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

SCM Repository

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

Annotation of /sml/trunk/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1098 - (view) (download)
Original Path: sml/trunk/HISTORY

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 : george 1003 Date: yyyy/mm/dd
12 : dbm 570 Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 1025
15 : george 1003 ----------------------------------------------------------------------
16 : blume 1096 Name: Matthias Blume
17 : blume 1098 Date: 2002/02/27 12:29:00 EST
18 :     Tag: blume-20020227-cdebug
19 :     Description:
20 :    
21 :     - made types in structure C and C_Debug to be equal
22 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
23 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
24 :    
25 :     ----------------------------------------------------------------------
26 :     Name: Matthias Blume
27 : blume 1096 Date: 2002/02/26 12:00:00 EST
28 :     Tag: blume-20020226-ffi
29 :     Description:
30 :    
31 :     1. Fixed a minor bug in CM's "noweb" tool:
32 :     If numbering is turned off, then truly don't number (i.e., do not
33 :     supply the -L option to noweb). The previous behavior was to supply
34 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
35 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
36 :    
37 :     2. Once again, I reworked some aspects of the FFI:
38 :    
39 :     A. The incomplete/complete type business:
40 :    
41 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
42 :     gone!
43 :     - ML types representing an incomplete type are now *equal* to
44 :     ML types representing their corresponding complete types (just like
45 :     in C). This is still safe because ml-nlffigen will not generate
46 :     RTTI for incomplete types, nor will it generate functions that
47 :     require access to such RTTI. But when ML code generated from both
48 :     incomplete and complete versions of the C type meet, the ML types
49 :     are trivially interoperable.
50 :    
51 :     NOTE: These changes restore the full generality of the translation
52 :     (which was previously lost when I eliminated functorization)!
53 :    
54 :     B. Enum types:
55 :    
56 :     - Structure C now has a type constructor "enum" that is similar to
57 :     how the "su" constructor works. However, "enum" is not a phantom
58 :     type because each "T enum" has values (and is isomorphic to
59 :     MLRep.Signed.int).
60 :     - There are generic access operations for enum objects (using
61 :     MLRep.Signed.int).
62 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
63 :     * The structure contains the definition of type "mlrep" (the ML-side
64 :     representation type of the enum). Normally, mlrep is the same
65 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
66 :     then mlrep will be defined as a datatype -- thus facilitating
67 :     pattern matching on mlrep values.
68 :     ("-ec" will be suppressed if there are duplicate values in an
69 :     enumeration.)
70 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
71 :     will be generated for each C enum constant xxx.
72 :     * Conversion functions m2i and i2m convert between mlrep and
73 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
74 :     * Coversion functions c and ml convert between mlrep and "tag enum".
75 :     * Access functions (get/set) fetch and store mlrep values.
76 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
77 :     enumerations are merged into one single enumeration represented by
78 :     structure E_'.
79 :    
80 :     ----------------------------------------------------------------------
81 : leunga 1080 Name: Allen Leung
82 : leunga 1094 Date: 2002/02/25 04:45:00 EST
83 :     Tag: leunga-20020225-cps-spill
84 :    
85 :     This is a new implementation of the CPS spill phase.
86 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
87 :     In case of problems, replace it with the old file spill.sml
88 :    
89 :     The current compiler runs into some serious performance problems when
90 :     constructing a large record. This can happen when we try to compile a
91 :     structure with many items. Even a very simple structure like the following
92 :     makes the compiler slow down.
93 :    
94 :     structure Foo = struct
95 :     val x_1 = 0w1 : Word32.int
96 :     val x_2 = 0w2 : Word32.int
97 :     val x_3 = 0w3 : Word32.int
98 :     ...
99 :     val x_N = 0wN : Word32.int
100 :     end
101 :    
102 :     The following table shows the compile time, from N=1000 to N=4000,
103 :     with the old compiler:
104 :    
105 :     N
106 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
107 :     MLRISC ra 0.06u 0.00s 0.05g
108 :     (spills = 0 reloads = 0)
109 :     TOTAL 0.63u 0.07s 0.21g
110 :    
111 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
112 :     MLRISC ra 5.68u 0.59s 3.93g
113 :     (spills = 0 reloads = 0)
114 :     TOTAL 14.71u 0.99s 4.81g
115 :    
116 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
117 :     MLRISC ra 5.54u 0.65s 3.91g
118 :     (spills = 543 reloads = 1082)
119 :     TOTAL 65.40u 3.13s 6.00g
120 :    
121 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
122 :     MLRISC ra 0.80u 0.10s 0.55g
123 :     (spills = 42 reloads = 84)
124 :     TOTAL 129.42u 5.10s 4.13g
125 :    
126 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
127 :     MLRISC ra 2.69u 0.27s 1.38g
128 :     (spills = 62 reloads = 124)
129 :     TOTAL 682.48u 19.61s 13.99g
130 :    
131 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
132 :     MLRISC ra 4.96u 0.27s 2.72g
133 :     (spills = 85 reloads = 170)
134 :     TOTAL 2375.26u 57.21s 48.00g
135 :    
136 :     As you can see the old cps spill module suffers from some serious
137 :     performance problem but since I cannot decipher the old code fully,
138 :     innstead of patching the problems up, I'm reimplementing it
139 :     with a different algorithm. The new code is more modular,
140 :     smaller when compiled, and substantially faster
141 :     (O(n log n) time and O(n) space). Timing of the new spill module:
142 :    
143 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
144 :     MLRISC ra 0.25u 0.02s 0.15g
145 :     (spills=1 reloads=3)
146 :     TOTAL 7.74u 0.34s 1.62g
147 :    
148 :     Implementation details:
149 :    
150 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
151 :     number of live variables at any program point (the bandwidth)
152 :     does not exceed a certain limit, which is determined by the
153 :     size of the spill area.
154 :    
155 :     When the bandwidth is too large, we decrease the register pressure by
156 :     packing live variables into spill records. How we achieve this is
157 :     completely different than what we did in the old code.
158 :    
159 :     First, there is something about the MLRiscGen code generator
160 :     that we should be aware of:
161 :    
162 :     o MLRiscGen performs code motion!
163 :    
164 :     In particular, it will move floating point computations and
165 :     address computations involving only the heap pointer to
166 :     their use sites (if there is only a single use).
167 :     What this means is that if we have a CPS record construction
168 :     statement
169 :    
170 :     RECORD(k,vl,w,e)
171 :    
172 :     we should never count the new record address w as live if w
173 :     has only one use (which is often the case).
174 :    
175 :     We should do something similar to floating point, but the transformation
176 :     there is much more complex, so I won't deal with that.
177 :    
178 :     Secondly, there are now two new cps primops at our disposal:
179 :    
180 :     1. rawrecord of record_kind option
181 :     This pure operator allocates some uninitialized storage from the heap.
182 :     There are two forms:
183 :    
184 :     rawrecord NONE [INT n] allocates a tagless record of length n
185 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
186 :     and initializes the tag.
187 :    
188 :     2. rawupdate of cty
189 :     rawupdate cty (v,i,x)
190 :     Assigns to x to the ith component of record v.
191 :     The storelist is not updated.
192 :    
193 :     We use these new primops for both spilling and increment record construction.
194 :    
195 :     1. Spilling.
196 :    
197 :     This is implemented with a linear scan algorithm (but generalized
198 :     to trees). The algorithm will create a single spill record at the
199 :     beginning of the cps function and use rawupdate to spill to it,
200 :     and SELECT or SELp to reload from it. So both spills and reloads
201 :     are fine-grain operations. In contrast, in the old algorithm
202 :     "spills" have to be bundled together in records.
203 :    
204 :     Ideally, we should sink the spill record construction to where
205 :     it is needed. We can even split the spill record into multiple ones
206 :     at the places where they are needed. But CPS is not a good
207 :     representation for global code motion, so I'll keep it simple and
208 :     am not attempting this.
209 :    
210 :     2. Incremental record construction (aka record splitting).
211 :    
212 :     Long records with many component values which are simulatenously live
213 :     (recall that single use record addresses are not considered to
214 :     be live) are constructed with rawrecord and rawupdate.
215 :     We allocate space on the heap with rawrecord first, then gradually
216 :     fill it in with rawupdate. This is the technique suggested to me
217 :     by Matthias.
218 :    
219 :     Some restrictions on when this is applicable:
220 :     1. It is not a VECTOR record. The code generator currently does not handle
221 :     this case. VECTOR record uses double indirection like arrays.
222 :     2. All the record component values are defined in the same "basic block"
223 :     as the record constructor. This is to prevent speculative
224 :     record construction.
225 :    
226 :     ----------------------------------------------------------------------
227 :     Name: Allen Leung
228 : leunga 1086 Date: 2002/02/22 01:02:00 EST
229 :     Tag: leunga-20020222-mlrisc-tools
230 :    
231 :     Minor bug fixes in the parser and rewriter
232 :    
233 :     ----------------------------------------------------------------------
234 :     Name: Allen Leung
235 : leunga 1085 Date: 2002/02/21 20:20:00 EST
236 :     Tag: leunga-20020221-peephole
237 :    
238 :     Regenerated the peephole files. Some contained typos in the specification
239 :     and some didn't compile because of pretty printing bugs in the old version
240 :     of 'nowhere'.
241 :    
242 :     ----------------------------------------------------------------------
243 :     Name: Allen Leung
244 : leunga 1079 Date: 2002/02/19 20:20:00 EST
245 :     Tag: leunga-20020219-mlrisc-tools
246 :     Description:
247 :    
248 :     Minor bug fixes to the mlrisc-tools library:
249 :    
250 :     1. Fixed up parsing colon suffixed keywords
251 :     2. Added the ability to shut the error messages up
252 :     3. Reimplemented the pretty printer and fixed up/improved
253 :     the pretty printing of handle and -> types.
254 :     4. Fixed up generation of literal symbols in the nowhere tool.
255 :     5. Added some SML keywords to to sml.sty
256 :    
257 :     ----------------------------------------------------------------------
258 :     Name: Matthias Blume
259 : blume 1078 Date: 2002/02/19 16:20:00 EST
260 :     Tag: blume-20020219-cmffi
261 :     Description:
262 :    
263 :     A wild mix of changes, some minor, some major:
264 :    
265 :     * All C FFI-related libraries are now anchored under $c:
266 :     $/c.cm --> $c/c.cm
267 :     $/c-int.cm --> $c/internals/c-int.cm
268 :     $/memory.cm --> $c/memory/memory.cm
269 :    
270 :     * "make" tool (in CM) now treats its argument pathname slightly
271 :     differently:
272 :     1. If the native expansion is an absolute name, then before invoking
273 :     the "make" command on it, CM will apply OS.Path.mkRelative
274 :     (with relativeTo = OS.FileSys.getDir()) to it.
275 :     2. The argument will be passed through to subsequent phases of CM
276 :     processing without "going native". In particular, if the argument
277 :     was an anchored path, then "make" will not lose track of that anchor.
278 :    
279 :     * Compiler backends now "know" their respective C calling conventions
280 :     instead of having to be told about it by ml-nlffigen. This relieves
281 :     ml-nlffigen from one of its burdens.
282 :    
283 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
284 :    
285 :     * Export C_DEBUG and C_Debug from $c/c.cm.
286 :    
287 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
288 :     subtyping relationship between incomplete pointers and their complete
289 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
290 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
291 :     the pointer target type is complete. In the incomplete case, 'o
292 :     will be instantiated with some "'c iobj" -- a type obtained by
293 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
294 :    
295 :     Operations that work on both incomplete and complete pointer types are
296 :     typed as taking an 'o ptr while operations that require the target to
297 :     be known are typed as taking some ('t, 'c) obj ptr.
298 :    
299 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
300 :     where void is an eqtype without any values. This makes it possible
301 :     to work on voidptr values using functions meant to operate on light
302 :     incomplete pointers.
303 :    
304 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
305 :     been vastly simplified.
306 :    
307 :     ----------------------------------------------------------------------
308 :     Name: Matthias Blume
309 : blume 1076 Date: 2002/02/19 10:48:00 EST
310 :     Tag: blume-20020219-pqfix
311 :     Description:
312 :    
313 :     Applied Chris Okasaki's bug fix for priority queues.
314 :    
315 :     ----------------------------------------------------------------------
316 :     Name: Matthias Blume
317 : blume 1073 Date: 2002/02/15 17:05:00
318 : blume 1069 Tag: Release_110_39
319 :     Description:
320 :    
321 : blume 1073 Last-minute retagging is becoming a tradition... :-(
322 :    
323 :     This is the working release 110.39.
324 :    
325 :     ----------------------------------------------------------------------
326 :     Name: Matthias Blume
327 :     Date: 2002/02/15 16:00:00 EST
328 :     Tag: Release_110_39-orig
329 :     Description:
330 :    
331 : blume 1069 Working release 110.39. New bootfiles.
332 :    
333 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
334 :     with all shells. So I retagged. -Matthias)
335 :    
336 : blume 1069 ----------------------------------------------------------------------
337 :     Name: Matthias Blume
338 : blume 1068 Date: 2002/02/15 14:17:00 EST
339 :     Tag: blume-20020215-showbindings
340 :     Description:
341 :    
342 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
343 :     the latter can be useful for exploring what bindings are available at
344 :     the interactive prompt. (The first function returns only the list
345 :     of symbols that are really bound, the second prints those but also the
346 :     ones that CM's autoloading mechanism knows about.)
347 :    
348 :     ----------------------------------------------------------------------
349 :     Name: Matthias Blume
350 : blume 1067 Date: 2002/02/15 12:08:00 EST
351 :     Tag: blume-20020215-iptrs
352 :     Description:
353 :    
354 :     Two improvements to ml-nlffigen:
355 :    
356 :     1. Write files only if they do not exist or if their current contents
357 :     do not coincide with what's being written. (That is, avoid messing
358 :     with the time stamps unless absolutely necessary.)
359 :    
360 :     2. Implement a "repository" mechanism for generated files related
361 :     to "incomplete pointer types". See the README file for details.
362 :    
363 :     ----------------------------------------------------------------------
364 :     Name: Matthias Blume
365 : blume 1066 Date: 2002/02/14 11:50:00 EST
366 :     Tag: blume-20020214-quote
367 :     Description:
368 :    
369 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
370 :     because of the new and improved tag generation scheme. (Thanks to Allen
371 :     Leung for pointing it out.)
372 :    
373 :     ----------------------------------------------------------------------
374 : george 1065 Name: Lal George
375 :     Date: 2002/02/14 09:55:27 EST 2002
376 :     Tag: george-20020214-isabelle-bug
377 :     Description:
378 :    
379 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
380 :     of Isabelle on the x86.
381 :    
382 :     From Allen:
383 :     -----------
384 :     I've found the problem:
385 :    
386 :     in ra-core.sml, I use the counter "blocked" to keep track of the
387 :     true number of elements in the freeze queue. When the counter goes
388 :     to zero, I skip examining the queue. But I've messed up the
389 :     bookkeeping in combine():
390 :    
391 :     else ();
392 :     case !ucol of
393 :     PSEUDO => (if !cntv > 0 then
394 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
395 :     ^^^^^^^^^^^^^^^^^^^^^^^
396 :     moveu := mergeMoveList(!movev, !moveu)
397 :     )
398 :     else ();
399 :    
400 :     combine() is called to coalesce two nodes u and v.
401 :     I think I was thinking that if the move counts of u and v are both
402 :     greater than zero then after they are coalesced then one node is
403 :     removed from the freeze queue. Apparently I was thinking that
404 :     both u and v are of low degree, but that's clearly not necessarily true.
405 :    
406 :    
407 :     02/12/2002:
408 :     Here's the patch. HOL now compiles.
409 :    
410 :     I don't know how this impact on performance (compile
411 :     time or runtime). This bug caused the RA (especially on the x86)
412 :     to go thru the potential spill phase when there are still nodes on the
413 :     freeze queue.
414 :    
415 :    
416 :    
417 :    
418 :     ----------------------------------------------------------------------
419 : blume 1058 Name: Matthias Blume
420 : blume 1064 Date: 2002/02/13 22:40:00 EST
421 :     Tag: blume-20020213-fptr-rtti
422 :     Description:
423 :    
424 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
425 :     updates.
426 :    
427 :     ----------------------------------------------------------------------
428 :     Name: Matthias Blume
429 : blume 1063 Date: 2002/02/13 16:41:00 EST
430 :     Tag: blume-20020213-cmlpq
431 :     Description:
432 :    
433 :     Added new priority queue export symbols (which have just been added to
434 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
435 :     not compile and the installer would choke.)
436 :    
437 :     ----------------------------------------------------------------------
438 :     Name: Matthias Blume
439 : blume 1062 Date: 2002/02/13 16:15:00 EST
440 :     Tag: blume-20020213-various
441 :     Description:
442 :    
443 :     1. More tweaks to ml-nlffigen:
444 :    
445 :     - better internal datastructures (resulting in slight speedup)
446 :     - "-match" option requires exact match
447 :     - "localized" gensym counters (untagged structs/unions nested within
448 :     other structs/unions or within typedefs get a fresh counter; their
449 :     tag will be prefixed by a concatenation of their parents' tags)
450 :     - bug fixes (related to calculation of transitive closure of types
451 :     to be included in the output)
452 :    
453 :     2. Minor Basis updates:
454 :    
455 :     - added implementations for List.collate and Option.app
456 :    
457 :     ----------------------------------------------------------------------
458 :     Name: Matthias Blume
459 : blume 1060 Date: 2002/02/11 15:55:00 EST
460 :     Tag: blume-20020211-gensym
461 :     Description:
462 :    
463 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
464 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
465 :     names (ML structure names that correspond to unnamed C structs, unions,
466 :     and enums), so that separate runs of ml-nlffigen do not clash.
467 :    
468 :     ----------------------------------------------------------------------
469 :     Name: Matthias Blume
470 : blume 1059 Date: 2002/02/11 12:05:00 EST
471 :     Tag: blume-20020211-gensml
472 :     Description:
473 :    
474 :     A quick fix for a problem with GenSML (in the pgraph-util library):
475 :     Make generation of toplevel "local" optional. (Strictly speaking,
476 :     signature definitions within "local" are not legal SML.)
477 :    
478 :     Other than that: updates to INSTALL and cm/TODO.
479 :    
480 :     ----------------------------------------------------------------------
481 :     Name: Matthias Blume
482 : blume 1058 Date: 2002/02/08 15:00:00 EST
483 :     Tag: blume-20020208-uniquepid
484 :     Description:
485 :    
486 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
487 :    
488 :     1. The installer (config/install.sh) has gotten smarter:
489 :    
490 :     - Configuration options are a bit easier to specify now
491 :     (in config/targets).
492 :     - Bug in recognizing .tar.bz2 files fixed.
493 :     - Installer automatically resolves dependencies between
494 :     configuration options (e.g., if you ask for eXene, you will
495 :     also get cml -- regardless whether you asked for it or not).
496 :     - Installer can run in "quieter mode" by setting the environment
497 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
498 :     "completely silent", though.
499 :     - Build HashCons library as part of smlnj-lib.
500 :    
501 :     2. A new scheme for assigning persistent identifiers to compilation
502 :     units (and, by extension, to types etc.) has been put into place.
503 :     This fixes a long-standing bug where types and even dynamic values
504 :     can get internally confused, thereby compromising type safety
505 :     (abstraction) and dynamic correctness. See
506 :    
507 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
508 :    
509 :     for an example of how things could go wrong until now.
510 :    
511 :     The downside of the new scheme is that pids are not quite as
512 :     persistent as they used to be: CM will generate a fresh pid
513 :     for every compilation unit that it thinks it sees for the first
514 :     time. That means that if you compile starting from a clean, fresh
515 :     source tree at two different times, you end up with different
516 :     binaries.
517 :    
518 :     Cutoff recompilation, however, has not been compromised because
519 :     CM keeps pid information in special caches between runs.
520 :    
521 :     ----------------------------------------------------------------------
522 : george 1053 Name: Lal George
523 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
524 :     Tag: <none>
525 :     Description:
526 :    
527 :     Compilers that generate assembly code may produce global labels
528 :     whose value is resolved at link time. The various peephole optimization
529 :     modules did not take this in account.
530 :    
531 :     TODO. The Labels.addrOf function should really return an option
532 :     type so that clients are forced to deal with this issue, rather
533 :     than an exception being raised.
534 :    
535 :     ----------------------------------------------------------------------
536 :     Name: Lal George
537 : george 1053 Date: 2002/02/06 13:55:02 EST
538 :     Tag: george-20020206-ra-breakup
539 :     Description:
540 :    
541 :     1. A bug fix from Allen.
542 :    
543 :     A typo causes extra fstp %st(0)'s to be generated at compensation
544 :     edges, which might cause stack underflow traps at runtime. This
545 :     occurs in fft where there are extraneous fstps right before the 'into'
546 :     trap instruction (in this case they are harmless since none of the
547 :     integers overflow.)
548 :    
549 : george 1054 2. Pulled out various utility modules that were embedded in the modules
550 : george 1053 of the register allocator. I need these modules for other purposes, but
551 : george 1054 they are not complete enough to put into a library (just yet).
552 : george 1053 ----------------------------------------------------------------------
553 : blume 1049 Name: Matthias Blume
554 :     Date: 2002/01/31 16:05:00 EST
555 :     Tag: blume-20020131-sparc-ccalls
556 :     Description:
557 :    
558 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
559 :     of extra stack space by mistake. Fixed.
560 :    
561 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
562 :    
563 :     ----------------------------------------------------------------------
564 : leunga 1048 Name: Allen Leung
565 :     Date: 2002/01/30
566 :     Tag: leunga-20020130-nowhere-bug-fix
567 :     Description:
568 :    
569 :     MLRISC bug fixes:
570 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
571 :     2. MachineInt.fromString was negating its value.
572 :    
573 :     ----------------------------------------------------------------------
574 : blume 1039 Name: Matthias Blume
575 : blume 1047 Date: 2002/01/29
576 :     Tag: blume-20020129-INSTALL
577 :     Description:
578 :    
579 :     - Added somewhat detailed installation instructions (file INSTALL).
580 :     - Fixed curl-detection bug in config/install.sh.
581 :     - It is now possible to select the URL getter using the URLGETTER
582 :     environment variable:
583 :    
584 :     not set / "unknown" --> automatic detection (script tries wget,
585 :     curl, and lynx)
586 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
587 :     how to properly invoke them)
588 :     other --> use $URLGETTER directly, it must take
589 :     precisely two command-line arguments
590 :     (source URL and destination file name)
591 :    
592 :     ----------------------------------------------------------------------
593 :     Name: Matthias Blume
594 : blume 1044 Date: 2002/01/28
595 : blume 1045 Tag: blume-20020128-sparc-ccalls
596 :     Description:
597 :    
598 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
599 :     - Make use of the allocParam argument in sparc-c-calls.
600 :    
601 :     ----------------------------------------------------------------------
602 :     Name: Matthias Blume
603 :     Date: 2002/01/28
604 : blume 1044 Tag: blume-20020128-allocParam
605 :     Description:
606 :    
607 :     John Reppy: Changes c-calls API to accept client-callback for
608 :     allocating extra stack space.
609 :     me: Corresponding changes to mlriscGen (using a dummy argument that
610 :     does not change the current behavior).
611 :    
612 :     ----------------------------------------------------------------------
613 :     Name: Matthias Blume
614 : blume 1040 Date: 2002/01/28 12:00:00
615 :     Tag: Release_110_38
616 :     Description:
617 :    
618 :     This time for real!!!
619 :    
620 :     ----------------------------------------------------------------------
621 :     Name: Matthias Blume
622 : blume 1039 Date: 2002/01/28 10:56:00 EST
623 :     Tag: blume-20020128-retraction
624 :     Description:
625 :    
626 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
627 :     with blume-Release_110_38-retracted.)
628 :    
629 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
630 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
631 :    
632 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
633 :     and enums fixed. The naming of corresponding ML identifiers should
634 :     now be consistent again.
635 :    
636 :     ----------------------------------------------------------------------
637 : leunga 1038 Name: Allen Leung
638 :     Date: 2002/01/27
639 :     Tag: leunga-20020127-nowhere
640 :     Description:
641 :    
642 :     Added a target called nowhere in the configuration scripts.
643 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
644 :     programs with where-clauses into legal SML code) during installation.
645 :    
646 :     ----------------------------------------------------------------------
647 : blume 1036 Name: Matthias Blume
648 : leunga 1038 Date: 2002/01/25 21:27:00 EST
649 : blume 1039 Tag: blume-Release_110_38-retracted
650 : blume 1037 Description:
651 :    
652 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
653 :    
654 :     README will be added later.
655 :    
656 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
657 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
658 :    
659 : blume 1037 ----------------------------------------------------------------------
660 :     Name: Matthias Blume
661 : blume 1036 Date: 2002/01/25
662 :     Tag: blume-20020125-ffi
663 :     Description:
664 :    
665 :     A large number of tweaks and improvements to ml-nlffi-lib and
666 :     ml-nlffigen:
667 :    
668 :     - ML represenation types have been streamlined
669 :     - getter and setter functions work with concrete values, not abstract
670 :     ones where possible
671 :     - ml-nlffigen command line more flexible (see README file there)
672 :     - some bugs have been fixed (hopefully)
673 :    
674 :     ----------------------------------------------------------------------
675 : george 1034 Name: Lal George
676 :     Date: 2002/01/24
677 :     Tag: george-20020124-risc-ra-interface
678 :     Description:
679 :    
680 :     There is a dramatic simplification in the interface to the
681 :     register allocator for RISC architectures as a result of making
682 :     parallel copy instructions explicit.
683 :    
684 :     ----------------------------------------------------------------------
685 : blume 1028 Name: Matthias Blume
686 : blume 1032 Date: 2002/01/22
687 :     Tag: blume-20020122-x86-ccalls
688 :     Description:
689 :    
690 :     Bug fix for c-calls on x86 (having to do with how char- and
691 :     short-arguments are being handled).
692 :    
693 :     ----------------------------------------------------------------------
694 :     Name: Matthias Blume
695 : blume 1031 Date: 2002/01/21
696 :     Tag: blume-20020121-ff
697 :     Description:
698 :    
699 :     Another day of fiddling with the FFI...
700 :    
701 :     1. Bug fix/workaround: CKIT does not complain about negative array
702 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
703 :     (Otherwise a negative dimension would send it into an infinite loop.)
704 :    
705 :     2. Some of the abstract types (light objects, light pointers, most "base"
706 :     types) in structure C are now eqtypes.
707 :    
708 :     3. Added constructors and test functions for NULL function pointers.
709 :    
710 :     ----------------------------------------------------------------------
711 :     Name: Matthias Blume
712 : blume 1028 Date: 2002/01/18
713 : blume 1030 Tag: blume-20020118-ready-for-new-release
714 :     Description:
715 :    
716 :     Made config/srcarchiveurl point to a new place. (Will provide boot
717 :     files shortly.)
718 :    
719 :     Maybe we christen this to be 110.38?
720 :    
721 :     ----------------------------------------------------------------------
722 :     Name: Matthias Blume
723 :     Date: 2002/01/18
724 : blume 1029 Tag: blume-20020118-more-ffifiddle
725 :     Description:
726 :    
727 :     Today's FFI fiddling:
728 :    
729 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
730 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
731 :     of abstract ones.
732 :    
733 :     - Provided word-style bit operations etc. for "int" representation
734 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
735 :     and Long.
736 :    
737 :     ----------------------------------------------------------------------
738 :     Name: Matthias Blume
739 :     Date: 2002/01/18
740 : blume 1028 Tag: blume-20020118-use-x86-fp
741 :     Description:
742 :    
743 :     Now that x86-fast-fp seems to be working, I turned it back on again
744 :     by default. (Seems to work fine now, even with the FFI.)
745 :    
746 :     Other than that, I added some documentation about the FFI to
747 :     src/ml-nlffigen/README and updated the FFI test examples in
748 :     src/ml-nlffi-lib/Tests/*.
749 :    
750 :     ----------------------------------------------------------------------
751 : leunga 1025 Name: Allen Leung
752 :     Date: 2002/01/17
753 :     Tag: leunga-20020117-x86-fast-fp-call
754 :     Description:
755 :    
756 :     1. Fixed a problem with handling return fp values when x86's fast fp
757 :     mode is turned on.
758 :    
759 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
760 :     of %f32.
761 :    
762 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
763 :    
764 :     ----------------------------------------------------------------------
765 : blume 1021 Name: Matthias Blume
766 :     Date: 2002/01/16
767 :     Tag: blume-20020116-ffifiddle
768 :     Description:
769 :    
770 :     More fiddling with the FFI interface:
771 :    
772 :     - Make constness 'c instead of rw wherever possible. This eliminates
773 :     the need for certain explicit coercions. (However, due to ML's
774 :     value polymorphism, there will still be many cases where explicit
775 :     coercions are necessary. Phantom types are not the whole answer
776 :     to modeling a subtyping relationship in ML.)
777 :    
778 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
779 :    
780 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
781 :     clumsy workaround for varargs problem.)
782 :    
783 :     ----------------------------------------------------------------------
784 : george 1016 Name: Lal George
785 :     Date: 2002/01/15
786 :     Tag: <none>
787 :     Description:
788 :    
789 :     1. Since COPY instructions are no longer native to the architecture,
790 :     a generic functor can be used to implement the expandCopies function.
791 :    
792 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
793 :     TEXT segment.
794 :    
795 :     ----------------------------------------------------------------------
796 : blume 1011 Name: Matthias Blume
797 : blume 1015 Date: 2002/01/15
798 :     Tag: blume-20020115-ffiupdates
799 :     Description:
800 :    
801 :     1. Fix for bug resulting in single-precision float values being returned
802 :     incorrectly from FFI calls.
803 :    
804 :     2. Small modifications to C FFI API:
805 :    
806 :     - memory-allocation routines return straight objects (no options)
807 :     and raise an exception in out-of-memory situations
808 :     - unsafe extensions to cast between function pointers and pointers
809 :     from/to ints
810 :     - added structure C_Debug as an alternative to structure C where
811 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
812 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
813 :     but also takes a (possibly empty) list of existing library handles
814 :     that the current library depends on
815 :    
816 :     ----------------------------------------------------------------------
817 :     Name: Matthias Blume
818 : blume 1011 Date: 2002/01/10
819 :     Tag: blume-20020110-newffigen
820 :     Description:
821 :    
822 :     1. Updates to portable graph code.
823 :    
824 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
825 :     more scalable now so that even huge interfaces such as the one
826 :     for GTK compile in finite time and space. :-)
827 :     See src/ml-nlffigen/README for details on what's new.
828 :    
829 :     ----------------------------------------------------------------------
830 : george 1003 Name: Lal George
831 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
832 :     Tag: george-20011206-rm-native-copy
833 :     Description:
834 :    
835 :     Removed the native COPY and FCOPY instructions
836 :     from all the architectures and replaced it with the
837 :     explicit COPY instruction from the previous commit.
838 :    
839 :     It is now possible to simplify many of the optimizations
840 :     modules that manipulate copies. This has not been
841 :     done in this change.
842 :    
843 :     ----------------------------------------------------------------------
844 :     Name: Lal George
845 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
846 :     Tag: george-20011206-mlrisc-instruction
847 :     Description:
848 : george 824
849 : george 1003 Changed the representation of instructions from being fully abstract
850 :     to being partially concrete. That is to say:
851 :    
852 :     from
853 :     type instruction
854 :    
855 :     to
856 :     type instr (* machine instruction *)
857 :    
858 :     datatype instruction =
859 :     LIVE of {regs: C.cellset, spilled: C.cellset}
860 :     | KILL of {regs: C.cellset, spilled: C.cellset}
861 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
862 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
863 :     | INSTR of instr
864 :    
865 :     This makes the handling of certain special instructions that appear on
866 :     all architectures easier and uniform.
867 :    
868 :     LIVE and KILL say that a list of registers are live or killed at the
869 :     program point where they appear. No spill code is generated when an
870 :     element of the 'regs' field is spilled, but the register is moved to
871 :     the 'spilled' (which is present, more for debugging than anything else).
872 :    
873 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
874 :     We used to generate:
875 :    
876 :     DEFFREG f1
877 :     f1 := f2 + f3
878 :     trapb
879 :    
880 :     but now generate:
881 :    
882 :     f1 := f2 + f3
883 :     trapb
884 :     LIVE {regs=[f1,f2,f3], spilled=[]}
885 :    
886 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
887 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
888 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
889 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
890 :     this architecture).
891 :    
892 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
893 :     all the architectures. This will result in further simplification of the
894 :     register allocator that must be aware of them for coalescing purposes, and
895 :     will also simplify certain aspects of the machine description that provides
896 :     callbacks related to parallel copies.
897 :    
898 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
899 :     machine instruction set!
900 :    
901 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
902 :     functions for making porting easier -- essentially converting upper case
903 :     to lower case. All machine instructions (of type instr) are in upper case,
904 :     and the lower case form generates an MLRISC instruction. For example on
905 :     the alpha we have:
906 :    
907 :     datatype instr =
908 :     LDA of {r:cell, b:cell, d:operand}
909 :     | ...
910 :    
911 :     val lda : {r:cell, b:cell, d:operand} -> instruction
912 :     ...
913 :    
914 :     where lda is just (INSTR o LDA), etc.
915 :    
916 : leunga 815 ----------------------------------------------------------------------
917 : blume 986 Name: Matthias Blume
918 : blume 993 Date: 2001/11/22 21:40:00 EST
919 :     Tag: Release_110_37
920 :     Description:
921 :    
922 :     Release 110.37. This time for real.
923 :    
924 :     ----------------------------------------------------------------------
925 :     Name: Matthias Blume
926 : blume 988 Date: 2001/11/21 16:35:00 EST
927 :     Tag: blume-20011121-foot-in-mouth
928 :     Description:
929 :    
930 :     Removed the "Release_110_37" tag because of a serious bug.
931 :     This will be re-tagged once the bug is fixed.
932 :    
933 :     ----------------------------------------------------------------------
934 :     Name: Matthias Blume
935 : blume 987 Date: 2001/11/21 16:14:00 EST
936 :     Tag: blume-20011121-forgottenfile
937 :     Description:
938 :    
939 :     Forgot to add a file. (Just a .tex-file -- part of
940 :     the CM manual source.)
941 :    
942 :     ----------------------------------------------------------------------
943 :     Name: Matthias Blume
944 : blume 986 Date: 2001/11/21 16:10:00 EST
945 : blume 988 Tag: blume-20011121-invalid_110_37
946 : blume 986 Description:
947 :    
948 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
949 :     because we found a serious bug in all non-x86 backends.
950 :     - Matthias
951 :    
952 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
953 :     so that code object name strings are directly inserted into code
954 :     objects at code generation time. The only business the runtime system
955 :     has with this is now to read the name strings on occasions.
956 :     (The encoding of the name string has also changed somewhat.)
957 :    
958 :     2. CM now implements a simple "set calculus" for specifying export lists.
959 :     In particular, it is now possible to refer to the export lists of
960 :     other libraries/groups/sources and form unions as well as differences.
961 :     See the latest CM manual for details.
962 :    
963 :     3. An separate notion of "proxy" libraries has again be eliminated from
964 :     CM's model. (Proxy libraries are now simply a special case of using
965 :     the export list calculus.)
966 :    
967 :     4. Some of the existing libraries now take advantage of the new set
968 :     calculus.
969 :     (Notice that not all libraries have been converted because some
970 :     of the existing .cm-files are supposed to be backward compatible
971 :     with 110.0.x.)
972 :    
973 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
974 :     "General.exnMessage"! The former relies on a certain hook to be
975 :     initialized, and that often does not happen in the stand-alone case.)
976 :    
977 :     ----------------------------------------------------------------------
978 : george 984 Name: Lal George
979 :     Date: 2001/11/21 13:56:18 EST
980 :     Tag: george-2001121-pseudo-ops
981 :     Description:
982 :    
983 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
984 :     ought to never be any question of incompatabilities with
985 :     pseudo-op syntax expected by host assemblers.
986 :    
987 :     For now, only modules supporting GAS syntax are implemented
988 :     but more should follow, such as MASM, and vendor assembler
989 :     syntax, e.g. IBM as, Sun as, etc.
990 :    
991 :     ----------------------------------------------------------------------
992 : blume 977 Name: Matthias Blume
993 :     Date: 2001/11/14 11:52:00 EST
994 :     Tag: blume-20011114-srcname
995 :     Description:
996 :    
997 :     1. Routed the name of the current source file to mlriscgen where it
998 :     should be directly emitted into the code object. (This last part
999 :     is yet to be done.)
1000 :    
1001 :     2. Some cleanup of the pgraph code to make it match the proposal that
1002 :     I put out the other day. (The proposal notwithstanding, things are
1003 :     still in flux here.)
1004 :    
1005 :     ----------------------------------------------------------------------
1006 : george 976 Name: Lal George
1007 :     Date: 2001/11/14 09:44:04 EST
1008 :     Tag:
1009 :     Description:
1010 :    
1011 :     Fix for a backpatching bug reported by Allen.
1012 :    
1013 :     Because the boundary between short and long span-dependent
1014 :     instructions is +/- 128, there are an astounding number of
1015 :     span-dependent instructions whose size is over estimated.
1016 :    
1017 :     Allen came up with the idea of letting the size of span
1018 :     dependent instructions be non-monotonic, for a maxIter
1019 :     number of times, after which the size must be monotonically
1020 :     increasing.
1021 :    
1022 :     This table shows the number of span-dependent instructions
1023 :     whose size was over-estimated as a function of maxIter, for the
1024 :     file Parse/parse/ml.grm.sml:
1025 :    
1026 :     maxIter # of instructions:
1027 :     10 687
1028 :     20 438
1029 :     30 198
1030 :     40 0
1031 :    
1032 :     In compiling the compiler, there is no significant difference in
1033 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1034 :     my measurements showed that maxIter=40 was a tad faster than
1035 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1036 :     point within 13 iterations, so fixing maxIter at 40, while high,
1037 :     is okay.
1038 :    
1039 :     ----------------------------------------------------------------------
1040 : blume 975 Name: Matthias Blume
1041 :     Date: 2001/10/31 15:25:00 EST
1042 :     Tag: blume-20011031-pgraph
1043 :     Description:
1044 :    
1045 :     CKIT:
1046 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1047 :     argument identifiers.
1048 :     * Changed the return type of TypeUtil.getFunction accordingly.
1049 :     * Type equality ignores the argument names.
1050 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1051 :     if there is a mismatch.
1052 :    
1053 :     installation script:
1054 :     * attempts to use "curl" if available (unless "wget" is available as well)
1055 :    
1056 :     CM:
1057 :     * has an experimental implementation of "portable graphs" which I will
1058 :     soon propose as an implementation-independent library format
1059 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1060 :    
1061 :     NLFFI-LIB:
1062 :     * some cleanup (all cosmetic)
1063 :    
1064 :     NLFFIGEN:
1065 :     * temporarily disabled the mechanism that suppresses ML output for
1066 :     C definitions whose identifiers start with an underscore character
1067 :     * generate val bindings for enum constants
1068 :     * user can request that only one style (light or heavy) is being used;
1069 :     default is to use both (command-line arguments: -heavy and -light)
1070 :     * fixed bug in handling of function types involving incomplete pointers
1071 :     * generate ML entry points that take record arguments (i.e., using
1072 :     named arguments) for C functions that have a prototype with named
1073 :     arguments
1074 :     (see changes to CKIT)
1075 :    
1076 :     ----------------------------------------------------------------------
1077 : leunga 972 Name: Allen Leung
1078 :     Date: 2001/10/27 20:34:00 EDT
1079 :     Tag: leunga-20011027-x86-fast-fp-call
1080 :     Description:
1081 :    
1082 :     Fixed the bug described in blume-20010920-slowfp.
1083 :    
1084 :     The fix involves
1085 :     1. generating FCOPYs in FSTP in ia32-svid
1086 :     2. marking a CALL with the appropriate annotation
1087 :    
1088 :     ----------------------------------------------------------------------
1089 : blume 961 Name: Matthias Blume
1090 :     Date: 2001/10/16 11:32:00 EDT
1091 :     Tag: blume-20011016-netbsd
1092 :     Description:
1093 :    
1094 :     Underscore patch from Chris Richards (fixing problem with compiling
1095 :     runtime system under recent NetBSD).
1096 :    
1097 :     ----------------------------------------------------------------------
1098 : george 959 Name: Allen Leung
1099 :     Date: 2001/10/12 17:18:32 EDT 2001
1100 :     Tag: leung-20011012-x86-printflowgraph
1101 :     Description:
1102 :    
1103 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1104 :    
1105 :     ----------------------------------------------------------------------
1106 : george 952 Name: Lal George
1107 : george 958 Date: 2001/10/11 23:51:34 EDT
1108 :     Tag: george-20011011-too-many-instrs
1109 :     Description:
1110 :    
1111 :     The representation of a program point never expected to see more
1112 :     than 65536 instructions in a basic block!
1113 :    
1114 :     ----------------------------------------------------------------------
1115 :     Name: Lal George
1116 : george 952 Date: 2001/10/09 09:41:37 EDT
1117 :     Tag: george-20011008-mlrisc-labels
1118 :     Description:
1119 :    
1120 :     Changed the machine description files to support printing of
1121 :     local and global labels in assembly code, based on host assembler
1122 :     conventions.
1123 :    
1124 :     ----------------------------------------------------------------------
1125 : blume 916 Name: Matthias Blume
1126 : blume 939 Date: 2001/09/25 15:25:00 EDT
1127 :     Tag: blume-20010925-exninfo
1128 :     Description:
1129 :    
1130 :     I provided a non-hook implementation of exnName (at the toplevel) and
1131 :     made the "dummy" implementation of exnMessage (at the toplevel) more
1132 :     useful: if nothing gets "hooked in", then at least you are going to
1133 :     see the exception name and a message indicating why you don't see more.
1134 :    
1135 :     [For the time being, programs that need exnMessage and want to use
1136 :     ml-build should either use General.exnMessage (strongly recommended) or
1137 :     refer to structure General at some other point so that CM sees a
1138 :     static dependency.]
1139 :    
1140 :     [Similar remarks go for "print" and "use": If you want to use their
1141 :     functionality in stand-alone programs generated by ml-build, then use
1142 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1143 :    
1144 :     ----------------------------------------------------------------------
1145 :     Name: Matthias Blume
1146 : blume 938 Date: 2001/09/20 17:28:00 EDT
1147 :     Tag: blume-20010920-slowfp
1148 :     Description:
1149 :    
1150 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1151 :    
1152 :     ----------------------------------------------------------------------
1153 :     Name: Matthias Blume
1154 : blume 937 Date: 2001/09/20 17:20:00 EDT
1155 :     Tag: blume-20010920-canonicalpaths
1156 :     Description:
1157 :    
1158 :     0. Updated the BOOT file (something that I forgot to do earlier).
1159 :    
1160 :     1. Small internal change to CM so that it avoids "/../" in filenames
1161 :     as much as possible (but only where it is safe).
1162 :    
1163 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1164 :     that arguments that contain delimiters are passed through correctly.
1165 :     This change also means that all "special" arguments of the form
1166 :     @SMLxxx... must come first.
1167 :    
1168 :     3. Changed install script to put relative anchor names for tool commands
1169 :     into pathconfig.
1170 :    
1171 :     ----------------------------------------------------------------------
1172 :     Name: Matthias Blume
1173 : blume 932 Date: 2001/09/18 15:35:00 EDT
1174 :     Tag: blume-20010918-readme11036
1175 :     Description:
1176 :    
1177 :     Added README files.
1178 :    
1179 :     ----------------------------------------------------------------------
1180 :     Name: Matthias Blume
1181 : blume 929 Date: 2001/09/18 11:45:00 EDT
1182 :     Tag: Release_110_36 (retag)
1183 :     Description:
1184 :    
1185 :     Fixed mistake in config/preloads. Retagged as 110.36.
1186 :    
1187 :     ----------------------------------------------------------------------
1188 :     Name: Matthias Blume
1189 : blume 928 Date: 2001/09/18 09:40:00 EDT
1190 : blume 929 Tag: Release_110_36_orig (tag changed)
1191 : blume 928 Description:
1192 :    
1193 :     New version (110.36). New bootfiles.
1194 :    
1195 :     ----------------------------------------------------------------------
1196 :     Name: Matthias Blume
1197 : blume 927 Date: 2001/09/14 16:15:00 EDT
1198 :     Tag: blume-20010914-x86fastfp
1199 :     Description:
1200 :    
1201 :     John committed some changes that Allen made, in particular a (hopefully)
1202 :     correctly working version of the x86-fp module.
1203 :    
1204 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1205 :     flag to "true". Everything seems to compile to a fixpoint ok, and
1206 :     "mandelbrot" speeds up by about 15%.
1207 :    
1208 :     ----------------------------------------------------------------------
1209 :     Name: Matthias Blume
1210 : blume 923 Date: 2001/09/13 11:20:00 EDT
1211 :     Tag: blume-20010913-minimal
1212 :     Description:
1213 :    
1214 :     1. Stefan Monnier's patch to fix a miscompilation problem that
1215 :     was brought to light by John Reppy's work on Moby.
1216 :    
1217 :     2. Implemented a minimal "structure Compiler" that contains just
1218 :     "version" and "architecture". The minimal version will be
1219 :     available when the full version is not. This is for backward-
1220 :     compatibility with code that wants to test Compiler.version.
1221 :    
1222 :     ----------------------------------------------------------------------
1223 :     Name: Matthias Blume
1224 : blume 920 Date: 2001/08/28 14:03:00 EDT
1225 :     Tag: blume-20010828-ml-lex
1226 :     Description:
1227 :    
1228 :     Fix for bug 1581, received from Neophytos Michael.
1229 :    
1230 :     ----------------------------------------------------------------------
1231 :     Name: Matthias Blume
1232 : blume 918 Date: 2001/08/27 11:20:00 EDT
1233 :     Tag: blume-20010827-readme11035
1234 :     Description:
1235 :    
1236 :     Fleshed out the README file for 110.35.
1237 :    
1238 :     ----------------------------------------------------------------------
1239 :     Name: Matthias Blume
1240 : blume 916 Date: 2001/08/24 17:10:00 EDT
1241 :     Tag: Release_110_35
1242 :     Description:
1243 :    
1244 :     New version number (110.35). New bootfiles.
1245 :    
1246 :     ----------------------------------------------------------------------
1247 : george 915 Name: Lal George
1248 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1249 :     Tag: george-20010824-MLRISC-graphs
1250 :     Description:
1251 :    
1252 :     removed clusters from MLRISC completely and replaced with graphs.
1253 :    
1254 :     ----------------------------------------------------------------------
1255 :     Name: Matthias Blume
1256 : blume 905 Date: 2001/08/23 17:50:00 EDT
1257 :     Tag: blume-20010823-toplevel
1258 :     Description:
1259 :    
1260 :     - some reorganization of the code that implements various kinds of
1261 :     environments in the compiler (static, dynamic, symbolic, combined)
1262 :     - re-implemented the EnvRef module so that evalStream works properly
1263 :     (if the stream contains references to "use", "CM.make", etc.)
1264 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1265 :    
1266 :     ----------------------------------------------------------------------
1267 :     Name: Matthias Blume
1268 : blume 904 Date: 2001/08/20 15:50 EDT
1269 :     Tag: blume20010820-slipup
1270 :     Description:
1271 :    
1272 :     I forgot to commit a few files. Here they are...
1273 :    
1274 :     ----------------------------------------------------------------------
1275 :     Name: Matthias Blume
1276 : blume 903 Date: 2001/08/20 15:35:00 EDT
1277 :     Tag: blume-20010820-debugprof
1278 :     Description:
1279 :    
1280 :     !!!! NEW BOOTFILES !!!!
1281 :    
1282 :     This is another round of reorganizing the compiler sources. This
1283 :     time the main goal was to factor out all the "instrumentation"
1284 :     passes (for profiling and backtracing) into their own library.
1285 :     The difficulty was to do it in such a way that it does not depend
1286 :     on elaborate.cm but only on elabdata.cm.
1287 :    
1288 :     Therefore there have been further changes to both elaborate.cm and
1289 :     elabdata.cm -- more "generic" things have been moved from the former
1290 :     to the latter. As a result, I was forced to split the assignment
1291 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1292 :     SML/NJ-specific. Since it would have been awkward to maintain,
1293 :     I bit the bullet and actually _changed_ the mapping between these
1294 :     numbers and primtycs. The bottom line of this is that you need
1295 :     a new set of bin- and bootfiles.
1296 :    
1297 :     I have built new bootfiles for all architectures, so doing a fresh
1298 :     checkout and config/install.sh should be all you need.
1299 :    
1300 :     The newly created library's name is
1301 :    
1302 :     $smlnj/viscomp/debugprof.cm
1303 :    
1304 :     and its sources live under
1305 :    
1306 :     src/compiler/DebugProf
1307 :    
1308 :     ----------------------------------------------------------------------
1309 :     Name: Matthias Blume
1310 : blume 902 Date: 2001/08/15 17:15:00 EDT
1311 :     Tag: blume-20010815-compreorg
1312 :     Description:
1313 :    
1314 :     This is a first cut at reorganizing the CM libraries that make up the
1315 :     core of the compiler. The idea is to separate out pieces that could
1316 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1317 :    
1318 :     The current status is a step in this direction, but it is not quite
1319 :     satisfactory yet. Expect more changes in the future.
1320 :    
1321 :     Here is the current (new) organization...
1322 :    
1323 :     What used to be $smlnj/viscomp/core.cm is now divided into
1324 :     six CM libraries:
1325 :    
1326 :     $smlnj/viscomp/basics.cm
1327 :     /parser.cm
1328 :     /elabdata.cm
1329 :     /elaborate.cm
1330 :     /execute.cm
1331 :     /core.cm
1332 :    
1333 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1334 :     All these libraries are proxy libraries that contain precisely
1335 :     one CM library component. Here are the locations of the components
1336 :     (all within the src/compiler tree):
1337 :    
1338 :     Basics/basics.cm
1339 :     Parse/parser.cm
1340 :     ElabData/elabdata.cm
1341 :     Elaborator/elaborate.cm
1342 :     Execution/execute.cm
1343 :     core.cm
1344 :    
1345 :     [This organization is the same that has been used already
1346 :     for a while for the architecture-specific parts of the visible
1347 :     compiler and for the old version of core.cm.]
1348 :    
1349 :     As you will notice, many source files have been moved from their
1350 :     respective original locations to a new home in one of the above
1351 :     subtrees.
1352 :    
1353 :     The division of labor between the new libraries is the following:
1354 :    
1355 :     basics.cm:
1356 :     - Simple, basic definitions that pertain to many (or all) of
1357 :     the other libraries.
1358 :     parser.cm:
1359 :     - The SML parser, producing output of type Ast.dec.
1360 :     - The type family for Ast is also defined and exported here.
1361 :     elabdata.cm:
1362 :     - The datatypes that describe input and output of the elaborator.
1363 :     This includes types, absyn, and static environments.
1364 :     elaborator.cm:
1365 :     - The SML/NJ type checker and elaborator.
1366 :     This maps an Ast.dec (with a given static environment) to
1367 :     an Absyn.dec (with a new static environment).
1368 :     - This libraries implements certain modules that used to be
1369 :     structures as functors (to remove dependencies on FLINT).
1370 :     execute.cm:
1371 :     - Everything having to do with executing binary code objects.
1372 :     - Dynamic environments.
1373 :     core.cm:
1374 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1375 :     - Top-level modules.
1376 :     - FLINT (this should eventually become its own library)
1377 :    
1378 :     Notes:
1379 :    
1380 :     I am not 100% happy with the way I separated the elaborator (and its
1381 :     data structures) from FLINT. Two instances of the same problem:
1382 :    
1383 :     1. Data structures contain certain fields that carry FLINT-specific
1384 :     information. I hacked around this using exn and the property list
1385 :     module from smlnj-lib. But the fact that there are middle-end
1386 :     specific fields around at all is a bit annoying.
1387 :    
1388 :     2. The elaborator calculates certain FLINT-related information. I tried
1389 :     to make this as abstract as I could using functorization, but, again,
1390 :     the fact that the elaborator has to perform calculations on behalf
1391 :     of the middle-end at all is not nice.
1392 :    
1393 :     3. Having to used exn and property lists is unfortunate because it
1394 :     weakens type checking. The other alternative (parameterizing
1395 :     nearly *everything*) is not appealing, though.
1396 :    
1397 :     I removed the "rebinding =" warning hack because due to the new organization
1398 :     it was awkward to maintain it. As a result, the compiler now issues some of
1399 :     these warnings when compiling init.cmi during bootstrap compilation. On
1400 :     the plus side, you also get a warning when you do, for example:
1401 :     val op = = Int32.+
1402 :     which was not the case up to now.
1403 :    
1404 :     I placed "assign" and "deref" into the _Core structure so that the
1405 :     code that deals with the "lazy" keyword can find them there. This
1406 :     removes the need for having access to the primitive environment
1407 :     during elaboration.
1408 :    
1409 :     ----------------------------------------------------------------------
1410 :     Name: Matthias Blume
1411 : blume 898 Date: 2001/08/13
1412 :     Tag: blume-20010813-closures
1413 :     Description:
1414 :    
1415 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1416 :     performance of certain loops by avoiding needless closure allocation.
1417 :    
1418 :     ----------------------------------------------------------------------
1419 : george 890 Name: Lal George
1420 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1421 :     Tag: george-20010731-x86-fmalloc
1422 :     Description: Fixed bug in x86 calls
1423 :    
1424 :     There was a bug where call instructions would mysteriously
1425 :     vanish. The call instruction had to be one that returned
1426 :     a floating point value.
1427 :    
1428 :     ----------------------------------------------------------------------
1429 :     Name: Lal George
1430 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1431 :     Tag: george-20010719-simple-cells
1432 :     Description:
1433 :    
1434 :     I have dramatically simplified the interface for CELLS in MLRISC.
1435 :    
1436 :     In summary, the cells interface is broken up into three parts:
1437 :    
1438 :     1. CellsBasis : CELLS_BASIS
1439 :    
1440 :     CellsBasis is a top level structure and common for all
1441 :     architectures. it contains the definitions of basic datatypes
1442 :     and utility functions over these types.
1443 :    
1444 :     2. functor Cells() : CELLS
1445 :    
1446 :     Cells generates an interface for CELLS that incorporates the
1447 :     specific resources on the target architecture, such as the
1448 :     presence of special register classes, their number and size,
1449 :     and various useful substructures.
1450 :    
1451 :     3. <ARCH>CELLS
1452 :    
1453 :     e.g. SparcCells: SPARCCELLS
1454 :    
1455 :     <ARCH>CELLS usually contains additional bindings for special
1456 :     registers on the architecture, such as:
1457 :    
1458 :     val r0 : cell (* register zero *)
1459 :     val y : cell (* Y register *)
1460 :     val psr : cell (* processor status register *)
1461 :     ...
1462 :    
1463 :     The structure returned by applying the Cells functor is opened
1464 :     in this interface.
1465 :    
1466 :     The main implication of all this is that the datatypes for cells is
1467 :     split between CellsBasis and CELLS -- a fairly simple change for user
1468 :     code.
1469 :    
1470 :     In the old scheme the CELLS interface had a definitional binding of
1471 :     the form:
1472 :    
1473 :     signature CELLS = sig
1474 :    
1475 :     structure CellsBasis = CellsBasis
1476 :    
1477 :     ...
1478 :    
1479 :     end
1480 :    
1481 :     With all the sharing constraints that goes on in MLRISC, this old
1482 :     design quickly leads to errors such as:
1483 :    
1484 :     "structure definition spec inside of sharing ... "
1485 :    
1486 :    
1487 :     and appears to require an unacceptable amount of sharing and where
1488 :     constraint hackery.
1489 :    
1490 :     I think this error message (the interaction of definitional specs and
1491 :     sharing) requires more explanation on our web page.
1492 :    
1493 :     ----------------------------------------------------------------------
1494 : blume 826 Name: Matthias Blume
1495 : blume 879 Date: 2001/07/19 15:00:00 EDT
1496 :     Tag: blume-20010719-libreorg
1497 :     Description:
1498 :    
1499 :     This update puts together a fairly extensive but straightforward change
1500 :     to the way the libraries that implement the interactive system are
1501 :     organized:
1502 :    
1503 :     The biggest change is the elimination of structure Compiler. As a
1504 :     replacement for this structure, there is now a CM library
1505 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1506 :     that exports all the substructures of the original structure Compiler
1507 :     directly. So instead of saying Compiler.Foo.bar one now simply
1508 :     says Foo.bar. (The CM libraries actually export a collection of
1509 :     structures that is richer than the collection of substructures of
1510 :     structure Compiler.)
1511 :    
1512 :     To make the transition smooth, there is a separate library called
1513 :     $smlnj/compiler/compiler.cm which puts together and exports the
1514 :     original structure Compiler (or at least something very close to it).
1515 :    
1516 :     There are five members of the original structure Compiler
1517 :     that are not exported directly but which instead became members
1518 :     of a new structure Backend (described by signature BACKEND). These are:
1519 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1520 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
1521 :     architecture (: string).
1522 :    
1523 :     Structure Compiler.Version has become structure CompilerVersion.
1524 :    
1525 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1526 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1527 :     or x86, respectively.
1528 :     Each of these exports the same frontend structures that
1529 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
1530 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1531 :     Hppa, PPC, Sparc, or X86, respectively.
1532 :    
1533 :     Library $smlnj/compiler/all.cm exports the union of the exports of
1534 :     $smlnj/compiler/<arch>.cm
1535 :    
1536 :     There are no structures <Arch>Compiler anymore, use
1537 :     $smlnj/compiler/<arch>.cm instead.
1538 :    
1539 :     Library host-compiler-0.cm is gone. Instead, the internal library
1540 :     that instantiates CM is now called cm0.cm. Selection of the host
1541 :     compiler (backend) is no longer done here but. (Responsibility for it
1542 :     now lies with $smlnj/compiler/current.cm. This seems to be more
1543 :     logical.)
1544 :    
1545 :     Many individual files have been moved or renamed. Some files have
1546 :     been split into multiple files, and some "dead" files have been deleted.
1547 :    
1548 :     Aside from these changes to library organization, there are also changes
1549 :     to the way the code itself is organized:
1550 :    
1551 :     Structure Binfile has been re-implemented in such a way that it no
1552 :     longer needs any knowledge of the compiler. It exclusively deals
1553 :     with the details of binfile layout. It no longer invokes the
1554 :     compiler (for the purpose of creating new prospective binfile
1555 :     content), and it no longer has any knowledge of how to interpret
1556 :     pickles.
1557 :    
1558 :     Structure Compile (: COMPILE) has been stripped down to the bare
1559 :     essentials of compilation. It no longer deals with linking/execution.
1560 :     The interface has been cleaned up considerably.
1561 :    
1562 :     Utility routines for dealing with linking and execution have been
1563 :     moved into their own substructures.
1564 :    
1565 :     (The ultimate goal of these changes is to provide a light-weight
1566 :     binfile loader/linker (at least for, e.g., stable libraries) that
1567 :     does not require CM or the compiler to be present.)
1568 :    
1569 :     CM documentation has been updated to reflect the changes to library
1570 :     organization.
1571 :    
1572 :     ----------------------------------------------------------------------
1573 :     Name: Matthias Blume
1574 : blume 875 Date: 2001/07/10 17:30:00 EDT
1575 :     Tag: Release_110_34
1576 :     Description:
1577 :    
1578 : blume 876 Minor tweak to 110.34 (re-tagged):
1579 :    
1580 :     - README.html file added to CVS repository
1581 :     - runtime compiles properly under FreeBSD 3.X and 4.X
1582 :    
1583 :     ----------------------------------------------------------------------
1584 :     Name: Matthias Blume
1585 :     Date: 2001/07/10 17:30:00 EDT
1586 :     Tag: Release_110_34
1587 :     Description:
1588 :    
1589 : blume 875 New version number (110.34). New bootfiles.
1590 :    
1591 :     ----------------------------------------------------------------------
1592 :     Name: Matthias Blume
1593 : blume 874 Date: 2001/07/09 16:00:00 EDT
1594 :     Tag: blume-20010709-more-varargs
1595 :     Description:
1596 :    
1597 :     I changed the handling of varargs in ml-nlffigen again:
1598 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
1599 :    
1600 :     The immediate effect is that you can actually call a varargs function
1601 :     from ML -- but you can't actually supply any arguments beyond the ones
1602 :     specified explicitly. (For example, you can call printf with its format
1603 :     string, but you cannot pass additional arguments.)
1604 :    
1605 :     This behavior is only marginally more useful than the one before, but
1606 :     it has the advantage that a function or, more importantly, a function
1607 :     type never gets dropped on the floor, thus avoiding follow-up problems with
1608 :     other types that refer to the offending one.
1609 :    
1610 :     ----------------------------------------------------------------------
1611 :     Name: Matthias Blume
1612 : blume 873 Date: 2001/07/09 11:25:00 EDT
1613 :     Tag: blume-20010709-varargs
1614 :     Description:
1615 :    
1616 :     1. ckit-lib.cm now exports structure Error
1617 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1618 :     with a warning accompanied by a source location. Moreover, it
1619 :     merely skips the offending function or type and proceeds with the
1620 :     rest of its work.u As a result, one can safely feed C code containing
1621 :     "..." to ml-nlffigen.
1622 :     3. There are some internal improvements to CM, providing slightly
1623 :     more general string substitutions in the tools subsystem.
1624 :    
1625 :     ----------------------------------------------------------------------
1626 :     Name: Matthias Blume
1627 : blume 854 Date: 2001/06/27 15:10:00 EDT
1628 :     Tag: blume-20010627-concur
1629 :     Description:
1630 :    
1631 :     Fixed a small bug in CM's handling of parallel compilation.
1632 :     (You could observe the bug by Control-C-interrupting an ordinary
1633 :     CMB.make or CM.stabilize and then attaching some compile servers.
1634 :     The result was that all of a sudden the previously interrupted
1635 :     compilation would continue on its own. This was because of
1636 :     an over-optimization: CM did not bother to clean out certain queues
1637 :     when no servers were attached "anyway", resulting in the contents
1638 :     of these queues to grab control when new servers did get attached.)
1639 :    
1640 :     There is also another minor update to the CM manual.
1641 :    
1642 :     ----------------------------------------------------------------------
1643 :     Name: Matthias Blume
1644 : blume 853 Date: 2001/06/26 16:15:00 EDT
1645 :     Tag: blume-20010626-cmdoc
1646 :     Description:
1647 :    
1648 :     Minor typo fixed in CM manual (syntax diagram for libraries).
1649 :    
1650 :     ----------------------------------------------------------------------
1651 :     Name: Matthias Blume
1652 : blume 852 Date: 2001/06/25 22:55:00 EDT
1653 :     Tag: blume-20010625-x86pc
1654 :     Description:
1655 :    
1656 :     Fixed a nasty bug in the X86 assembly code that caused signal
1657 :     handlers to fail (crash) randomly.
1658 :    
1659 :     ----------------------------------------------------------------------
1660 :     Name: Matthias Blume
1661 : blume 846 Date: 2001/06/25 12:05:00 EDT
1662 :     Tag: blume-20010625-nlffigen
1663 :     Description:
1664 :    
1665 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
1666 :     Nick Carter <nbc@andrew.cmu.edu>.
1667 :    
1668 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1669 :     2. Default names for generated files are now derived from the name of
1670 :     the C file *without its directory*. In particular, this causes generated
1671 :     files to be placed locally even if the C file is in some system directory.
1672 :     3. Default names for generated signatures and structures are also derived
1673 :     from the C file name without its directory. This avoids silly things
1674 :     like "structure GL/GL".
1675 :     (Other silly names are still possible because ml-nlffigen does not do
1676 : blume 847 a thorough check of whether generated names are legal ML identifiers.
1677 :     When in doubt, use command line arguments to force particular names.)
1678 : blume 846
1679 :     ----------------------------------------------------------------------
1680 :     Name: Matthias Blume
1681 : blume 845 Date: 2001/06/21 12:25:00 EDT
1682 :     Tag: blume-20010621-eXene
1683 :     Description:
1684 :    
1685 :     eXene now compiles and (sort of) works again.
1686 :    
1687 :     The library name (for version > 110.33) is $/eXene.cm.
1688 :    
1689 :     I also added an new example in src/eXene/examples/nbody. See the
1690 :     README file there for details.
1691 :    
1692 :     ----------------------------------------------------------------------
1693 :     Name: Matthias Blume
1694 : blume 844 Date: 2001/06/20 16:40:00 EDT
1695 :     Tag: blume-20010620-cml
1696 :     Description:
1697 :    
1698 :     CML now compiles and works again.
1699 :    
1700 :     Libraries (for version > 110.33):
1701 :    
1702 :     $cml/cml.cm Main CML library.
1703 :     $cml/basis.cm CML's version of $/basis.cm.
1704 :     $cml/cml-internal.cm Internal helper library.
1705 :     $cml/core-cml.cm Internal helper library.
1706 :     $cml-lib/trace-cml.cm Tracing facility.
1707 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
1708 :    
1709 :     The installer (config/install.sh) has been taught how to properly
1710 :     install this stuff.
1711 :    
1712 :     ----------------------------------------------------------------------
1713 :     Name: Matthias Blume
1714 : blume 843 Date: 2001/06/19 17:55:00 EDT
1715 :     Tag: blume-20010619-instantiate
1716 :     Description:
1717 :    
1718 :     This un-breaks the fix for bug 1432.
1719 :     (The bug was originally fixed in 110.9 but I broke it again some
1720 :     time after that.)
1721 :    
1722 :     ----------------------------------------------------------------------
1723 :     Name: Matthias Blume
1724 : blume 842 Date: 2001/06/19 17:25:00 EDT
1725 :     Tag: blume-20010619-signals
1726 :     Description:
1727 :    
1728 :     This should (hopefully) fix the long-standing signal handling bug.
1729 :     (The runtime system was constructing a continuation record with an
1730 :     incorrect descriptor which would cause the GC to drop data on the floor...)
1731 :    
1732 :     ----------------------------------------------------------------------
1733 :     Name: Matthias Blume
1734 : blume 840 Date: 2001/06/15 15:05:00 EDT
1735 : blume 841 Tag: blume-20010615-moresparc
1736 :     Description:
1737 :    
1738 :     Here is a short late-hour update related to Sparc c-calls:
1739 :    
1740 :     -- made handling of double-word arguments a bit smarter
1741 :    
1742 :     -- instruction selection phase tries to collapse certain clumsily
1743 :     constructed ML-Trees; typical example:
1744 :    
1745 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
1746 :    
1747 :     This currently has no further impact on SML/NJ since mlriscGen does
1748 :     not seem to generate such patterns in the first place, and c-calls
1749 :     (which did generate them in the beginning) has meanwhile been fixed
1750 :     so as to avoid them as well.
1751 :    
1752 :     ----------------------------------------------------------------------
1753 :     Name: Matthias Blume
1754 :     Date: 2001/06/15 15:05:00 EDT
1755 : blume 840 Tag: blume-20010615-sparc
1756 :     Description:
1757 :    
1758 :     The purpose of this update is to provide an implementation of NLFFI
1759 :     on Sparc machines.
1760 :    
1761 :     Here are the changes in detail:
1762 :    
1763 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
1764 :     the Sparc implementation of the c-calls API.
1765 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
1766 :     for accessing the ML frame. Thus, we have a real frame pointer and
1767 :     can freely modify %sp without need for an omit-frame-ptr phase.
1768 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
1769 :     this case.
1770 :     * ml-nlffigen has been taught to produce code for different architectures
1771 :     and calling conventions.
1772 :     * In a way similar to what was done in the x86 case, the Sparc
1773 :     backend uses its own specific extension to mltree. (For example,
1774 :     it needs to be able to generate UNIMP instructions which are part
1775 :     of the calling convention.)
1776 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
1777 :     to make it easier to plug in new machine- and os-dependent parts).
1778 :    
1779 :     There are some other fairly unrelated bug fixes and cleanups as well:
1780 :    
1781 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
1782 :     that they properly share their libraries with existing SML/NJ libraries.
1783 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
1784 :     follow-up error messages.
1785 :     * Updates to CM/CMB documentation.
1786 :    
1787 :     TODO items:
1788 :    
1789 :     * MLRISC should use a different register as its asmTemp on the Sparc.
1790 :     (The current %o2 is a really bad choice because it is part of the
1791 :     calling conventions, so things might interfere in unexpected ways.)
1792 :    
1793 :     ----------------------------------------------------------------------
1794 :     Name: Matthias Blume
1795 : blume 839 Date: 2001/06/07
1796 :     Tag: blume-20010607-calls
1797 :     Description:
1798 :    
1799 :     A number of internal changes related to C calls and calling conventions:
1800 :    
1801 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
1802 :     number of bytes popped implicitly (by the callee). In most cases
1803 :     this field is 0 but on x86/win32 it is some non-zero value. This
1804 :     is information provided for the benefit of the "omit-frameptr" pass.
1805 :     2. The CALL instruction on the x86 carries a similar "pops" field.
1806 :     The instruction selection phase copies its value from the ML-Tree
1807 :     CALL statement.
1808 :     3. On all other architectures, the instruction selection phase checks
1809 :     whether "pops=0" and complains if not.
1810 :     4. The c-calls implementation for x86 now accepts two calling conventions:
1811 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
1812 :     up after the call and pops is set to 0. For "stdcall", the caller
1813 :     does nothing, leaving the cleanup to the callee; pops is set to
1814 :     the number of bytes that were pushed onto the stack.
1815 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1816 :     distinguish between "ccall" and "stdcall".
1817 :     6. The UNIMP instruction has been added to the supported Sparc instruction
1818 :     set. (This is needed for implementing the official C calling convention
1819 :     on this architecture.)
1820 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1821 :     work with the latest CM.
1822 :    
1823 :     ----------------------------------------------------------------------
1824 :     Name: Matthias Blume
1825 : blume 838 Date: 2001/06/05 15:10:00 EDT
1826 :     Tag: blume-20010605-cm-index
1827 :     Description:
1828 :    
1829 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1830 :    
1831 :     1. CM can now generate "index files". These are human-readable files
1832 :     that list on a per-.cm-file basis each toplevel symbol defined or
1833 :     imported. The location of the index file for
1834 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1835 :     To enable index-file generation, set CM.Control.generate_index to true
1836 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
1837 :    
1838 :     The CM manual has been updated accordingly.
1839 :    
1840 :     2. I made some slight modifications to the c-calls API in MLRISC.
1841 :    
1842 :     a) There is now a callback to support saving/restoring of
1843 :     dedicated but caller-save registers around the actual call
1844 :     instruction.
1845 :     b) One can optionally specify a comment-annotation for the
1846 :     call instruction.
1847 :    
1848 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1849 :     (For example, the comment annotation shows the C prototype of
1850 :     the function being called.)
1851 :    
1852 :     ----------------------------------------------------------------------
1853 :     Name: Matthias Blume
1854 : blume 837 Date: 2001/06/01 13:30:00 EDT
1855 :     Tag: blume-20010601-nlffi-cleanup
1856 :     Description:
1857 :    
1858 :     This is mostly a cleanup of MLFFI stuff:
1859 :    
1860 :     - some signature files have been put into a more exposed place
1861 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
1862 :     - ml-nlffigen changed accordingly
1863 :     - tutorial updated
1864 :    
1865 :     Other changes:
1866 :    
1867 :     - author's affiliation in CM manual(s) updated
1868 :     - some more recognized keywords added to Allen's sml.sty
1869 :    
1870 :     ----------------------------------------------------------------------
1871 :     Name: Matthias Blume
1872 : blume 836 Date: 2001/05/25 15:30:00 EDT
1873 :     Tag: blume-20010525-iptr
1874 :     Description:
1875 :    
1876 :     - put the official 110.33-README (as it appears on the ftp server) under
1877 :     CVS
1878 :     - fixed a small bug related to incomplete pointer types in
1879 :     ml-nlffigen
1880 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1881 :     (it does not need the 'f type parameter)
1882 :    
1883 :     ----------------------------------------------------------------------
1884 :     Name: Matthias Blume
1885 : blume 834 Date: 2001/05/23 14:30:00 EDT
1886 :     Tag: Release_110_33
1887 :     Description:
1888 :    
1889 :     New version number (110.33). New bootfiles.
1890 :    
1891 :     ----------------------------------------------------------------------
1892 :     Name: Matthias Blume
1893 : blume 833 Date: 2001/05/22 18:06:00 EDT
1894 :     Tag: blume-20010522-targets
1895 :     Description:
1896 :    
1897 :     Made install.sh use file config/targets.customized if it exists, falling
1898 :     back to config/targets if it doesn't. This way one can have a customized
1899 :     version of the targets file without touching the "real thing", thus
1900 :     eliminating the constant fear of accidentally checking something bogus
1901 :     back into the CVS repository... (File config/targets.customized must
1902 :     not be added to the repository!)
1903 :    
1904 :     ----------------------------------------------------------------------
1905 :     Name: Matthias Blume
1906 : blume 832 Date: 2001/05/22 16:30:00 EDT
1907 :     Tag: blume-20010522-minitut
1908 :     Description:
1909 :    
1910 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1911 :     struct returns.
1912 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
1913 :     incomplete, preliminary documentation for NLFFI.
1914 :    
1915 :     ----------------------------------------------------------------------
1916 :     Name: Matthias Blume
1917 : blume 829 Date: 2001/05/14 11:30:00 EDT
1918 :     Tag: blume-20010514-script
1919 :     Description:
1920 :    
1921 :     Some bugs in install script fixed.
1922 :    
1923 :     In addition to that I also made a slight change to the NLFFI API:
1924 :     Functors generated by ml-nlffigen now take the dynamic library as a
1925 :     straight functor argument, not as a suspended one. (The original
1926 :     functor code used to force the suspension right away anyway, so there
1927 :     was nothing gained by this complication of the interface.)
1928 :    
1929 :     ----------------------------------------------------------------------
1930 :     Name: Matthias Blume
1931 : blume 828 Date: 2001/05/11 14:35:00 EDT
1932 :     Tag: blume-20010511-ml-nlffi
1933 :     Description:
1934 :    
1935 :     I finally took the plunge and added my new FFI code to the main
1936 :     repository. For x86-linux it is now ready for prime-time.
1937 :    
1938 :     There are two new subdirectories of "src":
1939 :    
1940 :     - ml-nlffi-lib:
1941 :     The utility library for programs using the FFI interface.
1942 :     Here is the implementation of $/c.cm and its associated low-level
1943 :     partners $/c-int.cm and $/memory.cm.
1944 :     - ml-nlffigen:
1945 :     A stand-alone program for generating ML glue code from C source
1946 :     code.
1947 :    
1948 :     Building ml-nlffigen requires $/ckit-lib.cm.
1949 :    
1950 :     The config/install.sh script has been updates to do the Right Thing
1951 :     (hopefully).
1952 :    
1953 :     Notice that the source tree for the C-Kit will not be put under "src"
1954 :     but directly under the installation root directory. (This is the
1955 :     structure that currently exists on the CVS server when you check out
1956 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
1957 :    
1958 :     Bugs: No documentation yet.
1959 :    
1960 :     ----------------------------------------------------------------------
1961 :     Name: Matthias Blume
1962 : blume 826 Date: 2001/05/09 16:35:00 EDT
1963 :     Tag: blume-20010509-cpscontract
1964 :     Description:
1965 :    
1966 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
1967 :     wrapper/unwrapper elimination did not decrement usage counts and some
1968 :     dead variables got overlooked by the dead-up logic.)
1969 :    
1970 :     ----------------------------------------------------------------------
1971 : george 824 Name: Lal George
1972 :     Date: 2001/05/08 17:26:09 EDT
1973 :     Tag: george-20010508-omit-frameptr
1974 :     Description:
1975 :    
1976 :     Changes to implement the omit-frame-pointer optimization to support
1977 :     raw C calls. For now, there is only support on the Intel x86, but
1978 :     other architectures will follow as more experience is gained with this.
1979 :    
1980 :    
1981 :     ----------------------------------------------------------------------
1982 : blume 818 Name: Matthias Blume
1983 : blume 819 Date: 2001/05/07 14:40:00 EDT
1984 :     Tag: blume-20010507-proxies
1985 :     Description:
1986 :    
1987 :     I made into "proxy libraries" all libraries that qualify for such a
1988 :     change. (A qualifying library is a library that has another library or
1989 :     groups as its sole member and repeats that member's export list
1990 :     verbatim. A proxy library avoids this repetition by omitting its export
1991 :     list, effectively inheriting the list that its (only) member exports.
1992 :     See the CM manual for more explanation.)
1993 :     The main effect is that explicit export lists for these libraries
1994 :     do not have to be kepts in sync, making maintenance a bit easier.
1995 :    
1996 :     I also added copyright notices to many .cm-files.
1997 :    
1998 :     Last but not least, I made a new set of bootfiles.
1999 :    
2000 :     ----------------------------------------------------------------------
2001 :     Name: Matthias Blume
2002 : blume 818 Date: 2001/05/04 17:00:00 EDT
2003 :     Tag: blume-20010504-cm-lsplit
2004 :     Description:
2005 :    
2006 :     0. John merged pending changes to $/smlnj-lib.cm
2007 :    
2008 :     1. Allen's previous change accidentally backed out of one of Lal's
2009 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2010 :    
2011 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2012 :     within the compiler where applicable. There is some code simplification
2013 :     because of that.
2014 :    
2015 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2016 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2017 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2018 :     it can now also be controlled on a per-source basis: CM has been taught
2019 :     a new tool parameter applicable to ML source files.
2020 :    
2021 :     - To turn lambda-splitting off completely:
2022 :     local open Compiler.Control.LambdaSplitting in
2023 :     val _ = set Off
2024 :     end
2025 :     - To make "no lambda-splitting" the global default (but allow per-source
2026 :     overriding); this is the initial setting:
2027 :     local open Compiler.Control.LambdaSplitting in
2028 :     val _ = set (Default NONE)
2029 :     end
2030 :     - To make "lambda-splitting with aggressiveness a" the global default
2031 :     (and allow per-source overriding):
2032 :     local open Compiler.Control.LambdaSplitting in
2033 :     val _ = set (Default (SOME a))
2034 :     end
2035 :    
2036 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2037 :     write (in the respective .cm-file):
2038 :     a.sml (lambdasplitting:off)
2039 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2040 :     a.sml (lambdasplitting:on)
2041 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2042 :     <a> is a decimal non-negative integer):
2043 :     a.sml (lambdasplitting:<a>)
2044 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2045 :     a.sml (lambdasplitting:infinity)
2046 :     - To use the global default for a.sml:
2047 :     a.sml (lambdasplitting:default)
2048 :     or simply
2049 :     a.sml
2050 :    
2051 :     ----------------------------------------------------------------------
2052 : leunga 815 Name: Allen Leung
2053 :     Date: 2001/05/04 01:57:00 EDT
2054 :     Tag: leunga-20010504-sync
2055 :     Description:
2056 : leunga 788
2057 : leunga 815 MLRISC features.
2058 :    
2059 :     1. Fix to CMPXCHG instructions.
2060 :     2. Changed RA interface to allow annotations in callbacks.
2061 :     3. Added a new method to the stream interface to allow annotations updates.
2062 :    
2063 : george 749 ----------------------------------------------------------------------
2064 : blume 812 Name: Matthias Blume
2065 : blume 814 Date: 2001/05/01 11:45:00 EDT
2066 :     Tag: blume-20010501-pcedittmp
2067 :     Description:
2068 :    
2069 :     Changed install.sh to use the current working directory instead of
2070 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2071 :     of /usr/tmp caused trouble with MacOS X because of file premission
2072 :     problems.
2073 :    
2074 :     ----------------------------------------------------------------------
2075 :     Name: Matthias Blume
2076 : blume 812 Date: 2001/04/20 11:10:00 EDT
2077 :     Tag: blume-20010420-inMLflag
2078 :     Description:
2079 :    
2080 :     - added vp_limitPtrMask to vproc-state.h
2081 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2082 :     handling)
2083 :     - made the ML compiler aware of various data-structure offsets so it
2084 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2085 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2086 :     raw C-calls
2087 :    
2088 :     ----------------------------------------------------------------------
2089 : george 810 Name: Lal George
2090 :     Date: 2001/04/20 09:15:28 EDT
2091 :     Tag: george-20010420-macosX
2092 :     Description:
2093 :    
2094 :     - Changes to port to Mac OS X; Darwin.
2095 :    
2096 :     - In the process I found that sqrt was broken on the PPC, because the
2097 :     fsqrt instruction is not implemented.
2098 :    
2099 :     ----------------------------------------------------------------------
2100 : blume 799 Name: Matthias Blume
2101 : blume 808 Date: 2001/04/18 12:45:00 EDT
2102 :     Tag: blume-20010418-ccalls
2103 :     Description:
2104 :    
2105 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2106 :     (this bug prevented structure arguments containing pointers from being
2107 :     passed correctly)
2108 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
2109 :     structure arguments are represented as a pointer to the beginning
2110 :     of the structure (instead of having a series of synthesized arguments,
2111 :     one for each structure member)
2112 :    
2113 :     - made makeml script's verbosity level configurable via environment
2114 :     variable (MAKEML_VERBOSITY)
2115 :    
2116 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2117 : blume 808 in rawmem-x86.sml; we are now using the real thing
2118 :    
2119 :     ----------------------------------------------------------------------
2120 :     Name: Matthias Blume
2121 : blume 806 Date: 2001/03/22 16:25:00 EST
2122 :     Tag: blume-20010322-bootfiles
2123 :     Description:
2124 :    
2125 :     Created a new set of bootfiles (for your automatic installation convenience).
2126 :    
2127 :     ----------------------------------------------------------------------
2128 :     Name: Matthias Blume
2129 : blume 805 Date: 2001/03/22 15:10:00 EST
2130 :     Tag: blume-20010322-rawmem-parcm
2131 :     Description:
2132 :    
2133 :     1. All "raw memory access" primitives for the new FFI are implemented now
2134 :     (at least on the x86).
2135 :     2. Some further cleanup of CM's parallel make mechanism.
2136 :    
2137 :     ----------------------------------------------------------------------
2138 :     Name: Matthias Blume
2139 : blume 801 Date: 2001/03/19 17:53:00 EST
2140 :     Tag: blume-20010319-parallel
2141 :     Description:
2142 :    
2143 :     Parallel make (using compile servers) now works again.
2144 :    
2145 :     To this end, CM.stabilize and CMB.make have been modified to work in
2146 :     two passes when compile servers are attached:
2147 :     1. Compile everything, do not perform stabilization; this pass
2148 :     uses compile servers
2149 :     2. Stabilize everything; this pass does not use compile servers
2150 :     If there are no compile servers, the two passes are combined into one
2151 :     (as before). Splitting the passes increases the inherent parallelism
2152 :     in the dependency graph because the entire graph including all
2153 :     libraries is available at the same time. This, in turn, improves
2154 :     server utilization. The downside is that the master process will
2155 :     have to do some extra work after compilation is done (because for
2156 :     technical reasons it must re-read all the binfiles during stabilization).
2157 :    
2158 :     ----------------------------------------------------------------------
2159 :     Name: Matthias Blume
2160 : blume 800 Date: 2001/03/16 12:22:00 EST
2161 :     Tag: blume-20010316-bootfiles
2162 :     Description:
2163 :    
2164 :     Created a new set of bootfiles (for your automatic installation convenience).
2165 :    
2166 :     ----------------------------------------------------------------------
2167 :     Name: Matthias Blume
2168 : blume 799 Date: 2001/03/16 11:00:00 EST
2169 :     Tag: blume-20010316-MLTREE-fixup
2170 :     Description:
2171 :    
2172 :     This is a minor fixup for an (untagged) earlier commit by Allen.
2173 :     (A file was missing).
2174 :    
2175 :     ----------------------------------------------------------------------
2176 : leunga 788 Name: Allen Leung
2177 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
2178 :     Tag: leunga-20010305-cut-support
2179 :    
2180 :     1. New support for alternative control-flow in MLTREE.
2181 :     Currently we support
2182 :    
2183 :     FLOW_TO(CALL ...., [k1,...,kn])
2184 :    
2185 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
2186 :     in Moby
2187 :    
2188 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2189 :    
2190 :     2. Register Allocator
2191 :    
2192 :     Changes in interface [from Fermin, John]
2193 :    
2194 :     3. Alpha 8-bit SLL support [Fermin]
2195 :    
2196 :     4. All architectures
2197 :    
2198 :     A new module (ClusterExpandCopies) for expanding parallel copies.
2199 :    
2200 :     ----------------------------------------------------------------------
2201 :     Name: Allen Leung
2202 : leunga 788 Date: 2001/02/27 23:07:00 EST
2203 :     Tag: leunga-20010227-minor-stuff
2204 :    
2205 :     1. Alpha bug fix for CMOVNE
2206 :     2. Handle mltree COND(..,FCMP ...,...)
2207 :     3. Bug fix in simplifier
2208 :    
2209 :     ----------------------------------------------------------------------
2210 : blume 777 Name: Matthias Blume
2211 : blume 781 Date: 2001/01/30 17:50:00 EST
2212 :     Tag: blume-20010130-sync
2213 :     Description:
2214 :    
2215 :     This is just a minor update to sync my devel branch with the main brach.
2216 :     The only visible change is the addition of some README files.
2217 :    
2218 :     ----------------------------------------------------------------------
2219 :     Name: Matthias Blume
2220 : blume 778 Date: 2001/01/12 23:30:00 JST
2221 :     Tag: blume-20010112-bootfiles
2222 :     Description:
2223 :    
2224 :     Made a new set of bootfiles that goes with the current state of the
2225 :     repository.
2226 :    
2227 :     ----------------------------------------------------------------------
2228 :     Name: Matthias Blume
2229 : blume 777 Date: 2001/01/12 21:20:00 JST
2230 :     Tag: blume-20010112-sync
2231 :     Description:
2232 :    
2233 :     I am just flushing out some minor changes that had accumulated in
2234 :     my private branch in order to sync with the main tree. (This is
2235 :     mainly because I had CVS trouble when trying to merge _into_ my
2236 :     private branch.)
2237 :    
2238 :     Most people should be completely unaffected by this.
2239 :    
2240 :     ----------------------------------------------------------------------
2241 : leunga 775 Name: Allen Leung
2242 :     Date: Thu Jan 11 21:03:00 EST 2001
2243 :     Tag: leunga-20010111-labexp=mltree
2244 :     Description:
2245 :    
2246 :     1. Removed the type LabelExp and replace it by MLTree.
2247 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2248 :     3. There were some bugs in alpha code generator which would break
2249 :     64-bit code generation.
2250 :     4. Redo the tools to generate code with the
2251 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2252 :     are now generated by perl scripts.
2253 :    
2254 :     ----------------------------------------------------------------------
2255 : blume 771 Name: Matthias Blume
2256 : blume 774 Date: 2001/01/10 21:55:00 JST
2257 :     Tag: blume-20010110-rcc
2258 :     Description:
2259 :    
2260 :     The RCC stuff now seems to work (but only on the x86).
2261 :     This required hacking of the c-calls interface (and -implementation) in
2262 :     MLRISC.
2263 :    
2264 :     Normal compiler users should be unaffected.
2265 :    
2266 :     ----------------------------------------------------------------------
2267 :     Name: Matthias Blume
2268 : blume 773 Date: 2001/01/09 01:20:00 JST
2269 :     Tag: blume-20010109-rcc
2270 :     Description:
2271 :    
2272 :     This is a fairly big patch, flushing out a large number of pending
2273 :     changes that I made to my development copy over the last couple of days.
2274 :    
2275 :     Of practical relevance at this moment is a workaround for a pickling
2276 :     bug that Allen ran into the other day. The cause of the bug itself is
2277 :     still unknown and it might be hard to fix it properly, but the
2278 :     workaround has some merits of its own (namely somewhat reducing pickling
2279 :     overhead for certain libraries). Therefore, I think this solution should
2280 :     be satisfactory at this time.
2281 :    
2282 :     The rest of the changes (i.e., the vast majority) has to do with my
2283 :     ongoing efforts of providing direct support for C function calls from
2284 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2285 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2286 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2287 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2288 :     handle RCC.
2289 :    
2290 :     The last part (the changes to mlriscGen.sml) are still known to be
2291 :     wrong on the x86 and not implemented on all other architectures. But
2292 :     the infrastructure is in place. I had to change a few functor
2293 :     signatures in the backend to be able to route the CCalls interface
2294 :     from MLRISC there, and I had to specialize the mltree type (on the
2295 :     x86) to include the necessary extensions. (The extensions themselves
2296 :     were already there and redy to go in MLRISC/x86).
2297 :    
2298 :     Everything should be very happy as soon as someone helps me with
2299 :     mlriscGen.sml...
2300 :    
2301 :     In any case, nothing of this should matter to anyone as long as the
2302 :     new primop is not being used (which is going to be the case unless you
2303 :     find it where I hid it :). The rest of the compiler is completely
2304 :     unaffected.
2305 :    
2306 :     ----------------------------------------------------------------------
2307 :     Name: Matthias Blume
2308 : blume 772 Date: 2001/01/05 00:30:00 JST
2309 :     Tag: blume-20010105-primops
2310 :     Description:
2311 :    
2312 :     Added some experimental support for work that I am doing right now.
2313 :     These changes mostly concern added primops, but there is also a new
2314 :     experimental C library in the runtime system (but currently not enabled
2315 :     anywhere except on Linux/X86).
2316 :    
2317 :     In the course of adding primops (and playing with them), I discovered that
2318 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2319 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2320 :     To recover, I made type information in INL_PRIM mandatory and changed
2321 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2322 :     has complete, correct type information (i.e., no bottom types).
2323 :    
2324 :     Since all these changes mean that we need new binfiles, I also bumped the
2325 :     version number to 110.32.1.
2326 :    
2327 :     ----------------------------------------------------------------------
2328 :     Name: Matthias Blume
2329 : blume 771 Date: 2000/12/30 22:10:00 JST
2330 :     Tag: blume-20001230-various
2331 :     Description:
2332 :    
2333 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2334 :     to each other using path anchors. (See CM manual for explanation.)
2335 :    
2336 :     Updated CM documentation.
2337 :    
2338 :     Fixed some bugs in CM.
2339 :    
2340 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2341 :    
2342 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2343 :    
2344 :     Added cleanup handler for exportML to reset timers and compiler stats.
2345 :    
2346 :     ----------------------------------------------------------------------
2347 : george 761 Name: Lal George
2348 :     Date: 2000/12/22 22:22:58 EST 2000
2349 :     Tag: Release_110_32
2350 :     Description:
2351 :    
2352 :     Infinite precision used throughout MLRISC.
2353 :     see MLRISC/mltree/machine-int.sig
2354 :    
2355 :     ----------------------------------------------------------------------
2356 : blume 756 Name: Matthias Blume
2357 : blume 760 Date: 2000/12/22 23:16:00 JST
2358 :     Tag: blume-20001222-warn
2359 :     Description:
2360 : blume 759
2361 : blume 760 Corrected wording and formatting of some CM warning message which I
2362 :     broke in my previous patch.
2363 : blume 759
2364 :     ----------------------------------------------------------------------
2365 :     Name: Matthias Blume
2366 : blume 763 Date: 2000/12/22 21:20:00 JST
2367 :     Tag: blume-20001222-anchorenv
2368 :     Description:
2369 :    
2370 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2371 :    
2372 :     ----------------------------------------------------------------------
2373 :     Name: Matthias Blume
2374 : blume 758 Date: 2000/12/22 13:15:00 JST
2375 :     Tag: blume-20001222-cleanup
2376 :     Description:
2377 :    
2378 :     Removed src/cm/ffi which does not (and did not) belong here.
2379 :    
2380 :     ----------------------------------------------------------------------
2381 :     Name: Matthias Blume
2382 : blume 757 Date: 2000/12/21 23:55:00 JST
2383 :     Tag: blume-20001221-exn
2384 :     Description:
2385 :    
2386 :     Probably most important: CM no longer silently swallows all exceptions
2387 :     in the compiler.
2388 :     Plus: some other minor CM changes. For example, CM now reports some
2389 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2390 :    
2391 :     ----------------------------------------------------------------------
2392 :     Name: Matthias Blume
2393 : blume 756 Date: 2000/12/15 00:01:05 JST
2394 :     Tag: blume-20001215-dirtool
2395 :     Description:
2396 :    
2397 :     - "dir" tool added.
2398 :     - improvements and cleanup to Tools structure
2399 :     - documentation updates
2400 :    
2401 :     ----------------------------------------------------------------------
2402 : leunga 752 Name: Allen Leung
2403 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2404 :     Description:
2405 :     Tag: leunga-20001214-int-inf
2406 :     Description:
2407 :    
2408 :     In IntInf, added these standard functions, which are missing from our
2409 :     implementation:
2410 :    
2411 :     andb : int * int -> int
2412 :     xorb : int * int -> int
2413 :     orb : int * int -> int
2414 :     notb : int -> int
2415 :     << : int * word -> int
2416 :     ~>> : int * word -> int
2417 :    
2418 :     Not tested, I hope they are correct.
2419 :    
2420 :     ----------------------------------------------------------------------
2421 :     Name: Allen Leung
2422 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2423 :     Description:
2424 :     Tag: leunga-20001208-nowhere
2425 :     Description:
2426 :    
2427 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2428 :     to generate better error messages etc. Plus a brief manual.
2429 :    
2430 :     ----------------------------------------------------------------------
2431 : george 749 Name: Lal George
2432 :     Date: 2000/12/08 09:54:02 EST 2000
2433 :     Tag: Release_110_31
2434 :     Description:
2435 : george 705
2436 : george 749 - Version 110.31
2437 : leunga 591 ----------------------------------------------------------------------
2438 : leunga 744 Name: Allen Leung
2439 :     Date: Thu Dec 7 22:01:04 EST 2000
2440 :     Tag: leunga-20001207-cell-monster-hack
2441 :     Description:
2442 :    
2443 :     Major MLRISC internal changes. Affect all clients.
2444 :     Summary:
2445 :    
2446 :     1. Type CELLS.cell = int is now replaced by a datatype.
2447 :     As a result, the old regmap is now gone. Almost all interfaces
2448 :     in MLRISC change as a consequence.
2449 :    
2450 :     2. A new brand version of machine description tool (v3.0) that generates
2451 :     modules expecting the new interface. The old version is removed.
2452 :    
2453 :     3. The RA interface has been further abstracted into two new functors.
2454 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2455 :     [See also directory MLRISC/demo.]
2456 :    
2457 :     4. Some other new source->source code generation tools are available:
2458 :    
2459 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2460 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2461 :     I use this tool to generate the peephole optimizers---with the new
2462 :     cell type changes, peephole rules are becoming difficult to write
2463 :     without conditional pattern matching.
2464 :    
2465 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2466 :     cover the entire MLRISC source tree so many things broke.
2467 :    
2468 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2469 :     They are moved because there are a lot of them and they clutter up the
2470 :     root dir.
2471 :    
2472 :     7. More detailed documentation to come...
2473 :    
2474 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2475 :     a makeml -rebuild first. This is because of other other
2476 :     changes that Matthias has made (see below).
2477 :    
2478 :    
2479 :     ----------------------------------------------------------------------
2480 : blume 733 Name: Matthias Blume
2481 : blume 742 Date: 2000/11/30 23:12:00 JST
2482 :     Tag: blume-20001130-filereorg
2483 :     Description:
2484 :    
2485 :     Some manual updates and some file reorganizations in CM.
2486 :    
2487 :     ----------------------------------------------------------------------
2488 :     Name: Matthias Blume
2489 : blume 737 Date: 2000/11/24 17:45:00 JST
2490 :     Tag: blume-20001124-link
2491 :     Description:
2492 :    
2493 :     Drastically improved link traversal code for the case that the dynamic
2494 :     value was already loaded at bootstrap time. As a result, CM and CMB
2495 :     now both load blazingly fast -- even on a very slow machine. Also,
2496 :     memory consumption has been further reduced by this.
2497 :    
2498 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2499 :     bootstrap you have to do this:
2500 :    
2501 :     1. Run CMB.make
2502 :     2. Make a symbolic link for the boot directory:
2503 :     ln -s sml.boot.ARCH-OS xxx
2504 :     3. "Rebuild" the boot directory:
2505 :     ./makeml -boot xxx -rebuild sml ; rm xxx
2506 :     4. Boot normally:
2507 :     ./makeml
2508 :    
2509 :     ----------------------------------------------------------------------
2510 :     Name: Matthias Blume
2511 : blume 735 Date: 2000/11/21 21:20:00 JST
2512 :     Tag: blume-20001121-tools
2513 :     Description:
2514 :    
2515 :     Continued hacking on autoloading problem -- with success this time.
2516 :     Also changed tool-plugin mechanism. See new CM manual.
2517 :    
2518 :     ----------------------------------------------------------------------
2519 :     Name: Matthias Blume
2520 : blume 734 Date: 2000/11/19 14:30:00 JST
2521 :     Tag: blume-20001119-autoload
2522 :     Description:
2523 :    
2524 :     Some hacking to make autoloading faster. Success for CMB, no success
2525 :     so far for CM. There is a reduced structure CM' that autoloads faster.
2526 :     (This is a temporary, non-documented hack to be eliminated again when
2527 :     the general problem is solved.)
2528 :    
2529 :     ----------------------------------------------------------------------
2530 :     Name: Matthias Blume
2531 : blume 733 Date: 2000/11/17 14:10:00 JST
2532 :     Tag: blume-20001117-pickle-lib
2533 :     Description:
2534 :    
2535 :     1. Eliminated comp-lib.cm
2536 :     2. Made pickle-lib.cm
2537 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
2538 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
2539 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
2540 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
2541 :     Duplicated intmap modules thrown out.
2542 :     6. Hunted down all duplicated SCC code and replaced it with
2543 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
2544 :     7. Rewrote Feedback module.
2545 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
2546 :     should be thrown out and equivalent modules from smlnj-lib.cm
2547 :     should be used (IntRedBlackSet, IntListSet, ...).
2548 :    
2549 :     Confirmed that compiler compiles to fixpoint.
2550 :    
2551 :     ----------------------------------------------------------------------
2552 : leunga 731 Name: Allen Leung
2553 :     Date: 2000/11/10 18:00:00
2554 :     Tag: leunga-20001110-new-x86-fp
2555 :    
2556 :     A new x86 floating point code generator has been added.
2557 :     By default this is turned off. To turn this on, do:
2558 :    
2559 :     CM.autoload "$smlnj/compiler.cm";
2560 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
2561 :    
2562 :     Changes:
2563 :    
2564 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
2565 :     2. Changed the extension callback for FTANGENT to generate:
2566 :    
2567 :     fptan
2568 :     fstp %st(0)
2569 :     instead of
2570 :     fptan
2571 :     fstpl ftempmem
2572 :    
2573 :     3. Numerous assembly fixes for x86.
2574 :    
2575 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
2576 :     support for a whole bunch of instructions and addressing modes:
2577 :    
2578 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
2579 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
2580 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
2581 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
2582 :     fxch %st(n)
2583 :     fld %st(n)
2584 :     fst %st(n)
2585 :     fst mem
2586 :     fstp %st(n)
2587 :     fucom %st(n)
2588 :     fucomp %st(n)
2589 :    
2590 :     All these are now generated when the fast fp mode is turned on.
2591 :    
2592 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
2593 :    
2594 :     ----------------------------------------------------------------------
2595 : blume 729 Name: Matthias Blume
2596 :     Date: 2000/11/09 11:20:00 JST
2597 :     Tag: blume-20001109-scc
2598 :     Description:
2599 :    
2600 :     Eliminated some code duplication:
2601 :    
2602 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
2603 :     (Otherwise the functor is useless.)
2604 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
2605 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
2606 :    
2607 :     ----------------------------------------------------------------------
2608 : george 721 Name: Lal George
2609 :     Date: 2000/11/06 09:02:21 EST 2000
2610 :     Tag: Release_110_30
2611 :     Description:
2612 :    
2613 :     - Version 110.30
2614 :     ----------------------------------------------------------------------
2615 : blume 715 Name: Matthias Blume
2616 : blume 716 Date: 2000/11/04 14:45:00
2617 :     Tag: blume-20001104-mlbuild
2618 :     Description:
2619 :    
2620 :     - Made ml-build faster on startup.
2621 :     - Documentation fixes.
2622 :    
2623 :     ----------------------------------------------------------------------
2624 :     Name: Matthias Blume
2625 : blume 715 Date: 2000/11/02 17:00:00 JST
2626 :     Tag: blume-20001102-condcomp
2627 :     Description:
2628 :    
2629 :     - Small tweaks to pickler -- new BOOTFILES!
2630 :     - Version bumped to 110.29.2.
2631 :     - Added conditional compilation facility to init.cmi (see comment there).
2632 :     ----------------------------------------------------------------------
2633 : leunga 713 Name: Allen Leung
2634 :     Date: 2000/10/23 19:31:00
2635 :     Tag: leunga-20001023-demo-ra
2636 :    
2637 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
2638 :     2. Test programs for the graph library updated
2639 :     3. Some new MLRISC demo programs added
2640 :    
2641 :     ----------------------------------------------------------------------
2642 : blume 710 Name: Matthias Blume
2643 : blume 711 Date: 2000/08/31 22:15:00 JST
2644 :     Tag: blume-20001017-errmsg
2645 :     Description:
2646 :    
2647 :     More error message grief: Where there used to be no messages, there
2648 :     now were some that had bogus error regions. Fixed.
2649 :    
2650 :     ----------------------------------------------------------------------
2651 :     Name: Matthias Blume
2652 : blume 710 Date: 2000/08/31 17:30:00 JST
2653 :     Tag: blume-20001017-v110p29p1
2654 :     Description:
2655 :    
2656 :     I made a version 110.29.1 with new bootfiles.
2657 :    
2658 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
2659 :     CM documentation changes and a small bugfix in CM's error reporting.
2660 :    
2661 :     ----------------------------------------------------------------------
2662 : george 705 Name: Lal George
2663 :     Date: 2000/09/27 14:42:35 EDT
2664 :     Tag: george-20000927-nodestatus
2665 :     Description:
2666 :    
2667 :     Changed the type of the nodestatus, so that:
2668 :    
2669 :     SPILLED(~1) is now SPILLED
2670 :     SPILLED(m) where m>=0 is now MEMREG(m)
2671 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
2672 :    
2673 :     ----------------------------------------------------------------------
2674 : blume 697 Name: Matthias Blume
2675 : blume 703 Date: 2000/09/07 14:45:00 JST
2676 :     Tag: blume-20000907-cmerrmsg
2677 :     Description:
2678 :    
2679 :     Small tweak to CM to avoid getting ML syntax error messages twice.
2680 :    
2681 :     ----------------------------------------------------------------------
2682 :     Name: Matthias Blume
2683 : blume 701 Date: 2000/08/31 18:00:00 JST
2684 :     Tag: blume-20000831-cvsbootfiles
2685 :     Description:
2686 :    
2687 :     New URL for boot files (because the 110.29 files on the BL server do
2688 :     now work correctly with my updated install scripts for yacc and lex).
2689 :    
2690 :     ----------------------------------------------------------------------
2691 :     Name: Matthias Blume
2692 : blume 697 Date: 2000/08/08 12:33:00 JST
2693 :     Tag: blume-20000808-manual
2694 :     Description:
2695 :    
2696 :     Tiny update to CM manual.
2697 :    
2698 :     ----------------------------------------------------------------------
2699 : leunga 695 Name: Allen Leung
2700 :     Date: 2000/08/7 19:31:00
2701 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
2702 :    
2703 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
2704 :     and Mobdy.
2705 :    
2706 :     1. x86
2707 :    
2708 :     a. Fixes to peephole module by John and Dan.
2709 :     b. Assembly fix to SETcc by Allen.
2710 :     c. Fix to c-call by John.
2711 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
2712 :     e. Instruction selection optimization to SETcc as suggested by John.
2713 :    
2714 :     For example,
2715 :    
2716 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
2717 :    
2718 :     should generate:
2719 :    
2720 :     MOVL a, x
2721 :     SUBL b, x
2722 :     SHRL 31, x
2723 :    
2724 :     2. IR stuff
2725 :    
2726 :     A bunch of new DJ-graph related algorithms added. These
2727 :     speed up SSA construction.
2728 :    
2729 :     3. SSA + Scheduling
2730 :    
2731 :     Added code for SSA and scheduling to the repository
2732 :    
2733 :     ----------------------------------------------------------------------
2734 : george 694 Name: Lal George
2735 :     Date: 2000/07/27 11:53:14 EDT
2736 :    
2737 :     Tag: lal-20000727-linux-ppc
2738 :     Description:
2739 :    
2740 :     Made changes to support Linux PPC.
2741 :     p.s. I have confirmation that the 110.29 boot files work fine.
2742 :    
2743 :     ----------------------------------------------------------------------
2744 : blume 684 Name: Matthias Blume
2745 : blume 692 Date: 2000/07/27 17:40:00 JST
2746 :     Tag: blume-20000727-scripts
2747 :     Description:
2748 :    
2749 :     !!!! WARNING !!!!
2750 :     You must recompile the runtime system!
2751 :     !!!! WARNING !!!!
2752 :    
2753 :     This is basically another round of script-enhancements:
2754 :    
2755 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
2756 :     and undefine CM preprocessor symbols.
2757 :    
2758 :     2. ml-build avoids generating a new heap image if it finds that the
2759 :     existing one is still ok. (The condition is that no ML file had to
2760 :     be recompiled and all ML files are found to be older that the heap
2761 :     file.)
2762 :    
2763 :     To make this work smoothly, I also hacked the runtime system as
2764 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
2765 :     (.sparc-solaris, ...) that is currently being used.
2766 :    
2767 :     Moreover, the signature of CM.mk_standalone has changed. See the
2768 :     CM manual.
2769 :    
2770 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
2771 :     CM manual for details.)
2772 :    
2773 :     4. More CM manual updates:
2774 :     - all of the above has been documented.
2775 :     - there is now a section describing the (CM-related) command line
2776 :     arguments that are accepted by the "sml" command
2777 :    
2778 :     ----------------------------------------------------------------------
2779 :     Name: Matthias Blume
2780 : blume 691 Date: 2000/07/25 16:20:00 JST
2781 :     Tag: blume-20000725-makedepend
2782 :     Description:
2783 :    
2784 :     Added a script called ml-makedepend. This can be used in makefiles
2785 :     for Unix' make in a way very similar to the "makedepend" command for
2786 :     C.
2787 :    
2788 :     The script internally uses function CM.sources.
2789 :    
2790 :     Synopsis:
2791 :    
2792 :     ml-makedepend [-f makefile] cmfile targetname
2793 :    
2794 :     The default for the makefile is "makefile" (or "Makefile" should
2795 :     "makefile" not exist).
2796 :    
2797 :     ml-makedepend adds a cmfile/targetname-specific section to this
2798 :     makefile (after removing the previous version of this section). The
2799 :     section contains a single dependency specification with targetname on
2800 :     the LHS (targetname is an arbitrary name), and a list of files derived
2801 :     from the cmfile on the RHS. Some of the files on the RHS are
2802 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
2803 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
2804 :     path names. The makefile writer is responsible for making sure that
2805 :     these variables have correct at the time "make" is invoked.
2806 :    
2807 :     ----------------------------------------------------------------------
2808 :     Name: Matthias Blume
2809 : blume 690 Date: 2000/07/22 23:30:00 JST
2810 :     Tag: blume-20000722-urlupdate
2811 :     Description:
2812 :    
2813 :     Changed BOOT and config/srcarchiveurl to point to BL server:
2814 :    
2815 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
2816 :    
2817 :     ----------------------------------------------------------------------
2818 :     Name: Matthias Blume
2819 : blume 686 Date: 2000/07/18 18:00:00 JST
2820 :     Tag: blume-20000718-Version_110_29
2821 :     Description:
2822 :    
2823 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
2824 :    
2825 :     2. Updated config/version to 110.29
2826 :    
2827 :     3. Updated config/srcarchiveurl
2828 :    
2829 :     3. New boot files!
2830 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
2831 :    
2832 :     ----------------------------------------------------------------------
2833 :     Name: Matthias Blume
2834 : blume 684 Date: 2000/07/11 13:58:00 JST
2835 :     Tag: blume-20000711-doctypo
2836 :     Description:
2837 :    
2838 :     Fixed a few typos in CM manual.
2839 :    
2840 :     ----------------------------------------------------------------------
2841 : leunga 683 Name: Allen Leung
2842 :     Date: 2000/06/15 00:38:00
2843 :     Tag: leunga-20000704-sparc-x86
2844 :    
2845 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
2846 :     2. fix to x86 RET bug [found by Dan Grossman]
2847 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
2848 :    
2849 :     Affects c-- and moby only
2850 :    
2851 :     ----------------------------------------------------------------------
2852 : blume 680 Name: Matthias Blume
2853 : blume 682 Date: 2000/07/04 15:26:00
2854 :     Tag: blume-20000704-trigger
2855 :     Description:
2856 :    
2857 :     1. Improvements to CM manual.
2858 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
2859 :     of getting a back-trace. The function, when called, raises an
2860 :     internal exception which explicitly carries the full back-trace history,
2861 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
2862 :     or not). The interactive loop will print that history once it arrives
2863 :     at top level.
2864 :     Short of having all exceptions implicitly carry the full history, the
2865 :     recommended way of using this facility is:
2866 :     - compile your program with instrumentation "on"
2867 :     - run it, when it raises an exception, look at the history
2868 :     - if the history is "cut off" because of some handler, go and modify
2869 :     your program so that it explicitly calls BTrace.trigger
2870 :     - recompile (still instrumented), and rerun; look at the full history
2871 :    
2872 :     ----------------------------------------------------------------------
2873 :     Name: Matthias Blume
2874 :     Date: 2000/07/03 15:36:00 JST
2875 : blume 680 Tag: blume-20000702-manual
2876 :     Description:
2877 :    
2878 :     Small corrections and updates to CM manual.
2879 :    
2880 :     ----------------------------------------------------------------------
2881 : blume 682 Name: Matthias Blume
2882 : blume 679 Date: 2000/06/29 16:04:00 JST
2883 :     Tag: blume-20000629-yacctool
2884 :     Description:
2885 :    
2886 :     Changes:
2887 :    
2888 :     1. Class "mlyacc" now takes separate arguments to pass options to
2889 :     generated .sml- and .sig-files independently.
2890 :     2. Corresponding CM manual updates.
2891 :     3. BTrace module now also reports call sites. (However, for loop clusters
2892 :     it only shows from where the cluster was entered.) There are associated
2893 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
2894 :    
2895 :     ----------------------------------------------------------------------
2896 : blume 675 Name: Matthias Blume
2897 : blume 678 Date: 2000/06/27 16:51:00 JST
2898 :     Tag: blume-20000627-noweb
2899 :     Description:
2900 :    
2901 :     Changes:
2902 :    
2903 :     1. Implemented "subdir" and "witness" options for noweb tool.
2904 :     This caused some slight internal changes in CM's tool implementation.
2905 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
2906 :     some remaining issues from earlier path anchor changes.
2907 :     3. Updated CM manual accordingly.
2908 :    
2909 :     4. Changed implementation of back-tracing so that I now consider it
2910 :     ready for prime-time.
2911 :    
2912 :     In particular, you don't have to explicitly trigger the back-trace
2913 :     anymore. Instead, if you are running BTrace-instrumented code and
2914 :     there is an uncaught exception (regardless of whether or not it was
2915 :     raised in instrumented code), the top-level evalloop will print
2916 :     the back-trace.
2917 :    
2918 :     Features:
2919 :    
2920 :     - Instrumented and uninstrumented code work together seemlessly.
2921 :     (Of course, uninstrumented code is never mentioned in actual
2922 :     back-traces.)
2923 :    
2924 :     - Asymptotic time- and space-complexity of instrumented code is
2925 :     equal to that of uninstrumented code. (This means that
2926 :     tail-recursion is preserved by the instrumentation phase.)
2927 :    
2928 :     - Modules whose code has been instrumented in different sessions
2929 :     work together without problem.
2930 :    
2931 :     - There is no penalty whatsoever on uninstrumented code.
2932 :    
2933 :     - There is no penalty on "raise" expressions, even in
2934 :     instrumented code.
2935 :    
2936 :     A potential bug (or perhaps it is a feature, too):
2937 :    
2938 :     A back-trace reaches no further than the outermost instrumented
2939 :     non-trivial "raise". Here, a "trivial" raise is one that is the
2940 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
2941 :    
2942 :     <exp> handle e => raise e
2943 :    
2944 :     and even
2945 :    
2946 :     <exp> handle Foo => raise Bar
2947 :    
2948 :     and, of course, through
2949 :    
2950 :     <exp> handle Foo => ...
2951 :    
2952 :     if the exception was not Foo.
2953 :    
2954 :     Back-traces always reach right through any un-instrumented code
2955 :     including any of its "handle" expressions, trivial or not.
2956 :    
2957 :     To try this out, do the following:
2958 :    
2959 :     - Erase all existing binfiles for your program.
2960 :     (You may keep binfiles for those modules where you think you
2961 :     definitely don't need back-tracing.)
2962 :     - Turn on back-trace instrumentation:
2963 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2964 :     - Recompile your program. (I.e., run "CM.make" or "use".)
2965 :     - You may now turn instrumentation off again (if you want):
2966 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2967 :     - Run your program as usual. If it raises an exception that
2968 :     reaches the interactive toplevel, then a back-trace will
2969 :     automatically be printed. After that, the toplevel loop
2970 :     will print the exception history as usual.
2971 :    
2972 :     ----------------------------------------------------------------------
2973 :     Name: Matthias Blume
2974 : blume 677 Date: 2000/06/26 09:56:46 JST
2975 :     Tag: blume-20000626-setup
2976 :     Description:
2977 :    
2978 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2979 :     ML code before and after compiling a file (e.g., to set compiler
2980 :     flags)
2981 :    
2982 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2983 :     - associated changes to btrace.sml (BTrace instrumentation pass)
2984 :     - cleaner implementation of btimp.sml (BTrace tracing and report
2985 :     module)
2986 :    
2987 :     CM manual: * new path encoding documented
2988 :     * description of setup-parameter to "sml" added
2989 :    
2990 :     The biggest user-visible change to back-tracing is that it is no
2991 :     longer necessary to compile all traced modules within the same
2992 :     session. (This was a real limitation.)
2993 :    
2994 :     ----------------------------------------------------------------------
2995 :     Name: Matthias Blume
2996 : blume 676 Date: 2000/06/24 12:40:00 JST
2997 :     Tag: blume-20000624-startup
2998 :     Description:
2999 :    
3000 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
3001 :     bit too often -- to put it mildly. :)
3002 :    
3003 :     ----------------------------------------------------------------------
3004 :     Name: Matthias Blume
3005 : blume 675 Date: 2000/06/23 18:20:00 JST
3006 :     Tag: blume-20000623-btrace
3007 :     Description:
3008 :    
3009 :     This updates adds a backtrace facility to aid programmers in debugging
3010 :     their programs. This involves the following changes:
3011 :    
3012 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3013 :     keeping track of the current call stack. When programs are compiled
3014 :     in a special mode, the compiler will insert calls to these hooks
3015 :     into the user program.
3016 :     "Hook" means that it is possible for different implementations of
3017 :     back-tracing to register themselves (at different times).
3018 :    
3019 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3020 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3021 :     It can be turned on using this call:
3022 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3023 :     Turning it off again:
3024 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3025 :     Querying the current status:
3026 :     SMLofNJ.Internals.BTrace.mode NONE;
3027 :     Annotated programs are about twice as big as normal ones, and they
3028 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3029 :     where all hooks do nothing). The slowdown with a plugin that is
3030 :     actually useful (such as the one supplied by default) is even greater,
3031 :     but in the case of the default plugin it is still only an constant
3032 :     factor (amortized).
3033 :    
3034 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3035 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3036 :     the above-mentioned function "mode" controls whether the annotation
3037 :     phase is invoked by the compiler. Another important function is
3038 :     "trigger": when called it aborts the current execution and causes
3039 :     the top-level loop to print a full back-trace.
3040 :    
3041 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3042 :     for back-tracing. It keeps track of the dynamic call stack and in
3043 :     addition to that it keeps a partial history at each "level" of that
3044 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3045 :     d and b (at separate times, dynamically), then the report will show:
3046 :    
3047 :     GOTO d
3048 :     /c
3049 :     GOTO \b
3050 :     CALL a
3051 :    
3052 :     This shows that there was an initial non-tail call of a, then a
3053 :     tail-call to b or c, looping behavior in a cluster of functions that
3054 :     consist of b and c, and then a goto from that cluster (i.e., either from
3055 :     b or from c) to d.
3056 :    
3057 :     Note that (depending on the user program) the amount of information
3058 :     that the back-trace module has to keep track of at each level is bounded
3059 :     by a constant. Thus, the whole implementation has the same asymptotical
3060 :     complexity as the original program (both in space and in time).
3061 :    
3062 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3063 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3064 :     which is raised by the "trigger" function mentioned above.
3065 :    
3066 :     Notes on usage:
3067 :    
3068 :     - Annotated code works well together with unannotated code:
3069 :     Unannotated calls simply do not show up at all in the backtrace.
3070 :    
3071 :     - It is not a good idea to let modules that were annotated during
3072 :     different sessions run at the same time. This is because the compiler
3073 :     chooses small integers to identify individual functions, and there
3074 :     will be clashes if different modules were compiled in separate sessions.
3075 :     (Nothing will crash, and you will even be told about the clashes, but
3076 :     back-trace information will in general not be useful.)
3077 :    
3078 :     - Back-tracing can be confused by callcc and capture.
3079 :    
3080 :     - The only way of getting a back-trace right now is to explicitly
3081 :     invoke the "trigger" function from your user program. Eventually, we
3082 :     should make every exception carry back-trace information (if
3083 :     available). But since this creates more overhead at "raise"-time
3084 :     (similar to the current exnHistory overhead), I have not yet
3085 :     implemented this. (The implementation will be rather easy.) With
3086 :     exceptions carrying back-trace information, this facility will be even
3087 :     more useful because users don't need to modify their programs...
3088 :    
3089 :     - While it is possible to compile the compiler with back-trace
3090 :     annotations turned on (I did it to get some confidence in
3091 :     correctness), you must make absolutely sure that core.sml and
3092 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3093 :     be compiled with annotation because there is no core access yet, but
3094 :     if you compile btimp.sml with annotation, then the system will go into
3095 :     an infinite recursion and crash.)
3096 :     Since CM currently does not know about BTrace, the only way to turn
3097 :     annotations on and off for different modules of the compiler is to
3098 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3099 :     this is awkward and clumsy.
3100 :    
3101 :     Sample sessions:
3102 :    
3103 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3104 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3105 :     [autoloading]
3106 :     [autoloading done]
3107 :     val it = false : bool
3108 :     - structure X = struct
3109 :     - fun main n = let
3110 :     - fun a (x, 0) = d x
3111 :     - | a (x, n) = b (x, n - 1)
3112 :     - and b (x, n) = c (x, n)
3113 :     - and c (x, n) = a (x, n)
3114 :     - and d x = e (x, 3)
3115 :     - and e (x, 0) = f x
3116 :     - | e (x, n) = e (x, n - 1)
3117 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
3118 :     - | f n = n * g (n - 1)
3119 :     - and g n = a (n, 3)
3120 :     - in
3121 :     - f n
3122 :     - end
3123 :     - end;
3124 :     structure X : sig val main : int -> int end
3125 :     - X.main 3;
3126 :     *** BACK-TRACE ***
3127 :     GOTO stdIn:4.2-13.20: X.main[2].f
3128 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3129 :     GOTO stdIn:4.2-13.20: X.main[2].d
3130 :     / stdIn:4.2-13.20: X.main[2].a
3131 :     | stdIn:4.2-13.20: X.main[2].b
3132 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3133 :     CALL stdIn:4.2-13.20: X.main[2].g
3134 :     GOTO stdIn:4.2-13.20: X.main[2].f
3135 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3136 :     GOTO stdIn:4.2-13.20: X.main[2].d
3137 :     / stdIn:4.2-13.20: X.main[2].a
3138 :     | stdIn:4.2-13.20: X.main[2].b
3139 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3140 :     CALL stdIn:4.2-13.20: X.main[2].g
3141 :     GOTO stdIn:4.2-13.20: X.main[2].f
3142 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3143 :     GOTO stdIn:4.2-13.20: X.main[2].d
3144 :     / stdIn:4.2-13.20: X.main[2].a
3145 :     | stdIn:4.2-13.20: X.main[2].b
3146 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3147 :     CALL stdIn:4.2-13.20: X.main[2].g
3148 :     GOTO stdIn:4.2-13.20: X.main[2].f
3149 :     CALL stdIn:2.15-17.4: X.main[2]
3150 :     -
3151 :    
3152 :     (Note that because of a FLINt bug the above code currently does not
3153 :     compile without BTrace turned on.)
3154 :    
3155 :     Here is another example, using my modified Tiger compiler:
3156 :    
3157 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3158 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3159 :     [autoloading]
3160 :     [autoloading done]
3161 :     val it = false : bool
3162 :     - CM.make "sources.cm";
3163 :     [autoloading]
3164 :     ...
3165 :     [autoloading done]
3166 :     [scanning sources.cm]
3167 :     [parsing (sources.cm):parse.sml]
3168 :     [creating directory CM/SKEL ...]
3169 :     [parsing (sources.cm):tiger.lex.sml]
3170 :     ...
3171 :     [wrote CM/sparc-unix/semant.sml]
3172 :     [compiling (sources.cm):main.sml]
3173 :     [wrote CM/sparc-unix/main.sml]
3174 :     [New bindings added.]
3175 :     val it = true : bool
3176 :     - Main.compile ("../testcases/merge.tig", "foo.out");
3177 :     *** BACK-TRACE ***
3178 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
3179 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3180 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3181 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3182 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3183 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3184 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
3185 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
3186 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
3187 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3188 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
3189 :     CALL main.sml:1.18-118.4: Main.compile[2]
3190 :     -
3191 :    
3192 :     ----------------------------------------------------------------------
3193 : blume 673 Name: Matthias Blumen
3194 :     Date: 2000/06/21 18:00:00 JST
3195 :     Tag: blume-20000621-manual
3196 :     Description:
3197 :    
3198 :     CM manual update: Path environments documented.
3199 :    
3200 :     ----------------------------------------------------------------------
3201 : blume 666 Name: Matthias Blume
3202 : blume 672 Date: 2000/06/19 13:40:00
3203 :     Tag: blume-20000619-manual
3204 :     Description:
3205 :    
3206 :     CM manual and system/README update. This only covers the fact that
3207 :     there are no more implicit anchors. (Path environments and the "bind"
3208 :     option to "cm" have yet to be documented.)
3209 :    
3210 :     ----------------------------------------------------------------------
3211 :     Name: Matthias Blume
3212 : blume 671 Date: 2000/06/19 11:05:00 JST
3213 :     Tag: blume-20000619-chdir-bugfix
3214 :     Description:
3215 :    
3216 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
3217 :    
3218 :     ----------------------------------------------------------------------
3219 :     Name: Matthias Blume
3220 : blume 670 Date: 2000/06/18 22:00:10 JST
3221 :     Tag: blume-20000618-implicit-anchors-really-gone
3222 :     Description:
3223 :    
3224 :     I updates the previous HISTORY entry where I forgot to mention that
3225 :     implicit anchors are no longer with us.
3226 :    
3227 :     The current update also gets rid of the (now useless) controller
3228 :     CM.Control.implicit_anchors.
3229 :    
3230 :     ----------------------------------------------------------------------
3231 :     Name: Matthias Blume
3232 : blume 666 Date: 2000/06/16 17:30:00 JST
3233 :     Tag: blume-20000616-anchorenv
3234 :     Description:
3235 :    
3236 :     This patch implements the long anticipated (just kidding :) "anchor
3237 :     environment" mechanism. In the course of doing this, I also
3238 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3239 :     one should be more robust in certain boundary cases. In any case, it
3240 :     is a lot cleaner than its predecessor (IMHO).
3241 :    
3242 :     This time, although there is yet another boot file format change, I
3243 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3244 :     are necessary and bootstrapping is straightforward. (You cannot read
3245 :     new bootfiles into an old system, but the other way around is no
3246 :     problem.)
3247 :    
3248 :     Visible changes:
3249 :    
3250 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3251 :     longer recognized at all. This means that such path names are not
3252 :     illegal either. For example, the name basis.cm simply refers to a
3253 :     local file called "basis.cm" (i.e, the name is an ordinary path
3254 :     relative to .cm-files directory). Or, to put it differently, only
3255 :     names that start with $ are anchored paths.
3256 : blume 666
3257 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3258 :     vanished.
3259 :    
3260 :     John (Reppy) had critizised this as soon as I originally proposed and
3261 : blume 666 implemented it, but at that time I did not really deeply believe
3262 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3263 :     in another place where it cannot be seen as an abbreviation for
3264 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3265 :     seems to mean (i.e., it "expands" into the corresponding anchor
3266 :     value).
3267 :    
3268 :     However, when paths are used as members in CM description files, it
3269 :     continues to be true that there must be at least another arc after the
3270 :     anchor. This is now enforced separately during semantic analysis
3271 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3272 : blume 666
3273 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3274 :     is a sub-option list of precisely two items -- one labeled "anchor"
3275 :     and the other one labeled "value". As you might expect, "anchor" is
3276 :     used to specify an anchor name to be bound, and "value" specifies what
3277 :     the anchor is being bound to.
3278 :    
3279 :     The value must be a directory name and can be given in either standard
3280 :     syntax (including the possibility that it is itself an anchored path)
3281 :     or native syntax.
3282 :    
3283 :     Examples:
3284 :    
3285 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3286 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3287 :    
3288 :     and so on.
3289 :    
3290 :     The meaning of this is that the .cm-file will be processed with an
3291 :     augmented anchor environment where the given anchor(s) is/are bound to
3292 :     the given values(s).
3293 :    
3294 :     The rationale for having this feature is this: Suppose you are trying
3295 :     to use two different (already stable) libraries a.cm and b.cm (that
3296 :     you perhaps didn't write yourself). Further, suppose each of these
3297 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3298 :     Normally you would now have a problem because the anchor "lib" can not
3299 :     be bound to more than one value globally. Therefore, the project that
3300 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3301 :     place:
3302 :    
3303 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3304 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3305 :    
3306 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3307 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3308 :    
3309 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3310 :     you when you do so at the time of CM.stabilize). Therefore, you can
3311 :     also use an anchored path as the value:
3312 :    
3313 :     a.cm (bind:(anchor:lib value:$a-lib))
3314 :     b.cm (bind:(anchor:lib value:$b-lib))
3315 :    
3316 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3317 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3318 :     itself is always locally bound, setting it globally is no longer
3319 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3320 :     can still be used as a global anchor for separate purposes. As a
3321 :     matter of fact, one can locally define "lib" in terms of a global
3322 :     "lib":
3323 :    
3324 :     a.cm (bind:(anchor:lib value:$lib/a))
3325 :     b.cm (bind:(anchor:lib value:$lib/b))
3326 :    
3327 :     ** 3: The encoding of path names has changed. This affects the way
3328 :     path names are shown in CM's progress report and also the internal
3329 :     protocol encoding used for parallel make.
3330 :    
3331 :     The encoding now uses one or more ':'-separated segments. Each
3332 :     segments corresponds to a file that has been specified relative to the
3333 :     file given by its preceding segment. The first segment is either
3334 :     relative to the CWD, absolute, or anchored. Each segment itself is
3335 :     basically a Unix pathname; all segments but the first are relative.
3336 :    
3337 :     Example:
3338 :    
3339 :     $foo/bar/baz.cm:a/b/c.sml
3340 :    
3341 :     This path denotes the file bar/a/b/c.sml relative to the directory
3342 :     denoted by anchor "foo". Notice that the encoding also includes
3343 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3344 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3345 :     be ignored to get the "real" pathname.
3346 :    
3347 :     To make this fact more obvious, CM puts the names of such "virtual
3348 :     arcs" into parentheses when they appear in progress reports. (No
3349 :     parentheses will appear in the internal protocol encoding.) Thus,
3350 :     what you really see is:
3351 :    
3352 :     $foo/bar/(baz.cm):a/b/c.sml
3353 :    
3354 :     I find this notation to be much more informative than before.
3355 :    
3356 :     Another new feature of the encoding is that special characters
3357 :     including parentheses, colons, (back)slashes, and white space are
3358 :     written as \ddd (where ddd is the decimal encoding of the character).
3359 :    
3360 : blume 670 *** The CM manual still needs to be updated.
3361 :    
3362 : blume 666 ----------------------------------------------------------------------
3363 : leunga 660 Name: Allen Leung
3364 :     Date: 2000/06/15 00:38:00
3365 :     Tag: leunga-20000615-x86-peephole
3366 :    
3367 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3368 :    
3369 :     ----------------------------------------------------------------------
3370 : blume 658 Name: Matthias Blume
3371 :     Date: 2000/06/12 11:40:00
3372 :     Tag: blume-20000612-parmakefix
3373 :     Description:
3374 :    
3375 :     More cleanup after changing the file naming scheme: This time I
3376 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3377 :    
3378 :     ----------------------------------------------------------------------
3379 : leunga 657 Name: Allen Leung
3380 :     Date: 2000/06/09 01:25:00
3381 :     Tag: leunga-20000609-various
3382 :    
3383 :     None of these things should affect normal SML/NJ operations
3384 :    
3385 :     1. Peephole improvements provided by Fermin (c--)
3386 :     2. New annotation DEFUSE for adding extra dependence (moby)
3387 :     3. New X86 LOCK instructions (moby)
3388 :     4. New machine description language for reservation tables (scheduling)
3389 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3390 :     trees etc.)
3391 :     6. I've changed the CM files so that they can work with versions
3392 :     110.0.6, 110.25 and 110.28
3393 :    
3394 :     ----------------------------------------------------------------------
3395 : blume 652 Name: Matthias Blume
3396 : blume 656 Date: 2000/06/09 12:40:00
3397 :     Tag: blume-20000609-log
3398 :     Description:
3399 :    
3400 :     - Removed all(?) remaining RCS Log entries from sources.
3401 :    
3402 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3403 :     anchored paths).
3404 :    
3405 :     ----------------------------------------------------------------------
3406 :     Name: Matthias Blume
3407 : blume 653 Date: 2000/06/07 17:00:00 JST
3408 :     Tag: blume-20000607-no-implicit-anchors
3409 :     Description:
3410 :    
3411 :     1. This update changes the default setting for
3412 :     CM.Control.implicit_anchors from true to false. This means that
3413 :     implicit anchors are no longer permitted by default. I also tried to
3414 :     make sure that nothing else still relies on implicit anchors.
3415 :     (This is the next step on the schedule towards a CM that does not even
3416 :     have the notion of implicit anchors anymore.)
3417 :    
3418 :     2. More CM manual updates.
3419 :    
3420 :     3. I managed to track down and fix the pickling bug I mentioned last
3421 :     time. Because of the previously existing workaround, this entails no
3422 :     immediate practical changes.
3423 :    
3424 :     ----------------------------------------------------------------------
3425 :     Name: Matthias Blume
3426 : blume 652 Date: 2000/06/06 11:15:00 JST
3427 :     Tag: blume-20000606-lazierpickle
3428 :     Description:
3429 :    
3430 :     !!!! NEW BOOT FILES !!!!
3431 :    
3432 :     * The main purpose of this update is to make library pickles lazier in
3433 :     order to reduce the initial space penalty for autoloading a library.
3434 :     As a result, it is now possible to have $smlnj/compiler.cm
3435 :     pre-registered. This should take care of the many complaints or
3436 :     inquiries about missing structure Compiler. This required changes to
3437 :     CM's internal data structures and small tweaks to some algorithms.
3438 :    
3439 :     As a neat additional effect, it is no longer necessary (for the sake
3440 :     of lean heap image files) to distinguish between a "minimal" CM and a
3441 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3442 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3443 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3444 :    
3445 :     To make the life of compiler-hackers easier, "makeml" now also
3446 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3447 :     words, after you bootstrap a new sml for the first time, you will not
3448 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3449 :     around you will still have to do it, though.)
3450 :    
3451 :     * A second change consists of major updates to the CM manual. There
3452 :     are now several appendices with summary information and also a full
3453 :     specification of the CM description file syntax.
3454 :    
3455 :     * In directory src/system I added the script "allcross". This script
3456 :     invokes sml and cross-compiles the compiler for all supported
3457 :     architectures. (Useful when providing a new set of boot files.)
3458 :    
3459 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3460 :     added a small tweak to pickle-util.sml to work around this problem,
3461 :     but it is not a proper fix yet. I will investigate further. (The
3462 :     effect of the bug was an inflation of library pickle size.)
3463 :    
3464 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3465 :    
3466 :     ----------------------------------------------------------------------
3467 : leunga 646 Name: Allen Leung
3468 : leunga 648 Date: 2000/05/25 17:28 EDT
3469 :     Tag: leunga-20000525-ra
3470 :     Description:
3471 :    
3472 :     Fixed a bug in freezing phase of the register allocator.
3473 :    
3474 :     ----------------------------------------------------------------------
3475 :     Name: Allen Leung
3476 : leunga 647 Date: 2000/05/15 22:53 EDT
3477 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3478 :     Description:
3479 :    
3480 :     1. Alpha
3481 :    
3482 :     Slight cleanup. Removed the instruction SGNXL
3483 :    
3484 :     2. X86
3485 :    
3486 :     Added the following instructions to the instruction set:
3487 :    
3488 :     ROLx, RORx,
3489 :     BTx, BTSx, BTLx, BTRx,
3490 :     XCHGx, and variants with the LOCK prefix
3491 :    
3492 :     3. Register Allocation
3493 :    
3494 :     The module ra-rewrite-with-renaming has been improved.
3495 :    
3496 : leunga 647 These have no effect on SML/NJ.
3497 :    
3498 : leunga 646 ----------------------------------------------------------------------
3499 : blume 642 Name: Matthias Blume
3500 : blume 645 Date: 2000/05/15 16:20:00 JST
3501 :     Tag: blume-20000515-lightrebuild
3502 :     Description:
3503 :    
3504 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
3505 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
3506 :     defined. In effect, the command will not build any cross-compiler
3507 :     backends and therefore finish more quickly.
3508 :    
3509 :     The "fixpt" script also takes a "-light" switch to be able to use
3510 :     this new facility while compiling for a fixpoint.
3511 :    
3512 :     2. I replaced all mentions of anchored paths in group owner specifications
3513 :     with simple relative paths (usually starting with "..").
3514 :     The rationale is that a library's internal workings should not be
3515 :     compromised by the lack of some anchor. (An anchor is necessary
3516 :     for someone who wants to refer to the library by an anchored path,
3517 :     but it should not be necessary to build the same library in the first
3518 :     place.)
3519 :    
3520 :     3. I changed the way CM's tool mechanism determines the shell command
3521 :     string used for things like ml-yacc etc. so that it does not break
3522 :     when CM.Control.implicit_anchors is turned off.
3523 :    
3524 :     ----------------------------------------------------------------------
3525 :     Name: Matthias Blume
3526 : blume 644 Date: 2000/05/12 18:20:00 JST
3527 :     Tag: blume-20000512-ml-build
3528 :     Description:
3529 :    
3530 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
3531 :     getting installed properly (by config/install.sh).
3532 :    
3533 :     ----------------------------------------------------------------------
3534 :     Name: Matthias Blume
3535 : blume 643 Date: 2000/05/12 17:30:00 JST
3536 :     Tag: blume-20000512-anchors
3537 :     Description:
3538 :    
3539 :     !!! NEW BOOT FILES !!!
3540 :    
3541 :     This change is in preparation of fading out support for "implicitly
3542 :     anchored path names". I went through all sources and used the
3543 :     explicit (and relatively new) $-notation. See system/README and the
3544 :     CM manual for more info on this.
3545 :    
3546 :     I also modified the anchoring scheme for some things such as "smlnj",
3547 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
3548 :     anchors are more expressive: anchor name and first arc do not have to
3549 :     coincide. This entails the following user-visible change:
3550 :    
3551 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
3552 :     particular, when you fire up sml with a command-line argument, say,
3553 :     e.g.:
3554 :    
3555 :     sml '$smlnj/cmb.cm'
3556 :    
3557 :     At the ML toplevel prompt:
3558 :    
3559 :     CM.autoload "$smlnj/cmb.cm";
3560 :    
3561 :     There is also a new controller in CM.Control that can be used to turn
3562 :     off all remaining support for implicit anchors by saying:
3563 :    
3564 :     CM.autoload "$smlnj/
3565 :     #set CM.Control.implicit_anchors false;
3566 :    
3567 :     This causes CM to reject implicitly anchored paths. This is (for the
3568 :     time being) less permissive than the "final" version where there will
3569 :     be no more such implicit anchors and relative paths will be just that:
3570 :     relative.
3571 :    
3572 :     The next step (version after next version?) will be to make the
3573 :     default for CM.Control.implicit_anchors false. After the dust has
3574 :     settled, I can then produce the "final" version of this...
3575 :    
3576 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
3577 :    
3578 :     ----------------------------------------------------------------------
3579 :     Name: Matthias Blume
3580 : blume 642 Date: 2000/05/11 16:30:00 JST
3581 :     Tag: blume-20000511-sources
3582 :     Description:
3583 :    
3584 :     The main change is that I added function CM.sources as a generalized
3585 :     version of the earlier CM.makedepend. This entails the following
3586 :     additional changes:
3587 :    
3588 :     - CM.makedepend has been dropped.
3589 :    
3590 :     - CM manual has been updated.
3591 :    
3592 :     - TOOLS signature and API have been changed.
3593 :    
3594 :     ----------------------------------------------------------------------
3595 : leunga 640 Name: Allen Leung
3596 : leunga 641 Date: 2000/05/10 21:17 EDT
3597 :     Tag: leunga-20000510-moby-c--ssa
3598 :     Description:
3599 :    
3600 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
3601 :     None of these affect SML/NJ.
3602 :    
3603 :     1. Register Allocation
3604 :    
3605 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
3606 :     This module tries to remove local (i.e. basic block level) redundancies
3607 :     during spilling.
3608 :    
3609 :     b. A new framework for performing region based register allocation.
3610 :     Not yet entirely functional.
3611 :    
3612 :     2. X86
3613 :    
3614 :     a. DefUse for POP was missing the stack pointer [found by Lal]
3615 :     b. Reload for CALL was incorrect in X86Spill [found by John]
3616 :     c. Various fixes in X86Spill so that it can be used correctly for
3617 :     the new spilling module.
3618 :    
3619 :     3. SSA/IR
3620 :    
3621 :     a. New module ir/dj-dataflow.sml implements elimination based
3622 :     data flow analysis.
3623 :    
3624 :     4. MLRiscGen
3625 :    
3626 :     a. Fix for gc type annotation
3627 :    
3628 :     5. MDGen
3629 :    
3630 :     Various fixes for machine description -> ml code translation. For ssa
3631 :     only.
3632 :    
3633 :     ----------------------------------------------------------------------
3634 :     Name: Allen Leung
3635 : leunga 640 Date: 2000/05/08 22:17 EDT
3636 :     Tag: leunga-20000508-labexp
3637 :     Description:
3638 :    
3639 :     Fermin has found a few assembly problems with constant expressions
3640 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
3641 :     which choke on dumb assemblers. This is his fix.
3642 :    
3643 :     ----------------------------------------------------------------------
3644 : dbm 635 Name: Dave MacQueen
3645 :     Date: 2000/04/09 14:00 EDT
3646 :     Tag: dbm-20000502-Version_110_28
3647 :     Description:
3648 :    
3649 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
3650 :    
3651 :     2. Updated config/version to 110.28
3652 :    
3653 :     3. Updated config/srcarchiveurl
3654 :    
3655 :     3. New boot files!
3656 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
3657 :    
3658 :     ----------------------------------------------------------------------
3659 : blume 634 Name: Matthias Blume
3660 :     Date: 2000/05/01 19:05:00 JST
3661 :     Tag: blume-20000501-noweb
3662 :     Description:
3663 :    
3664 :     A new noweb tool has been added. The existing system is entirely
3665 :     unaffected by this, but some CM users have asked for renewed noweb
3666 :     support. Everything is documented in the CM manual.
3667 :    
3668 :     New (plugin) libraries:
3669 :    
3670 :     noweb-tool.cm
3671 :     nw-ext.cm
3672 :    
3673 :     ----------------------------------------------------------------------
3674 : dbm 633 Name: Dave MacQueen
3675 :     Date: 2000/04/30 12:40PM EDT
3676 :     Tag: dbm-20000430-bug_fixes
3677 :     Description:
3678 :    
3679 :     1. Fix for bug 1498
3680 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
3681 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
3682 :     added toRealArray function
3683 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
3684 :     added check for tag Obj.RealArray to array printing case in ppObj
3685 :    
3686 :     2. Fix for bug 1510
3687 :     smlnj/src/compiler/Semant/types/typesutil.sml
3688 :     fixed definition of dummyargs (used by equalTycon) so that
3689 :     dummy args are distinct types
3690 :    
3691 :     ----------------------------------------------------------------------
3692 : blume 629 Name: Matthias Blume
3693 : blume 632 Date: 2000/04/30 01:00:00 JST
3694 :     Tag: blume-20000430-versions
3695 :     Description:
3696 :    
3697 :     1. CM version numbering added. This is an implementation of Lal's
3698 :     proposal for adding version numbers and version checking to .cm
3699 :     files. Lal said that his proposal was just that -- a proposal.
3700 :     For the time being I went ahead and implemented it so that people
3701 :     can comment on it. Everything is completely backward-compatible
3702 :     (except for the stable library format, i.e., new bootfiles!).
3703 :    
3704 :     As usual, see the CM manual for details.
3705 :    
3706 :     2. An alternative syntax for anchored paths has been implemented.
3707 :     Dave has recently voiced the same concerns that I had when I did
3708 :     this, so there should be some support. My take is that eventually
3709 :     I will let support for the current syntax (where anchors are
3710 :     "implicit") fade out in favor of the new, explicit syntax.
3711 :     In order to be backward-compatible, both old and new syntax are
3712 :     currently supported.
3713 :    
3714 :     Again, see the CM manual for details.
3715 :    
3716 :     3. Parallel make is trying to be slightly smarter: When the master
3717 :     process finds a "bottleneck", i.e., when there is only one
3718 :     compilation unit that can be compiled and everybody else is
3719 :     waiting on it, then it will simply compile it directly instead
3720 :     of clumsily telling one of the slaves to do it.
3721 :    
3722 :     4. Support for "unsharing" added. This is necessary in order to be
3723 :     able to have two different versions of the same library running
3724 :     at the same time (e.g., for trying out a new MLRISC while still
3725 :     having the old MLRISC linked into the current compiler, etc.)
3726 :     See the CM manual.
3727 :    
3728 :     5. Simple "makedepend" functionality added for generating Makefile
3729 :     dependency information. (This is rather crude at the moment.
3730 :     Expect some changes here in the future.)
3731 :    
3732 :     6. ".fun" added as a recognized suffix for ML files. Also documented
3733 :     explicitly in the manual that the fallback behavior (unknown suffix
3734 :     -> ML file) is not an official feature!
3735 :    
3736 :     7. Small changes to the pickler for stable libraries.
3737 :    
3738 :     8. Several internal changes to CM (for cleanup/improvement).
3739 :    
3740 :    
3741 :     !!!! NEW BINFILES !!!!
3742 :    
3743 :     ----------------------------------------------------------------------
3744 :     Name: Matthias Blume
3745 : blume 631 Date: 2000/04/28 17:30:00 JST
3746 :     Tag: blume-20000428-pathconfig
3747 :     Description:
3748 :    
3749 :     1. I changed config/install.sh to remove duplicate entries from the
3750 :     lib/pathconfig file at the end. Moreover, the final version of
3751 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
3752 :     in src/system/installml.
3753 :    
3754 :     2. The config/install.sh script now consistently uses relative
3755 :     pathnames in lib/pathconfig whenever the anchor is in the lib
3756 :     directory. (So far this was true for the libraries that come
3757 :     pre-compiled and bundled as part of the bootfiles but not for
3758 :     libraries that are compiled by the script itself.)
3759 :    
3760 :     ----------------------------------------------------------------------
3761 :     Name: Matthias Blume
3762 : blume 629 Date: 2000/04/26 13:10:00 JST
3763 :     Tag: blume-20000426-fun_suffix
3764 :     Description:
3765 :    
3766 :     Added ".fun" as a recognized file name suffix (for ML code).
3767 :    
3768 :     ----------------------------------------------------------------------
3769 : leunga 628 Name: Allen Leung
3770 :     Date: 2000/04/25 17:00:00 EST
3771 :     Tag: leunga-20000425-alpha-ra
3772 :     Description:
3773 :    
3774 :     1. Alpha
3775 :    
3776 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
3777 :     in C--.
3778 :    
3779 :     2. RA
3780 :    
3781 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
3782 :     and the interference graph.
3783 :    
3784 :     ----------------------------------------------------------------------
3785 : dbm 627 Name: Dave MacQueen
3786 :     Date: 2000/04/25/
3787 :     Tag: dbm-20000425-mlyacc_doc_examples
3788 :     Description:
3789 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
3790 :     Updated all three examples in src/ml-yacc/examples to run
3791 :     under 110.* using CM.make.
3792 :    
3793 :     ----------------------------------------------------------------------
3794 : leunga 624 Name: Allen Leung
3795 :     Date: 2000/04/20 23:04:00 EST
3796 :     Tag: leunga-20000420-ssa-c---stuff
3797 :     Description:
3798 :    
3799 :     This update synchronizes my repository with Yale's. Most of these
3800 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
3801 :    
3802 :     1. Register Allocator
3803 :    
3804 :     a. An improvement in the interference graph construction:
3805 :     Given a copy
3806 :    
3807 :     s <- t
3808 :    
3809 : leunga 628 no interference edge between s and t is added for this definition of s.
3810 : leunga 624
3811 :     b. I've added two new spill heuristic modules that Fermin and I developed
3812 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
3813 :     useful for others (Moby?)
3814 :    
3815 :     2. X86
3816 :    
3817 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
3818 :    
3819 :     3. Alpha
3820 :    
3821 :     a. Added the BSR instruction and code generation that goes with it [C--]
3822 :     b. Other fixes too numerous to recount provided by Fermin [C--]
3823 :    
3824 :     4. Regmaps
3825 :    
3826 :     a. The regmaps are not initialized with the identity physical bindings
3827 :     at creation time. This is unneeded.
3828 :    
3829 :     5. MLRISC Optimizations
3830 :    
3831 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
3832 :     intersects with liveness incrementally in linear time! Woohoo!
3833 :     This is now used in my new SSA construction algorithm.
3834 :    
3835 :     b. THe branch reorganization module is now smarter about linear chains of
3836 :     basic blocks.
3837 :    
3838 :    
3839 :     ----------------------------------------------------------------------
3840 : blume 615 Name: Matthias Blume
3841 :     Date: 2000/04/12 13:52:00 JST
3842 :     Tag: blume_main_v110p27_1
3843 :     Description:
3844 :    
3845 :     Changed install.sh script to handle archive files without version number
3846 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
3847 :     name of the boot file archive.
3848 :    
3849 :     ----------------------------------------------------------------------
3850 : dbm 607 Name: Dave MacQueen
3851 :     Date: 2000/04/09 14:00 EDT
3852 :     Tag: dbm-20000410-Version_110_27
3853 :     Description:
3854 :    
3855 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
3856 :    
3857 :     2. Updated src/config/version to 110.27
3858 :    
3859 :     3. New boot files!
3860 :    
3861 :     ----------------------------------------------------------------------
3862 : leunga 606 Name: Allen Leung
3863 :     Date: 2000/04/09 19:09:00 EST
3864 :     Tag: leunga-20000409-misc
3865 :     Description:
3866 :    
3867 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
3868 :    
3869 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
3870 :    
3871 :     ----------------------------------------------------------------------
3872 : monnier 605 Name: Stefan
3873 :     Date: 2000/04/07 10:00:00 EDT
3874 :     Tag: monnier-20000406-branch-handling
3875 :     Description:
3876 :    
3877 :     Improved handling of branches (mostly those generated from
3878 :     polymorphic equality), removed switchoff and changed the
3879 :     default optimization settings (more cpsopt and less flintopt).
3880 :    
3881 :     ----------------------------------------------------------------------
3882 : leunga 601 Name: Allen Leung
3883 : leunga 602 Date: 2000/04/06 01:30:00 EST
3884 :     Tag: leunga-20000406-peephole-x86-SSA-2
3885 :     Description:
3886 :    
3887 :     Forgot a few files.
3888 :    
3889 :     ----------------------------------------------------------------------
3890 :     Name: Allen Leung
3891 : leunga 601 Date: 2000/04/06 00:36:00 EST
3892 :     Tag: leunga-20000406-peephole-x86-SSA
3893 :     Description:
3894 :    
3895 :     1. New Peephole code
3896 :    
3897 :     2. Minor improvement to X86 instruction selection
3898 :    
3899 :     3. Various fixes to SSA and machine description -> code translator
3900 :    
3901 :     ----------------------------------------------------------------------
3902 : blume 595 Name: Matthias Blume
3903 :     Date: 2000/04/05 12:30:00 JST
3904 :     Tag: blume_main_v110p26p2_3
3905 :     Description:
3906 :    
3907 :     This update just merges three minor cosmetic updates to CM's sources
3908 :     to get ready for the 110.27 code freeze on Friday. No functionality
3909 :     has changed.
3910 :    
3911 :     ----------------------------------------------------------------------
3912 : leunga 593 Name: Allen Leung
3913 : leunga 594 Date: 2000/04/04 19:39:00 EST
3914 :     Tag: leunga-20000404-x86-asm
3915 :     Description:
3916 :    
3917 :     1. Fixed a problem in X86 assembly.
3918 :    
3919 :     Things like
3920 :    
3921 :     jmp %eax
3922 :     jmp (%eax)
3923 :    
3924 :     should be output as
3925 :    
3926 :     jmp *%eax
3927 :     jmp *(%eax)
3928 :    
3929 :     2. Assembly output
3930 :    
3931 :     Added a new flag
3932 :    
3933 :     "asm-indent-copies" (default to false)
3934 :    
3935 :     When this flag is on, parallel copies will be indented an extra level.
3936 :    
3937 :     ----------------------------------------------------------------------
3938 :     Name: Allen Leung
3939 : leunga 593 Date: 2000/04/04 03:18:00 EST
3940 :     Tag: leunga-20000404-C--Moby
3941 :     Description:
3942 :    
3943 :     All of these fixes are related to C--, Moby, and my own optimization
3944 :     stuff; so they shouldn't affect SML/NJ.
3945 :    
3946 :     1. X86
3947 :    
3948 :     Various fixes related floating point, and extensions.
3949 :    
3950 :     2. Alpha
3951 :    
3952 :     Some extra patterns related to loads with signed/zero extension
3953 :     provided by Fermin.
3954 :    
3955 :     3. Assembly
3956 :    
3957 : monnier 659 When generating assembly, resolve the value of client defined constants,
3958 : leunga 593 instead of generating symbolic values. This is controlled by the
3959 :     new flag "asm-resolve-constants", which is default to true.
3960 :    
3961 :     4. Machine Descriptions
3962 :    
3963 :     a. The precedence parser was slightly broken when parsing infixr symbols.
3964 :     b. The type generalizing code had the bound variables reversed, resulting
3965 :     in a problem during arity raising.
3966 :     c. Various fixes in machine descriptions.
3967 :    
3968 :     ----------------------------------------------------------------------
3969 : blume 592 Name: Matthias Blume
3970 :     Date: 2000/04/03 16:05:00 JST
3971 :     Tag: blume_main_v110p26p2_2
3972 :     Description:
3973 :    
3974 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
3975 :     now done via the ordinary static environment that is context to each
3976 :     compilation unit.
3977 :    
3978 :     To this end, I arranged that instead of "structure Core" as "structure
3979 :     _Core" is bound in the pervasive environment. Core access is done via
3980 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
3981 : blume 592 legal surface-syntax symbol).
3982 :    
3983 :     The current solution is much cleaner because the core environment is
3984 :     now simply part of the pervasive environment which is part of every
3985 :     compilation unit's context anyway. In particular, this eliminates all
3986 :     special-case handling that was necessary until now in order to deal
3987 :     with dynamic and symbolic parts of the core environment.
3988 :    
3989 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3990 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
3991 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
3992 :     more details.
3993 :    
3994 :     I also tried to track down all mentions of "Core" (as string argument
3995 :     to Symbol.strSymbol) in the compiler and replaced them with a
3996 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
3997 :     name appears in one place only.
3998 :    
3999 :     Binfile and bootfile format have not changed, but the switchover from
4000 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
4001 :     bootfiles anyway.
4002 :    
4003 :     ----------------------------------------------------------------------
4004 : leunga 591 Name: Allen Leung
4005 :     Date: 2000/04/02 21:17:00 EST
4006 :     Tag: leunga-20000402-mltree
4007 :     Description:
4008 : leunga 585
4009 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
4010 :     2. Added a bunch of files from my repository. These are currently
4011 :     used by other non-SMLNJ backends.
4012 :    
4013 : leunga 576 ----------------------------------------------------------------------
4014 : leunga 590 Name: Allen Leung
4015 :     Date: 2000/03/31 21:15:00 EST
4016 :     Tag: leunga-20000331-aliasing
4017 :     Description:
4018 :    
4019 :     This update contains a rewritten (and hopefully more correct) module
4020 :     for extracting aliasing information from CPS.
4021 :    
4022 :     To turn on this feature:
4023 :    
4024 :     Compiler.Control.CG.memDisambiguate := true
4025 :    
4026 :     To pretty print the region information with assembly
4027 :    
4028 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
4029 :    
4030 :     To control how many levels of aliasing information are printed, use:
4031 :    
4032 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
4033 :    
4034 :     The default of n is 3.
4035 :    
4036 :     ----------------------------------------------------------------------
4037 : dbm 589 Name: David MacQueen
4038 :     Date: 2000/03/31 11:15:00 EST
4039 :     Tag: dbm-20000331-runtime_fix
4040 :     Description:
4041 :    
4042 :     This update contains:
4043 :    
4044 :     1. runtime/c-lib/c-libraries.c
4045 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
4046 :    
4047 :     2. fix for bug 1556
4048 :     system/Basis/Implementation/NJ/internal-signals.sml
4049 :    
4050 :     ----------------------------------------------------------------------
4051 : blume 587 Name: Matthias Blume
4052 : blume 588 Date: 2000/03/31 18:00:00 JST
4053 :     Tag: blume_main_v110p26p2_1
4054 :     Description:
4055 :    
4056 :     This update contains:
4057 :    
4058 :     1. A small change to CM's handling of stable libraries:
4059 :     CM now maintains one "global" modmap that is used for all stable
4060 :     libraries. The use of such a global modmap maximizes sharing and
4061 :     minimizes the need for re-traversing parts of environments during
4062 :     modmap construction. (However, this has minor impact since modmap
4063 :     construction seems to account for just one percent or less of total
4064 :     compile time.)
4065 :    
4066 :     2. I added a "genmap" phase to the statistics. This is where I got the
4067 :     "one percent" number (see above).
4068 :    
4069 :     3. CM's new tool parameter mechanism just became _even_ better. :)
4070 :     - The parser understands named parameters and recursive options.
4071 :     - The "make" and "shell" tools use these new features.
4072 :     (This makes it a lot easier to cascade these tools.)
4073 :     - There is a small syntax change: named parameters use a
4074 :    
4075 :     <name> : ( <option> ... ) or
4076 :     <name> : <string>
4077 :    
4078 :     syntax. Previously, named parameters were implemented in an
4079 :     ad-hoc fashion by each tool individually (by parsing strings)
4080 :     and had the form
4081 :    
4082 :     <name>=<string>
4083 :    
4084 :     See the CM manual for a full description of these issues.
4085 :    
4086 :     ----------------------------------------------------------------------
4087 :     Name: Matthias Blume
4088 : blume 587 Date: 2000/03/30 18:00:00 JST
4089 :     Tag: blume_main_v110p26p2_0
4090 :     Description:
4091 :    
4092 :     !!!!! WARNING !!!!!!
4093 :     !! New binfiles !!
4094 :     !!!!!!!!!!!!!!!!!!!!
4095 :    
4096 :     This update contains:
4097 :    
4098 :     1. Moderate changes to CM:
4099 :    
4100 :     - Changes to CM's tools mechanism. In particular, it is now possible
4101 :     to have tools that accept additional "command line" parameters
4102 :     (specified in the .cm file at each instance where the tool's class is
4103 :     used).
4104 :    
4105 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
4106 :     facilitate fairly seamless hookup to portions of code managed using
4107 : blume 587 Makefiles or Shell scripts.
4108 :    
4109 :     There are no classes "shared" or "private" anymore. Instead, the
4110 :     sharing annotation is now a parameter to the "sml" class.
4111 :    
4112 :     There is a bit of generic machinery for implementing one's own
4113 :     tools that accept command-line parameters. However, I am not yet fully
4114 :     satisfied with that part, so expect changes here in the future.
4115 :    
4116 :     All existing tools are described in the CM manual.
4117 :    
4118 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
4119 : blume 587 error messages that tended to be more annoying than helpful.)
4120 :    
4121 :     2. Major changes to the compiler's static environment data structures.
4122 :    
4123 :     - no CMStaticEnv anymore.
4124 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
4125 :     but it is called Environment), no conversions between different
4126 :     kinds of static environments
4127 :    
4128 :     - There is still a notion of a "modmap", but such modmaps are generated
4129 :     on demand at the time when they are needed. This sounds slow, but I
4130 :     sped up the code that generates modmaps enough for this not to lead to
4131 :     a slowdown of the compiler (at least I didn't detect any).
4132 :    
4133 :     - To facilitate rapid modmap generation, static environments now
4134 :     contain an (optional) "modtree" structure. Modtree annotations are
4135 :     constructed by the unpickler during unpickling. (This means that
4136 :     the elaborator does not have to worry about modtrees at all.)
4137 :     Modtrees have the advantage that they are compositional in the same
4138 :     way as the environment data structure itself is compositional.
4139 :     As a result, modtrees never hang on to parts of an environment that
4140 :     has already been rendered "stale" by filtering or rebinding.
4141 :    
4142 :     - I went through many, many trials and errors before arriving at the
4143 :     current solution. (The initial idea of "linkpaths" did not work.)
4144 :     But the result of all this is that I have touched a lot of files that
4145 :     depend on the "modules" and "types" data structures (most of the
4146 :     elaborator). There were a lot of changes during my "linkpath" trials
4147 :     that could have been reverted to their original state but weren't.
4148 :     Please, don't be too harsh on me for messing with this code a bit more
4149 : monnier 921 than what was strictly necessary... (I _did_ resist the temptation
4150 : blume 587 of doing any "global reformatting" to avoid an untimely death at
4151 :     Dave's hands. :)
4152 :    
4153 :     - One positive aspect of the previous point: At least I made sure that
4154 :     all files that I touched now compile without warnings (other than
4155 :     "polyEqual").
4156 :    
4157 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
4158 :     redundant modmaps)
4159 :    
4160 :     ----------------------------------------------------------------------
4161 : leunga 580 Name: Allen Leung
4162 : leunga 585 Date: 2000/03/29 18:00:00
4163 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
4164 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
4165 : leunga 585 Description:
4166 :    
4167 :     This update contains *MAJOR* changes to the way code is generated from CPS
4168 :     in the module mlriscGen, and in various backend modules.
4169 :    
4170 :     CHANGES
4171 :     =======
4172 :    
4173 :     1. MLRiscGen: forward propagation fix.
4174 :    
4175 :     There was a bug in forward propagation introduced at about the same time
4176 :     as the MLRISC x86 backend, which prohibits coalescing to be
4177 :     performed effectively in loops.
4178 :    
4179 :     Effect: speed up of loops in RISC architectures.
4180 :     By itself, this actually slowed down certain benchmarks on the x86.
4181 :    
4182 :     2. MLRiscGen: forward propagating addresses from consing.
4183 :    
4184 :     I've changed the way consing code is generated. Basically I separated
4185 :     out the initialization part:
4186 :    
4187 :     store tag, offset(allocptr)
4188 :     store elem1, offset+4(allocptr)
4189 :     store elem2, offset+8(allocptr)
4190 :     ...
4191 :     store elemn, offset+4n(allocptr)
4192 :    
4193 :     and the address computation part:
4194 :    
4195 :     celladdr <- offset+4+alloctpr
4196 :    
4197 :     and move the address computation part
4198 :    
4199 :     Effect: register pressure is generally lower as a result. This
4200 :     makes compilation of certain expressions much faster, such as
4201 :     long lists with non-trivial elements.
4202 :    
4203 :     [(0,0), (0,0), .... (0,0)]
4204 :    
4205 :     3. MLRiscGen: base pointer elimination.
4206 :    
4207 :     As part of the linkage mechanism, we generate the sequence:
4208 :    
4209 :     L: ... <- start of the code fragment
4210 :    
4211 :     L1:
4212 :     base pointer <- linkreg - L1 + L
4213 :    
4214 :     The base pointer was then used for computing relocatable addresses
4215 :     in the code fragment. Frequently (such as in lots of continuations)
4216 :     this is not needed. We now eliminate this sequence whenever possible.
4217 :    
4218 :     For compile time efficiency, I'm using a very stupid local heuristic.
4219 :     But in general, this should be done as a control flow analysis.
4220 :    
4221 :     Effect: Smaller code size. Speed up of most programs.
4222 :    
4223 :     4. Hppa back end
4224 :    
4225 :     Long jumps in span dependence resolution used to depend on the existence
4226 :     of the base pointer.
4227 :    
4228 :     A jump to a long label L was expanded into the following sequence:
4229 :    
4230 :     LDIL %hi(L-8192), %r29
4231 :     LDO %lo(L-8192)(%r29), %r29
4232 :     ADD %r29, baseptr, %r29
4233 :     BV,n %r0(%r29)
4234 :    
4235 :     In the presence of change (3) above, this will not work. I've changed
4236 :     it so that the following sequence of instructions are generated, which
4237 :     doesn't mention the base pointer at all:
4238 :    
4239 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
4240 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
4241 :     BV,n %r0(%r29) /* Jump */
4242 :    
4243 :     5. Alpha back end
4244 :    
4245 :     New alpha instructions LDB/LDW have been added, as per Fermin's
4246 :     suggestions. This is unrelated to all other changes.
4247 :    
4248 :     6. X86 back end
4249 :    
4250 :     I've changed andl to testl in the floating point test sequence
4251 :     whenever appropriate. The Intel optimization guide states that
4252 : monnier 659 testl is preferable to andl.
4253 : leunga 585
4254 :     7. RA (x86 only)
4255 :    
4256 :     I've improved the spill propagation algorithm, using an approximation
4257 :     of maximal weighted independent sets. This seems to be necessary to
4258 :     alleviate the negative effect in light of the slow down in (1).
4259 :    
4260 :     I'll write down the algorithm one of these days.
4261 :    
4262 :     8. MLRiscGen: frequencies
4263 :    
4264 :     I've added an annotation that states that all call gc blocks have zero
4265 :     execution frequencies. This improves register allocation on the x86.
4266 :    
4267 :     BENCHMARKS
4268 :     ==========
4269 :    
4270 :     I've only perform the comparison on 110.25.
4271 :    
4272 :     The platforms are:
4273 :    
4274 :     HPPA A four processor HP machine (E9000) with 5G of memory.
4275 :     X86 A 300Hhz Pentium II with 128M of memory, and
4276 :     SPARC An Ultra sparc 2 with 512M of memory.
4277 :    
4278 :     I used the following parameters for the SML benchmarks:
4279 :    
4280 :     @SMLalloc
4281 :     HPPA 256k
4282 :     SPARC 512k
4283 :     X86 256k
4284 :    
4285 :     COMPILATION TIME
4286 :     ----------------
4287 :     Here are the numbers comparing the compilation times of the compilers.
4288 :     I've only compared 110.25 compiling the new sources versus
4289 :     a fixpoint version of the new compiler compiling the same.
4290 :    
4291 :     110.25 New
4292 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
4293 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
4294 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
4295 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
4296 :    
4297 :     110.25 New
4298 :     Code Size Code Size
4299 :     HPPA 8596736 8561421
4300 :     SPARC 8974299 8785143
4301 :     X86 9029180 8716783
4302 :    
4303 :     So in summary, things are at least as good as before. Dramatic
4304 :     reduction in compilation is obtained on the Sparc; I can't explain it,
4305 :     but it is reproducible. Perhaps someone should try to reproduce this
4306 :     on their own machines.
4307 :    
4308 :     SML BENCHMARKS
4309 :     --------------
4310 :    
4311 :     On the average, all benchmarks perform at least as well as before.
4312 :    
4313 :     HPPA Compilation Time Spill+Reload Run Time
4314 :     110.25 New 110.25 New 110.25 New
4315 :    
4316 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
4317 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
4318 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
4319 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
4320 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
4321 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
4322 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
4323 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%