     fragile nevertheless. Please, report any strange things that you might
177 :     see wrt. div/mod/quot/rem...
178 :    
179 :     ----------------------------------------------------------------------
180 :     Name: Matthias Blume
181 : blume 1184 Date: 2002/03/29 17:22:00
182 :     Tag: blume-20020329-div
183 :     Description:
184 :    
185 :     Fixed my broken div/mod logic. Unfortunately, this means that the
186 :     inline code for div/mod now has one more comparison than before.
187 :     Fast paths (quotient > 0 or remainder = 0) are not affected, though.
188 :     The problem was with quotient = 0, because that alone does not tell
189 :     us which way the rounding went. One then has to look at whether
190 :     remainder and divisor have the same sign... :(
191 :    
192 :     Anyway, I replaced the bootfiles with fresh ones...
193 :    
194 :     ----------------------------------------------------------------------
195 :     Name: Matthias Blume
196 : blume 1183 Date: 2002/03/29 14:10:00 EST
197 :     Tag: blume-20020329-inlprims
198 :     Description:
199 :    
200 :     NEW BOOTFILES!!! Version number bumped to 110.39.3.
201 :    
202 :     Primops have changed. This means that the bin/boot-file formats have
203 :     changed as well.
204 :    
205 :     To make sure that there is no confusion, I made a new version.
206 :    
207 :    
208 :     CHANGES:
209 :    
210 :     * removed REMT from mltree (remainder should never overflow).
211 :    
212 :     * added primops to deal with divisions of all flavors to the frontend
213 :    
214 :     * handled these primops all the way through so they map to their respective
215 :     MLRISC support
216 :    
217 :     * used these primops in the implementation of Int, Int32, Word, Word32
218 :    
219 :     * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
220 :    
221 :     * parameterized INLMIN, INLMAX, and INLABS by a numkind
222 :    
223 :     * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
224 :     floating point
225 :    
226 :     * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
227 :     and Real (but Real.abs maps to a separate floating-point-only primop)
228 :    
229 :    
230 :     TODO items:
231 :    
232 :     * Hacked Alpha32 instruction selection, disabling the selection of REMx
233 :     instructions because the machine instruction encoder cannot handle
234 :     them. (Hppa, PPC, and Sparc instruction selection did not handle
235 :     REM in the first place, and REM is supported by the x86 machine coder.)
236 :    
237 :     * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
238 :     selection phase. (The two can be streamlined because the hardware
239 :     delivers both quotient and remainder at the same time anyway.)
240 :    
241 :     * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
242 :     (Currently the behavior is inconsistent both across architectures and
243 :     wrt. the draft Basis spec.)
244 :    
245 :     * Word8 should eventually be handled natively, too.
246 :    
247 :     * There seems to be one serious bug in mltree-gen.sml. It appears, though,
248 :     as if there currently is no execution path that could trigger it in
249 :     SML/NJ. (The assumptions underlying functions arith and promotable do not
250 :     hold for things like multiplication and division.)
251 :    
252 :     ----------------------------------------------------------------------
253 :     Name: Matthias Blume
254 : blume 1181 Date: 2002/03/27 16:27:00 EST
255 :     Tag: blume-20020327-mlrisc-divisions
256 :     Description:
257 :    
258 :     Added support for all four division operations (ML's div, mod, quot,
259 :     and rem) to MLRISC. In the course of doing so, I also rationalized
260 :     the naming (no more annoying switch-around of DIV and QUOT), by
261 :     parameterizing the operation by div_rounding_mode (which can be either
262 :     DIV_TO_ZERO or DIV_TO_NEGINF).
263 :    
264 :     The generic MLTreeGen functor takes care of compiling all four
265 :     operations down to only round-to-zero div.
266 :    
267 :     Missing pieces:
268 :    
269 :     * Doing something smarter than relying on MLTreeGen on architectures
270 :     like, e.g., the x86 where hardware division delivers both quotient and
271 :     remainder at the same time. With this, the implementation of the
272 :     round-to-neginf operations could be further streamlined.
273 :    
274 :     * Remove inlining support for div/mod/rem from the frontend and replace it
275 :     with primops that get carried through to the backend. Do this for all
276 :     int and word types.
277 :    
278 :     ----------------------------------------------------------------------
279 :     Name: Matthias Blume
280 : blume 1180 Date: 2002/03/25 17:25:00 EST
281 :     Tag: blume-20020325-divmod
282 :     Description:
283 :    
284 :     I improved (hopefully without breaking them) the implementation of Int.div,
285 :     Int.mod, and Int.rem. For this, the code in translate.sml now takes
286 :     advantage of the following observations:
287 :    
288 :     Let q = x quot y r = x rem y
289 :     d = x div y m = x mod y
290 :    
291 :     where "quot" is the round-to-zero version of integer division that
292 :     hardware usually provides. Then we have:
293 :    
294 :     r = x - q * y where neither the * nor the - will overflow
295 :     d = if q >= 0 orelse x = q * y then q else q - 1
296 :     where neither the * nor the - will overflow
297 :     m = if q >= 0 orelse r = 0 then r else r + y
298 :     where the + will not overflow
299 :    
300 :     This results in substantial simplification of the generated code.
301 :     The following table shows the number of CFG nodes and edges generated
302 :     for
303 :     fun f (x, y) = x OPER y
304 :     (* with OPER \in div, mod, quot, rem *)
305 :    
306 :    
307 :     OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
308 :     --------------------------------------------------------
309 :     div | 24 | 39 | 12 | 16
310 :     mod | 41 | 71 | 12 | 16
311 :     quot | 8 | 10 | 8 | 10
312 :     rem | 10 | 14 | 8 | 10
313 :    
314 :    
315 :     ----------------------------------------------------------------------
316 :     Name: Matthias Blume
317 : blume 1179 Date: 2002/03/25 22:06:00 EST
318 :     Tag: blume-20020325-cprotobug
319 :     Description:
320 :    
321 :     Fixed a bug in cproto (c prototype decoder).
322 :    
323 :     ----------------------------------------------------------------------
324 :     Name: Matthias Blume
325 : blume 1178 Date: 2002/03/25 16:00:00 EST
326 :     Tag: blume-20020325-raw-primops
327 :     Description:
328 :    
329 :     I did some cleanup to Allen's new primop code and
330 :     replaced yesterday's bootfiles with new ones.
331 :     (But they are stored in the same place.)
332 :    
333 :     ----------------------------------------------------------------------
334 :     Name: Matthias Blume
335 : blume 1175 Date: 2002/03/24 22:40:00 EST
336 :     Tag: blume-20020324-bootfiles
337 :     Description:
338 :    
339 :     Made the bootfiles that Allen asked for.
340 :    
341 :     ----------------------------------------------------------------------
342 : leunga 1174 Name: Allen Leung
343 :     Date: 2002/03/23 15:50:00 EST
344 :     Tag: leunga-20020323-flint-cps-rcc-primops
345 :     Description:
346 :    
347 :     1. Changes to FLINT primops:
348 :    
349 :     (* make a call to a C-function;
350 :     * The primop carries C function prototype information and specifies
351 :     * which of its (ML-) arguments are floating point. C prototype
352 :     * information is for use by the backend, ML information is for
353 :     * use by the CPS converter. *)
354 :     | RAW_CCALL of { c_proto: CTypes.c_proto,
355 :     ml_args: ccall_type list,
356 :     ml_res_opt: ccall_type option,
357 :     reentrant : bool
358 :     } option
359 :     (* Allocate uninitialized storage on the heap.
360 :     * The record is meant to hold short-lived C objects, i.e., they
361 :     * are not ML pointers. With the tag, the representation is
362 :     * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
363 :     *)
364 :     | RAW_RECORD of {tag:bool,sz:int}
365 :     and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
366 :    
367 :     2. These CPS primops are now overloaded:
368 :    
369 :     rawload of {kind:numkind}
370 :     rawstore of {kind:numkind}
371 :    
372 :     The one argument form is:
373 :    
374 :     rawload {kind} address
375 :    
376 :     The two argument form is:
377 :    
378 :     rawload {kind} [ml object, byte-offset]
379 :    
380 :     3. RAW_CCALL/RCC now takes two extra arguments:
381 :    
382 :     a. The first is whether the C call is reentrant, i.e., whether
383 :     ML state should be saved and restored.
384 :     b. The second argument is a string argument specifying the name of
385 :     library and the C function.
386 :    
387 :     These things are currently not handled in the code generator, yet.
388 :    
389 :     4. In CProto,
390 :    
391 :     An encoding type of "bool" means "ml object" and is mapped into
392 :     C prototype of PTR. Note that "bool" is different than "string",
393 :     even though "string" is also mapped into PTR, because "bool"
394 :     is assigned an CPS type of BOGt, while "string" is assigned INT32t.
395 :    
396 :     5. Pickler/unpicker
397 :    
398 :     Changed to handle RAW_RECORD and newest RAW_CCALL
399 :    
400 :     6. MLRiscGen,
401 :    
402 :     1. Changed to handle the new rawload/rawstore/rawrecord operators.
403 :     2. Code for handling C Calls has been moved to a new module CPSCCalls,
404 :     in the file CodeGen/cpscompile/cps-c-calls.sml
405 :    
406 :     7. Added the conditional move operator
407 :    
408 :     condmove of branch
409 :    
410 :     to cps. Generation of this is still buggy so it is currently
411 :     disabled.
412 :    
413 :     ----------------------------------------------------------------------
414 : george 1168 Name: Lal George
415 :     Date: 2002/03/22 14:18:25 EST
416 : george 1171 Tag: george-20020322-cps-branch-prob
417 : george 1168 Description:
418 :    
419 : george 1171 Implemented the Ball-Larus branch prediction-heuristics, and
420 : george 1168 incorporated graphical viewers for control flow graphs.
421 :    
422 : george 1171 Ball-Larus Heuristics:
423 : george 1168 ---------------------
424 :     See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
425 :    
426 :     By design it uses the Dempster-Shafer theory for combining
427 :     probabilities. For example, in the function:
428 :    
429 :     fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
430 :    
431 :     the ball-larus heuristics predicts that the n=0 is unlikely
432 :     (OH-heuristic), and the 'then' branch is unlikely because of the
433 :     RH-heuristic -- giving the 'then' branch an even lower combined
434 : george 1171 probability using the Dempster-Shafer theory.
435 : george 1168
436 :     Finally, John Reppy's loop analysis in MLRISC, further lowers the
437 :     probability of the 'then' branch because of the loop in the else
438 :     branch.
439 :    
440 :    
441 :     Graphical Viewing:
442 :     ------------------
443 :     I merely plugged in Allen's graphical viewers into the compiler. The
444 :     additional code is not much. At the top level, saying:
445 :    
446 :     Control.MLRISC.getFlag "cfg-graphical-view" := true;
447 :    
448 :     will display the graphical view of the control flow graph just before
449 :     back-patching. daVinci must be in your path for this to work. If
450 :     daVinci is not available, then the default viewer can be changed
451 :     using:
452 :    
453 :     Control.MLRISC.getString "viewer"
454 :    
455 :     which can be set to "dot" or "vcg" for the corresponding viewers. Of
456 :     course, these viewers must be in your path.
457 :    
458 :     The above will display the compilation unit at the level of clusters,
459 :     many of which are small, boring, and un-interesting. Also setting:
460 :    
461 :     Control.MLRISC.getInt "cfg-graphical-view_size"
462 :    
463 :     will display clusters that are larger than the value set by the above.
464 :    
465 :    
466 :     ----------------------------------------------------------------------
467 : blume 1159 Name: Matthias Blume
468 :     Date: 2002/03/21 22:20:00 EST
469 :     Tag: blume-20020321-kmp-bugfix
470 :     Description:
471 :    
472 :     Changed the interface to the KMP routine in PreString and fixed
473 :     a minor bug in one place where it was used.
474 :    
475 :     ----------------------------------------------------------------------
476 : leunga 1156 Name: Allen Leung
477 : leunga 1158 Date: 2002/03/21 20:30:00 EST
478 :     Tag: leunga-20020321-cfg
479 :     Description:
480 :    
481 :     Fixed a potential problem in cfg edge splitting.
482 :    
483 :     ----------------------------------------------------------------------
484 :     Name: Allen Leung
485 : leunga 1156 Date: 2002/03/21 17:15:00 EST
486 :     Tag: leunga-20020321-x86-fp-cfg
487 :     Description:
488 :    
489 :     1. Recoded the buggy parts of x86-fp.
490 :    
491 :     a. All the block reordering code has been removed.
492 :     We now depend on the block placement phases to do this work.
493 :    
494 :     b. Critical edge splitting code has been simplified and moved into the
495 :     CFG modules, as where they belong.
496 :    
497 :     Both of these were quite buggy and complex. The code is now much, much
498 :     simpler.
499 :    
500 :     2. X86 backend.
501 :    
502 :     a. Added instructions for 64-bit support. Instruction selection for
503 :     64-bit has not been committed, however, since that
504 :     requires changes to MLTREE which haven't been approved by
505 :     Lal and John.
506 :    
507 :     b. Added support for FUCOMI and FUCOMIP when generating code for
508 :     PentiumPro and above. We only generate these instructions in
509 :     the fast-fp mode.
510 :    
511 :     c. Added cases for JP and JNP in X86FreqProps.
512 :    
513 :     3. CFG
514 :    
515 :     CFG now has a bunch of methods for edge splitting and merging.
516 :    
517 :     4. Machine description.
518 :    
519 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
520 :     description things:
521 :    
522 :     rtl-build.{sig,sml} and hppa.mdl fixed.
523 :    
524 :     NOTE: the machine description stuff in the repository is still broken.
525 :     Again, I can't put my fixes in because that involves
526 :     changes to MLTREE.
527 :    
528 :     ----------------------------------------------------------------------
529 : blume 1145 Name: Matthias Blume
530 : blume 1155 Date: 2002/03/20 15:55:00 EST
531 :     Tag: blume-20020320-kmp
532 :     Description:
533 :    
534 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
535 :     it for String.isSubstring, Substring.isSubstring, and
536 :     Substring.position.
537 :    
538 :     (Might need some stress-testing. Simple examples worked fine.)
539 :    
540 :     ----------------------------------------------------------------------
541 :     Name: Matthias Blume
542 : blume 1152 Date: 2002/03/19 16:37:00 EST
543 :     Tag: blume-20020319-witnesses
544 :     Description:
545 :    
546 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
547 :    
548 :     This implements a generic mechanism for changing constness qualifiers
549 :     anywhere within big C types without resorting to outright "casts".
550 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
551 :     modify the constness at the outermost level.)
552 :     The implementation of "convert" is based on the idea of "witness"
553 :     values -- values that are not used by the operation but whose types
554 :     "testify" to their applicability. On the implementation side, "convert"
555 :     is simply a projection (returning its second curried argument). With
556 :     cross-module inlining, it should not result in any machine code being
557 :     generated.
558 :    
559 :     ----------------------------------------------------------------------
560 :     Name: Matthias Blume
561 : blume 1148 Date: 2002/03/15 16:40:00 EST
562 :     Tag: blume-20020315-basis
563 :     Description:
564 :    
565 :     Provided (preliminary?) implementations for
566 :    
567 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
568 :    
569 :     and
570 :    
571 :     Substring.full
572 :    
573 :     Those are in the Basis spec but they were missing in SML/NJ.
574 :    
575 :     ----------------------------------------------------------------------
576 :     Name: Matthias Blume
577 : blume 1145 Date: 2002/03/14 21:30:00 EST
578 :     Tag: blume-20020314-controls
579 :     Description:
580 :    
581 :     Controls:
582 :     ---------
583 :    
584 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
585 :     it into its own library $/controls-lib.cm. The source tree for
586 :     this is under src/smlnj-lib/Controls.
587 :    
588 :     2. Changed the names of types and functions in this interface, so they
589 :     make a bit more "sense":
590 :    
591 :     module -> registry
592 :     'a registry -> 'a group
593 :    
594 :     3. The interface now deals in ref cells only. The getter/setter interface
595 :     is (mostly) gone.
596 :    
597 :     4. Added a function that lets one register an already-existing ref cell.
598 :    
599 :     5. Made the corresponding modifications to the rest of the code so that
600 :     everything compiles again.
601 :    
602 :     6. Changed the implementation of Controls.MLRISC back to something closer
603 :     to the original. In particular, this module (and therefore MLRISC)
604 :     does not depend on Controls. There now is some link-time code in
605 :     int-sys.sml that registers the MLRISC controls with the Controls
606 :     module.
607 :    
608 :     CM:
609 :     ---
610 :    
611 :     * One can now specify the lambda-split aggressiveness in init.cmi.
612 :    
613 :     ----------------------------------------------------------------------
614 : leunga 1142 Name: Allen Leung
615 :     Date: 2002/03/13 17:30:00 EST
616 :     Tag: leunga-20020313-x86-fp-unary
617 :     Description:
618 :    
619 :     Bug fix for:
620 :    
621 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
622 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
623 :     > - fun f(x,(y,z)) = Real.~ y;
624 :     > [autoloading]
625 :     > [autoloading done]
626 :     > fchsl (%eax), 184(%esp)
627 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
628 :     >
629 :     > uncaught exception Error
630 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
631 :    
632 :     The problem was that the code generator did not generate any fp registers
633 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
634 :     translate the pseudo fp instruction. This only happened with unary fp
635 :     operators in certain situations.
636 :    
637 :     ----------------------------------------------------------------------
638 : blume 1137 Name: Matthias Blume
639 : blume 1140 Date: 2002/03/13 14:00:00 EST
640 :     Tag: blume-20020313-overload-etc
641 :     Description:
642 :    
643 :     1. Added _overload as a synonym for overload for backward compatibility.
644 :     (Control.overloadKW must be true for either version to be accepted.)
645 :    
646 :     2. Fixed bug in install script that caused more things to be installed
647 :     than what was requested in config/targets.
648 :    
649 :     3. Made CM aware of the (_)overload construct so that autoloading
650 :     works.
651 :    
652 :     ----------------------------------------------------------------------
653 :     Name: Matthias Blume
654 : blume 1138 Date: 2002/03/12 22:03:00 EST
655 :     Tag: blume-20020312-url
656 :     Description:
657 :    
658 :     Forgot to update BOOT and srcarchiveurl.
659 :    
660 :     ----------------------------------------------------------------------
661 :     Name: Matthias Blume
662 : blume 1137 Date: 2002/03/12 17:30:00 EST
663 :     Tag: blume-20020312-version110392
664 :     Description:
665 :    
666 :     Yet another version number bump (because of small changes to the
667 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
668 :    
669 :     Changes:
670 :    
671 :     The new pid generation scheme described a few weeks ago was overly
672 :     complicated. I implemented a new mechanism that is simpler and
673 :     provides a bit more "stability": Once CM has seen a compilation
674 :     unit, it keeps its identity constant (as long as you do not delete
675 :     those crucial CM/GUID/* files). This means that when you change
676 : blume 1145 an interface, compile, then go back to the old interface, and
677 : blume 1137 compile again, you arrive at the original pid.
678 :    
679 :     There now also is a mechanism that instructs CM to use the plain
680 :     environment hash as a module's pid (effectively making its GUID
681 :     the empty string). For this, "noguid" must be specified as an
682 :     option to the .sml file in question within its .cm file.
683 :     This is most useful for code that is being generated by tools such
684 :     as ml-nlffigen (because during development programmers tend to
685 :     erase the tool's entire output directory tree including CM's cached
686 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
687 :     revert to the old, broken behavior of SML/NJ, but in specific cases
688 :     where there is no danger of interface confusion, its use is ok
689 :     (I think).
690 :    
691 :     ml-nlffigen by default generates "noguid" annotations. They can be
692 :     turned off by specifying -guid in its command line.
693 :    
694 :     ----------------------------------------------------------------------
695 : george 1136 Name: Lal George
696 :     Date: 2002/03/12 12 14:42:36 EST
697 :     Tag: george-20020312-frequency-computation
698 :     Description:
699 :    
700 :     Integrated jump chaining and static block frequency into the
701 :     compiler. More details and numbers later.
702 : leunga 1025
703 : george 1003 ----------------------------------------------------------------------
704 : george 1133 Name: Lal George
705 :     Date: 2002/03/11 11 22:38:53 EST
706 :     Tag: george-20020311-jump-chain-elim
707 :     Description:
708 :    
709 :     Tested the jump chain elimination on all architectures (except the
710 :     hppa). This is on by default right now and is profitable for the
711 :     alpha and x86, however, it may not be profitable for the sparc and ppc
712 :     when compiling the compiler.
713 :    
714 :     The gc test will typically jump to a label at the end of the cluster,
715 :     where there is another jump to an external cluster containing the actual
716 :     code to invoke gc. This is to allow factoring of common gc invocation
717 :     sequences. That is to say, we generate:
718 :    
719 :     f:
720 :     testgc
721 :     ja L1 % jump if above to L1
722 :    
723 :     L1:
724 :     jmp L2
725 :    
726 :    
727 :     After jump chain elimination the 'ja L1' instructions is converted to
728 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
729 :     up being implemented in their long form (if L2 is far away) using:
730 :    
731 :     jbe L3 % jump if below or equal to L3
732 :     jmp L2
733 :     L3:
734 :     ...
735 :    
736 :    
737 :     For large compilation units L2 may be far away.
738 :    
739 :    
740 :     ----------------------------------------------------------------------
741 : blume 1132 Name: Matthias Blume
742 :     Date: 2002/03/11 13:30:00 EST
743 :     Tag: blume-20020311-mltreeeval
744 :     Description:
745 :    
746 :     A functor parameter was missing.
747 :    
748 :     ----------------------------------------------------------------------
749 : leunga 1130 Name: Allen Leung
750 : leunga 1131 Date: 2002/03/11 10:30:00 EST
751 : leunga 1142 Tag: leunga-20020311-runtime-string0
752 : leunga 1131 Description:
753 :    
754 :     The representation of the empty string now points to a
755 :     legal null terminated C string instead of unit. It is now possible
756 :     to convert an ML string into C string with InlineT.CharVector.getData.
757 :     This compiles into one single machine instruction.
758 :    
759 :     ----------------------------------------------------------------------
760 :     Name: Allen Leung
761 : leunga 1130 Date: 2002/03/10 23:55:00 EST
762 :     Tag: leunga-20020310-x86-call
763 :     Description:
764 :    
765 :     Added machine generation for CALL instruction (relative displacement mode)
766 :    
767 :     ----------------------------------------------------------------------
768 : blume 1128 Name: Matthias Blume
769 :     Date: 2002/03/08 16:05:00
770 :     Tag: blume-20020308-entrypoints
771 :     Description:
772 :    
773 :     Version number bumped to 110.39.1. NEW BOOTFILES!
774 :    
775 :     Entrypoints: non-zero offset into a code object where execution should begin.
776 :    
777 :     - Added the notion of an entrypoint to CodeObj.
778 :     - Added reading/writing of entrypoint info to Binfile.
779 :     - Made runtime system bootloader aware of entrypoints.
780 :     - Use the address of the label of the first function given to mlriscGen
781 :     as the entrypoint. This address is currently always 0, but it will
782 :     not be 0 once we turn on block placement.
783 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
784 :     with entry points) from mlriscGen.
785 :    
786 :     ----------------------------------------------------------------------
787 : leunga 1127 Name: Allen Leung
788 :     Date: 2002/03/07 20:45:00 EST
789 :     Tag: leunga-20020307-x86-cmov
790 :     Description:
791 :    
792 :     Bug fixes for CMOVcc on x86.
793 :    
794 :     1. Added machine code generation for CMOVcc
795 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
796 :     3. CMOVcc cannot have an immediate operand as argument.
797 :    
798 :     ----------------------------------------------------------------------
799 : blume 1126 Name: Matthias Blume
800 :     Date: 2002/03/07 16:15:00 EST
801 :     Tag: blume-20020307-controls
802 :     Description:
803 :    
804 :     This is a very large but mostly boring patch which makes (almost)
805 :     every tuneable compiler knob (i.e., pretty much everything under
806 :     Control.* plus a few other things) configurable via both the command
807 :     line and environment variables in the style CM did its configuration
808 :     until now.
809 :    
810 :     Try starting sml with '-h' (or, if you are brave, '-H')
811 :    
812 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
813 :     implements the underlying generic mechanism.
814 :    
815 :     The interface to some of the existing such facilities has changed somewhat.
816 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
817 :     (The getFoo interface is still there for backward-compatibility, but its
818 :     use is deprecated.)
819 :    
820 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
821 :     that one can now twiddle the compiler settings when using this "batch"
822 :     compiler.
823 :    
824 :     TODO items:
825 :    
826 :     We should go through and throw out all controls that are no longer
827 :     connected to anything. Moreover, we should go through and provide
828 :     meaningful (and correct!) documentation strings for those controls
829 :     that still are connected.
830 :    
831 :     Currently, multiple calls to Controls.new are accepted (only the first
832 :     has any effect). Eventually we should make sure that every control
833 :     is being made (via Controls.new) exactly once. Future access can then
834 :     be done using Controls.acc.
835 :    
836 :     Finally, it would probably be a good idea to use the getter-setter
837 :     interface to controls rather than ref cells. For the time being, both
838 :     styles are provided by the Controls module, but getter-setter pairs are
839 :     better if thread-safety is of any concern because they can be wrapped.
840 :    
841 :     *****************************************
842 :    
843 :     One bug fix: The function blockPlacement in three of the MLRISC
844 :     backpatch files used to be hard-wired to one of two possibilities at
845 :     link time (according to the value of the placementFlag). But (I
846 :     think) it should rather sense the flag every time.
847 :    
848 :     *****************************************
849 :    
850 :     Other assorted changes (by other people who did not supply a HISTORY entry):
851 :    
852 :     1. the cross-module inliner now works much better (Monnier)
853 :     2. representation of weights, frequencies, and probabilities in MLRISC
854 :     changed in preparation of using those for weighted block placement
855 :     (Reppy, George)
856 :    
857 :     ----------------------------------------------------------------------
858 : george 1116 Name: Lal George
859 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
860 :     Tag: george-20020307-weighted-block-placement
861 :    
862 :     Tested the weighted block placement optimization on all architectures
863 :     (except the hppa) using AMPL to generate the block and edge frequencies.
864 :     Changes were required in the machine properties to correctly
865 :     categorize trap instructions. There is an MLRISC flag
866 :     "weighted-block-placement" that can be used to enable weighted block
867 :     placement, but this will be ineffective without block/edge
868 :     frequencies (coming soon).
869 :    
870 :    
871 :     ----------------------------------------------------------------------
872 :     Name: Lal George
873 : george 1116 Date: 2002/03/05 17:24:48 EST
874 :     Tag: george-20020305-linkage-cluster
875 :    
876 : george 1124 In order to support the block placement optimization, a new cluster
877 :     is generated as the very first cluster (called the linkage cluster).
878 :     It contains a single jump to the 'real' entry point for the compilation
879 :     unit. Block placement has no effect on the linkage cluster itself, but
880 :     all the other clusters have full freedom in the manner in which they
881 :     reorder blocks or functions.
882 : george 1116
883 :     On the x86 the typical linkage code that is generated is:
884 :     ----------------------
885 :     .align 2
886 :     L0:
887 :     addl $L1-L0, 72(%esp)
888 : george 1124 jmp L1
889 : george 1116
890 :    
891 :     .align 2
892 :     L1:
893 :     ----------------------
894 :    
895 :     72(%esp) is the memory location for the stdlink register. This
896 :     must contain the address of the CPS function being called. In the
897 : george 1124 above example, it contains the address of L0; before
898 : george 1116 calling L1 (the real entry point for the compilation unit), it
899 :     must contain the address for L1, and hence
900 :    
901 :     addl $L1-L0, 72(%esp)
902 :    
903 : george 1124 I have tested this on all architectures except the hppa.The increase
904 :     in code size is of course negligible
905 : george 1116
906 :     ----------------------------------------------------------------------
907 : leunga 1115 Name: Allen Leung
908 :     Date: 2002/03/03 13:20:00 EST
909 :     Tag: leunga-20020303-mlrisc-tools
910 :    
911 :     Added #[ ... ] expressions to mlrisc tools
912 :    
913 :     ----------------------------------------------------------------------
914 : blume 1096 Name: Matthias Blume
915 : blume 1098 Date: 2002/02/27 12:29:00 EST
916 :     Tag: blume-20020227-cdebug
917 :     Description:
918 :    
919 :     - made types in structure C and C_Debug to be equal
920 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
921 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
922 :    
923 :     ----------------------------------------------------------------------
924 :     Name: Matthias Blume
925 : blume 1096 Date: 2002/02/26 12:00:00 EST
926 :     Tag: blume-20020226-ffi
927 :     Description:
928 :    
929 :     1. Fixed a minor bug in CM's "noweb" tool:
930 :     If numbering is turned off, then truly don't number (i.e., do not
931 :     supply the -L option to noweb). The previous behavior was to supply
932 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
933 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
934 :    
935 :     2. Once again, I reworked some aspects of the FFI:
936 :    
937 :     A. The incomplete/complete type business:
938 :    
939 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
940 :     gone!
941 :     - ML types representing an incomplete type are now *equal* to
942 :     ML types representing their corresponding complete types (just like
943 :     in C). This is still safe because ml-nlffigen will not generate
944 :     RTTI for incomplete types, nor will it generate functions that
945 :     require access to such RTTI. But when ML code generated from both
946 :     incomplete and complete versions of the C type meet, the ML types
947 :     are trivially interoperable.
948 :    
949 :     NOTE: These changes restore the full generality of the translation
950 :     (which was previously lost when I eliminated functorization)!
951 :    
952 :     B. Enum types:
953 :    
954 :     - Structure C now has a type constructor "enum" that is similar to
955 :     how the "su" constructor works. However, "enum" is not a phantom
956 :     type because each "T enum" has values (and is isomorphic to
957 :     MLRep.Signed.int).
958 :     - There are generic access operations for enum objects (using
959 :     MLRep.Signed.int).
960 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
961 :     * The structure contains the definition of type "mlrep" (the ML-side
962 :     representation type of the enum). Normally, mlrep is the same
963 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
964 :     then mlrep will be defined as a datatype -- thus facilitating
965 :     pattern matching on mlrep values.
966 :     ("-ec" will be suppressed if there are duplicate values in an
967 :     enumeration.)
968 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
969 :     will be generated for each C enum constant xxx.
970 :     * Conversion functions m2i and i2m convert between mlrep and
971 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
972 :     * Coversion functions c and ml convert between mlrep and "tag enum".
973 :     * Access functions (get/set) fetch and store mlrep values.
974 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
975 :     enumerations are merged into one single enumeration represented by
976 :     structure E_'.
977 :    
978 :     ----------------------------------------------------------------------
979 : leunga 1080 Name: Allen Leung
980 : leunga 1094 Date: 2002/02/25 04:45:00 EST
981 :     Tag: leunga-20020225-cps-spill
982 :    
983 :     This is a new implementation of the CPS spill phase.
984 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
985 :     In case of problems, replace it with the old file spill.sml
986 :    
987 :     The current compiler runs into some serious performance problems when
988 :     constructing a large record. This can happen when we try to compile a
989 :     structure with many items. Even a very simple structure like the following
990 :     makes the compiler slow down.
991 :    
992 :     structure Foo = struct
993 :     val x_1 = 0w1 : Word32.int
994 :     val x_2 = 0w2 : Word32.int
995 :     val x_3 = 0w3 : Word32.int
996 :     ...
997 :     val x_N = 0wN : Word32.int
998 :     end
999 :    
1000 :     The following table shows the compile time, from N=1000 to N=4000,
1001 :     with the old compiler:
1002 :    
1003 :     N
1004 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
1005 :     MLRISC ra 0.06u 0.00s 0.05g
1006 :     (spills = 0 reloads = 0)
1007 :     TOTAL 0.63u 0.07s 0.21g
1008 :    
1009 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
1010 :     MLRISC ra 5.68u 0.59s 3.93g
1011 :     (spills = 0 reloads = 0)
1012 :     TOTAL 14.71u 0.99s 4.81g
1013 :    
1014 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
1015 :     MLRISC ra 5.54u 0.65s 3.91g
1016 :     (spills = 543 reloads = 1082)
1017 :     TOTAL 65.40u 3.13s 6.00g
1018 :    
1019 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
1020 :     MLRISC ra 0.80u 0.10s 0.55g
1021 :     (spills = 42 reloads = 84)
1022 :     TOTAL 129.42u 5.10s 4.13g
1023 :    
1024 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
1025 :     MLRISC ra 2.69u 0.27s 1.38g
1026 :     (spills = 62 reloads = 124)
1027 :     TOTAL 682.48u 19.61s 13.99g
1028 :    
1029 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
1030 :     MLRISC ra 4.96u 0.27s 2.72g
1031 :     (spills = 85 reloads = 170)
1032 :     TOTAL 2375.26u 57.21s 48.00g
1033 :    
1034 :     As you can see the old cps spill module suffers from some serious
1035 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
1036 :     instead of patching the problems up, I'm reimplementing it
1037 : leunga 1094 with a different algorithm. The new code is more modular,
1038 :     smaller when compiled, and substantially faster
1039 :     (O(n log n) time and O(n) space). Timing of the new spill module:
1040 :    
1041 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
1042 :     MLRISC ra 0.25u 0.02s 0.15g
1043 :     (spills=1 reloads=3)
1044 :     TOTAL 7.74u 0.34s 1.62g
1045 :    
1046 :     Implementation details:
1047 :    
1048 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
1049 :     number of live variables at any program point (the bandwidth)
1050 :     does not exceed a certain limit, which is determined by the
1051 :     size of the spill area.
1052 :    
1053 :     When the bandwidth is too large, we decrease the register pressure by
1054 :     packing live variables into spill records. How we achieve this is
1055 :     completely different than what we did in the old code.
1056 :    
1057 :     First, there is something about the MLRiscGen code generator
1058 :     that we should be aware of:
1059 :    
1060 :     o MLRiscGen performs code motion!
1061 :    
1062 :     In particular, it will move floating point computations and
1063 :     address computations involving only the heap pointer to
1064 :     their use sites (if there is only a single use).
1065 :     What this means is that if we have a CPS record construction
1066 :     statement
1067 :    
1068 :     RECORD(k,vl,w,e)
1069 :    
1070 :     we should never count the new record address w as live if w
1071 :     has only one use (which is often the case).
1072 :    
1073 :     We should do something similar to floating point, but the transformation
1074 :     there is much more complex, so I won't deal with that.
1075 :    
1076 :     Secondly, there are now two new cps primops at our disposal:
1077 :    
1078 :     1. rawrecord of record_kind option
1079 :     This pure operator allocates some uninitialized storage from the heap.
1080 :     There are two forms:
1081 :    
1082 :     rawrecord NONE [INT n] allocates a tagless record of length n
1083 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1084 :     and initializes the tag.
1085 :    
1086 :     2. rawupdate of cty
1087 :     rawupdate cty (v,i,x)
1088 :     Assigns to x to the ith component of record v.
1089 :     The storelist is not updated.
1090 :    
1091 :     We use these new primops for both spilling and increment record construction.
1092 :    
1093 :     1. Spilling.
1094 :    
1095 :     This is implemented with a linear scan algorithm (but generalized
1096 :     to trees). The algorithm will create a single spill record at the
1097 :     beginning of the cps function and use rawupdate to spill to it,
1098 :     and SELECT or SELp to reload from it. So both spills and reloads
1099 :     are fine-grain operations. In contrast, in the old algorithm
1100 :     "spills" have to be bundled together in records.
1101 :    
1102 :     Ideally, we should sink the spill record construction to where
1103 :     it is needed. We can even split the spill record into multiple ones
1104 :     at the places where they are needed. But CPS is not a good
1105 :     representation for global code motion, so I'll keep it simple and
1106 :     am not attempting this.
1107 :    
1108 :     2. Incremental record construction (aka record splitting).
1109 :    
1110 :     Long records with many component values which are simulatenously live
1111 :     (recall that single use record addresses are not considered to
1112 :     be live) are constructed with rawrecord and rawupdate.
1113 :     We allocate space on the heap with rawrecord first, then gradually
1114 :     fill it in with rawupdate. This is the technique suggested to me
1115 :     by Matthias.
1116 :    
1117 :     Some restrictions on when this is applicable:
1118 :     1. It is not a VECTOR record. The code generator currently does not handle
1119 :     this case. VECTOR record uses double indirection like arrays.
1120 :     2. All the record component values are defined in the same "basic block"
1121 :     as the record constructor. This is to prevent speculative
1122 :     record construction.
1123 :    
1124 :     ----------------------------------------------------------------------
1125 :     Name: Allen Leung
1126 : leunga 1086 Date: 2002/02/22 01:02:00 EST
1127 :     Tag: leunga-20020222-mlrisc-tools
1128 :    
1129 :     Minor bug fixes in the parser and rewriter
1130 :    
1131 :     ----------------------------------------------------------------------
1132 :     Name: Allen Leung
1133 : leunga 1085 Date: 2002/02/21 20:20:00 EST
1134 :     Tag: leunga-20020221-peephole
1135 :    
1136 :     Regenerated the peephole files. Some contained typos in the specification
1137 :     and some didn't compile because of pretty printing bugs in the old version
1138 :     of 'nowhere'.
1139 :    
1140 :     ----------------------------------------------------------------------
1141 :     Name: Allen Leung
1142 : leunga 1079 Date: 2002/02/19 20:20:00 EST
1143 :     Tag: leunga-20020219-mlrisc-tools
1144 :     Description:
1145 :    
1146 :     Minor bug fixes to the mlrisc-tools library:
1147 :    
1148 :     1. Fixed up parsing colon suffixed keywords
1149 :     2. Added the ability to shut the error messages up
1150 :     3. Reimplemented the pretty printer and fixed up/improved
1151 :     the pretty printing of handle and -> types.
1152 :     4. Fixed up generation of literal symbols in the nowhere tool.
1153 :     5. Added some SML keywords to to sml.sty
1154 :    
1155 :     ----------------------------------------------------------------------
1156 :     Name: Matthias Blume
1157 : blume 1078 Date: 2002/02/19 16:20:00 EST
1158 :     Tag: blume-20020219-cmffi
1159 :     Description:
1160 :    
1161 :     A wild mix of changes, some minor, some major:
1162 :    
1163 :     * All C FFI-related libraries are now anchored under $c:
1164 :     $/c.cm --> $c/c.cm
1165 :     $/c-int.cm --> $c/internals/c-int.cm
1166 :     $/memory.cm --> $c/memory/memory.cm
1167 :    
1168 :     * "make" tool (in CM) now treats its argument pathname slightly
1169 :     differently:
1170 :     1. If the native expansion is an absolute name, then before invoking
1171 :     the "make" command on it, CM will apply OS.Path.mkRelative
1172 :     (with relativeTo = OS.FileSys.getDir()) to it.
1173 :     2. The argument will be passed through to subsequent phases of CM
1174 :     processing without "going native". In particular, if the argument
1175 :     was an anchored path, then "make" will not lose track of that anchor.
1176 :    
1177 :     * Compiler backends now "know" their respective C calling conventions
1178 :     instead of having to be told about it by ml-nlffigen. This relieves
1179 :     ml-nlffigen from one of its burdens.
1180 :    
1181 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1182 :    
1183 :     * Export C_DEBUG and C_Debug from $c/c.cm.
1184 :    
1185 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1186 :     subtyping relationship between incomplete pointers and their complete
1187 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
1188 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
1189 :     the pointer target type is complete. In the incomplete case, 'o
1190 :     will be instantiated with some "'c iobj" -- a type obtained by
1191 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
1192 :    
1193 :     Operations that work on both incomplete and complete pointer types are
1194 :     typed as taking an 'o ptr while operations that require the target to
1195 :     be known are typed as taking some ('t, 'c) obj ptr.
1196 :    
1197 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1198 :     where void is an eqtype without any values. This makes it possible
1199 :     to work on voidptr values using functions meant to operate on light
1200 :     incomplete pointers.
1201 :    
1202 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1203 :     been vastly simplified.
1204 :    
1205 :     ----------------------------------------------------------------------
1206 :     Name: Matthias Blume
1207 : blume 1076 Date: 2002/02/19 10:48:00 EST
1208 :     Tag: blume-20020219-pqfix
1209 :     Description:
1210 :    
1211 :     Applied Chris Okasaki's bug fix for priority queues.
1212 :    
1213 :     ----------------------------------------------------------------------
1214 :     Name: Matthias Blume
1215 : blume 1073 Date: 2002/02/15 17:05:00
1216 : blume 1069 Tag: Release_110_39
1217 :     Description:
1218 :    
1219 : blume 1073 Last-minute retagging is becoming a tradition... :-(
1220 :    
1221 :     This is the working release 110.39.
1222 :    
1223 :     ----------------------------------------------------------------------
1224 :     Name: Matthias Blume
1225 :     Date: 2002/02/15 16:00:00 EST
1226 :     Tag: Release_110_39-orig
1227 :     Description:
1228 :    
1229 : blume 1069 Working release 110.39. New bootfiles.
1230 :    
1231 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
1232 :     with all shells. So I retagged. -Matthias)
1233 :    
1234 : blume 1069 ----------------------------------------------------------------------
1235 :     Name: Matthias Blume
1236 : blume 1068 Date: 2002/02/15 14:17:00 EST
1237 :     Tag: blume-20020215-showbindings
1238 :     Description:
1239 :    
1240 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
1241 :     the latter can be useful for exploring what bindings are available at
1242 :     the interactive prompt. (The first function returns only the list
1243 :     of symbols that are really bound, the second prints those but also the
1244 :     ones that CM's autoloading mechanism knows about.)
1245 :    
1246 :     ----------------------------------------------------------------------
1247 :     Name: Matthias Blume
1248 : blume 1067 Date: 2002/02/15 12:08:00 EST
1249 :     Tag: blume-20020215-iptrs
1250 :     Description:
1251 :    
1252 :     Two improvements to ml-nlffigen:
1253 :    
1254 :     1. Write files only if they do not exist or if their current contents
1255 :     do not coincide with what's being written. (That is, avoid messing
1256 :     with the time stamps unless absolutely necessary.)
1257 :    
1258 :     2. Implement a "repository" mechanism for generated files related
1259 :     to "incomplete pointer types". See the README file for details.
1260 :    
1261 :     ----------------------------------------------------------------------
1262 :     Name: Matthias Blume
1263 : blume 1066 Date: 2002/02/14 11:50:00 EST
1264 :     Tag: blume-20020214-quote
1265 :     Description:
1266 :    
1267 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
1268 :     because of the new and improved tag generation scheme. (Thanks to Allen
1269 :     Leung for pointing it out.)
1270 :    
1271 :     ----------------------------------------------------------------------
1272 : george 1065 Name: Lal George
1273 :     Date: 2002/02/14 09:55:27 EST 2002
1274 :     Tag: george-20020214-isabelle-bug
1275 :     Description:
1276 :    
1277 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1278 :     of Isabelle on the x86.
1279 :    
1280 :     From Allen:
1281 :     -----------
1282 :     I've found the problem:
1283 :    
1284 :     in ra-core.sml, I use the counter "blocked" to keep track of the
1285 :     true number of elements in the freeze queue. When the counter goes
1286 :     to zero, I skip examining the queue. But I've messed up the
1287 :     bookkeeping in combine():
1288 :    
1289 :     else ();
1290 :     case !ucol of
1291 :     PSEUDO => (if !cntv > 0 then
1292 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
1293 :     ^^^^^^^^^^^^^^^^^^^^^^^
1294 :     moveu := mergeMoveList(!movev, !moveu)
1295 :     )
1296 :     else ();
1297 :    
1298 :     combine() is called to coalesce two nodes u and v.
1299 :     I think I was thinking that if the move counts of u and v are both
1300 :     greater than zero then after they are coalesced then one node is
1301 :     removed from the freeze queue. Apparently I was thinking that
1302 :     both u and v are of low degree, but that's clearly not necessarily true.
1303 :    
1304 :    
1305 :     02/12/2002:
1306 :     Here's the patch. HOL now compiles.
1307 :    
1308 :     I don't know how this impact on performance (compile
1309 :     time or runtime). This bug caused the RA (especially on the x86)
1310 :     to go thru the potential spill phase when there are still nodes on the
1311 :     freeze queue.
1312 :    
1313 :    
1314 :    
1315 :    
1316 :     ----------------------------------------------------------------------
1317 : blume 1058 Name: Matthias Blume
1318 : blume 1064 Date: 2002/02/13 22:40:00 EST
1319 :     Tag: blume-20020213-fptr-rtti
1320 :     Description:
1321 :    
1322 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
1323 :     updates.
1324 :    
1325 :     ----------------------------------------------------------------------
1326 :     Name: Matthias Blume
1327 : blume 1063 Date: 2002/02/13 16:41:00 EST
1328 :     Tag: blume-20020213-cmlpq
1329 :     Description:
1330 :    
1331 :     Added new priority queue export symbols (which have just been added to
1332 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
1333 :     not compile and the installer would choke.)
1334 :    
1335 :     ----------------------------------------------------------------------
1336 :     Name: Matthias Blume
1337 : blume 1062 Date: 2002/02/13 16:15:00 EST
1338 :     Tag: blume-20020213-various
1339 :     Description:
1340 :    
1341 :     1. More tweaks to ml-nlffigen:
1342 :    
1343 :     - better internal datastructures (resulting in slight speedup)
1344 :     - "-match" option requires exact match
1345 :     - "localized" gensym counters (untagged structs/unions nested within
1346 :     other structs/unions or within typedefs get a fresh counter; their
1347 :     tag will be prefixed by a concatenation of their parents' tags)
1348 :     - bug fixes (related to calculation of transitive closure of types
1349 :     to be included in the output)
1350 :    
1351 :     2. Minor Basis updates:
1352 :    
1353 :     - added implementations for List.collate and Option.app
1354 :    
1355 :     ----------------------------------------------------------------------
1356 :     Name: Matthias Blume
1357 : blume 1060 Date: 2002/02/11 15:55:00 EST
1358 :     Tag: blume-20020211-gensym
1359 :     Description:
1360 :    
1361 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
1362 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
1363 :     names (ML structure names that correspond to unnamed C structs, unions,
1364 :     and enums), so that separate runs of ml-nlffigen do not clash.
1365 :    
1366 :     ----------------------------------------------------------------------
1367 :     Name: Matthias Blume
1368 : blume 1059 Date: 2002/02/11 12:05:00 EST
1369 :     Tag: blume-20020211-gensml
1370 :     Description:
1371 :    
1372 :     A quick fix for a problem with GenSML (in the pgraph-util library):
1373 :     Make generation of toplevel "local" optional. (Strictly speaking,
1374 :     signature definitions within "local" are not legal SML.)
1375 :    
1376 :     Other than that: updates to INSTALL and cm/TODO.
1377 :    
1378 :     ----------------------------------------------------------------------
1379 :     Name: Matthias Blume
1380 : blume 1058 Date: 2002/02/08 15:00:00 EST
1381 :     Tag: blume-20020208-uniquepid
1382 :     Description:
1383 :    
1384 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
1385 :    
1386 :     1. The installer (config/install.sh) has gotten smarter:
1387 :    
1388 :     - Configuration options are a bit easier to specify now
1389 :     (in config/targets).
1390 :     - Bug in recognizing .tar.bz2 files fixed.
1391 :     - Installer automatically resolves dependencies between
1392 :     configuration options (e.g., if you ask for eXene, you will
1393 :     also get cml -- regardless whether you asked for it or not).
1394 :     - Installer can run in "quieter mode" by setting the environment
1395 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
1396 :     "completely silent", though.
1397 :     - Build HashCons library as part of smlnj-lib.
1398 :    
1399 :     2. A new scheme for assigning persistent identifiers to compilation
1400 :     units (and, by extension, to types etc.) has been put into place.
1401 :     This fixes a long-standing bug where types and even dynamic values
1402 :     can get internally confused, thereby compromising type safety
1403 :     (abstraction) and dynamic correctness. See
1404 :    
1405 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1406 :    
1407 :     for an example of how things could go wrong until now.
1408 :    
1409 :     The downside of the new scheme is that pids are not quite as
1410 :     persistent as they used to be: CM will generate a fresh pid
1411 :     for every compilation unit that it thinks it sees for the first
1412 :     time. That means that if you compile starting from a clean, fresh
1413 :     source tree at two different times, you end up with different
1414 :     binaries.
1415 :    
1416 :     Cutoff recompilation, however, has not been compromised because
1417 :     CM keeps pid information in special caches between runs.
1418 :    
1419 :     ----------------------------------------------------------------------
1420 : george 1053 Name: Lal George
1421 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
1422 :     Tag: <none>
1423 :     Description:
1424 :    
1425 :     Compilers that generate assembly code may produce global labels
1426 :     whose value is resolved at link time. The various peephole optimization
1427 :     modules did not take this in account.
1428 :    
1429 :     TODO. The Labels.addrOf function should really return an option
1430 :     type so that clients are forced to deal with this issue, rather
1431 :     than an exception being raised.
1432 :    
1433 :     ----------------------------------------------------------------------
1434 :     Name: Lal George
1435 : george 1053 Date: 2002/02/06 13:55:02 EST
1436 :     Tag: george-20020206-ra-breakup
1437 :     Description:
1438 :    
1439 :     1. A bug fix from Allen.
1440 :    
1441 :     A typo causes extra fstp %st(0)'s to be generated at compensation
1442 :     edges, which might cause stack underflow traps at runtime. This
1443 :     occurs in fft where there are extraneous fstps right before the 'into'
1444 :     trap instruction (in this case they are harmless since none of the
1445 :     integers overflow.)
1446 :    
1447 : george 1054 2. Pulled out various utility modules that were embedded in the modules
1448 : george 1053 of the register allocator. I need these modules for other purposes, but
1449 : george 1054 they are not complete enough to put into a library (just yet).
1450 : george 1053 ----------------------------------------------------------------------
1451 : blume 1049 Name: Matthias Blume
1452 :     Date: 2002/01/31 16:05:00 EST
1453 :     Tag: blume-20020131-sparc-ccalls
1454 :     Description:
1455 :    
1456 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1457 :     of extra stack space by mistake. Fixed.
1458 :    
1459 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1460 :    
1461 :     ----------------------------------------------------------------------
1462 : leunga 1048 Name: Allen Leung
1463 :     Date: 2002/01/30
1464 :     Tag: leunga-20020130-nowhere-bug-fix
1465 :     Description:
1466 :    
1467 :     MLRISC bug fixes:
1468 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1469 :     2. MachineInt.fromString was negating its value.
1470 :    
1471 :     ----------------------------------------------------------------------
1472 : blume 1039 Name: Matthias Blume
1473 : blume 1047 Date: 2002/01/29
1474 :     Tag: blume-20020129-INSTALL
1475 :     Description:
1476 :    
1477 :     - Added somewhat detailed installation instructions (file INSTALL).
1478 :     - Fixed curl-detection bug in config/install.sh.
1479 :     - It is now possible to select the URL getter using the URLGETTER
1480 :     environment variable:
1481 :    
1482 :     not set / "unknown" --> automatic detection (script tries wget,
1483 :     curl, and lynx)
1484 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1485 :     how to properly invoke them)
1486 :     other --> use $URLGETTER directly, it must take
1487 :     precisely two command-line arguments
1488 :     (source URL and destination file name)
1489 :    
1490 :     ----------------------------------------------------------------------
1491 :     Name: Matthias Blume
1492 : blume 1044 Date: 2002/01/28
1493 : blume 1045 Tag: blume-20020128-sparc-ccalls
1494 :     Description:
1495 :    
1496 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
1497 :     - Make use of the allocParam argument in sparc-c-calls.
1498 :    
1499 :     ----------------------------------------------------------------------
1500 :     Name: Matthias Blume
1501 :     Date: 2002/01/28
1502 : blume 1044 Tag: blume-20020128-allocParam
1503 :     Description:
1504 :    
1505 :     John Reppy: Changes c-calls API to accept client-callback for
1506 :     allocating extra stack space.
1507 :     me: Corresponding changes to mlriscGen (using a dummy argument that
1508 :     does not change the current behavior).
1509 :    
1510 :     ----------------------------------------------------------------------
1511 :     Name: Matthias Blume
1512 : blume 1040 Date: 2002/01/28 12:00:00
1513 :     Tag: Release_110_38
1514 :     Description:
1515 :    
1516 :     This time for real!!!
1517 :    
1518 :     ----------------------------------------------------------------------
1519 :     Name: Matthias Blume
1520 : blume 1039 Date: 2002/01/28 10:56:00 EST
1521 :     Tag: blume-20020128-retraction
1522 :     Description:
1523 :    
1524 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
1525 :     with blume-Release_110_38-retracted.)
1526 :    
1527 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
1528 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1529 :    
1530 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1531 :     and enums fixed. The naming of corresponding ML identifiers should
1532 :     now be consistent again.
1533 :    
1534 :     ----------------------------------------------------------------------
1535 : leunga 1038 Name: Allen Leung
1536 :     Date: 2002/01/27
1537 :     Tag: leunga-20020127-nowhere
1538 :     Description:
1539 :    
1540 :     Added a target called nowhere in the configuration scripts.
1541 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
1542 :     programs with where-clauses into legal SML code) during installation.
1543 :    
1544 :     ----------------------------------------------------------------------
1545 : blume 1036 Name: Matthias Blume
1546 : leunga 1038 Date: 2002/01/25 21:27:00 EST
1547 : blume 1039 Tag: blume-Release_110_38-retracted
1548 : blume 1037 Description:
1549 :    
1550 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
1551 :    
1552 :     README will be added later.
1553 :    
1554 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
1555 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
1556 :    
1557 : blume 1037 ----------------------------------------------------------------------
1558 :     Name: Matthias Blume
1559 : blume 1036 Date: 2002/01/25
1560 :     Tag: blume-20020125-ffi
1561 :     Description:
1562 :    
1563 :     A large number of tweaks and improvements to ml-nlffi-lib and
1564 :     ml-nlffigen:
1565 :    
1566 :     - ML represenation types have been streamlined
1567 :     - getter and setter functions work with concrete values, not abstract
1568 :     ones where possible
1569 :     - ml-nlffigen command line more flexible (see README file there)
1570 :     - some bugs have been fixed (hopefully)
1571 :    
1572 :     ----------------------------------------------------------------------
1573 : george 1034 Name: Lal George
1574 :     Date: 2002/01/24
1575 :     Tag: george-20020124-risc-ra-interface
1576 :     Description:
1577 :    
1578 :     There is a dramatic simplification in the interface to the
1579 :     register allocator for RISC architectures as a result of making
1580 :     parallel copy instructions explicit.
1581 :    
1582 :     ----------------------------------------------------------------------
1583 : blume 1028 Name: Matthias Blume
1584 : blume 1032 Date: 2002/01/22
1585 :     Tag: blume-20020122-x86-ccalls
1586 :     Description:
1587 :    
1588 :     Bug fix for c-calls on x86 (having to do with how char- and
1589 :     short-arguments are being handled).
1590 :    
1591 :     ----------------------------------------------------------------------
1592 :     Name: Matthias Blume
1593 : blume 1031 Date: 2002/01/21
1594 :     Tag: blume-20020121-ff
1595 :     Description:
1596 :    
1597 :     Another day of fiddling with the FFI...
1598 :    
1599 :     1. Bug fix/workaround: CKIT does not complain about negative array
1600 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1601 :     (Otherwise a negative dimension would send it into an infinite loop.)
1602 :    
1603 :     2. Some of the abstract types (light objects, light pointers, most "base"
1604 :     types) in structure C are now eqtypes.
1605 :    
1606 :     3. Added constructors and test functions for NULL function pointers.
1607 :    
1608 :     ----------------------------------------------------------------------
1609 :     Name: Matthias Blume
1610 : blume 1028 Date: 2002/01/18
1611 : blume 1030 Tag: blume-20020118-ready-for-new-release
1612 :     Description:
1613 :    
1614 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1615 :     files shortly.)
1616 :    
1617 :     Maybe we christen this to be 110.38?
1618 :    
1619 :     ----------------------------------------------------------------------
1620 :     Name: Matthias Blume
1621 :     Date: 2002/01/18
1622 : blume 1029 Tag: blume-20020118-more-ffifiddle
1623 :     Description:
1624 :    
1625 :     Today's FFI fiddling:
1626 :    
1627 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1628 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1629 :     of abstract ones.
1630 :    
1631 :     - Provided word-style bit operations etc. for "int" representation
1632 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1633 :     and Long.
1634 :    
1635 :     ----------------------------------------------------------------------
1636 :     Name: Matthias Blume
1637 :     Date: 2002/01/18
1638 : blume 1028 Tag: blume-20020118-use-x86-fp
1639 :     Description:
1640 :    
1641 :     Now that x86-fast-fp seems to be working, I turned it back on again
1642 :     by default. (Seems to work fine now, even with the FFI.)
1643 :    
1644 :     Other than that, I added some documentation about the FFI to
1645 :     src/ml-nlffigen/README and updated the FFI test examples in
1646 :     src/ml-nlffi-lib/Tests/*.
1647 :    
1648 :     ----------------------------------------------------------------------
1649 : leunga 1025 Name: Allen Leung
1650 :     Date: 2002/01/17
1651 :     Tag: leunga-20020117-x86-fast-fp-call
1652 :     Description:
1653 :    
1654 :     1. Fixed a problem with handling return fp values when x86's fast fp
1655 :     mode is turned on.
1656 :    
1657 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1658 :     of %f32.
1659 :    
1660 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1661 :    
1662 :     ----------------------------------------------------------------------
1663 : blume 1021 Name: Matthias Blume
1664 :     Date: 2002/01/16
1665 :     Tag: blume-20020116-ffifiddle
1666 :     Description:
1667 :    
1668 :     More fiddling with the FFI interface:
1669 :    
1670 :     - Make constness 'c instead of rw wherever possible. This eliminates
1671 :     the need for certain explicit coercions. (However, due to ML's
1672 :     value polymorphism, there will still be many cases where explicit
1673 :     coercions are necessary. Phantom types are not the whole answer
1674 :     to modeling a subtyping relationship in ML.)
1675 :    
1676 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1677 :    
1678 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1679 :     clumsy workaround for varargs problem.)
1680 :    
1681 :     ----------------------------------------------------------------------
1682 : george 1016 Name: Lal George
1683 :     Date: 2002/01/15
1684 :     Tag: <none>
1685 :     Description:
1686 :    
1687 :     1. Since COPY instructions are no longer native to the architecture,
1688 :     a generic functor can be used to implement the expandCopies function.
1689 :    
1690 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1691 :     TEXT segment.
1692 :    
1693 :     ----------------------------------------------------------------------
1694 : blume 1011 Name: Matthias Blume
1695 : blume 1015 Date: 2002/01/15
1696 :     Tag: blume-20020115-ffiupdates
1697 :     Description:
1698 :    
1699 :     1. Fix for bug resulting in single-precision float values being returned
1700 :     incorrectly from FFI calls.
1701 :    
1702 :     2. Small modifications to C FFI API:
1703 :    
1704 :     - memory-allocation routines return straight objects (no options)
1705 :     and raise an exception in out-of-memory situations
1706 :     - unsafe extensions to cast between function pointers and pointers
1707 :     from/to ints
1708 :     - added structure C_Debug as an alternative to structure C where
1709 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1710 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1711 :     but also takes a (possibly empty) list of existing library handles
1712 :     that the current library depends on
1713 :    
1714 :     ----------------------------------------------------------------------
1715 :     Name: Matthias Blume
1716 : blume 1011 Date: 2002/01/10
1717 :     Tag: blume-20020110-newffigen
1718 :     Description:
1719 :    
1720 :     1. Updates to portable graph code.
1721 :    
1722 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1723 :     more scalable now so that even huge interfaces such as the one
1724 :     for GTK compile in finite time and space. :-)
1725 :     See src/ml-nlffigen/README for details on what's new.
1726 :    
1727 :     ----------------------------------------------------------------------
1728 : george 1003 Name: Lal George
1729 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1730 :     Tag: george-20011206-rm-native-copy
1731 :     Description:
1732 :    
1733 :     Removed the native COPY and FCOPY instructions
1734 :     from all the architectures and replaced it with the
1735 :     explicit COPY instruction from the previous commit.
1736 :    
1737 :     It is now possible to simplify many of the optimizations
1738 :     modules that manipulate copies. This has not been
1739 :     done in this change.
1740 :    
1741 :     ----------------------------------------------------------------------
1742 :     Name: Lal George
1743 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1744 :     Tag: george-20011206-mlrisc-instruction
1745 :     Description:
1746 : george 824
1747 : george 1003 Changed the representation of instructions from being fully abstract
1748 :     to being partially concrete. That is to say:
1749 :    
1750 :     from
1751 :     type instruction
1752 :    
1753 :     to
1754 :     type instr (* machine instruction *)
1755 :    
1756 :     datatype instruction =
1757 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1758 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1759 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1760 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1761 :     | INSTR of instr
1762 :    
1763 :     This makes the handling of certain special instructions that appear on
1764 :     all architectures easier and uniform.
1765 :    
1766 :     LIVE and KILL say that a list of registers are live or killed at the
1767 :     program point where they appear. No spill code is generated when an
1768 :     element of the 'regs' field is spilled, but the register is moved to
1769 :     the 'spilled' (which is present, more for debugging than anything else).
1770 :    
1771 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1772 :     We used to generate:
1773 :    
1774 :     DEFFREG f1
1775 :     f1 := f2 + f3
1776 :     trapb
1777 :    
1778 :     but now generate:
1779 :    
1780 :     f1 := f2 + f3
1781 :     trapb
1782 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1783 :    
1784 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1785 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1786 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1787 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1788 :     this architecture).
1789 :    
1790 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1791 :     all the architectures. This will result in further simplification of the
1792 :     register allocator that must be aware of them for coalescing purposes, and
1793 :     will also simplify certain aspects of the machine description that provides
1794 :     callbacks related to parallel copies.
1795 :    
1796 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1797 :     machine instruction set!
1798 :    
1799 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1800 :     functions for making porting easier -- essentially converting upper case
1801 :     to lower case. All machine instructions (of type instr) are in upper case,
1802 :     and the lower case form generates an MLRISC instruction. For example on
1803 :     the alpha we have:
1804 :    
1805 :     datatype instr =
1806 :     LDA of {r:cell, b:cell, d:operand}
1807 :     | ...
1808 :    
1809 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1810 :     ...
1811 :    
1812 :     where lda is just (INSTR o LDA), etc.
1813 :    
1814 : leunga 815 ----------------------------------------------------------------------
1815 : blume 986 Name: Matthias Blume
1816 : blume 993 Date: 2001/11/22 21:40:00 EST
1817 :     Tag: Release_110_37
1818 :     Description:
1819 :    
1820 :     Release 110.37. This time for real.
1821 :    
1822 :     ----------------------------------------------------------------------
1823 :     Name: Matthias Blume
1824 : blume 988 Date: 2001/11/21 16:35:00 EST
1825 :     Tag: blume-20011121-foot-in-mouth
1826 :     Description:
1827 :    
1828 :     Removed the "Release_110_37" tag because of a serious bug.
1829 :     This will be re-tagged once the bug is fixed.
1830 :    
1831 :     ----------------------------------------------------------------------
1832 :     Name: Matthias Blume
1833 : blume 987 Date: 2001/11/21 16:14:00 EST
1834 :     Tag: blume-20011121-forgottenfile
1835 :     Description:
1836 :    
1837 :     Forgot to add a file. (Just a .tex-file -- part of
1838 :     the CM manual source.)
1839 :    
1840 :     ----------------------------------------------------------------------
1841 :     Name: Matthias Blume
1842 : blume 986 Date: 2001/11/21 16:10:00 EST
1843 : blume 988 Tag: blume-20011121-invalid_110_37
1844 : blume 986 Description:
1845 :    
1846 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
1847 :     because we found a serious bug in all non-x86 backends.
1848 :     - Matthias
1849 :    
1850 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
1851 :     so that code object name strings are directly inserted into code
1852 :     objects at code generation time. The only business the runtime system
1853 :     has with this is now to read the name strings on occasions.
1854 :     (The encoding of the name string has also changed somewhat.)
1855 :    
1856 :     2. CM now implements a simple "set calculus" for specifying export lists.
1857 :     In particular, it is now possible to refer to the export lists of
1858 :     other libraries/groups/sources and form unions as well as differences.
1859 :     See the latest CM manual for details.
1860 :    
1861 :     3. An separate notion of "proxy" libraries has again be eliminated from
1862 :     CM's model. (Proxy libraries are now simply a special case of using
1863 :     the export list calculus.)
1864 :    
1865 :     4. Some of the existing libraries now take advantage of the new set
1866 :     calculus.
1867 :     (Notice that not all libraries have been converted because some
1868 :     of the existing .cm-files are supposed to be backward compatible
1869 :     with 110.0.x.)
1870 :    
1871 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
1872 :     "General.exnMessage"! The former relies on a certain hook to be
1873 :     initialized, and that often does not happen in the stand-alone case.)
1874 :    
1875 :     ----------------------------------------------------------------------
1876 : george 984 Name: Lal George
1877 :     Date: 2001/11/21 13:56:18 EST
1878 :     Tag: george-2001121-pseudo-ops
1879 :     Description:
1880 :    
1881 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
1882 :     ought to never be any question of incompatabilities with
1883 :     pseudo-op syntax expected by host assemblers.
1884 :    
1885 :     For now, only modules supporting GAS syntax are implemented
1886 :     but more should follow, such as MASM, and vendor assembler
1887 :     syntax, e.g. IBM as, Sun as, etc.
1888 :    
1889 :     ----------------------------------------------------------------------
1890 : blume 977 Name: Matthias Blume
1891 :     Date: 2001/11/14 11:52:00 EST
1892 :     Tag: blume-20011114-srcname
1893 :     Description:
1894 :    
1895 :     1. Routed the name of the current source file to mlriscgen where it
1896 :     should be directly emitted into the code object. (This last part
1897 :     is yet to be done.)
1898 :    
1899 :     2. Some cleanup of the pgraph code to make it match the proposal that
1900 :     I put out the other day. (The proposal notwithstanding, things are
1901 :     still in flux here.)
1902 :    
1903 :     ----------------------------------------------------------------------
1904 : george 976 Name: Lal George
1905 :     Date: 2001/11/14 09:44:04 EST
1906 :     Tag:
1907 :     Description:
1908 :    
1909 :     Fix for a backpatching bug reported by Allen.
1910 :    
1911 :     Because the boundary between short and long span-dependent
1912 :     instructions is +/- 128, there are an astounding number of
1913 :     span-dependent instructions whose size is over estimated.
1914 :    
1915 :     Allen came up with the idea of letting the size of span
1916 :     dependent instructions be non-monotonic, for a maxIter
1917 :     number of times, after which the size must be monotonically
1918 :     increasing.
1919 :    
1920 :     This table shows the number of span-dependent instructions
1921 :     whose size was over-estimated as a function of maxIter, for the
1922 :     file Parse/parse/ml.grm.sml:
1923 :    
1924 :     maxIter # of instructions:
1925 :     10 687
1926 :     20 438
1927 :     30 198
1928 :     40 0
1929 :    
1930 :     In compiling the compiler, there is no significant difference in
1931 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1932 :     my measurements showed that maxIter=40 was a tad faster than
1933 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1934 :     point within 13 iterations, so fixing maxIter at 40, while high,
1935 :     is okay.
1936 :    
1937 :     ----------------------------------------------------------------------
1938 : blume 975 Name: Matthias Blume
1939 :     Date: 2001/10/31 15:25:00 EST
1940 :     Tag: blume-20011031-pgraph
1941 :     Description:
1942 :    
1943 :     CKIT:
1944 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1945 :     argument identifiers.
1946 :     * Changed the return type of TypeUtil.getFunction accordingly.
1947 :     * Type equality ignores the argument names.
1948 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1949 :     if there is a mismatch.
1950 :    
1951 :     installation script:
1952 :     * attempts to use "curl" if available (unless "wget" is available as well)
1953 :    
1954 :     CM:
1955 :     * has an experimental implementation of "portable graphs" which I will
1956 :     soon propose as an implementation-independent library format
1957 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1958 :    
1959 :     NLFFI-LIB:
1960 :     * some cleanup (all cosmetic)
1961 :    
1962 :     NLFFIGEN:
1963 :     * temporarily disabled the mechanism that suppresses ML output for
1964 :     C definitions whose identifiers start with an underscore character
1965 :     * generate val bindings for enum constants
1966 :     * user can request that only one style (light or heavy) is being used;
1967 :     default is to use both (command-line arguments: -heavy and -light)
1968 :     * fixed bug in handling of function types involving incomplete pointers
1969 :     * generate ML entry points that take record arguments (i.e., using
1970 :     named arguments) for C functions that have a prototype with named
1971 :     arguments
1972 :     (see changes to CKIT)
1973 :    
1974 :     ----------------------------------------------------------------------
1975 : leunga 972 Name: Allen Leung
1976 :     Date: 2001/10/27 20:34:00 EDT
1977 :     Tag: leunga-20011027-x86-fast-fp-call
1978 :     Description:
1979 :    
1980 :     Fixed the bug described in blume-20010920-slowfp.
1981 :    
1982 :     The fix involves
1983 :     1. generating FCOPYs in FSTP in ia32-svid
1984 :     2. marking a CALL with the appropriate annotation
1985 :    
1986 :     ----------------------------------------------------------------------
1987 : blume 961 Name: Matthias Blume
1988 :     Date: 2001/10/16 11:32:00 EDT
1989 :     Tag: blume-20011016-netbsd
1990 :     Description:
1991 :    
1992 :     Underscore patch from Chris Richards (fixing problem with compiling
1993 :     runtime system under recent NetBSD).
1994 :    
1995 :     ----------------------------------------------------------------------
1996 : george 959 Name: Allen Leung
1997 :     Date: 2001/10/12 17:18:32 EDT 2001
1998 :     Tag: leung-20011012-x86-printflowgraph
1999 :     Description:
2000 :    
2001 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2002 :    
2003 :     ----------------------------------------------------------------------
2004 : george 952 Name: Lal George
2005 : george 958 Date: 2001/10/11 23:51:34 EDT
2006 :     Tag: george-20011011-too-many-instrs
2007 :     Description:
2008 :    
2009 :     The representation of a program point never expected to see more
2010 :     than 65536 instructions in a basic block!
2011 :    
2012 :     ----------------------------------------------------------------------
2013 :     Name: Lal George
2014 : george 952 Date: 2001/10/09 09:41:37 EDT
2015 :     Tag: george-20011008-mlrisc-labels
2016 :     Description:
2017 :    
2018 :     Changed the machine description files to support printing of
2019 :     local and global labels in assembly code, based on host assembler
2020 :     conventions.
2021 :    
2022 :     ----------------------------------------------------------------------
2023 : blume 916 Name: Matthias Blume
2024 : blume 939 Date: 2001/09/25 15:25:00 EDT
2025 :     Tag: blume-20010925-exninfo
2026 :     Description:
2027 :    
2028 :     I provided a non-hook implementation of exnName (at the toplevel) and
2029 :     made the "dummy" implementation of exnMessage (at the toplevel) more
2030 :     useful: if nothing gets "hooked in", then at least you are going to
2031 :     see the exception name and a message indicating why you don't see more.
2032 :    
2033 :     [For the time being, programs that need exnMessage and want to use
2034 :     ml-build should either use General.exnMessage (strongly recommended) or
2035 :     refer to structure General at some other point so that CM sees a
2036 :     static dependency.]
2037 :    
2038 :     [Similar remarks go for "print" and "use": If you want to use their
2039 :     functionality in stand-alone programs generated by ml-build, then use
2040 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2041 :    
2042 :     ----------------------------------------------------------------------
2043 :     Name: Matthias Blume
2044 : blume 938 Date: 2001/09/20 17:28:00 EDT
2045 :     Tag: blume-20010920-slowfp
2046 :     Description:
2047 :    
2048 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2049 :    
2050 :     ----------------------------------------------------------------------
2051 :     Name: Matthias Blume
2052 : blume 937 Date: 2001/09/20 17:20:00 EDT
2053 :     Tag: blume-20010920-canonicalpaths
2054 :     Description:
2055 :    
2056 :     0. Updated the BOOT file (something that I forgot to do earlier).
2057 :    
2058 :     1. Small internal change to CM so that it avoids "/../" in filenames
2059 :     as much as possible (but only where it is safe).
2060 :    
2061 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2062 :     that arguments that contain delimiters are passed through correctly.
2063 :     This change also means that all "special" arguments of the form
2064 :     @SMLxxx... must come first.
2065 :    
2066 :     3. Changed install script to put relative anchor names for tool commands
2067 :     into pathconfig.
2068 :    
2069 :     ----------------------------------------------------------------------
2070 :     Name: Matthias Blume
2071 : blume 932 Date: 2001/09/18 15:35:00 EDT
2072 :     Tag: blume-20010918-readme11036
2073 :     Description:
2074 :    
2075 :     Added README files.
2076 :    
2077 :     ----------------------------------------------------------------------
2078 :     Name: Matthias Blume
2079 : blume 929 Date: 2001/09/18 11:45:00 EDT
2080 :     Tag: Release_110_36 (retag)
2081 :     Description:
2082 :    
2083 :     Fixed mistake in config/preloads. Retagged as 110.36.
2084 :    
2085 :     ----------------------------------------------------------------------
2086 :     Name: Matthias Blume
2087 : blume 928 Date: 2001/09/18 09:40:00 EDT
2088 : blume 929 Tag: Release_110_36_orig (tag changed)
2089 : blume 928 Description:
2090 :    
2091 :     New version (110.36). New bootfiles.
2092 :    
2093 :     ----------------------------------------------------------------------
2094 :     Name: Matthias Blume
2095 : blume 927 Date: 2001/09/14 16:15:00 EDT
2096 :     Tag: blume-20010914-x86fastfp
2097 :     Description:
2098 :    
2099 :     John committed some changes that Allen made, in particular a (hopefully)
2100 :     correctly working version of the x86-fp module.
2101 :    
2102 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2103 :     flag to "true". Everything seems to compile to a fixpoint ok, and
2104 :     "mandelbrot" speeds up by about 15%.
2105 :    
2106 :     ----------------------------------------------------------------------
2107 :     Name: Matthias Blume
2108 : blume 923 Date: 2001/09/13 11:20:00 EDT
2109 :     Tag: blume-20010913-minimal
2110 :     Description:
2111 :    
2112 :     1. Stefan Monnier's patch to fix a miscompilation problem that
2113 :     was brought to light by John Reppy's work on Moby.
2114 :    
2115 :     2. Implemented a minimal "structure Compiler" that contains just
2116 :     "version" and "architecture". The minimal version will be
2117 :     available when the full version is not. This is for backward-
2118 :     compatibility with code that wants to test Compiler.version.
2119 :    
2120 :     ----------------------------------------------------------------------
2121 :     Name: Matthias Blume
2122 : blume 920 Date: 2001/08/28 14:03:00 EDT
2123 :     Tag: blume-20010828-ml-lex
2124 :     Description:
2125 :    
2126 :     Fix for bug 1581, received from Neophytos Michael.
2127 :    
2128 :     ----------------------------------------------------------------------
2129 :     Name: Matthias Blume
2130 : blume 918 Date: 2001/08/27 11:20:00 EDT
2131 :     Tag: blume-20010827-readme11035
2132 :     Description:
2133 :    
2134 :     Fleshed out the README file for 110.35.
2135 :    
2136 :     ----------------------------------------------------------------------
2137 :     Name: Matthias Blume
2138 : blume 916 Date: 2001/08/24 17:10:00 EDT
2139 :     Tag: Release_110_35
2140 :     Description:
2141 :    
2142 :     New version number (110.35). New bootfiles.
2143 :    
2144 :     ----------------------------------------------------------------------
2145 : george 915 Name: Lal George
2146 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
2147 :     Tag: george-20010824-MLRISC-graphs
2148 :     Description:
2149 :    
2150 :     removed clusters from MLRISC completely and replaced with graphs.
2151 :    
2152 :     ----------------------------------------------------------------------
2153 :     Name: Matthias Blume
2154 : blume 905 Date: 2001/08/23 17:50:00 EDT
2155 :     Tag: blume-20010823-toplevel
2156 :     Description:
2157 :    
2158 :     - some reorganization of the code that implements various kinds of
2159 :     environments in the compiler (static, dynamic, symbolic, combined)
2160 :     - re-implemented the EnvRef module so that evalStream works properly
2161 :     (if the stream contains references to "use", "CM.make", etc.)
2162 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2163 :    
2164 :     ----------------------------------------------------------------------
2165 :     Name: Matthias Blume
2166 : blume 904 Date: 2001/08/20 15:50 EDT
2167 :     Tag: blume20010820-slipup
2168 :     Description:
2169 :    
2170 :     I forgot to commit a few files. Here they are...
2171 :    
2172 :     ----------------------------------------------------------------------
2173 :     Name: Matthias Blume
2174 : blume 903 Date: 2001/08/20 15:35:00 EDT
2175 :     Tag: blume-20010820-debugprof
2176 :     Description:
2177 :    
2178 :     !!!! NEW BOOTFILES !!!!
2179 :    
2180 :     This is another round of reorganizing the compiler sources. This
2181 :     time the main goal was to factor out all the "instrumentation"
2182 :     passes (for profiling and backtracing) into their own library.
2183 :     The difficulty was to do it in such a way that it does not depend
2184 :     on elaborate.cm but only on elabdata.cm.
2185 :    
2186 :     Therefore there have been further changes to both elaborate.cm and
2187 :     elabdata.cm -- more "generic" things have been moved from the former
2188 :     to the latter. As a result, I was forced to split the assignment
2189 :     of numbers indicating "primtyc"s into two portions: SML-generic and
2190 :     SML/NJ-specific. Since it would have been awkward to maintain,
2191 :     I bit the bullet and actually _changed_ the mapping between these
2192 :     numbers and primtycs. The bottom line of this is that you need
2193 :     a new set of bin- and bootfiles.
2194 :    
2195 :     I have built new bootfiles for all architectures, so doing a fresh
2196 :     checkout and config/install.sh should be all you need.
2197 :    
2198 :     The newly created library's name is
2199 :    
2200 :     $smlnj/viscomp/debugprof.cm
2201 :    
2202 :     and its sources live under
2203 :    
2204 :     src/compiler/DebugProf
2205 :    
2206 :     ----------------------------------------------------------------------
2207 :     Name: Matthias Blume
2208 : blume 902 Date: 2001/08/15 17:15:00 EDT
2209 :     Tag: blume-20010815-compreorg
2210 :     Description:
2211 :    
2212 :     This is a first cut at reorganizing the CM libraries that make up the
2213 :     core of the compiler. The idea is to separate out pieces that could
2214 :     be used independently by tools, e.g., the parser, the typechecker, etc.
2215 :    
2216 :     The current status is a step in this direction, but it is not quite
2217 :     satisfactory yet. Expect more changes in the future.
2218 :    
2219 :     Here is the current (new) organization...
2220 :    
2221 :     What used to be $smlnj/viscomp/core.cm is now divided into
2222 :     six CM libraries:
2223 :    
2224 :     $smlnj/viscomp/basics.cm
2225 :     /parser.cm
2226 :     /elabdata.cm
2227 :     /elaborate.cm
2228 :     /execute.cm
2229 :     /core.cm
2230 :    
2231 :     The CM files for these libraries live under src/system/smlnj/viscomp.
2232 :     All these libraries are proxy libraries that contain precisely
2233 :     one CM library component. Here are the locations of the components
2234 :     (all within the src/compiler tree):
2235 :    
2236 :     Basics/basics.cm
2237 :     Parse/parser.cm
2238 :     ElabData/elabdata.cm
2239 :     Elaborator/elaborate.cm
2240 :     Execution/execute.cm
2241 :     core.cm
2242 :    
2243 :     [This organization is the same that has been used already
2244 :     for a while for the architecture-specific parts of the visible
2245 :     compiler and for the old version of core.cm.]
2246 :    
2247 :     As you will notice, many source files have been moved from their
2248 :     respective original locations to a new home in one of the above
2249 :     subtrees.
2250 :    
2251 :     The division of labor between the new libraries is the following:
2252 :    
2253 :     basics.cm:
2254 :     - Simple, basic definitions that pertain to many (or all) of
2255 :     the other libraries.
2256 :     parser.cm:
2257 :     - The SML parser, producing output of type Ast.dec.
2258 :     - The type family for Ast is also defined and exported here.
2259 :     elabdata.cm:
2260 :     - The datatypes that describe input and output of the elaborator.
2261 :     This includes types, absyn, and static environments.
2262 :     elaborator.cm:
2263 :     - The SML/NJ type checker and elaborator.
2264 :     This maps an Ast.dec (with a given static environment) to
2265 :     an Absyn.dec (with a new static environment).
2266 :     - This libraries implements certain modules that used to be
2267 :     structures as functors (to remove dependencies on FLINT).
2268 :     execute.cm:
2269 :     - Everything having to do with executing binary code objects.
2270 :     - Dynamic environments.
2271 :     core.cm:
2272 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
2273 :     - Top-level modules.
2274 :     - FLINT (this should eventually become its own library)
2275 :    
2276 :     Notes:
2277 :    
2278 :     I am not 100% happy with the way I separated the elaborator (and its
2279 :     data structures) from FLINT. Two instances of the same problem:
2280 :    
2281 :     1. Data structures contain certain fields that carry FLINT-specific
2282 :     information. I hacked around this using exn and the property list
2283 :     module from smlnj-lib. But the fact that there are middle-end
2284 :     specific fields around at all is a bit annoying.
2285 :    
2286 :     2. The elaborator calculates certain FLINT-related information. I tried
2287 :     to make this as abstract as I could using functorization, but, again,
2288 :     the fact that the elaborator has to perform calculations on behalf
2289 :     of the middle-end at all is not nice.
2290 :    
2291 :     3. Having to used exn and property lists is unfortunate because it
2292 :     weakens type checking. The other alternative (parameterizing
2293 :     nearly *everything*) is not appealing, though.
2294 :    
2295 :     I removed the "rebinding =" warning hack because due to the new organization
2296 :     it was awkward to maintain it. As a result, the compiler now issues some of
2297 :     these warnings when compiling init.cmi during bootstrap compilation. On
2298 :     the plus side, you also get a warning when you do, for example:
2299 :     val op = = Int32.+
2300 :     which was not the case up to now.
2301 :    
2302 :     I placed "assign" and "deref" into the _Core structure so that the
2303 :     code that deals with the "lazy" keyword can find them there. This
2304 :     removes the need for having access to the primitive environment
2305 :     during elaboration.
2306 :    
2307 :     ----------------------------------------------------------------------
2308 :     Name: Matthias Blume
2309 : blume 898 Date: 2001/08/13
2310 :     Tag: blume-20010813-closures
2311 :     Description:
2312 :    
2313 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
2314 :     performance of certain loops by avoiding needless closure allocation.
2315 :    
2316 :     ----------------------------------------------------------------------
2317 : george 890 Name: Lal George
2318 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
2319 :     Tag: george-20010731-x86-fmalloc
2320 :     Description: Fixed bug in x86 calls
2321 :    
2322 :     There was a bug where call instructions would mysteriously
2323 :     vanish. The call instruction had to be one that returned
2324 :     a floating point value.
2325 :    
2326 :     ----------------------------------------------------------------------
2327 :     Name: Lal George
2328 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
2329 :     Tag: george-20010719-simple-cells
2330 :     Description:
2331 :    
2332 :     I have dramatically simplified the interface for CELLS in MLRISC.
2333 :    
2334 :     In summary, the cells interface is broken up into three parts:
2335 :    
2336 :     1. CellsBasis : CELLS_BASIS
2337 :    
2338 :     CellsBasis is a top level structure and common for all
2339 :     architectures. it contains the definitions of basic datatypes
2340 :     and utility functions over these types.
2341 :    
2342 :     2. functor Cells() : CELLS
2343 :    
2344 :     Cells generates an interface for CELLS that incorporates the
2345 :     specific resources on the target architecture, such as the
2346 :     presence of special register classes, their number and size,
2347 :     and various useful substructures.
2348 :    
2349 :     3. <ARCH>CELLS
2350 :    
2351 :     e.g. SparcCells: SPARCCELLS
2352 :    
2353 :     <ARCH>CELLS usually contains additional bindings for special
2354 :     registers on the architecture, such as:
2355 :    
2356 :     val r0 : cell (* register zero *)
2357 :     val y : cell (* Y register *)
2358 :     val psr : cell (* processor status register *)
2359 :     ...
2360 :    
2361 :     The structure returned by applying the Cells functor is opened
2362 :     in this interface.
2363 :    
2364 :     The main implication of all this is that the datatypes for cells is
2365 :     split between CellsBasis and CELLS -- a fairly simple change for user
2366 :     code.
2367 :    
2368 :     In the old scheme the CELLS interface had a definitional binding of
2369 :     the form:
2370 :    
2371 :     signature CELLS = sig
2372 :    
2373 :     structure CellsBasis = CellsBasis
2374 :    
2375 :     ...
2376 :    
2377 :     end
2378 :    
2379 :     With all the sharing constraints that goes on in MLRISC, this old
2380 :     design quickly leads to errors such as:
2381 :    
2382 :     "structure definition spec inside of sharing ... "
2383 :    
2384 :    
2385 :     and appears to require an unacceptable amount of sharing and where
2386 :     constraint hackery.
2387 :    
2388 :     I think this error message (the interaction of definitional specs and
2389 :     sharing) requires more explanation on our web page.
2390 :    
2391 :     ----------------------------------------------------------------------
2392 : blume 826 Name: Matthias Blume
2393 : blume 879 Date: 2001/07/19 15:00:00 EDT
2394 :     Tag: blume-20010719-libreorg
2395 :     Description:
2396 :    
2397 :     This update puts together a fairly extensive but straightforward change
2398 :     to the way the libraries that implement the interactive system are
2399 :     organized:
2400 :    
2401 :     The biggest change is the elimination of structure Compiler. As a
2402 :     replacement for this structure, there is now a CM library
2403 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2404 :     that exports all the substructures of the original structure Compiler
2405 :     directly. So instead of saying Compiler.Foo.bar one now simply
2406 :     says Foo.bar. (The CM libraries actually export a collection of
2407 :     structures that is richer than the collection of substructures of
2408 :     structure Compiler.)
2409 :    
2410 :     To make the transition smooth, there is a separate library called
2411 :     $smlnj/compiler/compiler.cm which puts together and exports the
2412 :     original structure Compiler (or at least something very close to it).
2413 :    
2414 :     There are five members of the original structure Compiler
2415 :     that are not exported directly but which instead became members
2416 :     of a new structure Backend (described by signature BACKEND). These are:
2417 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2418 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
2419 :     architecture (: string).
2420 :    
2421 :     Structure Compiler.Version has become structure CompilerVersion.
2422 :    
2423 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2424 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2425 :     or x86, respectively.
2426 :     Each of these exports the same frontend structures that
2427 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
2428 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2429 :     Hppa, PPC, Sparc, or X86, respectively.
2430 :    
2431 :     Library $smlnj/compiler/all.cm exports the union of the exports of
2432 :     $smlnj/compiler/<arch>.cm
2433 :    
2434 :     There are no structures <Arch>Compiler anymore, use
2435 :     $smlnj/compiler/<arch>.cm instead.
2436 :    
2437 :     Library host-compiler-0.cm is gone. Instead, the internal library
2438 :     that instantiates CM is now called cm0.cm. Selection of the host
2439 :     compiler (backend) is no longer done here but. (Responsibility for it
2440 :     now lies with $smlnj/compiler/current.cm. This seems to be more
2441 :     logical.)
2442 :    
2443 :     Many individual files have been moved or renamed. Some files have
2444 :     been split into multiple files, and some "dead" files have been deleted.
2445 :    
2446 :     Aside from these changes to library organization, there are also changes
2447 :     to the way the code itself is organized:
2448 :    
2449 :     Structure Binfile has been re-implemented in such a way that it no
2450 :     longer needs any knowledge of the compiler. It exclusively deals
2451 :     with the details of binfile layout. It no longer invokes the
2452 :     compiler (for the purpose of creating new prospective binfile
2453 :     content), and it no longer has any knowledge of how to interpret
2454 :     pickles.
2455 :    
2456 :     Structure Compile (: COMPILE) has been stripped down to the bare
2457 :     essentials of compilation. It no longer deals with linking/execution.
2458 :     The interface has been cleaned up considerably.
2459 :    
2460 :     Utility routines for dealing with linking and execution have been
2461 :     moved into their own substructures.
2462 :    
2463 :     (The ultimate goal of these changes is to provide a light-weight
2464 :     binfile loader/linker (at least for, e.g., stable libraries) that
2465 :     does not require CM or the compiler to be present.)
2466 :    
2467 :     CM documentation has been updated to reflect the changes to library
2468 :     organization.
2469 :    
2470 :     ----------------------------------------------------------------------
2471 :     Name: Matthias Blume
2472 : blume 875 Date: 2001/07/10 17:30:00 EDT
2473 :     Tag: Release_110_34
2474 :     Description:
2475 :    
2476 : blume 876 Minor tweak to 110.34 (re-tagged):
2477 :    
2478 :     - README.html file added to CVS repository
2479 :     - runtime compiles properly under FreeBSD 3.X and 4.X
2480 :    
2481 :     ----------------------------------------------------------------------
2482 :     Name: Matthias Blume
2483 :     Date: 2001/07/10 17:30:00 EDT
2484 :     Tag: Release_110_34
2485 :     Description:
2486 :    
2487 : blume 875 New version number (110.34). New bootfiles.
2488 :    
2489 :     ----------------------------------------------------------------------
2490 :     Name: Matthias Blume
2491 : blume 874 Date: 2001/07/09 16:00:00 EDT
2492 :     Tag: blume-20010709-more-varargs
2493 :     Description:
2494 :    
2495 :     I changed the handling of varargs in ml-nlffigen again:
2496 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
2497 :    
2498 :     The immediate effect is that you can actually call a varargs function
2499 :     from ML -- but you can't actually supply any arguments beyond the ones
2500 :     specified explicitly. (For example, you can call printf with its format
2501 :     string, but you cannot pass additional arguments.)
2502 :    
2503 :     This behavior is only marginally more useful than the one before, but
2504 :     it has the advantage that a function or, more importantly, a function
2505 :     type never gets dropped on the floor, thus avoiding follow-up problems with
2506 :     other types that refer to the offending one.
2507 :    
2508 :     ----------------------------------------------------------------------
2509 :     Name: Matthias Blume
2510 : blume 873 Date: 2001/07/09 11:25:00 EDT
2511 :     Tag: blume-20010709-varargs
2512 :     Description:
2513 :    
2514 :     1. ckit-lib.cm now exports structure Error
2515 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2516 :     with a warning accompanied by a source location. Moreover, it
2517 :     merely skips the offending function or type and proceeds with the
2518 :     rest of its work.u As a result, one can safely feed C code containing
2519 :     "..." to ml-nlffigen.
2520 :     3. There are some internal improvements to CM, providing slightly
2521 :     more general string substitutions in the tools subsystem.
2522 :    
2523 :     ----------------------------------------------------------------------
2524 :     Name: Matthias Blume
2525 : blume 854 Date: 2001/06/27 15:10:00 EDT
2526 :     Tag: blume-20010627-concur
2527 :     Description:
2528 :    
2529 :     Fixed a small bug in CM's handling of parallel compilation.
2530 :     (You could observe the bug by Control-C-interrupting an ordinary
2531 :     CMB.make or CM.stabilize and then attaching some compile servers.
2532 :     The result was that all of a sudden the previously interrupted
2533 :     compilation would continue on its own. This was because of
2534 :     an over-optimization: CM did not bother to clean out certain queues
2535 :     when no servers were attached "anyway", resulting in the contents
2536 :     of these queues to grab control when new servers did get attached.)
2537 :    
2538 :     There is also another minor update to the CM manual.
2539 :    
2540 :     ----------------------------------------------------------------------
2541 :     Name: Matthias Blume
2542 : blume 853 Date: 2001/06/26 16:15:00 EDT
2543 :     Tag: blume-20010626-cmdoc
2544 :     Description:
2545 :    
2546 :     Minor typo fixed in CM manual (syntax diagram for libraries).
2547 :    
2548 :     ----------------------------------------------------------------------
2549 :     Name: Matthias Blume
2550 : blume 852 Date: 2001/06/25 22:55:00 EDT
2551 :     Tag: blume-20010625-x86pc
2552 :     Description:
2553 :    
2554 :     Fixed a nasty bug in the X86 assembly code that caused signal
2555 :     handlers to fail (crash) randomly.
2556 :    
2557 :     ----------------------------------------------------------------------
2558 :     Name: Matthias Blume
2559 : blume 846 Date: 2001/06/25 12:05:00 EDT
2560 :     Tag: blume-20010625-nlffigen
2561 :     Description:
2562 :    
2563 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
2564 :     Nick Carter <nbc@andrew.cmu.edu>.
2565 :    
2566 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2567 :     2. Default names for generated files are now derived from the name of
2568 :     the C file *without its directory*. In particular, this causes generated
2569 :     files to be placed locally even if the C file is in some system directory.
2570 :     3. Default names for generated signatures and structures are also derived
2571 :     from the C file name without its directory. This avoids silly things
2572 :     like "structure GL/GL".
2573 :     (Other silly names are still possible because ml-nlffigen does not do
2574 : blume 847 a thorough check of whether generated names are legal ML identifiers.
2575 :     When in doubt, use command line arguments to force particular names.)
2576 : blume 846
2577 :     ----------------------------------------------------------------------
2578 :     Name: Matthias Blume
2579 : blume 845 Date: 2001/06/21 12:25:00 EDT
2580 :     Tag: blume-20010621-eXene
2581 :     Description:
2582 :    
2583 :     eXene now compiles and (sort of) works again.
2584 :    
2585 :     The library name (for version > 110.33) is $/eXene.cm.
2586 :    
2587 :     I also added an new example in src/eXene/examples/nbody. See the
2588 :     README file there for details.
2589 :    
2590 :     ----------------------------------------------------------------------
2591 :     Name: Matthias Blume
2592 : blume 844 Date: 2001/06/20 16:40:00 EDT
2593 :     Tag: blume-20010620-cml
2594 :     Description:
2595 :    
2596 :     CML now compiles and works again.
2597 :    
2598 :     Libraries (for version > 110.33):
2599 :    
2600 :     $cml/cml.cm Main CML library.
2601 :     $cml/basis.cm CML's version of $/basis.cm.
2602 :     $cml/cml-internal.cm Internal helper library.
2603 :     $cml/core-cml.cm Internal helper library.
2604 :     $cml-lib/trace-cml.cm Tracing facility.
2605 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2606 :    
2607 :     The installer (config/install.sh) has been taught how to properly
2608 :     install this stuff.
2609 :    
2610 :     ----------------------------------------------------------------------
2611 :     Name: Matthias Blume
2612 : blume 843 Date: 2001/06/19 17:55:00 EDT
2613 :     Tag: blume-20010619-instantiate
2614 :     Description:
2615 :    
2616 :     This un-breaks the fix for bug 1432.
2617 :     (The bug was originally fixed in 110.9 but I broke it again some
2618 :     time after that.)
2619 :    
2620 :     ----------------------------------------------------------------------
2621 :     Name: Matthias Blume
2622 : blume 842 Date: 2001/06/19 17:25:00 EDT
2623 :     Tag: blume-20010619-signals
2624 :     Description:
2625 :    
2626 :     This should (hopefully) fix the long-standing signal handling bug.
2627 :     (The runtime system was constructing a continuation record with an
2628 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2629 :    
2630 :     ----------------------------------------------------------------------
2631 :     Name: Matthias Blume
2632 : blume 840 Date: 2001/06/15 15:05:00 EDT
2633 : blume 841 Tag: blume-20010615-moresparc
2634 :     Description:
2635 :    
2636 :     Here is a short late-hour update related to Sparc c-calls:
2637 :    
2638 :     -- made handling of double-word arguments a bit smarter
2639 :    
2640 :     -- instruction selection phase tries to collapse certain clumsily
2641 :     constructed ML-Trees; typical example:
2642 :    
2643 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2644 :    
2645 :     This currently has no further impact on SML/NJ since mlriscGen does
2646 :     not seem to generate such patterns in the first place, and c-calls
2647 :     (which did generate them in the beginning) has meanwhile been fixed
2648 :     so as to avoid them as well.
2649 :    
2650 :     ----------------------------------------------------------------------
2651 :     Name: Matthias Blume
2652 :     Date: 2001/06/15 15:05:00 EDT
2653 : blume 840 Tag: blume-20010615-sparc
2654 :     Description:
2655 :    
2656 :     The purpose of this update is to provide an implementation of NLFFI
2657 :     on Sparc machines.
2658 :    
2659 :     Here are the changes in detail:
2660 :    
2661 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2662 :     the Sparc implementation of the c-calls API.
2663 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2664 :     for accessing the ML frame. Thus, we have a real frame pointer and
2665 :     can freely modify %sp without need for an omit-frame-ptr phase.
2666 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2667 :     this case.
2668 :     * ml-nlffigen has been taught to produce code for different architectures
2669 :     and calling conventions.
2670 :     * In a way similar to what was done in the x86 case, the Sparc
2671 :     backend uses its own specific extension to mltree. (For example,
2672 :     it needs to be able to generate UNIMP instructions which are part
2673 :     of the calling convention.)
2674 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2675 :     to make it easier to plug in new machine- and os-dependent parts).
2676 :    
2677 :     There are some other fairly unrelated bug fixes and cleanups as well:
2678 :    
2679 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2680 :     that they properly share their libraries with existing SML/NJ libraries.
2681 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2682 :     follow-up error messages.
2683 :     * Updates to CM/CMB documentation.
2684 :    
2685 :     TODO items:
2686 :    
2687 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2688 :     (The current %o2 is a really bad choice because it is part of the
2689 :     calling conventions, so things might interfere in unexpected ways.)
2690 :    
2691 :     ----------------------------------------------------------------------
2692 :     Name: Matthias Blume
2693 : blume 839 Date: 2001/06/07
2694 :     Tag: blume-20010607-calls
2695 :     Description:
2696 :    
2697 :     A number of internal changes related to C calls and calling conventions:
2698 :    
2699 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2700 :     number of bytes popped implicitly (by the callee). In most cases
2701 :     this field is 0 but on x86/win32 it is some non-zero value. This
2702 :     is information provided for the benefit of the "omit-frameptr" pass.
2703 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2704 :     The instruction selection phase copies its value from the ML-Tree
2705 :     CALL statement.
2706 :     3. On all other architectures, the instruction selection phase checks
2707 :     whether "pops=0" and complains if not.
2708 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2709 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2710 :     up after the call and pops is set to 0. For "stdcall", the caller
2711 :     does nothing, leaving the cleanup to the callee; pops is set to
2712 :     the number of bytes that were pushed onto the stack.
2713 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2714 :     distinguish between "ccall" and "stdcall".
2715 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2716 :     set. (This is needed for implementing the official C calling convention
2717 :     on this architecture.)
2718 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2719 :     work with the latest CM.
2720 :    
2721 :     ----------------------------------------------------------------------
2722 :     Name: Matthias Blume
2723 : blume 838 Date: 2001/06/05 15:10:00 EDT
2724 :     Tag: blume-20010605-cm-index
2725 :     Description:
2726 :    
2727 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2728 :    
2729 :     1. CM can now generate "index files". These are human-readable files
2730 :     that list on a per-.cm-file basis each toplevel symbol defined or
2731 :     imported. The location of the index file for
2732 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2733 :     To enable index-file generation, set CM.Control.generate_index to true
2734 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2735 :    
2736 :     The CM manual has been updated accordingly.
2737 :    
2738 :     2. I made some slight modifications to the c-calls API in MLRISC.
2739 :    
2740 :     a) There is now a callback to support saving/restoring of
2741 :     dedicated but caller-save registers around the actual call
2742 :     instruction.
2743 :     b) One can optionally specify a comment-annotation for the
2744 :     call instruction.
2745 :    
2746 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2747 :     (For example, the comment annotation shows the C prototype of
2748 :     the function being called.)
2749 :    
2750 :     ----------------------------------------------------------------------
2751 :     Name: Matthias Blume
2752 : blume 837 Date: 2001/06/01 13:30:00 EDT
2753 :     Tag: blume-20010601-nlffi-cleanup
2754 :     Description:
2755 :    
2756 :     This is mostly a cleanup of MLFFI stuff:
2757 :    
2758 :     - some signature files have been put into a more exposed place
2759 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2760 :     - ml-nlffigen changed accordingly
2761 :     - tutorial updated
2762 :    
2763 :     Other changes:
2764 :    
2765 :     - author's affiliation in CM manual(s) updated
2766 :     - some more recognized keywords added to Allen's sml.sty
2767 :    
2768 :     ----------------------------------------------------------------------
2769 :     Name: Matthias Blume
2770 : blume 836 Date: 2001/05/25 15:30:00 EDT
2771 :     Tag: blume-20010525-iptr
2772 :     Description:
2773 :    
2774 :     - put the official 110.33-README (as it appears on the ftp server) under
2775 :     CVS
2776 :     - fixed a small bug related to incomplete pointer types in
2777 :     ml-nlffigen
2778 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2779 :     (it does not need the 'f type parameter)
2780 :    
2781 :     ----------------------------------------------------------------------
2782 :     Name: Matthias Blume
2783 : blume 834 Date: 2001/05/23 14:30:00 EDT
2784 :     Tag: Release_110_33
2785 :     Description:
2786 :    
2787 :     New version number (110.33). New bootfiles.
2788 :    
2789 :     ----------------------------------------------------------------------
2790 :     Name: Matthias Blume
2791 : blume 833 Date: 2001/05/22 18:06:00 EDT
2792 :     Tag: blume-20010522-targets
2793 :     Description:
2794 :    
2795 :     Made install.sh use file config/targets.customized if it exists, falling
2796 :     back to config/targets if it doesn't. This way one can have a customized
2797 :     version of the targets file without touching the "real thing", thus
2798 :     eliminating the constant fear of accidentally checking something bogus
2799 :     back into the CVS repository... (File config/targets.customized must
2800 :     not be added to the repository!)
2801 :    
2802 :     ----------------------------------------------------------------------
2803 :     Name: Matthias Blume
2804 : blume 832 Date: 2001/05/22 16:30:00 EDT
2805 :     Tag: blume-20010522-minitut
2806 :     Description:
2807 :    
2808 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2809 :     struct returns.
2810 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2811 :     incomplete, preliminary documentation for NLFFI.
2812 :    
2813 :     ----------------------------------------------------------------------
2814 :     Name: Matthias Blume
2815 : blume 829 Date: 2001/05/14 11:30:00 EDT
2816 :     Tag: blume-20010514-script
2817 :     Description:
2818 :    
2819 :     Some bugs in install script fixed.
2820 :    
2821 :     In addition to that I also made a slight change to the NLFFI API:
2822 :     Functors generated by ml-nlffigen now take the dynamic library as a
2823 :     straight functor argument, not as a suspended one. (The original
2824 :     functor code used to force the suspension right away anyway, so there
2825 :     was nothing gained by this complication of the interface.)
2826 :    
2827 :     ----------------------------------------------------------------------
2828 :     Name: Matthias Blume
2829 : blume 828 Date: 2001/05/11 14:35:00 EDT
2830 :     Tag: blume-20010511-ml-nlffi
2831 :     Description:
2832 :    
2833 :     I finally took the plunge and added my new FFI code to the main
2834 :     repository. For x86-linux it is now ready for prime-time.
2835 :    
2836 :     There are two new subdirectories of "src":
2837 :    
2838 :     - ml-nlffi-lib:
2839 :     The utility library for programs using the FFI interface.
2840 :     Here is the implementation of $/c.cm and its associated low-level
2841 :     partners $/c-int.cm and $/memory.cm.
2842 :     - ml-nlffigen:
2843 :     A stand-alone program for generating ML glue code from C source
2844 :     code.
2845 :    
2846 :     Building ml-nlffigen requires $/ckit-lib.cm.
2847 :    
2848 :     The config/install.sh script has been updates to do the Right Thing
2849 :     (hopefully).
2850 :    
2851 :     Notice that the source tree for the C-Kit will not be put under "src"
2852 :     but directly under the installation root directory. (This is the
2853 :     structure that currently exists on the CVS server when you check out
2854 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
2855 :    
2856 :     Bugs: No documentation yet.
2857 :    
2858 :     ----------------------------------------------------------------------
2859 :     Name: Matthias Blume
2860 : blume 826 Date: 2001/05/09 16:35:00 EDT
2861 :     Tag: blume-20010509-cpscontract
2862 :     Description:
2863 :    
2864 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
2865 :     wrapper/unwrapper elimination did not decrement usage counts and some
2866 :     dead variables got overlooked by the dead-up logic.)
2867 :    
2868 :     ----------------------------------------------------------------------
2869 : george 824 Name: Lal George
2870 :     Date: 2001/05/08 17:26:09 EDT
2871 :     Tag: george-20010508-omit-frameptr
2872 :     Description:
2873 :    
2874 :     Changes to implement the omit-frame-pointer optimization to support
2875 :     raw C calls. For now, there is only support on the Intel x86, but
2876 :     other architectures will follow as more experience is gained with this.
2877 :    
2878 :    
2879 :     ----------------------------------------------------------------------
2880 : blume 818 Name: Matthias Blume
2881 : blume 819 Date: 2001/05/07 14:40:00 EDT
2882 :     Tag: blume-20010507-proxies
2883 :     Description:
2884 :    
2885 :     I made into "proxy libraries" all libraries that qualify for such a
2886 :     change. (A qualifying library is a library that has another library or
2887 :     groups as its sole member and repeats that member's export list
2888 :     verbatim. A proxy library avoids this repetition by omitting its export
2889 :     list, effectively inheriting the list that its (only) member exports.
2890 :     See the CM manual for more explanation.)
2891 :     The main effect is that explicit export lists for these libraries
2892 :     do not have to be kepts in sync, making maintenance a bit easier.
2893 :    
2894 :     I also added copyright notices to many .cm-files.
2895 :    
2896 :     Last but not least, I made a new set of bootfiles.
2897 :    
2898 :     ----------------------------------------------------------------------
2899 :     Name: Matthias Blume
2900 : blume 818 Date: 2001/05/04 17:00:00 EDT
2901 :     Tag: blume-20010504-cm-lsplit
2902 :     Description:
2903 :    
2904 :     0. John merged pending changes to $/smlnj-lib.cm
2905 :    
2906 :     1. Allen's previous change accidentally backed out of one of Lal's
2907 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2908 :    
2909 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2910 :     within the compiler where applicable. There is some code simplification
2911 :     because of that.
2912 :    
2913 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2914 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2915 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2916 :     it can now also be controlled on a per-source basis: CM has been taught
2917 :     a new tool parameter applicable to ML source files.
2918 :    
2919 :     - To turn lambda-splitting off completely:
2920 :     local open Compiler.Control.LambdaSplitting in
2921 :     val _ = set Off
2922 :     end
2923 :     - To make "no lambda-splitting" the global default (but allow per-source
2924 :     overriding); this is the initial setting:
2925 :     local open Compiler.Control.LambdaSplitting in
2926 :     val _ = set (Default NONE)
2927 :     end
2928 :     - To make "lambda-splitting with aggressiveness a" the global default
2929 :     (and allow per-source overriding):
2930 :     local open Compiler.Control.LambdaSplitting in
2931 :     val _ = set (Default (SOME a))
2932 :     end
2933 :    
2934 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2935 :     write (in the respective .cm-file):
2936 :     a.sml (lambdasplitting:off)
2937 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2938 :     a.sml (lambdasplitting:on)
2939 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2940 :     <a> is a decimal non-negative integer):
2941 :     a.sml (lambdasplitting:<a>)
2942 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2943 :     a.sml (lambdasplitting:infinity)
2944 :     - To use the global default for a.sml:
2945 :     a.sml (lambdasplitting:default)
2946 :     or simply
2947 :     a.sml
2948 :    
2949 :     ----------------------------------------------------------------------
2950 : leunga 815 Name: Allen Leung
2951 :     Date: 2001/05/04 01:57:00 EDT
2952 :     Tag: leunga-20010504-sync
2953 :     Description:
2954 : leunga 788
2955 : leunga 815 MLRISC features.
2956 :    
2957 :     1. Fix to CMPXCHG instructions.
2958 :     2. Changed RA interface to allow annotations in callbacks.
2959 :     3. Added a new method to the stream interface to allow annotations updates.
2960 :    
2961 : george 749 ----------------------------------------------------------------------
2962 : blume 812 Name: Matthias Blume
2963 : blume 814 Date: 2001/05/01 11:45:00 EDT
2964 :     Tag: blume-20010501-pcedittmp
2965 :     Description:
2966 :    
2967 :     Changed install.sh to use the current working directory instead of
2968 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2969 :     of /usr/tmp caused trouble with MacOS X because of file premission
2970 :     problems.
2971 :    
2972 :     ----------------------------------------------------------------------
2973 :     Name: Matthias Blume
2974 : blume 812 Date: 2001/04/20 11:10:00 EDT
2975 :     Tag: blume-20010420-inMLflag
2976 :     Description:
2977 :    
2978 :     - added vp_limitPtrMask to vproc-state.h
2979 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2980 :     handling)
2981 :     - made the ML compiler aware of various data-structure offsets so it
2982 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2983 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2984 :     raw C-calls
2985 :    
2986 :     ----------------------------------------------------------------------
2987 : george 810 Name: Lal George
2988 :     Date: 2001/04/20 09:15:28 EDT
2989 :     Tag: george-20010420-macosX
2990 :     Description:
2991 :    
2992 :     - Changes to port to Mac OS X; Darwin.
2993 :    
2994 :     - In the process I found that sqrt was broken on the PPC, because the
2995 :     fsqrt instruction is not implemented.
2996 :    
2997 :     ----------------------------------------------------------------------
2998 : blume 799 Name: Matthias Blume
2999 : blume 808 Date: 2001/04/18 12:45:00 EDT
3000 :     Tag: blume-20010418-ccalls
3001 :     Description:
3002 :    
3003 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
3004 :     (this bug prevented structure arguments containing pointers from being
3005 :     passed correctly)
3006 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
3007 :     structure arguments are represented as a pointer to the beginning
3008 :     of the structure (instead of having a series of synthesized arguments,
3009 :     one for each structure member)
3010 :    
3011 :     - made makeml script's verbosity level configurable via environment
3012 :     variable (MAKEML_VERBOSITY)
3013 :    
3014 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3015 : blume 808 in rawmem-x86.sml; we are now using the real thing
3016 :    
3017 :     ----------------------------------------------------------------------
3018 :     Name: Matthias Blume
3019 : blume 806 Date: 2001/03/22 16:25:00 EST
3020 :     Tag: blume-20010322-bootfiles
3021 :     Description:
3022 :    
3023 :     Created a new set of bootfiles (for your automatic installation convenience).
3024 :    
3025 :     ----------------------------------------------------------------------
3026 :     Name: Matthias Blume
3027 : blume 805 Date: 2001/03/22 15:10:00 EST
3028 :     Tag: blume-20010322-rawmem-parcm
3029 :     Description:
3030 :    
3031 :     1. All "raw memory access" primitives for the new FFI are implemented now
3032 :     (at least on the x86).
3033 :     2. Some further cleanup of CM's parallel make mechanism.
3034 :    
3035 :     ----------------------------------------------------------------------
3036 :     Name: Matthias Blume
3037 : blume 801 Date: 2001/03/19 17:53:00 EST
3038 :     Tag: blume-20010319-parallel
3039 :     Description:
3040 :    
3041 :     Parallel make (using compile servers) now works again.
3042 :    
3043 :     To this end, CM.stabilize and CMB.make have been modified to work in
3044 :     two passes when compile servers are attached:
3045 :     1. Compile everything, do not perform stabilization; this pass
3046 :     uses compile servers
3047 :     2. Stabilize everything; this pass does not use compile servers
3048 :     If there are no compile servers, the two passes are combined into one
3049 :     (as before). Splitting the passes increases the inherent parallelism
3050 :     in the dependency graph because the entire graph including all
3051 :     libraries is available at the same time. This, in turn, improves
3052 :     server utilization. The downside is that the master process will
3053 :     have to do some extra work after compilation is done (because for
3054 :     technical reasons it must re-read all the binfiles during stabilization).
3055 :    
3056 :     ----------------------------------------------------------------------
3057 :     Name: Matthias Blume
3058 : blume 800 Date: 2001/03/16 12:22:00 EST
3059 :     Tag: blume-20010316-bootfiles
3060 :     Description:
3061 :    
3062 :     Created a new set of bootfiles (for your automatic installation convenience).
3063 :    
3064 :     ----------------------------------------------------------------------
3065 :     Name: Matthias Blume
3066 : blume 799 Date: 2001/03/16 11:00:00 EST
3067 :     Tag: blume-20010316-MLTREE-fixup
3068 :     Description:
3069 :    
3070 :     This is a minor fixup for an (untagged) earlier commit by Allen.
3071 :     (A file was missing).
3072 :    
3073 :     ----------------------------------------------------------------------
3074 : leunga 788 Name: Allen Leung
3075 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
3076 :     Tag: leunga-20010305-cut-support
3077 :    
3078 :     1. New support for alternative control-flow in MLTREE.
3079 :     Currently we support
3080 :    
3081 :     FLOW_TO(CALL ...., [k1,...,kn])
3082 :    
3083 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
3084 :     in Moby
3085 :    
3086 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3087 :    
3088 :     2. Register Allocator
3089 :    
3090 :     Changes in interface [from Fermin, John]
3091 :    
3092 :     3. Alpha 8-bit SLL support [Fermin]
3093 :    
3094 :     4. All architectures
3095 :    
3096 :     A new module (ClusterExpandCopies) for expanding parallel copies.
3097 :    
3098 :     ----------------------------------------------------------------------
3099 :     Name: Allen Leung
3100 : leunga 788 Date: 2001/02/27 23:07:00 EST
3101 :     Tag: leunga-20010227-minor-stuff
3102 :    
3103 :     1. Alpha bug fix for CMOVNE
3104 :     2. Handle mltree COND(..,FCMP ...,...)
3105 :     3. Bug fix in simplifier
3106 :    
3107 :     ----------------------------------------------------------------------
3108 : blume 777 Name: Matthias Blume
3109 : blume 781 Date: 2001/01/30 17:50:00 EST
3110 :     Tag: blume-20010130-sync
3111 :     Description:
3112 :    
3113 :     This is just a minor update to sync my devel branch with the main brach.
3114 :     The only visible change is the addition of some README files.
3115 :    
3116 :     ----------------------------------------------------------------------
3117 :     Name: Matthias Blume
3118 : blume 778 Date: 2001/01/12 23:30:00 JST
3119 :     Tag: blume-20010112-bootfiles
3120 :     Description:
3121 :    
3122 :     Made a new set of bootfiles that goes with the current state of the
3123 :     repository.
3124 :    
3125 :     ----------------------------------------------------------------------
3126 :     Name: Matthias Blume
3127 : blume 777 Date: 2001/01/12 21:20:00 JST
3128 :     Tag: blume-20010112-sync
3129 :     Description:
3130 :    
3131 :     I am just flushing out some minor changes that had accumulated in
3132 :     my private branch in order to sync with the main tree. (This is
3133 :     mainly because I had CVS trouble when trying to merge _into_ my
3134 :     private branch.)
3135 :    
3136 :     Most people should be completely unaffected by this.
3137 :    
3138 :     ----------------------------------------------------------------------
3139 : leunga 775 Name: Allen Leung
3140 :     Date: Thu Jan 11 21:03:00 EST 2001
3141 :     Tag: leunga-20010111-labexp=mltree
3142 :     Description:
3143 :    
3144 :     1. Removed the type LabelExp and replace it by MLTree.
3145 :     2. Rewritten mltree-simplify with the pattern matcher tool.
3146 :     3. There were some bugs in alpha code generator which would break
3147 :     64-bit code generation.
3148 :     4. Redo the tools to generate code with the
3149 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3150 :     are now generated by perl scripts.
3151 :    
3152 :     ----------------------------------------------------------------------
3153 : blume 771 Name: Matthias Blume
3154 : blume 774 Date: 2001/01/10 21:55:00 JST
3155 :     Tag: blume-20010110-rcc
3156 :     Description:
3157 :    
3158 :     The RCC stuff now seems to work (but only on the x86).
3159 :     This required hacking of the c-calls interface (and -implementation) in
3160 :     MLRISC.
3161 :    
3162 :     Normal compiler users should be unaffected.
3163 :    
3164 :     ----------------------------------------------------------------------
3165 :     Name: Matthias Blume
3166 : blume 773 Date: 2001/01/09 01:20:00 JST
3167 :     Tag: blume-20010109-rcc
3168 :     Description:
3169 :    
3170 :     This is a fairly big patch, flushing out a large number of pending
3171 :     changes that I made to my development copy over the last couple of days.
3172 :    
3173 :     Of practical relevance at this moment is a workaround for a pickling
3174 :     bug that Allen ran into the other day. The cause of the bug itself is
3175 :     still unknown and it might be hard to fix it properly, but the
3176 :     workaround has some merits of its own (namely somewhat reducing pickling
3177 :     overhead for certain libraries). Therefore, I think this solution should
3178 :     be satisfactory at this time.
3179 :    
3180 :     The rest of the changes (i.e., the vast majority) has to do with my
3181 :     ongoing efforts of providing direct support for C function calls from
3182 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
3183 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3184 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3185 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3186 :     handle RCC.
3187 :    
3188 :     The last part (the changes to mlriscGen.sml) are still known to be
3189 :     wrong on the x86 and not implemented on all other architectures. But
3190 :     the infrastructure is in place. I had to change a few functor
3191 :     signatures in the backend to be able to route the CCalls interface
3192 :     from MLRISC there, and I had to specialize the mltree type (on the
3193 :     x86) to include the necessary extensions. (The extensions themselves
3194 :     were already there and redy to go in MLRISC/x86).
3195 :    
3196 :     Everything should be very happy as soon as someone helps me with
3197 :     mlriscGen.sml...
3198 :    
3199 :     In any case, nothing of this should matter to anyone as long as the
3200 :     new primop is not being used (which is going to be the case unless you
3201 :     find it where I hid it :). The rest of the compiler is completely
3202 :     unaffected.
3203 :    
3204 :     ----------------------------------------------------------------------
3205 :     Name: Matthias Blume
3206 : blume 772 Date: 2001/01/05 00:30:00 JST
3207 :     Tag: blume-20010105-primops
3208 :     Description:
3209 :    
3210 :     Added some experimental support for work that I am doing right now.
3211 :     These changes mostly concern added primops, but there is also a new
3212 :     experimental C library in the runtime system (but currently not enabled
3213 :     anywhere except on Linux/X86).
3214 :    
3215 :     In the course of adding primops (and playing with them), I discovered that
3216 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3217 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
3218 :     To recover, I made type information in INL_PRIM mandatory and changed
3219 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
3220 :     has complete, correct type information (i.e., no bottom types).
3221 :    
3222 :     Since all these changes mean that we need new binfiles, I also bumped the
3223 :     version number to 110.32.1.
3224 :    
3225 :     ----------------------------------------------------------------------
3226 :     Name: Matthias Blume
3227 : blume 771 Date: 2000/12/30 22:10:00 JST
3228 :     Tag: blume-20001230-various
3229 :     Description:
3230 :    
3231 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
3232 :     to each other using path anchors. (See CM manual for explanation.)
3233 :    
3234 :     Updated CM documentation.
3235 :    
3236 :     Fixed some bugs in CM.
3237 :    
3238 :     Implemented "proxy" libraries (= syntactic sugar for CM).
3239 :    
3240 :     Added "-quiet" option to makeml and changed runtime system accordingly.
3241 :    
3242 :     Added cleanup handler for exportML to reset timers and compiler stats.
3243 :    
3244 :     ----------------------------------------------------------------------
3245 : george 761 Name: Lal George
3246 :     Date: 2000/12/22 22:22:58 EST 2000
3247 :     Tag: Release_110_32
3248 :     Description:
3249 :    
3250 :     Infinite precision used throughout MLRISC.
3251 :     see MLRISC/mltree/machine-int.sig
3252 :    
3253 :     ----------------------------------------------------------------------
3254 : blume 756 Name: Matthias Blume
3255 : blume 760 Date: 2000/12/22 23:16:00 JST
3256 :     Tag: blume-20001222-warn
3257 :     Description:
3258 : blume 759
3259 : blume 760 Corrected wording and formatting of some CM warning message which I
3260 :     broke in my previous patch.
3261 : blume 759
3262 :     ----------------------------------------------------------------------
3263 :     Name: Matthias Blume
3264 : blume 763 Date: 2000/12/22 21:20:00 JST
3265 :     Tag: blume-20001222-anchorenv
3266 :     Description:
3267 :    
3268 :     Fixed CM's handling of anchor environments in connection with CMB.make.
3269 :    
3270 :     ----------------------------------------------------------------------
3271 :     Name: Matthias Blume
3272 : blume 758 Date: 2000/12/22 13:15:00 JST
3273 :     Tag: blume-20001222-cleanup
3274 :     Description:
3275 :    
3276 :     Removed src/cm/ffi which does not (and did not) belong here.
3277 :    
3278 :     ----------------------------------------------------------------------
3279 :     Name: Matthias Blume
3280 : blume 757 Date: 2000/12/21 23:55:00 JST
3281 :     Tag: blume-20001221-exn
3282 :     Description:
3283 :    
3284 :     Probably most important: CM no longer silently swallows all exceptions
3285 :     in the compiler.
3286 :     Plus: some other minor CM changes. For example, CM now reports some
3287 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
3288 :    
3289 :     ----------------------------------------------------------------------
3290 :     Name: Matthias Blume
3291 : blume 756 Date: 2000/12/15 00:01:05 JST
3292 :     Tag: blume-20001215-dirtool
3293 :     Description:
3294 :    
3295 :     - "dir" tool added.
3296 :     - improvements and cleanup to Tools structure
3297 :     - documentation updates
3298 :    
3299 :     ----------------------------------------------------------------------
3300 : leunga 752 Name: Allen Leung
3301 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
3302 :     Description:
3303 :     Tag: leunga-20001214-int-inf
3304 :     Description:
3305 :    
3306 :     In IntInf, added these standard functions, which are missing from our
3307 :     implementation:
3308 :    
3309 :     andb : int * int -> int
3310 :     xorb : int * int -> int
3311 :     orb : int * int -> int
3312 :     notb : int -> int
3313 :     << : int * word -> int
3314 :     ~>> : int * word -> int
3315 :    
3316 :     Not tested, I hope they are correct.
3317 :    
3318 :     ----------------------------------------------------------------------
3319 :     Name: Allen Leung
3320 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
3321 :     Description:
3322 :     Tag: leunga-20001208-nowhere
3323 :     Description:
3324 :    
3325 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
3326 :     to generate better error messages etc. Plus a brief manual.
3327 :    
3328 :     ----------------------------------------------------------------------
3329 : george 749 Name: Lal George
3330 :     Date: 2000/12/08 09:54:02 EST 2000
3331 :     Tag: Release_110_31
3332 :     Description:
3333 : george 705
3334 : george 749 - Version 110.31
3335 : leunga 591 ----------------------------------------------------------------------
3336 : leunga 744 Name: Allen Leung
3337 :     Date: Thu Dec 7 22:01:04 EST 2000
3338 :     Tag: leunga-20001207-cell-monster-hack
3339 :     Description:
3340 :    
3341 :     Major MLRISC internal changes. Affect all clients.
3342 :     Summary:
3343 :    
3344 :     1. Type CELLS.cell = int is now replaced by a datatype.
3345 :     As a result, the old regmap is now gone. Almost all interfaces
3346 :     in MLRISC change as a consequence.
3347 :    
3348 :     2. A new brand version of machine description tool (v3.0) that generates
3349 :     modules expecting the new interface. The old version is removed.
3350 :    
3351 :     3. The RA interface has been further abstracted into two new functors.
3352 :     RISC_RA and X86RA. These functors have much simpler interfaces.
3353 :     [See also directory MLRISC/demo.]
3354 :    
3355 :     4. Some other new source->source code generation tools are available:
3356 :    
3357 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
3358 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
3359 :     I use this tool to generate the peephole optimizers---with the new
3360 :     cell type changes, peephole rules are becoming difficult to write
3361 :     without conditional pattern matching.
3362 :    
3363 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
3364 :     cover the entire MLRISC source tree so many things broke.
3365 :    
3366 :     6. CM files have been moved to the subdirectory MLRISC/cm.
3367 :     They are moved because there are a lot of them and they clutter up the
3368 :     root dir.
3369 :    
3370 :     7. More detailed documentation to come...
3371 :    
3372 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
3373 :     a makeml -rebuild first. This is because of other other
3374 :     changes that Matthias has made (see below).
3375 :    
3376 :    
3377 :     ----------------------------------------------------------------------
3378 : blume 733 Name: Matthias Blume
3379 : blume 742 Date: 2000/11/30 23:12:00 JST
3380 :     Tag: blume-20001130-filereorg
3381 :     Description:
3382 :    
3383 :     Some manual updates and some file reorganizations in CM.
3384 :    
3385 :     ----------------------------------------------------------------------
3386 :     Name: Matthias Blume
3387 : blume 737 Date: 2000/11/24 17:45:00 JST
3388 :     Tag: blume-20001124-link
3389 :     Description:
3390 :    
3391 :     Drastically improved link traversal code for the case that the dynamic
3392 :     value was already loaded at bootstrap time. As a result, CM and CMB
3393 :     now both load blazingly fast -- even on a very slow machine. Also,
3394 :     memory consumption has been further reduced by this.
3395 :    
3396 :     Warning: The format of the PIDMAP file has changed. THerefore, to
3397 :     bootstrap you have to do this:
3398 :    
3399 :     1. Run CMB.make
3400 :     2. Make a symbolic link for the boot directory:
3401 :     ln -s sml.boot.ARCH-OS xxx
3402 :     3. "Rebuild" the boot directory:
3403 :     ./makeml -boot xxx -rebuild sml ; rm xxx
3404 :     4. Boot normally:
3405 :     ./makeml
3406 :    
3407 :     ----------------------------------------------------------------------
3408 :     Name: Matthias Blume
3409 : blume 735 Date: 2000/11/21 21:20:00 JST
3410 :     Tag: blume-20001121-tools
3411 :     Description:
3412 :    
3413 :     Continued hacking on autoloading problem -- with success this time.
3414 :     Also changed tool-plugin mechanism. See new CM manual.
3415 :    
3416 :     ----------------------------------------------------------------------
3417 :     Name: Matthias Blume
3418 : blume 734 Date: 2000/11/19 14:30:00 JST
3419 :     Tag: blume-20001119-autoload
3420 :     Description:
3421 :    
3422 :     Some hacking to make autoloading faster. Success for CMB, no success
3423 :     so far for CM. There is a reduced structure CM' that autoloads faster.
3424 :     (This is a temporary, non-documented hack to be eliminated again when
3425 :     the general problem is solved.)
3426 :    
3427 :     ----------------------------------------------------------------------
3428 :     Name: Matthias Blume
3429 : blume 733 Date: 2000/11/17 14:10:00 JST
3430 :     Tag: blume-20001117-pickle-lib
3431 :     Description:
3432 :    
3433 :     1. Eliminated comp-lib.cm
3434 :     2. Made pickle-lib.cm
3435 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
3436 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
3437 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
3438 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
3439 :     Duplicated intmap modules thrown out.
3440 :     6. Hunted down all duplicated SCC code and replaced it with
3441 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
3442 :     7. Rewrote Feedback module.
3443 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
3444 :     should be thrown out and equivalent modules from smlnj-lib.cm
3445 :     should be used (IntRedBlackSet, IntListSet, ...).
3446 :    
3447 :     Confirmed that compiler compiles to fixpoint.
3448 :    
3449 :     ----------------------------------------------------------------------
3450 : leunga 731 Name: Allen Leung
3451 :     Date: 2000/11/10 18:00:00
3452 :     Tag: leunga-20001110-new-x86-fp
3453 :    
3454 :     A new x86 floating point code generator has been added.
3455 :     By default this is turned off. To turn this on, do:
3456 :    
3457 :     CM.autoload "$smlnj/compiler.cm";
3458 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3459 :    
3460 :     Changes:
3461 :    
3462 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
3463 :     2. Changed the extension callback for FTANGENT to generate:
3464 :    
3465 :     fptan
3466 :     fstp %st(0)
3467 :     instead of
3468 :     fptan
3469 :     fstpl ftempmem
3470 :    
3471 :     3. Numerous assembly fixes for x86.
3472 :    
3473 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
3474 :     support for a whole bunch of instructions and addressing modes:
3475 :    
3476 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
3477 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
3478 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
3479 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3480 :     fxch %st(n)
3481 :     fld %st(n)
3482 :     fst %st(n)
3483 :     fst mem
3484 :     fstp %st(n)
3485 :     fucom %st(n)
3486 :     fucomp %st(n)
3487 :    
3488 :     All these are now generated when the fast fp mode is turned on.
3489 :    
3490 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3491 :    
3492 :     ----------------------------------------------------------------------
3493 : blume 729 Name: Matthias Blume
3494 :     Date: 2000/11/09 11:20:00 JST
3495 :     Tag: blume-20001109-scc
3496 :     Description:
3497 :    
3498 :     Eliminated some code duplication:
3499 :    
3500 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
3501 :     (Otherwise the functor is useless.)
3502 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
3503 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
3504 :    
3505 :     ----------------------------------------------------------------------
3506 : george 721 Name: Lal George
3507 :     Date: 2000/11/06 09:02:21 EST 2000
3508 :     Tag: Release_110_30
3509 :     Description:
3510 :    
3511 :     - Version 110.30
3512 :     ----------------------------------------------------------------------
3513 : blume 715 Name: Matthias Blume
3514 : blume 716 Date: 2000/11/04 14:45:00
3515 :     Tag: blume-20001104-mlbuild
3516 :     Description:
3517 :    
3518 :     - Made ml-build faster on startup.
3519 :     - Documentation fixes.
3520 :    
3521 :     ----------------------------------------------------------------------
3522 :     Name: Matthias Blume
3523 : blume 715 Date: 2000/11/02 17:00:00 JST
3524 :     Tag: blume-20001102-condcomp
3525 :     Description:
3526 :    
3527 :     - Small tweaks to pickler -- new BOOTFILES!
3528 :     - Version bumped to 110.29.2.
3529 :     - Added conditional compilation facility to init.cmi (see comment there).
3530 :     ----------------------------------------------------------------------
3531 : leunga 713 Name: Allen Leung
3532 :     Date: 2000/10/23 19:31:00
3533 :     Tag: leunga-20001023-demo-ra
3534 :    
3535 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
3536 :     2. Test programs for the graph library updated
3537 :     3. Some new MLRISC demo programs added
3538 :    
3539 :     ----------------------------------------------------------------------
3540 : blume 710 Name: Matthias Blume
3541 : blume 711 Date: 2000/08/31 22:15:00 JST
3542 :     Tag: blume-20001017-errmsg
3543 :     Description:
3544 :    
3545 :     More error message grief: Where there used to be no messages, there
3546 :     now were some that had bogus error regions. Fixed.
3547 :    
3548 :     ----------------------------------------------------------------------
3549 :     Name: Matthias Blume
3550 : blume 710 Date: 2000/08/31 17:30:00 JST
3551 :     Tag: blume-20001017-v110p29p1
3552 :     Description:
3553 :    
3554 :     I made a version 110.29.1 with new bootfiles.
3555 :    
3556 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
3557 :     CM documentation changes and a small bugfix in CM's error reporting.
3558 :    
3559 :     ----------------------------------------------------------------------
3560 : george 705 Name: Lal George
3561 :     Date: 2000/09/27 14:42:35 EDT
3562 :     Tag: george-20000927-nodestatus
3563 :     Description:
3564 :    
3565 :     Changed the type of the nodestatus, so that:
3566 :    
3567 :     SPILLED(~1) is now SPILLED
3568 :     SPILLED(m) where m>=0 is now MEMREG(m)
3569 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
3570 :    
3571 :     ----------------------------------------------------------------------
3572 : blume 697 Name: Matthias Blume
3573 : blume 703 Date: 2000/09/07 14:45:00 JST
3574 :     Tag: blume-20000907-cmerrmsg
3575 :     Description:
3576 :    
3577 :     Small tweak to CM to avoid getting ML syntax error messages twice.
3578 :    
3579 :     ----------------------------------------------------------------------
3580 :     Name: Matthias Blume
3581 : blume 701 Date: 2000/08/31 18:00:00 JST
3582 :     Tag: blume-20000831-cvsbootfiles
3583 :     Description:
3584 :    
3585 :     New URL for boot files (because the 110.29 files on the BL server do
3586 :     now work correctly with my updated install scripts for yacc and lex).
3587 :    
3588 :     ----------------------------------------------------------------------
3589 :     Name: Matthias Blume
3590 : blume 697 Date: 2000/08/08 12:33:00 JST
3591 :     Tag: blume-20000808-manual
3592 :     Description:
3593 :    
3594 :     Tiny update to CM manual.
3595 :    
3596 :     ----------------------------------------------------------------------
3597 : leunga 695 Name: Allen Leung
3598 :     Date: 2000/08/7 19:31:00
3599 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3600 :    
3601 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3602 :     and Mobdy.
3603 :    
3604 :     1. x86
3605 :    
3606 :     a. Fixes to peephole module by John and Dan.
3607 :     b. Assembly fix to SETcc by Allen.
3608 :     c. Fix to c-call by John.
3609 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3610 :     e. Instruction selection optimization to SETcc as suggested by John.
3611 :    
3612 :     For example,
3613 :    
3614 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3615 :    
3616 :     should generate:
3617 :    
3618 :     MOVL a, x
3619 :     SUBL b, x
3620 :     SHRL 31, x
3621 :    
3622 :     2. IR stuff
3623 :    
3624 :     A bunch of new DJ-graph related algorithms added. These
3625 :     speed up SSA construction.
3626 :    
3627 :     3. SSA + Scheduling
3628 :    
3629 :     Added code for SSA and scheduling to the repository
3630 :    
3631 :     ----------------------------------------------------------------------
3632 : george 694 Name: Lal George
3633 :     Date: 2000/07/27 11:53:14 EDT
3634 :    
3635 :     Tag: lal-20000727-linux-ppc
3636 :     Description:
3637 :    
3638 :     Made changes to support Linux PPC.
3639 :     p.s. I have confirmation that the 110.29 boot files work fine.
3640 :    
3641 :     ----------------------------------------------------------------------
3642 : blume 684 Name: Matthias Blume
3643 : blume 692 Date: 2000/07/27 17:40:00 JST
3644 :     Tag: blume-20000727-scripts
3645 :     Description:
3646 :    
3647 :     !!!! WARNING !!!!
3648 :     You must recompile the runtime system!
3649 :     !!!! WARNING !!!!
3650 :    
3651 :     This is basically another round of script-enhancements:
3652 :    
3653 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3654 :     and undefine CM preprocessor symbols.
3655 :    
3656 :     2. ml-build avoids generating a new heap image if it finds that the
3657 :     existing one is still ok. (The condition is that no ML file had to
3658 :     be recompiled and all ML files are found to be older that the heap
3659 :     file.)
3660 :    
3661 :     To make this work smoothly, I also hacked the runtime system as
3662 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3663 :     (.sparc-solaris, ...) that is currently being used.
3664 :    
3665 :     Moreover, the signature of CM.mk_standalone has changed. See the
3666 :     CM manual.
3667 :    
3668 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3669 :     CM manual for details.)
3670 :    
3671 :     4. More CM manual updates:
3672 :     - all of the above has been documented.
3673 :     - there is now a section describing the (CM-related) command line
3674 :     arguments that are accepted by the "sml" command
3675 :    
3676 :     ----------------------------------------------------------------------
3677 :     Name: Matthias Blume
3678 : blume 691 Date: 2000/07/25 16:20:00 JST
3679 :     Tag: blume-20000725-makedepend
3680 :     Description:
3681 :    
3682 :     Added a script called ml-makedepend. This can be used in makefiles
3683 :     for Unix' make in a way very similar to the "makedepend" command for
3684 :     C.
3685 :    
3686 :     The script internally uses function CM.sources.
3687 :    
3688 :     Synopsis:
3689 :    
3690 :     ml-makedepend [-f makefile] cmfile targetname
3691 :    
3692 :     The default for the makefile is "makefile" (or "Makefile" should
3693 :     "makefile" not exist).
3694 :    
3695 :     ml-makedepend adds a cmfile/targetname-specific section to this
3696 :     makefile (after removing the previous version of this section). The
3697 :     section contains a single dependency specification with targetname on
3698 :     the LHS (targetname is an arbitrary name), and a list of files derived
3699 :     from the cmfile on the RHS. Some of the files on the RHS are
3700 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3701 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3702 :     path names. The makefile writer is responsible for making sure that
3703 :     these variables have correct at the time "make" is invoked.
3704 :    
3705 :     ----------------------------------------------------------------------
3706 :     Name: Matthias Blume
3707 : blume 690 Date: 2000/07/22 23:30:00 JST
3708 :     Tag: blume-20000722-urlupdate
3709 :     Description:
3710 :    
3711 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3712 :    
3713 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3714 :    
3715 :     ----------------------------------------------------------------------
3716 :     Name: Matthias Blume
3717 : blume 686 Date: 2000/07/18 18:00:00 JST
3718 :     Tag: blume-20000718-Version_110_29
3719 :     Description:
3720 :    
3721 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3722 :    
3723 :     2. Updated config/version to 110.29
3724 :    
3725 :     3. Updated config/srcarchiveurl
3726 :    
3727 :     3. New boot files!
3728 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3729 :    
3730 :     ----------------------------------------------------------------------
3731 :     Name: Matthias Blume
3732 : blume 684 Date: 2000/07/11 13:58:00 JST
3733 :     Tag: blume-20000711-doctypo
3734 :     Description:
3735 :    
3736 :     Fixed a few typos in CM manual.
3737 :    
3738 :     ----------------------------------------------------------------------
3739 : leunga 683 Name: Allen Leung
3740 :     Date: 2000/06/15 00:38:00
3741 :     Tag: leunga-20000704-sparc-x86
3742 :    
3743 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3744 :     2. fix to x86 RET bug [found by Dan Grossman]
3745 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3746 :    
3747 :     Affects c-- and moby only
3748 :    
3749 :     ----------------------------------------------------------------------
3750 : blume 680 Name: Matthias Blume
3751 : blume 682 Date: 2000/07/04 15:26:00
3752 :     Tag: blume-20000704-trigger
3753 :     Description:
3754 :    
3755 :     1. Improvements to CM manual.
3756 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3757 :     of getting a back-trace. The function, when called, raises an
3758 :     internal exception which explicitly carries the full back-trace history,
3759 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3760 :     or not). The interactive loop will print that history once it arrives
3761 :     at top level.
3762 :     Short of having all exceptions implicitly carry the full history, the
3763 :     recommended way of using this facility is:
3764 :     - compile your program with instrumentation "on"
3765 :     - run it, when it raises an exception, look at the history
3766 :     - if the history is "cut off" because of some handler, go and modify
3767 :     your program so that it explicitly calls BTrace.trigger
3768 :     - recompile (still instrumented), and rerun; look at the full history
3769 :    
3770 :     ----------------------------------------------------------------------
3771 :     Name: Matthias Blume
3772 :     Date: 2000/07/03 15:36:00 JST
3773 : blume 680 Tag: blume-20000702-manual
3774 :     Description:
3775 :    
3776 :     Small corrections and updates to CM manual.
3777 :    
3778 :     ----------------------------------------------------------------------
3779 : blume 682 Name: Matthias Blume
3780 : blume 679 Date: 2000/06/29 16:04:00 JST
3781 :     Tag: blume-20000629-yacctool
3782 :     Description:
3783 :    
3784 :     Changes:
3785 :    
3786 :     1. Class "mlyacc" now takes separate arguments to pass options to
3787 :     generated .sml- and .sig-files independently.
3788 :     2. Corresponding CM manual updates.
3789 :     3. BTrace module now also reports call sites. (However, for loop clusters
3790 :     it only shows from where the cluster was entered.) There are associated
3791 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3792 :    
3793 :     ----------------------------------------------------------------------
3794 : blume 675 Name: Matthias Blume
3795 : blume 678 Date: 2000/06/27 16:51:00 JST
3796 :     Tag: blume-20000627-noweb
3797 :     Description:
3798 :    
3799 :     Changes:
3800 :    
3801 :     1. Implemented "subdir" and "witness" options for noweb tool.
3802 :     This caused some slight internal changes in CM's tool implementation.
3803 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3804 :     some remaining issues from earlier path anchor changes.
3805 :     3. Updated CM manual accordingly.
3806 :    
3807 :     4. Changed implementation of back-tracing so that I now consider it
3808 :     ready for prime-time.
3809 :    
3810 :     In particular, you don't have to explicitly trigger the back-trace
3811 :     anymore. Instead, if you are running BTrace-instrumented code and
3812 :     there is an uncaught exception (regardless of whether or not it was
3813 :     raised in instrumented code), the top-level evalloop will print
3814 :     the back-trace.
3815 :    
3816 :     Features:
3817 :    
3818 :     - Instrumented and uninstrumented code work together seemlessly.
3819 :     (Of course, uninstrumented code is never mentioned in actual
3820 :     back-traces.)
3821 :    
3822 :     - Asymptotic time- and space-complexity of instrumented code is
3823 :     equal to that of uninstrumented code. (This means that
3824 :     tail-recursion is preserved by the instrumentation phase.)
3825 :    
3826 :     - Modules whose code has been instrumented in different sessions
3827 :     work together without problem.
3828 :    
3829 :     - There is no penalty whatsoever on uninstrumented code.
3830 :    
3831 :     - There is no penalty on "raise" expressions, even in
3832 :     instrumented code.
3833 :    
3834 :     A potential bug (or perhaps it is a feature, too):
3835 :    
3836 :     A back-trace reaches no further than the outermost instrumented
3837 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3838 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3839 :    
3840 :     <exp> handle e => raise e
3841 :    
3842 :     and even
3843 :    
3844 :     <exp> handle Foo => raise Bar
3845 :    
3846 :     and, of course, through
3847 :    
3848 :     <exp> handle Foo => ...
3849 :    
3850 :     if the exception was not Foo.
3851 :    
3852 :     Back-traces always reach right through any un-instrumented code
3853 :     including any of its "handle" expressions, trivial or not.
3854 :    
3855 :     To try this out, do the following:
3856 :    
3857 :     - Erase all existing binfiles for your program.
3858 :     (You may keep binfiles for those modules where you think you
3859 :     definitely don't need back-tracing.)
3860 :     - Turn on back-trace instrumentation:
3861 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3862 :     - Recompile your program. (I.e., run "CM.make" or "use".)
3863 :     - You may now turn instrumentation off again (if you want):
3864 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3865 :     - Run your program as usual. If it raises an exception that
3866 :     reaches the interactive toplevel, then a back-trace will
3867 :     automatically be printed. After that, the toplevel loop
3868 :     will print the exception history as usual.
3869 :    
3870 :     ----------------------------------------------------------------------
3871 :     Name: Matthias Blume
3872 : blume 677 Date: 2000/06/26 09:56:46 JST
3873 :     Tag: blume-20000626-setup
3874 :     Description:
3875 :    
3876 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
3877 :     ML code before and after compiling a file (e.g., to set compiler
3878 :     flags)
3879 :    
3880 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
3881 :     - associated changes to btrace.sml (BTrace instrumentation pass)
3882 :     - cleaner implementation of btimp.sml (BTrace tracing and report
3883 :     module)
3884 :    
3885 :     CM manual: * new path encoding documented
3886 :     * description of setup-parameter to "sml" added
3887 :    
3888 :     The biggest user-visible change to back-tracing is that it is no
3889 :     longer necessary to compile all traced modules within the same
3890 :     session. (This was a real limitation.)
3891 :    
3892 :     ----------------------------------------------------------------------
3893 :     Name: Matthias Blume
3894 : blume 676 Date: 2000/06/24 12:40:00 JST
3895 :     Tag: blume-20000624-startup
3896 :     Description:
3897 :    
3898 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
3899 :     bit too often -- to put it mildly. :)
3900 :    
3901 :     ----------------------------------------------------------------------
3902 :     Name: Matthias Blume
3903 : blume 675 Date: 2000/06/23 18:20:00 JST
3904 :     Tag: blume-20000623-btrace
3905 :     Description:
3906 :    
3907 :     This updates adds a backtrace facility to aid programmers in debugging
3908 :     their programs. This involves the following changes:
3909 :    
3910 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3911 :     keeping track of the current call stack. When programs are compiled
3912 :     in a special mode, the compiler will insert calls to these hooks
3913 :     into the user program.
3914 :     "Hook" means that it is possible for different implementations of
3915 :     back-tracing to register themselves (at different times).
3916 :    
3917 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3918 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3919 :     It can be turned on using this call:
3920 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3921 :     Turning it off again:
3922 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3923 :     Querying the current status:
3924 :     SMLofNJ.Internals.BTrace.mode NONE;
3925 :     Annotated programs are about twice as big as normal ones, and they
3926 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3927 :     where all hooks do nothing). The slowdown with a plugin that is
3928 :     actually useful (such as the one supplied by default) is even greater,
3929 :     but in the case of the default plugin it is still only an constant
3930 :     factor (amortized).
3931 :    
3932 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3933 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3934 :     the above-mentioned function "mode" controls whether the annotation
3935 :     phase is invoked by the compiler. Another important function is
3936 :     "trigger": when called it aborts the current execution and causes
3937 :     the top-level loop to print a full back-trace.
3938 :    
3939 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3940 :     for back-tracing. It keeps track of the dynamic call stack and in
3941 :     addition to that it keeps a partial history at each "level" of that
3942 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3943 :     d and b (at separate times, dynamically), then the report will show:
3944 :    
3945 :     GOTO d
3946 :     /c
3947 :     GOTO \b
3948 :     CALL a
3949 :    
3950 :     This shows that there was an initial non-tail call of a, then a
3951 :     tail-call to b or c, looping behavior in a cluster of functions that
3952 :     consist of b and c, and then a goto from that cluster (i.e., either from
3953 :     b or from c) to d.
3954 :    
3955 :     Note that (depending on the user program) the amount of information
3956 :     that the back-trace module has to keep track of at each level is bounded
3957 :     by a constant. Thus, the whole implementation has the same asymptotical
3958 :     complexity as the original program (both in space and in time).
3959 :    
3960 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3961 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3962 :     which is raised by the "trigger" function mentioned above.
3963 :    
3964 :     Notes on usage:
3965 :    
3966 :     - Annotated code works well together with unannotated code:
3967 :     Unannotated calls simply do not show up at all in the backtrace.
3968 :    
3969 :     - It is not a good idea to let modules that were annotated during
3970 :     different sessions run at the same time. This is because the compiler
3971 :     chooses small integers to identify individual functions, and there
3972 :     will be clashes if different modules were compiled in separate sessions.
3973 :     (Nothing will crash, and you will even be told about the clashes, but
3974 :     back-trace information will in general not be useful.)
3975 :    
3976 :     - Back-tracing can be confused by callcc and capture.
3977 :    
3978 :     - The only way of getting a back-trace right now is to explicitly
3979 :     invoke the "trigger" function from your user program. Eventually, we
3980 :     should make every exception carry back-trace information (if
3981 :     available). But since this creates more overhead at "raise"-time
3982 :     (similar to the current exnHistory overhead), I have not yet
3983 :     implemented this. (The implementation will be rather easy.) With
3984 :     exceptions carrying back-trace information, this facility will be even
3985 :     more useful because users don't need to modify their programs...
3986 :    
3987 :     - While it is possible to compile the compiler with back-trace
3988 :     annotations turned on (I did it to get some confidence in
3989 :     correctness), you must make absolutely sure that core.sml and
3990 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3991 :     be compiled with annotation because there is no core access yet, but
3992 :     if you compile btimp.sml with annotation, then the system will go into
3993 :     an infinite recursion and crash.)
3994 :     Since CM currently does not know about BTrace, the only way to turn
3995 :     annotations on and off for different modules of the compiler is to
3996 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3997 :     this is awkward and clumsy.
3998 :    
3999 :     Sample sessions:
4000 :    
4001 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4002 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4003 :     [autoloading]
4004 :     [autoloading done]
4005 :     val it = false : bool
4006 :     - structure X = struct
4007 :     - fun main n = let
4008 :     - fun a (x, 0) = d x
4009 :     - | a (x, n) = b (x, n - 1)
4010 :     - and b (x, n) = c (x, n)
4011 :     - and c (x, n) = a (x, n)
4012 :     - and d x = e (x, 3)
4013 :     - and e (x, 0) = f x
4014 :     - | e (x, n) = e (x, n - 1)
4015 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
4016 :     - | f n = n * g (n - 1)
4017 :     - and g n = a (n, 3)
4018 :     - in
4019 :     - f n
4020 :     - end
4021 :     - end;
4022 :     structure X : sig val main : int -> int end
4023 :     - X.main 3;
4024 :     *** BACK-TRACE ***
4025 :     GOTO stdIn:4.2-13.20: X.main[2].f
4026 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4027 :     GOTO stdIn:4.2-13.20: X.main[2].d
4028 :     / stdIn:4.2-13.20: X.main[2].a
4029 :     | stdIn:4.2-13.20: X.main[2].b
4030 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4031 :     CALL stdIn:4.2-13.20: X.main[2].g
4032 :     GOTO stdIn:4.2-13.20: X.main[2].f
4033 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4034 :     GOTO stdIn:4.2-13.20: X.main[2].d
4035 :     / stdIn:4.2-13.20: X.main[2].a
4036 :     | stdIn:4.2-13.20: X.main[2].b
4037 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4038 :     CALL stdIn:4.2-13.20: X.main[2].g
4039 :     GOTO stdIn:4.2-13.20: X.main[2].f
4040 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
4041 :     GOTO stdIn:4.2-13.20: X.main[2].d
4042 :     / stdIn:4.2-13.20: X.main[2].a
4043 :     | stdIn:4.2-13.20: X.main[2].b
4044 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
4045 :     CALL stdIn:4.2-13.20: X.main[2].g
4046 :     GOTO stdIn:4.2-13.20: X.main[2].f
4047 :     CALL stdIn:2.15-17.4: X.main[2]
4048 :     -
4049 :    
4050 :     (Note that because of a FLINt bug the above code currently does not
4051 :     compile without BTrace turned on.)
4052 :    
4053 :     Here is another example, using my modified Tiger compiler:
4054 :    
4055 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
4056 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
4057 :     [autoloading]
4058 :     [autoloading done]
4059 :     val it = false : bool
4060 :     - CM.make "sources.cm";
4061 :     [autoloading]
4062 :     ...
4063 :     [autoloading done]
4064 :     [scanning sources.cm]
4065 :     [parsing (sources.cm):parse.sml]
4066 :     [creating directory CM/SKEL ...]
4067 :     [parsing (sources.cm):tiger.lex.sml]
4068 :     ...
4069 :     [wrote CM/sparc-unix/semant.sml]
4070 :     [compiling (sources.cm):main.sml]
4071 :     [wrote CM/sparc-unix/main.sml]
4072 :     [New bindings added.]
4073 :     val it = true : bool
4074 :     - Main.compile ("../testcases/merge.tig", "foo.out");
4075 :     *** BACK-TRACE ***
4076 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
4077 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4078 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4079 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
4080 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4081 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4082 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
4083 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
4084 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
4085 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
4086 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
4087 :     CALL main.sml:1.18-118.4: Main.compile[2]
4088 :     -
4089 :    
4090 :     ----------------------------------------------------------------------
4091 : blume 673 Name: Matthias Blumen
4092 :     Date: 2000/06/21 18:00:00 JST
4093 :     Tag: blume-20000621-manual
4094 :     Description:
4095 :    
4096 :     CM manual update: Path environments documented.
4097 :    
4098 :     ----------------------------------------------------------------------
4099 : blume 666 Name: Matthias Blume
4100 : blume 672 Date: 2000/06/19 13:40:00
4101 :     Tag: blume-20000619-manual
4102 :     Description:
4103 :    
4104 :     CM manual and system/README update. This only covers the fact that
4105 :     there are no more implicit anchors. (Path environments and the "bind"
4106 :     option to "cm" have yet to be documented.)
4107 :    
4108 :     ----------------------------------------------------------------------
4109 :     Name: Matthias Blume
4110 : blume 671 Date: 2000/06/19 11:05:00 JST
4111 :     Tag: blume-20000619-chdir-bugfix
4112 :     Description:
4113 :    
4114 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
4115 :    
4116 :     ----------------------------------------------------------------------
4117 :     Name: Matthias Blume
4118 : blume 670 Date: 2000/06/18 22:00:10 JST
4119 :     Tag: blume-20000618-implicit-anchors-really-gone
4120 :     Description:
4121 :    
4122 :     I updates the previous HISTORY entry where I forgot to mention that
4123 :     implicit anchors are no longer with us.
4124 :    
4125 :     The current update also gets rid of the (now useless) controller
4126 :     CM.Control.implicit_anchors.
4127 :    
4128 :     ----------------------------------------------------------------------
4129 :     Name: Matthias Blume
4130 : blume 666 Date: 2000/06/16 17:30:00 JST
4131 :     Tag: blume-20000616-anchorenv
4132 :     Description:
4133 :    
4134 :     This patch implements the long anticipated (just kidding :) "anchor
4135 :     environment" mechanism. In the course of doing this, I also
4136 :     re-implemented CM's internal "SrcPath" module from scratch. The new
4137 :     one should be more robust in certain boundary cases. In any case, it
4138 :     is a lot cleaner than its predecessor (IMHO).
4139 :    
4140 :     This time, although there is yet another boot file format change, I
4141 :     kept the unpickler backward-compatible. As a result, no new bootfiles
4142 :     are necessary and bootstrapping is straightforward. (You cannot read
4143 :     new bootfiles into an old system, but the other way around is no
4144 :     problem.)
4145 :    
4146 :     Visible changes:
4147 :    
4148 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
4149 :     longer recognized at all. This means that such path names are not
4150 :     illegal either. For example, the name basis.cm simply refers to a
4151 :     local file called "basis.cm" (i.e, the name is an ordinary path
4152 :     relative to .cm-files directory). Or, to put it differently, only
4153 :     names that start with $ are anchored paths.
4154 : blume 666
4155 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
4156 :     vanished.
4157 :    
4158 :     John (Reppy) had critizised this as soon as I originally proposed and
4159 : blume 666 implemented it, but at that time I did not really deeply believe
4160 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
4161 :     in another place where it cannot be seen as an abbreviation for
4162 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
4163 :     seems to mean (i.e., it "expands" into the corresponding anchor
4164 :     value).
4165 :    
4166 :     However, when paths are used as members in CM description files, it
4167 :     continues to be true that there must be at least another arc after the
4168 :     anchor. This is now enforced separately during semantic analysis
4169 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
4170 : blume 666
4171 :     ** 2. The "cm" class now accepts an option "bind". The option's value
4172 :     is a sub-option list of precisely two items -- one labeled "anchor"
4173 :     and the other one labeled "value". As you might expect, "anchor" is
4174 :     used to specify an anchor name to be bound, and "value" specifies what
4175 :     the anchor is being bound to.
4176 :    
4177 :     The value must be a directory name and can be given in either standard
4178 :     syntax (including the possibility that it is itself an anchored path)
4179 :     or native syntax.
4180 :    
4181 :     Examples:
4182 :    
4183 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
4184 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
4185 :    
4186 :     and so on.
4187 :    
4188 :     The meaning of this is that the .cm-file will be processed with an
4189 :     augmented anchor environment where the given anchor(s) is/are bound to
4190 :     the given values(s).
4191 :    
4192 :     The rationale for having this feature is this: Suppose you are trying
4193 :     to use two different (already stable) libraries a.cm and b.cm (that
4194 :     you perhaps didn't write yourself). Further, suppose each of these
4195 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
4196 :     Normally you would now have a problem because the anchor "lib" can not
4197 :     be bound to more than one value globally. Therefore, the project that
4198 :     uses both a.cm and b.cm must locally redirect the anchor to some other
4199 :     place:
4200 :    
4201 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
4202 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
4203 :    
4204 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
4205 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
4206 :    
4207 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
4208 :     you when you do so at the time of CM.stabilize). Therefore, you can
4209 :     also use an anchored path as the value:
4210 :    
4211 :     a.cm (bind:(anchor:lib value:$a-lib))
4212 :     b.cm (bind:(anchor:lib value:$b-lib))
4213 :    
4214 :     Now you can globally configure (using the usual CM.Anchor.anchor or
4215 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
4216 :     itself is always locally bound, setting it globally is no longer
4217 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
4218 :     can still be used as a global anchor for separate purposes. As a
4219 :     matter of fact, one can locally define "lib" in terms of a global
4220 :     "lib":
4221 :    
4222 :     a.cm (bind:(anchor:lib value:$lib/a))
4223 :     b.cm (bind:(anchor:lib value:$lib/b))
4224 :    
4225 :     ** 3: The encoding of path names has changed. This affects the way
4226 :     path names are shown in CM's progress report and also the internal
4227 :     protocol encoding used for parallel make.
4228 :    
4229 :     The encoding now uses one or more ':'-separated segments. Each
4230 :     segments corresponds to a file that has been specified relative to the
4231 :     file given by its preceding segment. The first segment is either
4232 :     relative to the CWD, absolute, or anchored. Each segment itself is
4233 :     basically a Unix pathname; all segments but the first are relative.
4234 :    
4235 :     Example:
4236 :    
4237 :     $foo/bar/baz.cm:a/b/c.sml
4238 :    
4239 :     This path denotes the file bar/a/b/c.sml relative to the directory
4240 :     denoted by anchor "foo". Notice that the encoding also includes
4241 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
4242 :     paths are resolved relative to the .cm-files directory, so baz.cm must
4243 :     be ignored to get the "real" pathname.
4244 :    
4245 :     To make this fact more obvious, CM puts the names of such "virtual
4246 :     arcs" into parentheses when they appear in progress reports. (No
4247 :     parentheses will appear in the internal protocol encoding.) Thus,
4248 :     what you really see is:
4249 :    
4250 :     $foo/bar/(baz.cm):a/b/c.sml
4251 :    
4252 :     I find this notation to be much more informative than before.
4253 :    
4254 :     Another new feature of the encoding is that special characters
4255 :     including parentheses, colons, (back)slashes, and white space are
4256 :     written as \ddd (where ddd is the decimal encoding of the character).
4257 :    
4258 : blume 670 *** The CM manual still needs to be updated.
4259 :    
4260 : blume 666 ----------------------------------------------------------------------
4261 : leunga 660 Name: Allen Leung
4262 :     Date: 2000/06/15 00:38:00
4263 :     Tag: leunga-20000615-x86-peephole
4264 :    
4265 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
4266 :    
4267 :     ----------------------------------------------------------------------
4268 : blume 658 Name: Matthias Blume
4269 :     Date: 2000/06/12 11:40:00
4270 :     Tag: blume-20000612-parmakefix
4271 :     Description:
4272 :    
4273 :     More cleanup after changing the file naming scheme: This time I
4274 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
4275 :    
4276 :     ----------------------------------------------------------------------
4277 : leunga 657 Name: Allen Leung
4278 :     Date: 2000/06/09 01:25:00
4279 :     Tag: leunga-20000609-various
4280 :    
4281 :     None of these things should affect normal SML/NJ operations
4282 :    
4283 :     1. Peephole improvements provided by Fermin (c--)
4284 :     2. New annotation DEFUSE for adding extra dependence (moby)
4285 :     3. New X86 LOCK instructions (moby)
4286 :     4. New machine description language for reservation tables (scheduling)
4287 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
4288 :     trees etc.)
4289 :     6. I've changed the CM files so that they can work with versions
4290 :     110.0.6, 110.25 and 110.28
4291 :    
4292 :     ----------------------------------------------------------------------
4293 : blume 652 Name: Matthias Blume
4294 : blume 656 Date: 2000/06/09 12:40:00
4295 :     Tag: blume-20000609-log
4296 :     Description:
4297 :    
4298 :     - Removed all(?) remaining RCS Log entries from sources.
4299 :    
4300 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
4301 :     anchored paths).
4302 :    
4303 :     ----------------------------------------------------------------------
4304 :     Name: Matthias Blume
4305 : blume 653 Date: 2000/06/07 17:00:00 JST
4306 :     Tag: blume-20000607-no-implicit-anchors
4307 :     Description:
4308 :    
4309 :     1. This update changes the default setting for
4310 :     CM.Control.implicit_anchors from true to false. This means that
4311 :     implicit anchors are no longer permitted by default. I also tried to
4312 :     make sure that nothing else still relies on implicit anchors.
4313 :     (This is the next step on the schedule towards a CM that does not even
4314 :     have the notion of implicit anchors anymore.)
4315 :    
4316 :     2. More CM manual updates.
4317 :    
4318 :     3. I managed to track down and fix the pickling bug I mentioned last
4319 :     time. Because of the previously existing workaround, this entails no
4320 :     immediate practical changes.
4321 :    
4322 :     ----------------------------------------------------------------------
4323 :     Name: Matthias Blume
4324 : blume 652 Date: 2000/06/06 11:15:00 JST
4325 :     Tag: blume-20000606-lazierpickle
4326 :     Description:
4327 :    
4328 :     !!!! NEW BOOT FILES !!!!
4329 :    
4330 :     * The main purpose of this update is to make library pickles lazier in
4331 :     order to reduce the initial space penalty for autoloading a library.
4332 :     As a result, it is now possible to have $smlnj/compiler.cm
4333 :     pre-registered. This should take care of the many complaints or
4334 :     inquiries about missing structure Compiler. This required changes to
4335 :     CM's internal data structures and small tweaks to some algorithms.
4336 :    
4337 :     As a neat additional effect, it is no longer necessary (for the sake
4338 :     of lean heap image files) to distinguish between a "minimal" CM and a
4339 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
4340 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
4341 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
4342 :    
4343 :     To make the life of compiler-hackers easier, "makeml" now also
4344 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
4345 :     words, after you bootstrap a new sml for the first time, you will not
4346 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
4347 :     around you will still have to do it, though.)
4348 :    
4349 :     * A second change consists of major updates to the CM manual. There
4350 :     are now several appendices with summary information and also a full
4