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

SCM Repository

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

Annotation of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1096 - (view) (download)

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 : george 1003 Date: yyyy/mm/dd
12 : dbm 570 Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 1025
15 : george 1003 ----------------------------------------------------------------------
16 : blume 1096 Name: Matthias Blume
17 :     Date: 2002/02/26 12:00:00 EST
18 :     Tag: blume-20020226-ffi
19 :     Description:
20 :    
21 :     1. Fixed a minor bug in CM's "noweb" tool:
22 :     If numbering is turned off, then truly don't number (i.e., do not
23 :     supply the -L option to noweb). The previous behavior was to supply
24 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
25 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
26 :    
27 :     2. Once again, I reworked some aspects of the FFI:
28 :    
29 :     A. The incomplete/complete type business:
30 :    
31 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
32 :     gone!
33 :     - ML types representing an incomplete type are now *equal* to
34 :     ML types representing their corresponding complete types (just like
35 :     in C). This is still safe because ml-nlffigen will not generate
36 :     RTTI for incomplete types, nor will it generate functions that
37 :     require access to such RTTI. But when ML code generated from both
38 :     incomplete and complete versions of the C type meet, the ML types
39 :     are trivially interoperable.
40 :    
41 :     NOTE: These changes restore the full generality of the translation
42 :     (which was previously lost when I eliminated functorization)!
43 :    
44 :     B. Enum types:
45 :    
46 :     - Structure C now has a type constructor "enum" that is similar to
47 :     how the "su" constructor works. However, "enum" is not a phantom
48 :     type because each "T enum" has values (and is isomorphic to
49 :     MLRep.Signed.int).
50 :     - There are generic access operations for enum objects (using
51 :     MLRep.Signed.int).
52 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
53 :     * The structure contains the definition of type "mlrep" (the ML-side
54 :     representation type of the enum). Normally, mlrep is the same
55 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
56 :     then mlrep will be defined as a datatype -- thus facilitating
57 :     pattern matching on mlrep values.
58 :     ("-ec" will be suppressed if there are duplicate values in an
59 :     enumeration.)
60 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
61 :     will be generated for each C enum constant xxx.
62 :     * Conversion functions m2i and i2m convert between mlrep and
63 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
64 :     * Coversion functions c and ml convert between mlrep and "tag enum".
65 :     * Access functions (get/set) fetch and store mlrep values.
66 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
67 :     enumerations are merged into one single enumeration represented by
68 :     structure E_'.
69 :    
70 :     ----------------------------------------------------------------------
71 : leunga 1080 Name: Allen Leung
72 : leunga 1094 Date: 2002/02/25 04:45:00 EST
73 :     Tag: leunga-20020225-cps-spill
74 :    
75 :     This is a new implementation of the CPS spill phase.
76 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
77 :     In case of problems, replace it with the old file spill.sml
78 :    
79 :     The current compiler runs into some serious performance problems when
80 :     constructing a large record. This can happen when we try to compile a
81 :     structure with many items. Even a very simple structure like the following
82 :     makes the compiler slow down.
83 :    
84 :     structure Foo = struct
85 :     val x_1 = 0w1 : Word32.int
86 :     val x_2 = 0w2 : Word32.int
87 :     val x_3 = 0w3 : Word32.int
88 :     ...
89 :     val x_N = 0wN : Word32.int
90 :     end
91 :    
92 :     The following table shows the compile time, from N=1000 to N=4000,
93 :     with the old compiler:
94 :    
95 :     N
96 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
97 :     MLRISC ra 0.06u 0.00s 0.05g
98 :     (spills = 0 reloads = 0)
99 :     TOTAL 0.63u 0.07s 0.21g
100 :    
101 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
102 :     MLRISC ra 5.68u 0.59s 3.93g
103 :     (spills = 0 reloads = 0)
104 :     TOTAL 14.71u 0.99s 4.81g
105 :    
106 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
107 :     MLRISC ra 5.54u 0.65s 3.91g
108 :     (spills = 543 reloads = 1082)
109 :     TOTAL 65.40u 3.13s 6.00g
110 :    
111 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
112 :     MLRISC ra 0.80u 0.10s 0.55g
113 :     (spills = 42 reloads = 84)
114 :     TOTAL 129.42u 5.10s 4.13g
115 :    
116 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
117 :     MLRISC ra 2.69u 0.27s 1.38g
118 :     (spills = 62 reloads = 124)
119 :     TOTAL 682.48u 19.61s 13.99g
120 :    
121 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
122 :     MLRISC ra 4.96u 0.27s 2.72g
123 :     (spills = 85 reloads = 170)
124 :     TOTAL 2375.26u 57.21s 48.00g
125 :    
126 :     As you can see the old cps spill module suffers from some serious
127 :     performance problem but since I cannot decipher the old code fully,
128 :     innstead of patching the problems up, I'm reimplementing it
129 :     with a different algorithm. The new code is more modular,
130 :     smaller when compiled, and substantially faster
131 :     (O(n log n) time and O(n) space). Timing of the new spill module:
132 :    
133 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
134 :     MLRISC ra 0.25u 0.02s 0.15g
135 :     (spills=1 reloads=3)
136 :     TOTAL 7.74u 0.34s 1.62g
137 :    
138 :     Implementation details:
139 :    
140 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
141 :     number of live variables at any program point (the bandwidth)
142 :     does not exceed a certain limit, which is determined by the
143 :     size of the spill area.
144 :    
145 :     When the bandwidth is too large, we decrease the register pressure by
146 :     packing live variables into spill records. How we achieve this is
147 :     completely different than what we did in the old code.
148 :    
149 :     First, there is something about the MLRiscGen code generator
150 :     that we should be aware of:
151 :    
152 :     o MLRiscGen performs code motion!
153 :    
154 :     In particular, it will move floating point computations and
155 :     address computations involving only the heap pointer to
156 :     their use sites (if there is only a single use).
157 :     What this means is that if we have a CPS record construction
158 :     statement
159 :    
160 :     RECORD(k,vl,w,e)
161 :    
162 :     we should never count the new record address w as live if w
163 :     has only one use (which is often the case).
164 :    
165 :     We should do something similar to floating point, but the transformation
166 :     there is much more complex, so I won't deal with that.
167 :    
168 :     Secondly, there are now two new cps primops at our disposal:
169 :    
170 :     1. rawrecord of record_kind option
171 :     This pure operator allocates some uninitialized storage from the heap.
172 :     There are two forms:
173 :    
174 :     rawrecord NONE [INT n] allocates a tagless record of length n
175 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
176 :     and initializes the tag.
177 :    
178 :     2. rawupdate of cty
179 :     rawupdate cty (v,i,x)
180 :     Assigns to x to the ith component of record v.
181 :     The storelist is not updated.
182 :    
183 :     We use these new primops for both spilling and increment record construction.
184 :    
185 :     1. Spilling.
186 :    
187 :     This is implemented with a linear scan algorithm (but generalized
188 :     to trees). The algorithm will create a single spill record at the
189 :     beginning of the cps function and use rawupdate to spill to it,
190 :     and SELECT or SELp to reload from it. So both spills and reloads
191 :     are fine-grain operations. In contrast, in the old algorithm
192 :     "spills" have to be bundled together in records.
193 :    
194 :     Ideally, we should sink the spill record construction to where
195 :     it is needed. We can even split the spill record into multiple ones
196 :     at the places where they are needed. But CPS is not a good
197 :     representation for global code motion, so I'll keep it simple and
198 :     am not attempting this.
199 :    
200 :     2. Incremental record construction (aka record splitting).
201 :    
202 :     Long records with many component values which are simulatenously live
203 :     (recall that single use record addresses are not considered to
204 :     be live) are constructed with rawrecord and rawupdate.
205 :     We allocate space on the heap with rawrecord first, then gradually
206 :     fill it in with rawupdate. This is the technique suggested to me
207 :     by Matthias.
208 :    
209 :     Some restrictions on when this is applicable:
210 :     1. It is not a VECTOR record. The code generator currently does not handle
211 :     this case. VECTOR record uses double indirection like arrays.
212 :     2. All the record component values are defined in the same "basic block"
213 :     as the record constructor. This is to prevent speculative
214 :     record construction.
215 :    
216 :     ----------------------------------------------------------------------
217 :     Name: Allen Leung
218 : leunga 1086 Date: 2002/02/22 01:02:00 EST
219 :     Tag: leunga-20020222-mlrisc-tools
220 :    
221 :     Minor bug fixes in the parser and rewriter
222 :    
223 :     ----------------------------------------------------------------------
224 :     Name: Allen Leung
225 : leunga 1085 Date: 2002/02/21 20:20:00 EST
226 :     Tag: leunga-20020221-peephole
227 :    
228 :     Regenerated the peephole files. Some contained typos in the specification
229 :     and some didn't compile because of pretty printing bugs in the old version
230 :     of 'nowhere'.
231 :    
232 :     ----------------------------------------------------------------------
233 :     Name: Allen Leung
234 : leunga 1079 Date: 2002/02/19 20:20:00 EST
235 :     Tag: leunga-20020219-mlrisc-tools
236 :     Description:
237 :    
238 :     Minor bug fixes to the mlrisc-tools library:
239 :    
240 :     1. Fixed up parsing colon suffixed keywords
241 :     2. Added the ability to shut the error messages up
242 :     3. Reimplemented the pretty printer and fixed up/improved
243 :     the pretty printing of handle and -> types.
244 :     4. Fixed up generation of literal symbols in the nowhere tool.
245 :     5. Added some SML keywords to to sml.sty
246 :    
247 :     ----------------------------------------------------------------------
248 :     Name: Matthias Blume
249 : blume 1078 Date: 2002/02/19 16:20:00 EST
250 :     Tag: blume-20020219-cmffi
251 :     Description:
252 :    
253 :     A wild mix of changes, some minor, some major:
254 :    
255 :     * All C FFI-related libraries are now anchored under $c:
256 :     $/c.cm --> $c/c.cm
257 :     $/c-int.cm --> $c/internals/c-int.cm
258 :     $/memory.cm --> $c/memory/memory.cm
259 :    
260 :     * "make" tool (in CM) now treats its argument pathname slightly
261 :     differently:
262 :     1. If the native expansion is an absolute name, then before invoking
263 :     the "make" command on it, CM will apply OS.Path.mkRelative
264 :     (with relativeTo = OS.FileSys.getDir()) to it.
265 :     2. The argument will be passed through to subsequent phases of CM
266 :     processing without "going native". In particular, if the argument
267 :     was an anchored path, then "make" will not lose track of that anchor.
268 :    
269 :     * Compiler backends now "know" their respective C calling conventions
270 :     instead of having to be told about it by ml-nlffigen. This relieves
271 :     ml-nlffigen from one of its burdens.
272 :    
273 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
274 :    
275 :     * Export C_DEBUG and C_Debug from $c/c.cm.
276 :    
277 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
278 :     subtyping relationship between incomplete pointers and their complete
279 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
280 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
281 :     the pointer target type is complete. In the incomplete case, 'o
282 :     will be instantiated with some "'c iobj" -- a type obtained by
283 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
284 :    
285 :     Operations that work on both incomplete and complete pointer types are
286 :     typed as taking an 'o ptr while operations that require the target to
287 :     be known are typed as taking some ('t, 'c) obj ptr.
288 :    
289 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
290 :     where void is an eqtype without any values. This makes it possible
291 :     to work on voidptr values using functions meant to operate on light
292 :     incomplete pointers.
293 :    
294 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
295 :     been vastly simplified.
296 :    
297 :     ----------------------------------------------------------------------
298 :     Name: Matthias Blume
299 : blume 1076 Date: 2002/02/19 10:48:00 EST
300 :     Tag: blume-20020219-pqfix
301 :     Description:
302 :    
303 :     Applied Chris Okasaki's bug fix for priority queues.
304 :    
305 :     ----------------------------------------------------------------------
306 :     Name: Matthias Blume
307 : blume 1073 Date: 2002/02/15 17:05:00
308 : blume 1069 Tag: Release_110_39
309 :     Description:
310 :    
311 : blume 1073 Last-minute retagging is becoming a tradition... :-(
312 :    
313 :     This is the working release 110.39.
314 :    
315 :     ----------------------------------------------------------------------
316 :     Name: Matthias Blume
317 :     Date: 2002/02/15 16:00:00 EST
318 :     Tag: Release_110_39-orig
319 :     Description:
320 :    
321 : blume 1069 Working release 110.39. New bootfiles.
322 :    
323 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
324 :     with all shells. So I retagged. -Matthias)
325 :    
326 : blume 1069 ----------------------------------------------------------------------
327 :     Name: Matthias Blume
328 : blume 1068 Date: 2002/02/15 14:17:00 EST
329 :     Tag: blume-20020215-showbindings
330 :     Description:
331 :    
332 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
333 :     the latter can be useful for exploring what bindings are available at
334 :     the interactive prompt. (The first function returns only the list
335 :     of symbols that are really bound, the second prints those but also the
336 :     ones that CM's autoloading mechanism knows about.)
337 :    
338 :     ----------------------------------------------------------------------
339 :     Name: Matthias Blume
340 : blume 1067 Date: 2002/02/15 12:08:00 EST
341 :     Tag: blume-20020215-iptrs
342 :     Description:
343 :    
344 :     Two improvements to ml-nlffigen:
345 :    
346 :     1. Write files only if they do not exist or if their current contents
347 :     do not coincide with what's being written. (That is, avoid messing
348 :     with the time stamps unless absolutely necessary.)
349 :    
350 :     2. Implement a "repository" mechanism for generated files related
351 :     to "incomplete pointer types". See the README file for details.
352 :    
353 :     ----------------------------------------------------------------------
354 :     Name: Matthias Blume
355 : blume 1066 Date: 2002/02/14 11:50:00 EST
356 :     Tag: blume-20020214-quote
357 :     Description:
358 :    
359 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
360 :     because of the new and improved tag generation scheme. (Thanks to Allen
361 :     Leung for pointing it out.)
362 :    
363 :     ----------------------------------------------------------------------
364 : george 1065 Name: Lal George
365 :     Date: 2002/02/14 09:55:27 EST 2002
366 :     Tag: george-20020214-isabelle-bug
367 :     Description:
368 :    
369 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
370 :     of Isabelle on the x86.
371 :    
372 :     From Allen:
373 :     -----------
374 :     I've found the problem:
375 :    
376 :     in ra-core.sml, I use the counter "blocked" to keep track of the
377 :     true number of elements in the freeze queue. When the counter goes
378 :     to zero, I skip examining the queue. But I've messed up the
379 :     bookkeeping in combine():
380 :    
381 :     else ();
382 :     case !ucol of
383 :     PSEUDO => (if !cntv > 0 then
384 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
385 :     ^^^^^^^^^^^^^^^^^^^^^^^
386 :     moveu := mergeMoveList(!movev, !moveu)
387 :     )
388 :     else ();
389 :    
390 :     combine() is called to coalesce two nodes u and v.
391 :     I think I was thinking that if the move counts of u and v are both
392 :     greater than zero then after they are coalesced then one node is
393 :     removed from the freeze queue. Apparently I was thinking that
394 :     both u and v are of low degree, but that's clearly not necessarily true.
395 :    
396 :    
397 :     02/12/2002:
398 :     Here's the patch. HOL now compiles.
399 :    
400 :     I don't know how this impact on performance (compile
401 :     time or runtime). This bug caused the RA (especially on the x86)
402 :     to go thru the potential spill phase when there are still nodes on the
403 :     freeze queue.
404 :    
405 :    
406 :    
407 :    
408 :     ----------------------------------------------------------------------
409 : blume 1058 Name: Matthias Blume
410 : blume 1064 Date: 2002/02/13 22:40:00 EST
411 :     Tag: blume-20020213-fptr-rtti
412 :     Description:
413 :    
414 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
415 :     updates.
416 :    
417 :     ----------------------------------------------------------------------
418 :     Name: Matthias Blume
419 : blume 1063 Date: 2002/02/13 16:41:00 EST
420 :     Tag: blume-20020213-cmlpq
421 :     Description:
422 :    
423 :     Added new priority queue export symbols (which have just been added to
424 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
425 :     not compile and the installer would choke.)
426 :    
427 :     ----------------------------------------------------------------------
428 :     Name: Matthias Blume
429 : blume 1062 Date: 2002/02/13 16:15:00 EST
430 :     Tag: blume-20020213-various
431 :     Description:
432 :    
433 :     1. More tweaks to ml-nlffigen:
434 :    
435 :     - better internal datastructures (resulting in slight speedup)
436 :     - "-match" option requires exact match
437 :     - "localized" gensym counters (untagged structs/unions nested within
438 :     other structs/unions or within typedefs get a fresh counter; their
439 :     tag will be prefixed by a concatenation of their parents' tags)
440 :     - bug fixes (related to calculation of transitive closure of types
441 :     to be included in the output)
442 :    
443 :     2. Minor Basis updates:
444 :    
445 :     - added implementations for List.collate and Option.app
446 :    
447 :     ----------------------------------------------------------------------
448 :     Name: Matthias Blume
449 : blume 1060 Date: 2002/02/11 15:55:00 EST
450 :     Tag: blume-20020211-gensym
451 :     Description:
452 :    
453 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
454 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
455 :     names (ML structure names that correspond to unnamed C structs, unions,
456 :     and enums), so that separate runs of ml-nlffigen do not clash.
457 :    
458 :     ----------------------------------------------------------------------
459 :     Name: Matthias Blume
460 : blume 1059 Date: 2002/02/11 12:05:00 EST
461 :     Tag: blume-20020211-gensml
462 :     Description:
463 :    
464 :     A quick fix for a problem with GenSML (in the pgraph-util library):
465 :     Make generation of toplevel "local" optional. (Strictly speaking,
466 :     signature definitions within "local" are not legal SML.)
467 :    
468 :     Other than that: updates to INSTALL and cm/TODO.
469 :    
470 :     ----------------------------------------------------------------------
471 :     Name: Matthias Blume
472 : blume 1058 Date: 2002/02/08 15:00:00 EST
473 :     Tag: blume-20020208-uniquepid
474 :     Description:
475 :    
476 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
477 :    
478 :     1. The installer (config/install.sh) has gotten smarter:
479 :    
480 :     - Configuration options are a bit easier to specify now
481 :     (in config/targets).
482 :     - Bug in recognizing .tar.bz2 files fixed.
483 :     - Installer automatically resolves dependencies between
484 :     configuration options (e.g., if you ask for eXene, you will
485 :     also get cml -- regardless whether you asked for it or not).
486 :     - Installer can run in "quieter mode" by setting the environment
487 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
488 :     "completely silent", though.
489 :     - Build HashCons library as part of smlnj-lib.
490 :    
491 :     2. A new scheme for assigning persistent identifiers to compilation
492 :     units (and, by extension, to types etc.) has been put into place.
493 :     This fixes a long-standing bug where types and even dynamic values
494 :     can get internally confused, thereby compromising type safety
495 :     (abstraction) and dynamic correctness. See
496 :    
497 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
498 :    
499 :     for an example of how things could go wrong until now.
500 :    
501 :     The downside of the new scheme is that pids are not quite as
502 :     persistent as they used to be: CM will generate a fresh pid
503 :     for every compilation unit that it thinks it sees for the first
504 :     time. That means that if you compile starting from a clean, fresh
505 :     source tree at two different times, you end up with different
506 :     binaries.
507 :    
508 :     Cutoff recompilation, however, has not been compromised because
509 :     CM keeps pid information in special caches between runs.
510 :    
511 :     ----------------------------------------------------------------------
512 : george 1053 Name: Lal George
513 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
514 :     Tag: <none>
515 :     Description:
516 :    
517 :     Compilers that generate assembly code may produce global labels
518 :     whose value is resolved at link time. The various peephole optimization
519 :     modules did not take this in account.
520 :    
521 :     TODO. The Labels.addrOf function should really return an option
522 :     type so that clients are forced to deal with this issue, rather
523 :     than an exception being raised.
524 :    
525 :     ----------------------------------------------------------------------
526 :     Name: Lal George
527 : george 1053 Date: 2002/02/06 13:55:02 EST
528 :     Tag: george-20020206-ra-breakup
529 :     Description:
530 :    
531 :     1. A bug fix from Allen.
532 :    
533 :     A typo causes extra fstp %st(0)'s to be generated at compensation
534 :     edges, which might cause stack underflow traps at runtime. This
535 :     occurs in fft where there are extraneous fstps right before the 'into'
536 :     trap instruction (in this case they are harmless since none of the
537 :     integers overflow.)
538 :    
539 : george 1054 2. Pulled out various utility modules that were embedded in the modules
540 : george 1053 of the register allocator. I need these modules for other purposes, but
541 : george 1054 they are not complete enough to put into a library (just yet).
542 : george 1053 ----------------------------------------------------------------------
543 : blume 1049 Name: Matthias Blume
544 :     Date: 2002/01/31 16:05:00 EST
545 :     Tag: blume-20020131-sparc-ccalls
546 :     Description:
547 :    
548 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
549 :     of extra stack space by mistake. Fixed.
550 :    
551 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
552 :    
553 :     ----------------------------------------------------------------------
554 : leunga 1048 Name: Allen Leung
555 :     Date: 2002/01/30
556 :     Tag: leunga-20020130-nowhere-bug-fix
557 :     Description:
558 :    
559 :     MLRISC bug fixes:
560 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
561 :     2. MachineInt.fromString was negating its value.
562 :    
563 :     ----------------------------------------------------------------------
564 : blume 1039 Name: Matthias Blume
565 : blume 1047 Date: 2002/01/29
566 :     Tag: blume-20020129-INSTALL
567 :     Description:
568 :    
569 :     - Added somewhat detailed installation instructions (file INSTALL).
570 :     - Fixed curl-detection bug in config/install.sh.
571 :     - It is now possible to select the URL getter using the URLGETTER
572 :     environment variable:
573 :    
574 :     not set / "unknown" --> automatic detection (script tries wget,
575 :     curl, and lynx)
576 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
577 :     how to properly invoke them)
578 :     other --> use $URLGETTER directly, it must take
579 :     precisely two command-line arguments
580 :     (source URL and destination file name)
581 :    
582 :     ----------------------------------------------------------------------
583 :     Name: Matthias Blume
584 : blume 1044 Date: 2002/01/28
585 : blume 1045 Tag: blume-20020128-sparc-ccalls
586 :     Description:
587 :    
588 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
589 :     - Make use of the allocParam argument in sparc-c-calls.
590 :    
591 :     ----------------------------------------------------------------------
592 :     Name: Matthias Blume
593 :     Date: 2002/01/28
594 : blume 1044 Tag: blume-20020128-allocParam
595 :     Description:
596 :    
597 :     John Reppy: Changes c-calls API to accept client-callback for
598 :     allocating extra stack space.
599 :     me: Corresponding changes to mlriscGen (using a dummy argument that
600 :     does not change the current behavior).
601 :    
602 :     ----------------------------------------------------------------------
603 :     Name: Matthias Blume
604 : blume 1040 Date: 2002/01/28 12:00:00
605 :     Tag: Release_110_38
606 :     Description:
607 :    
608 :     This time for real!!!
609 :    
610 :     ----------------------------------------------------------------------
611 :     Name: Matthias Blume
612 : blume 1039 Date: 2002/01/28 10:56:00 EST
613 :     Tag: blume-20020128-retraction
614 :     Description:
615 :    
616 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
617 :     with blume-Release_110_38-retracted.)
618 :    
619 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
620 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
621 :    
622 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
623 :     and enums fixed. The naming of corresponding ML identifiers should
624 :     now be consistent again.
625 :    
626 :     ----------------------------------------------------------------------
627 : leunga 1038 Name: Allen Leung
628 :     Date: 2002/01/27
629 :     Tag: leunga-20020127-nowhere
630 :     Description:
631 :    
632 :     Added a target called nowhere in the configuration scripts.
633 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
634 :     programs with where-clauses into legal SML code) during installation.
635 :    
636 :     ----------------------------------------------------------------------
637 : blume 1036 Name: Matthias Blume
638 : leunga 1038 Date: 2002/01/25 21:27:00 EST
639 : blume 1039 Tag: blume-Release_110_38-retracted
640 : blume 1037 Description:
641 :    
642 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
643 :    
644 :     README will be added later.
645 :    
646 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
647 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
648 :    
649 : blume 1037 ----------------------------------------------------------------------
650 :     Name: Matthias Blume
651 : blume 1036 Date: 2002/01/25
652 :     Tag: blume-20020125-ffi
653 :     Description:
654 :    
655 :     A large number of tweaks and improvements to ml-nlffi-lib and
656 :     ml-nlffigen:
657 :    
658 :     - ML represenation types have been streamlined
659 :     - getter and setter functions work with concrete values, not abstract
660 :     ones where possible
661 :     - ml-nlffigen command line more flexible (see README file there)
662 :     - some bugs have been fixed (hopefully)
663 :    
664 :     ----------------------------------------------------------------------
665 : george 1034 Name: Lal George
666 :     Date: 2002/01/24
667 :     Tag: george-20020124-risc-ra-interface
668 :     Description:
669 :    
670 :     There is a dramatic simplification in the interface to the
671 :     register allocator for RISC architectures as a result of making
672 :     parallel copy instructions explicit.
673 :    
674 :     ----------------------------------------------------------------------
675 : blume 1028 Name: Matthias Blume
676 : blume 1032 Date: 2002/01/22
677 :     Tag: blume-20020122-x86-ccalls
678 :     Description:
679 :    
680 :     Bug fix for c-calls on x86 (having to do with how char- and
681 :     short-arguments are being handled).
682 :    
683 :     ----------------------------------------------------------------------
684 :     Name: Matthias Blume
685 : blume 1031 Date: 2002/01/21
686 :     Tag: blume-20020121-ff
687 :     Description:
688 :    
689 :     Another day of fiddling with the FFI...
690 :    
691 :     1. Bug fix/workaround: CKIT does not complain about negative array
692 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
693 :     (Otherwise a negative dimension would send it into an infinite loop.)
694 :    
695 :     2. Some of the abstract types (light objects, light pointers, most "base"
696 :     types) in structure C are now eqtypes.
697 :    
698 :     3. Added constructors and test functions for NULL function pointers.
699 :    
700 :     ----------------------------------------------------------------------
701 :     Name: Matthias Blume
702 : blume 1028 Date: 2002/01/18
703 : blume 1030 Tag: blume-20020118-ready-for-new-release
704 :     Description:
705 :    
706 :     Made config/srcarchiveurl point to a new place. (Will provide boot
707 :     files shortly.)
708 :    
709 :     Maybe we christen this to be 110.38?
710 :    
711 :     ----------------------------------------------------------------------
712 :     Name: Matthias Blume
713 :     Date: 2002/01/18
714 : blume 1029 Tag: blume-20020118-more-ffifiddle
715 :     Description:
716 :    
717 :     Today's FFI fiddling:
718 :    
719 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
720 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
721 :     of abstract ones.
722 :    
723 :     - Provided word-style bit operations etc. for "int" representation
724 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
725 :     and Long.
726 :    
727 :     ----------------------------------------------------------------------
728 :     Name: Matthias Blume
729 :     Date: 2002/01/18
730 : blume 1028 Tag: blume-20020118-use-x86-fp
731 :     Description:
732 :    
733 :     Now that x86-fast-fp seems to be working, I turned it back on again
734 :     by default. (Seems to work fine now, even with the FFI.)
735 :    
736 :     Other than that, I added some documentation about the FFI to
737 :     src/ml-nlffigen/README and updated the FFI test examples in
738 :     src/ml-nlffi-lib/Tests/*.
739 :    
740 :     ----------------------------------------------------------------------
741 : leunga 1025 Name: Allen Leung
742 :     Date: 2002/01/17
743 :     Tag: leunga-20020117-x86-fast-fp-call
744 :     Description:
745 :    
746 :     1. Fixed a problem with handling return fp values when x86's fast fp
747 :     mode is turned on.
748 :    
749 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
750 :     of %f32.
751 :    
752 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
753 :    
754 :     ----------------------------------------------------------------------
755 : blume 1021 Name: Matthias Blume
756 :     Date: 2002/01/16
757 :     Tag: blume-20020116-ffifiddle
758 :     Description:
759 :    
760 :     More fiddling with the FFI interface:
761 :    
762 :     - Make constness 'c instead of rw wherever possible. This eliminates
763 :     the need for certain explicit coercions. (However, due to ML's
764 :     value polymorphism, there will still be many cases where explicit
765 :     coercions are necessary. Phantom types are not the whole answer
766 :     to modeling a subtyping relationship in ML.)
767 :    
768 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
769 :    
770 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
771 :     clumsy workaround for varargs problem.)
772 :    
773 :     ----------------------------------------------------------------------
774 : george 1016 Name: Lal George
775 :     Date: 2002/01/15
776 :     Tag: <none>
777 :     Description:
778 :    
779 :     1. Since COPY instructions are no longer native to the architecture,
780 :     a generic functor can be used to implement the expandCopies function.
781 :    
782 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
783 :     TEXT segment.
784 :    
785 :     ----------------------------------------------------------------------
786 : blume 1011 Name: Matthias Blume
787 : blume 1015 Date: 2002/01/15
788 :     Tag: blume-20020115-ffiupdates
789 :     Description:
790 :    
791 :     1. Fix for bug resulting in single-precision float values being returned
792 :     incorrectly from FFI calls.
793 :    
794 :     2. Small modifications to C FFI API:
795 :    
796 :     - memory-allocation routines return straight objects (no options)
797 :     and raise an exception in out-of-memory situations
798 :     - unsafe extensions to cast between function pointers and pointers
799 :     from/to ints
800 :     - added structure C_Debug as an alternative to structure C where
801 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
802 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
803 :     but also takes a (possibly empty) list of existing library handles
804 :     that the current library depends on
805 :    
806 :     ----------------------------------------------------------------------
807 :     Name: Matthias Blume
808 : blume 1011 Date: 2002/01/10
809 :     Tag: blume-20020110-newffigen
810 :     Description:
811 :    
812 :     1. Updates to portable graph code.
813 :    
814 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
815 :     more scalable now so that even huge interfaces such as the one
816 :     for GTK compile in finite time and space. :-)
817 :     See src/ml-nlffigen/README for details on what's new.
818 :    
819 :     ----------------------------------------------------------------------
820 : george 1003 Name: Lal George
821 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
822 :     Tag: george-20011206-rm-native-copy
823 :     Description:
824 :    
825 :     Removed the native COPY and FCOPY instructions
826 :     from all the architectures and replaced it with the
827 :     explicit COPY instruction from the previous commit.
828 :    
829 :     It is now possible to simplify many of the optimizations
830 :     modules that manipulate copies. This has not been
831 :     done in this change.
832 :    
833 :     ----------------------------------------------------------------------
834 :     Name: Lal George
835 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
836 :     Tag: george-20011206-mlrisc-instruction
837 :     Description:
838 : george 824
839 : george 1003 Changed the representation of instructions from being fully abstract
840 :     to being partially concrete. That is to say:
841 :    
842 :     from
843 :     type instruction
844 :    
845 :     to
846 :     type instr (* machine instruction *)
847 :    
848 :     datatype instruction =
849 :     LIVE of {regs: C.cellset, spilled: C.cellset}
850 :     | KILL of {regs: C.cellset, spilled: C.cellset}
851 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
852 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
853 :     | INSTR of instr
854 :    
855 :     This makes the handling of certain special instructions that appear on
856 :     all architectures easier and uniform.
857 :    
858 :     LIVE and KILL say that a list of registers are live or killed at the
859 :     program point where they appear. No spill code is generated when an
860 :     element of the 'regs' field is spilled, but the register is moved to
861 :     the 'spilled' (which is present, more for debugging than anything else).
862 :    
863 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
864 :     We used to generate:
865 :    
866 :     DEFFREG f1
867 :     f1 := f2 + f3
868 :     trapb
869 :    
870 :     but now generate:
871 :    
872 :     f1 := f2 + f3
873 :     trapb
874 :     LIVE {regs=[f1,f2,f3], spilled=[]}
875 :    
876 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
877 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
878 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
879 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
880 :     this architecture).
881 :    
882 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
883 :     all the architectures. This will result in further simplification of the
884 :     register allocator that must be aware of them for coalescing purposes, and
885 :     will also simplify certain aspects of the machine description that provides
886 :     callbacks related to parallel copies.
887 :    
888 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
889 :     machine instruction set!
890 :    
891 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
892 :     functions for making porting easier -- essentially converting upper case
893 :     to lower case. All machine instructions (of type instr) are in upper case,
894 :     and the lower case form generates an MLRISC instruction. For example on
895 :     the alpha we have:
896 :    
897 :     datatype instr =
898 :     LDA of {r:cell, b:cell, d:operand}
899 :     | ...
900 :    
901 :     val lda : {r:cell, b:cell, d:operand} -> instruction
902 :     ...
903 :    
904 :     where lda is just (INSTR o LDA), etc.
905 :    
906 : leunga 815 ----------------------------------------------------------------------
907 : blume 986 Name: Matthias Blume
908 : blume 993 Date: 2001/11/22 21:40:00 EST
909 :     Tag: Release_110_37
910 :     Description:
911 :    
912 :     Release 110.37. This time for real.
913 :    
914 :     ----------------------------------------------------------------------
915 :     Name: Matthias Blume
916 : blume 988 Date: 2001/11/21 16:35:00 EST
917 :     Tag: blume-20011121-foot-in-mouth
918 :     Description:
919 :    
920 :     Removed the "Release_110_37" tag because of a serious bug.
921 :     This will be re-tagged once the bug is fixed.
922 :    
923 :     ----------------------------------------------------------------------
924 :     Name: Matthias Blume
925 : blume 987 Date: 2001/11/21 16:14:00 EST
926 :     Tag: blume-20011121-forgottenfile
927 :     Description:
928 :    
929 :     Forgot to add a file. (Just a .tex-file -- part of
930 :     the CM manual source.)
931 :    
932 :     ----------------------------------------------------------------------
933 :     Name: Matthias Blume
934 : blume 986 Date: 2001/11/21 16:10:00 EST
935 : blume 988 Tag: blume-20011121-invalid_110_37
936 : blume 986 Description:
937 :    
938 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
939 :     because we found a serious bug in all non-x86 backends.
940 :     - Matthias
941 :    
942 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
943 :     so that code object name strings are directly inserted into code
944 :     objects at code generation time. The only business the runtime system
945 :     has with this is now to read the name strings on occasions.
946 :     (The encoding of the name string has also changed somewhat.)
947 :    
948 :     2. CM now implements a simple "set calculus" for specifying export lists.
949 :     In particular, it is now possible to refer to the export lists of
950 :     other libraries/groups/sources and form unions as well as differences.
951 :     See the latest CM manual for details.
952 :    
953 :     3. An separate notion of "proxy" libraries has again be eliminated from
954 :     CM's model. (Proxy libraries are now simply a special case of using
955 :     the export list calculus.)
956 :    
957 :     4. Some of the existing libraries now take advantage of the new set
958 :     calculus.
959 :     (Notice that not all libraries have been converted because some
960 :     of the existing .cm-files are supposed to be backward compatible
961 :     with 110.0.x.)
962 :    
963 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
964 :     "General.exnMessage"! The former relies on a certain hook to be
965 :     initialized, and that often does not happen in the stand-alone case.)
966 :    
967 :     ----------------------------------------------------------------------
968 : george 984 Name: Lal George
969 :     Date: 2001/11/21 13:56:18 EST
970 :     Tag: george-2001121-pseudo-ops
971 :     Description:
972 :    
973 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
974 :     ought to never be any question of incompatabilities with
975 :     pseudo-op syntax expected by host assemblers.
976 :    
977 :     For now, only modules supporting GAS syntax are implemented
978 :     but more should follow, such as MASM, and vendor assembler
979 :     syntax, e.g. IBM as, Sun as, etc.
980 :    
981 :     ----------------------------------------------------------------------
982 : blume 977 Name: Matthias Blume
983 :     Date: 2001/11/14 11:52:00 EST
984 :     Tag: blume-20011114-srcname
985 :     Description:
986 :    
987 :     1. Routed the name of the current source file to mlriscgen where it
988 :     should be directly emitted into the code object. (This last part
989 :     is yet to be done.)
990 :    
991 :     2. Some cleanup of the pgraph code to make it match the proposal that
992 :     I put out the other day. (The proposal notwithstanding, things are
993 :     still in flux here.)
994 :    
995 :     ----------------------------------------------------------------------
996 : george 976 Name: Lal George
997 :     Date: 2001/11/14 09:44:04 EST
998 :     Tag:
999 :     Description:
1000 :    
1001 :     Fix for a backpatching bug reported by Allen.
1002 :    
1003 :     Because the boundary between short and long span-dependent
1004 :     instructions is +/- 128, there are an astounding number of
1005 :     span-dependent instructions whose size is over estimated.
1006 :    
1007 :     Allen came up with the idea of letting the size of span
1008 :     dependent instructions be non-monotonic, for a maxIter
1009 :     number of times, after which the size must be monotonically
1010 :     increasing.
1011 :    
1012 :     This table shows the number of span-dependent instructions
1013 :     whose size was over-estimated as a function of maxIter, for the
1014 :     file Parse/parse/ml.grm.sml:
1015 :    
1016 :     maxIter # of instructions:
1017 :     10 687
1018 :     20 438
1019 :     30 198
1020 :     40 0
1021 :    
1022 :     In compiling the compiler, there is no significant difference in
1023 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1024 :     my measurements showed that maxIter=40 was a tad faster than
1025 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1026 :     point within 13 iterations, so fixing maxIter at 40, while high,
1027 :     is okay.
1028 :    
1029 :     ----------------------------------------------------------------------
1030 : blume 975 Name: Matthias Blume
1031 :     Date: 2001/10/31 15:25:00 EST
1032 :     Tag: blume-20011031-pgraph
1033 :     Description:
1034 :    
1035 :     CKIT:
1036 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1037 :     argument identifiers.
1038 :     * Changed the return type of TypeUtil.getFunction accordingly.
1039 :     * Type equality ignores the argument names.
1040 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1041 :     if there is a mismatch.
1042 :    
1043 :     installation script:
1044 :     * attempts to use "curl" if available (unless "wget" is available as well)
1045 :    
1046 :     CM:
1047 :     * has an experimental implementation of "portable graphs" which I will
1048 :     soon propose as an implementation-independent library format
1049 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1050 :    
1051 :     NLFFI-LIB:
1052 :     * some cleanup (all cosmetic)
1053 :    
1054 :     NLFFIGEN:
1055 :     * temporarily disabled the mechanism that suppresses ML output for
1056 :     C definitions whose identifiers start with an underscore character
1057 :     * generate val bindings for enum constants
1058 :     * user can request that only one style (light or heavy) is being used;
1059 :     default is to use both (command-line arguments: -heavy and -light)
1060 :     * fixed bug in handling of function types involving incomplete pointers
1061 :     * generate ML entry points that take record arguments (i.e., using
1062 :     named arguments) for C functions that have a prototype with named
1063 :     arguments
1064 :     (see changes to CKIT)
1065 :    
1066 :     ----------------------------------------------------------------------
1067 : leunga 972 Name: Allen Leung
1068 :     Date: 2001/10/27 20:34:00 EDT
1069 :     Tag: leunga-20011027-x86-fast-fp-call
1070 :     Description:
1071 :    
1072 :     Fixed the bug described in blume-20010920-slowfp.
1073 :    
1074 :     The fix involves
1075 :     1. generating FCOPYs in FSTP in ia32-svid
1076 :     2. marking a CALL with the appropriate annotation
1077 :    
1078 :     ----------------------------------------------------------------------
1079 : blume 961 Name: Matthias Blume
1080 :     Date: 2001/10/16 11:32:00 EDT
1081 :     Tag: blume-20011016-netbsd
1082 :     Description:
1083 :    
1084 :     Underscore patch from Chris Richards (fixing problem with compiling
1085 :     runtime system under recent NetBSD).
1086 :    
1087 :     ----------------------------------------------------------------------
1088 : george 959 Name: Allen Leung
1089 :     Date: 2001/10/12 17:18:32 EDT 2001
1090 :     Tag: leung-20011012-x86-printflowgraph
1091 :     Description:
1092 :    
1093 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1094 :    
1095 :     ----------------------------------------------------------------------
1096 : george 952 Name: Lal George
1097 : george 958 Date: 2001/10/11 23:51:34 EDT
1098 :     Tag: george-20011011-too-many-instrs
1099 :     Description:
1100 :    
1101 :     The representation of a program point never expected to see more
1102 :     than 65536 instructions in a basic block!
1103 :    
1104 :     ----------------------------------------------------------------------
1105 :     Name: Lal George
1106 : george 952 Date: 2001/10/09 09:41:37 EDT
1107 :     Tag: george-20011008-mlrisc-labels
1108 :     Description:
1109 :    
1110 :     Changed the machine description files to support printing of
1111 :     local and global labels in assembly code, based on host assembler
1112 :     conventions.
1113 :    
1114 :     ----------------------------------------------------------------------
1115 : blume 916 Name: Matthias Blume
1116 : blume 939 Date: 2001/09/25 15:25:00 EDT
1117 :     Tag: blume-20010925-exninfo
1118 :     Description:
1119 :    
1120 :     I provided a non-hook implementation of exnName (at the toplevel) and
1121 :     made the "dummy" implementation of exnMessage (at the toplevel) more
1122 :     useful: if nothing gets "hooked in", then at least you are going to
1123 :     see the exception name and a message indicating why you don't see more.
1124 :    
1125 :     [For the time being, programs that need exnMessage and want to use
1126 :     ml-build should either use General.exnMessage (strongly recommended) or
1127 :     refer to structure General at some other point so that CM sees a
1128 :     static dependency.]
1129 :    
1130 :     [Similar remarks go for "print" and "use": If you want to use their
1131 :     functionality in stand-alone programs generated by ml-build, then use
1132 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1133 :    
1134 :     ----------------------------------------------------------------------
1135 :     Name: Matthias Blume
1136 : blume 938 Date: 2001/09/20 17:28:00 EDT
1137 :     Tag: blume-20010920-slowfp
1138 :     Description:
1139 :    
1140 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1141 :    
1142 :     ----------------------------------------------------------------------
1143 :     Name: Matthias Blume
1144 : blume 937 Date: 2001/09/20 17:20:00 EDT
1145 :     Tag: blume-20010920-canonicalpaths
1146 :     Description:
1147 :    
1148 :     0. Updated the BOOT file (something that I forgot to do earlier).
1149 :    
1150 :     1. Small internal change to CM so that it avoids "/../" in filenames
1151 :     as much as possible (but only where it is safe).
1152 :    
1153 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1154 :     that arguments that contain delimiters are passed through correctly.
1155 :     This change also means that all "special" arguments of the form
1156 :     @SMLxxx... must come first.
1157 :    
1158 :     3. Changed install script to put relative anchor names for tool commands
1159 :     into pathconfig.
1160 :    
1161 :     ----------------------------------------------------------------------
1162 :     Name: Matthias Blume
1163 : blume 932 Date: 2001/09/18 15:35:00 EDT
1164 :     Tag: blume-20010918-readme11036
1165 :     Description:
1166 :    
1167 :     Added README files.
1168 :    
1169 :     ----------------------------------------------------------------------
1170 :     Name: Matthias Blume
1171 : blume 929 Date: 2001/09/18 11:45:00 EDT
1172 :     Tag: Release_110_36 (retag)
1173 :     Description:
1174 :    
1175 :     Fixed mistake in config/preloads. Retagged as 110.36.
1176 :    
1177 :     ----------------------------------------------------------------------
1178 :     Name: Matthias Blume
1179 : blume 928 Date: 2001/09/18 09:40:00 EDT
1180 : blume 929 Tag: Release_110_36_orig (tag changed)
1181 : blume 928 Description:
1182 :    
1183 :     New version (110.36). New bootfiles.
1184 :    
1185 :     ----------------------------------------------------------------------
1186 :     Name: Matthias Blume
1187 : blume 927 Date: 2001/09/14 16:15:00 EDT
1188 :     Tag: blume-20010914-x86fastfp
1189 :     Description:
1190 :    
1191 :     John committed some changes that Allen made, in particular a (hopefully)
1192 :     correctly working version of the x86-fp module.
1193 :    
1194 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1195 :     flag to "true". Everything seems to compile to a fixpoint ok, and
1196 :     "mandelbrot" speeds up by about 15%.
1197 :    
1198 :     ----------------------------------------------------------------------
1199 :     Name: Matthias Blume
1200 : blume 923 Date: 2001/09/13 11:20:00 EDT
1201 :     Tag: blume-20010913-minimal
1202 :     Description:
1203 :    
1204 :     1. Stefan Monnier's patch to fix a miscompilation problem that
1205 :     was brought to light by John Reppy's work on Moby.
1206 :    
1207 :     2. Implemented a minimal "structure Compiler" that contains just
1208 :     "version" and "architecture". The minimal version will be
1209 :     available when the full version is not. This is for backward-
1210 :     compatibility with code that wants to test Compiler.version.
1211 :    
1212 :     ----------------------------------------------------------------------
1213 :     Name: Matthias Blume
1214 : blume 920 Date: 2001/08/28 14:03:00 EDT
1215 :     Tag: blume-20010828-ml-lex
1216 :     Description:
1217 :    
1218 :     Fix for bug 1581, received from Neophytos Michael.
1219 :    
1220 :     ----------------------------------------------------------------------
1221 :     Name: Matthias Blume
1222 : blume 918 Date: 2001/08/27 11:20:00 EDT
1223 :     Tag: blume-20010827-readme11035
1224 :     Description:
1225 :    
1226 :     Fleshed out the README file for 110.35.
1227 :    
1228 :     ----------------------------------------------------------------------
1229 :     Name: Matthias Blume
1230 : blume 916 Date: 2001/08/24 17:10:00 EDT
1231 :     Tag: Release_110_35
1232 :     Description:
1233 :    
1234 :     New version number (110.35). New bootfiles.
1235 :    
1236 :     ----------------------------------------------------------------------
1237 : george 915 Name: Lal George
1238 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1239 :     Tag: george-20010824-MLRISC-graphs
1240 :     Description:
1241 :    
1242 :     removed clusters from MLRISC completely and replaced with graphs.
1243 :    
1244 :     ----------------------------------------------------------------------
1245 :     Name: Matthias Blume
1246 : blume 905 Date: 2001/08/23 17:50:00 EDT
1247 :     Tag: blume-20010823-toplevel
1248 :     Description:
1249 :    
1250 :     - some reorganization of the code that implements various kinds of
1251 :     environments in the compiler (static, dynamic, symbolic, combined)
1252 :     - re-implemented the EnvRef module so that evalStream works properly
1253 :     (if the stream contains references to "use", "CM.make", etc.)
1254 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1255 :    
1256 :     ----------------------------------------------------------------------
1257 :     Name: Matthias Blume
1258 : blume 904 Date: 2001/08/20 15:50 EDT
1259 :     Tag: blume20010820-slipup
1260 :     Description:
1261 :    
1262 :     I forgot to commit a few files. Here they are...
1263 :    
1264 :     ----------------------------------------------------------------------
1265 :     Name: Matthias Blume
1266 : blume 903 Date: 2001/08/20 15:35:00 EDT
1267 :     Tag: blume-20010820-debugprof
1268 :     Description:
1269 :    
1270 :     !!!! NEW BOOTFILES !!!!
1271 :    
1272 :     This is another round of reorganizing the compiler sources. This
1273 :     time the main goal was to factor out all the "instrumentation"
1274 :     passes (for profiling and backtracing) into their own library.
1275 :     The difficulty was to do it in such a way that it does not depend
1276 :     on elaborate.cm but only on elabdata.cm.
1277 :    
1278 :     Therefore there have been further changes to both elaborate.cm and
1279 :     elabdata.cm -- more "generic" things have been moved from the former
1280 :     to the latter. As a result, I was forced to split the assignment
1281 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1282 :     SML/NJ-specific. Since it would have been awkward to maintain,
1283 :     I bit the bullet and actually _changed_ the mapping between these
1284 :     numbers and primtycs. The bottom line of this is that you need
1285 :     a new set of bin- and bootfiles.
1286 :    
1287 :     I have built new bootfiles for all architectures, so doing a fresh
1288 :     checkout and config/install.sh should be all you need.
1289 :    
1290 :     The newly created library's name is
1291 :    
1292 :     $smlnj/viscomp/debugprof.cm
1293 :    
1294 :     and its sources live under
1295 :    
1296 :     src/compiler/DebugProf
1297 :    
1298 :     ----------------------------------------------------------------------
1299 :     Name: Matthias Blume
1300 : blume 902 Date: 2001/08/15 17:15:00 EDT
1301 :     Tag: blume-20010815-compreorg
1302 :     Description:
1303 :    
1304 :     This is a first cut at reorganizing the CM libraries that make up the
1305 :     core of the compiler. The idea is to separate out pieces that could
1306 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1307 :    
1308 :     The current status is a step in this direction, but it is not quite
1309 :     satisfactory yet. Expect more changes in the future.
1310 :    
1311 :     Here is the current (new) organization...
1312 :    
1313 :     What used to be $smlnj/viscomp/core.cm is now divided into
1314 :     six CM libraries:
1315 :    
1316 :     $smlnj/viscomp/basics.cm
1317 :     /parser.cm
1318 :     /elabdata.cm
1319 :     /elaborate.cm
1320 :     /execute.cm
1321 :     /core.cm
1322 :    
1323 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1324 :     All these libraries are proxy libraries that contain precisely
1325 :     one CM library component. Here are the locations of the components
1326 :     (all within the src/compiler tree):
1327 :    
1328 :     Basics/basics.cm
1329 :     Parse/parser.cm
1330 :     ElabData/elabdata.cm
1331 :     Elaborator/elaborate.cm
1332 :     Execution/execute.cm
1333 :     core.cm
1334 :    
1335 :     [This organization is the same that has been used already
1336 :     for a while for the architecture-specific parts of the visible
1337 :     compiler and for the old version of core.cm.]
1338 :    
1339 :     As you will notice, many source files have been moved from their
1340 :     respective original locations to a new home in one of the above
1341 :     subtrees.
1342 :    
1343 :     The division of labor between the new libraries is the following:
1344 :    
1345 :     basics.cm:
1346 :     - Simple, basic definitions that pertain to many (or all) of
1347 :     the other libraries.
1348 :     parser.cm:
1349 :     - The SML parser, producing output of type Ast.dec.
1350 :     - The type family for Ast is also defined and exported here.
1351 :     elabdata.cm:
1352 :     - The datatypes that describe input and output of the elaborator.
1353 :     This includes types, absyn, and static environments.
1354 :     elaborator.cm:
1355 :     - The SML/NJ type checker and elaborator.
1356 :     This maps an Ast.dec (with a given static environment) to
1357 :     an Absyn.dec (with a new static environment).
1358 :     - This libraries implements certain modules that used to be
1359 :     structures as functors (to remove dependencies on FLINT).
1360 :     execute.cm:
1361 :     - Everything having to do with executing binary code objects.
1362 :     - Dynamic environments.
1363 :     core.cm:
1364 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1365 :     - Top-level modules.
1366 :     - FLINT (this should eventually become its own library)
1367 :    
1368 :     Notes:
1369 :    
1370 :     I am not 100% happy with the way I separated the elaborator (and its
1371 :     data structures) from FLINT. Two instances of the same problem:
1372 :    
1373 :     1. Data structures contain certain fields that carry FLINT-specific
1374 :     information. I hacked around this using exn and the property list
1375 :     module from smlnj-lib. But the fact that there are middle-end
1376 :     specific fields around at all is a bit annoying.
1377 :    
1378 :     2. The elaborator calculates certain FLINT-related information. I tried
1379 :     to make this as abstract as I could using functorization, but, again,
1380 :     the fact that the elaborator has to perform calculations on behalf
1381 :     of the middle-end at all is not nice.
1382 :    
1383 :     3. Having to used exn and property lists is unfortunate because it
1384 :     weakens type checking. The other alternative (parameterizing
1385 :     nearly *everything*) is not appealing, though.
1386 :    
1387 :     I removed the "rebinding =" warning hack because due to the new organization
1388 :     it was awkward to maintain it. As a result, the compiler now issues some of
1389 :     these warnings when compiling init.cmi during bootstrap compilation. On
1390 :     the plus side, you also get a warning when you do, for example:
1391 :     val op = = Int32.+
1392 :     which was not the case up to now.
1393 :    
1394 :     I placed "assign" and "deref" into the _Core structure so that the
1395 :     code that deals with the "lazy" keyword can find them there. This
1396 :     removes the need for having access to the primitive environment
1397 :     during elaboration.
1398 :    
1399 :     ----------------------------------------------------------------------
1400 :     Name: Matthias Blume
1401 : blume 898 Date: 2001/08/13
1402 :     Tag: blume-20010813-closures
1403 :     Description:
1404 :    
1405 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1406 :     performance of certain loops by avoiding needless closure allocation.
1407 :    
1408 :     ----------------------------------------------------------------------
1409 : george 890 Name: Lal George
1410 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1411 :     Tag: george-20010731-x86-fmalloc
1412 :     Description: Fixed bug in x86 calls
1413 :    
1414 :     There was a bug where call instructions would mysteriously
1415 :     vanish. The call instruction had to be one that returned
1416 :     a floating point value.
1417 :    
1418 :     ----------------------------------------------------------------------
1419 :     Name: Lal George
1420 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1421 :     Tag: george-20010719-simple-cells
1422 :     Description:
1423 :    
1424 :     I have dramatically simplified the interface for CELLS in MLRISC.
1425 :    
1426 :     In summary, the cells interface is broken up into three parts:
1427 :    
1428 :     1. CellsBasis : CELLS_BASIS
1429 :    
1430 :     CellsBasis is a top level structure and common for all
1431 :     architectures. it contains the definitions of basic datatypes
1432 :     and utility functions over these types.
1433 :    
1434 :     2. functor Cells() : CELLS
1435 :    
1436 :     Cells generates an interface for CELLS that incorporates the
1437 :     specific resources on the target architecture, such as the
1438 :     presence of special register classes, their number and size,
1439 :     and various useful substructures.
1440 :    
1441 :     3. <ARCH>CELLS
1442 :    
1443 :     e.g. SparcCells: SPARCCELLS
1444 :    
1445 :     <ARCH>CELLS usually contains additional bindings for special
1446 :     registers on the architecture, such as:
1447 :    
1448 :     val r0 : cell (* register zero *)
1449 :     val y : cell (* Y register *)
1450 :     val psr : cell (* processor status register *)
1451 :     ...
1452 :    
1453 :     The structure returned by applying the Cells functor is opened
1454 :     in this interface.
1455 :    
1456 :     The main implication of all this is that the datatypes for cells is
1457 :     split between CellsBasis and CELLS -- a fairly simple change for user
1458 :     code.
1459 :    
1460 :     In the old scheme the CELLS interface had a definitional binding of
1461 :     the form:
1462 :    
1463 :     signature CELLS = sig
1464 :    
1465 :     structure CellsBasis = CellsBasis
1466 :    
1467 :     ...
1468 :    
1469 :     end
1470 :    
1471 :     With all the sharing constraints that goes on in MLRISC, this old
1472 :     design quickly leads to errors such as:
1473 :    
1474 :     "structure definition spec inside of sharing ... "
1475 :    
1476 :    
1477 :     and appears to require an unacceptable amount of sharing and where
1478 :     constraint hackery.
1479 :    
1480 :     I think this error message (the interaction of definitional specs and
1481 :     sharing) requires more explanation on our web page.
1482 :    
1483 :     ----------------------------------------------------------------------
1484 : blume 826 Name: Matthias Blume
1485 : blume 879 Date: 2001/07/19 15:00:00 EDT
1486 :     Tag: blume-20010719-libreorg
1487 :     Description:
1488 :    
1489 :     This update puts together a fairly extensive but straightforward change
1490 :     to the way the libraries that implement the interactive system are
1491 :     organized:
1492 :    
1493 :     The biggest change is the elimination of structure Compiler. As a
1494 :     replacement for this structure, there is now a CM library
1495 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1496 :     that exports all the substructures of the original structure Compiler
1497 :     directly. So instead of saying Compiler.Foo.bar one now simply
1498 :     says Foo.bar. (The CM libraries actually export a collection of
1499 :     structures that is richer than the collection of substructures of
1500 :     structure Compiler.)
1501 :    
1502 :     To make the transition smooth, there is a separate library called
1503 :     $smlnj/compiler/compiler.cm which puts together and exports the
1504 :     original structure Compiler (or at least something very close to it).
1505 :    
1506 :     There are five members of the original structure Compiler
1507 :     that are not exported directly but which instead became members
1508 :     of a new structure Backend (described by signature BACKEND). These are:
1509 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1510 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
1511 :     architecture (: string).
1512 :    
1513 :     Structure Compiler.Version has become structure CompilerVersion.
1514 :    
1515 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1516 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1517 :     or x86, respectively.
1518 :     Each of these exports the same frontend structures that
1519 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
1520 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1521 :     Hppa, PPC, Sparc, or X86, respectively.
1522 :    
1523 :     Library $smlnj/compiler/all.cm exports the union of the exports of
1524 :     $smlnj/compiler/<arch>.cm
1525 :    
1526 :     There are no structures <Arch>Compiler anymore, use
1527 :     $smlnj/compiler/<arch>.cm instead.
1528 :    
1529 :     Library host-compiler-0.cm is gone. Instead, the internal library
1530 :     that instantiates CM is now called cm0.cm. Selection of the host
1531 :     compiler (backend) is no longer done here but. (Responsibility for it
1532 :     now lies with $smlnj/compiler/current.cm. This seems to be more
1533 :     logical.)
1534 :    
1535 :     Many individual files have been moved or renamed. Some files have
1536 :     been split into multiple files, and some "dead" files have been deleted.
1537 :    
1538 :     Aside from these changes to library organization, there are also changes
1539 :     to the way the code itself is organized:
1540 :    
1541 :     Structure Binfile has been re-implemented in such a way that it no
1542 :     longer needs any knowledge of the compiler. It exclusively deals
1543 :     with the details of binfile layout. It no longer invokes the
1544 :     compiler (for the purpose of creating new prospective binfile
1545 :     content), and it no longer has any knowledge of how to interpret
1546 :     pickles.
1547 :    
1548 :     Structure Compile (: COMPILE) has been stripped down to the bare
1549 :     essentials of compilation. It no longer deals with linking/execution.
1550 :     The interface has been cleaned up considerably.
1551 :    
1552 :     Utility routines for dealing with linking and execution have been
1553 :     moved into their own substructures.
1554 :    
1555 :     (The ultimate goal of these changes is to provide a light-weight
1556 :     binfile loader/linker (at least for, e.g., stable libraries) that
1557 :     does not require CM or the compiler to be present.)
1558 :    
1559 :     CM documentation has been updated to reflect the changes to library
1560 :     organization.
1561 :    
1562 :     ----------------------------------------------------------------------
1563 :     Name: Matthias Blume
1564 : blume 875 Date: 2001/07/10 17:30:00 EDT
1565 :     Tag: Release_110_34
1566 :     Description:
1567 :    
1568 : blume 876 Minor tweak to 110.34 (re-tagged):
1569 :    
1570 :     - README.html file added to CVS repository
1571 :     - runtime compiles properly under FreeBSD 3.X and 4.X
1572 :    
1573 :     ----------------------------------------------------------------------
1574 :     Name: Matthias Blume
1575 :     Date: 2001/07/10 17:30:00 EDT
1576 :     Tag: Release_110_34
1577 :     Description:
1578 :    
1579 : blume 875 New version number (110.34). New bootfiles.
1580 :    
1581 :     ----------------------------------------------------------------------
1582 :     Name: Matthias Blume
1583 : blume 874 Date: 2001/07/09 16:00:00 EDT
1584 :     Tag: blume-20010709-more-varargs
1585 :     Description:
1586 :    
1587 :     I changed the handling of varargs in ml-nlffigen again:
1588 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
1589 :    
1590 :     The immediate effect is that you can actually call a varargs function
1591 :     from ML -- but you can't actually supply any arguments beyond the ones
1592 :     specified explicitly. (For example, you can call printf with its format
1593 :     string, but you cannot pass additional arguments.)
1594 :    
1595 :     This behavior is only marginally more useful than the one before, but
1596 :     it has the advantage that a function or, more importantly, a function
1597 :     type never gets dropped on the floor, thus avoiding follow-up problems with
1598 :     other types that refer to the offending one.
1599 :    
1600 :     ----------------------------------------------------------------------
1601 :     Name: Matthias Blume
1602 : blume 873 Date: 2001/07/09 11:25:00 EDT
1603 :     Tag: blume-20010709-varargs
1604 :     Description:
1605 :    
1606 :     1. ckit-lib.cm now exports structure Error
1607 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1608 :     with a warning accompanied by a source location. Moreover, it
1609 :     merely skips the offending function or type and proceeds with the
1610 :     rest of its work.u As a result, one can safely feed C code containing
1611 :     "..." to ml-nlffigen.
1612 :     3. There are some internal improvements to CM, providing slightly
1613 :     more general string substitutions in the tools subsystem.
1614 :    
1615 :     ----------------------------------------------------------------------
1616 :     Name: Matthias Blume
1617 : blume 854 Date: 2001/06/27 15:10:00 EDT
1618 :     Tag: blume-20010627-concur
1619 :     Description:
1620 :    
1621 :     Fixed a small bug in CM's handling of parallel compilation.
1622 :     (You could observe the bug by Control-C-interrupting an ordinary
1623 :     CMB.make or CM.stabilize and then attaching some compile servers.
1624 :     The result was that all of a sudden the previously interrupted
1625 :     compilation would continue on its own. This was because of
1626 :     an over-optimization: CM did not bother to clean out certain queues
1627 :     when no servers were attached "anyway", resulting in the contents
1628 :     of these queues to grab control when new servers did get attached.)
1629 :    
1630 :     There is also another minor update to the CM manual.
1631 :    
1632 :     ----------------------------------------------------------------------
1633 :     Name: Matthias Blume
1634 : blume 853 Date: 2001/06/26 16:15:00 EDT
1635 :     Tag: blume-20010626-cmdoc
1636 :     Description:
1637 :    
1638 :     Minor typo fixed in CM manual (syntax diagram for libraries).
1639 :    
1640 :     ----------------------------------------------------------------------
1641 :     Name: Matthias Blume
1642 : blume 852 Date: 2001/06/25 22:55:00 EDT
1643 :     Tag: blume-20010625-x86pc
1644 :     Description:
1645 :    
1646 :     Fixed a nasty bug in the X86 assembly code that caused signal
1647 :     handlers to fail (crash) randomly.
1648 :    
1649 :     ----------------------------------------------------------------------
1650 :     Name: Matthias Blume
1651 : blume 846 Date: 2001/06/25 12:05:00 EDT
1652 :     Tag: blume-20010625-nlffigen
1653 :     Description:
1654 :    
1655 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
1656 :     Nick Carter <nbc@andrew.cmu.edu>.
1657 :    
1658 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1659 :     2. Default names for generated files are now derived from the name of
1660 :     the C file *without its directory*. In particular, this causes generated
1661 :     files to be placed locally even if the C file is in some system directory.
1662 :     3. Default names for generated signatures and structures are also derived
1663 :     from the C file name without its directory. This avoids silly things
1664 :     like "structure GL/GL".
1665 :     (Other silly names are still possible because ml-nlffigen does not do
1666 : blume 847 a thorough check of whether generated names are legal ML identifiers.
1667 :     When in doubt, use command line arguments to force particular names.)
1668 : blume 846
1669 :     ----------------------------------------------------------------------
1670 :     Name: Matthias Blume
1671 : blume 845 Date: 2001/06/21 12:25:00 EDT
1672 :     Tag: blume-20010621-eXene
1673 :     Description:
1674 :    
1675 :     eXene now compiles and (sort of) works again.
1676 :    
1677 :     The library name (for version > 110.33) is $/eXene.cm.
1678 :    
1679 :     I also added an new example in src/eXene/examples/nbody. See the
1680 :     README file there for details.
1681 :    
1682 :     ----------------------------------------------------------------------
1683 :     Name: Matthias Blume
1684 : blume 844 Date: 2001/06/20 16:40:00 EDT
1685 :     Tag: blume-20010620-cml
1686 :     Description:
1687 :    
1688 :     CML now compiles and works again.
1689 :    
1690 :     Libraries (for version > 110.33):
1691 :    
1692 :     $cml/cml.cm Main CML library.
1693 :     $cml/basis.cm CML's version of $/basis.cm.
1694 :     $cml/cml-internal.cm Internal helper library.
1695 :     $cml/core-cml.cm Internal helper library.
1696 :     $cml-lib/trace-cml.cm Tracing facility.
1697 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
1698 :    
1699 :     The installer (config/install.sh) has been taught how to properly
1700 :     install this stuff.
1701 :    
1702 :     ----------------------------------------------------------------------
1703 :     Name: Matthias Blume
1704 : blume 843 Date: 2001/06/19 17:55:00 EDT
1705 :     Tag: blume-20010619-instantiate
1706 :     Description:
1707 :    
1708 :     This un-breaks the fix for bug 1432.
1709 :     (The bug was originally fixed in 110.9 but I broke it again some
1710 :     time after that.)
1711 :    
1712 :     ----------------------------------------------------------------------
1713 :     Name: Matthias Blume
1714 : blume 842 Date: 2001/06/19 17:25:00 EDT
1715 :     Tag: blume-20010619-signals
1716 :     Description:
1717 :    
1718 :     This should (hopefully) fix the long-standing signal handling bug.
1719 :     (The runtime system was constructing a continuation record with an
1720 :     incorrect descriptor which would cause the GC to drop data on the floor...)
1721 :    
1722 :     ----------------------------------------------------------------------
1723 :     Name: Matthias Blume
1724 : blume 840 Date: 2001/06/15 15:05:00 EDT
1725 : blume 841 Tag: blume-20010615-moresparc
1726 :     Description:
1727 :    
1728 :     Here is a short late-hour update related to Sparc c-calls:
1729 :    
1730 :     -- made handling of double-word arguments a bit smarter
1731 :    
1732 :     -- instruction selection phase tries to collapse certain clumsily
1733 :     constructed ML-Trees; typical example:
1734 :    
1735 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
1736 :    
1737 :     This currently has no further impact on SML/NJ since mlriscGen does
1738 :     not seem to generate such patterns in the first place, and c-calls
1739 :     (which did generate them in the beginning) has meanwhile been fixed
1740 :     so as to avoid them as well.
1741 :    
1742 :     ----------------------------------------------------------------------
1743 :     Name: Matthias Blume
1744 :     Date: 2001/06/15 15:05:00 EDT
1745 : blume 840 Tag: blume-20010615-sparc
1746 :     Description:
1747 :    
1748 :     The purpose of this update is to provide an implementation of NLFFI
1749 :     on Sparc machines.
1750 :    
1751 :     Here are the changes in detail:
1752 :    
1753 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
1754 :     the Sparc implementation of the c-calls API.
1755 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
1756 :     for accessing the ML frame. Thus, we have a real frame pointer and
1757 :     can freely modify %sp without need for an omit-frame-ptr phase.
1758 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
1759 :     this case.
1760 :     * ml-nlffigen has been taught to produce code for different architectures
1761 :     and calling conventions.
1762 :     * In a way similar to what was done in the x86 case, the Sparc
1763 :     backend uses its own specific extension to mltree. (For example,
1764 :     it needs to be able to generate UNIMP instructions which are part
1765 :     of the calling convention.)
1766 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
1767 :     to make it easier to plug in new machine- and os-dependent parts).
1768 :    
1769 :     There are some other fairly unrelated bug fixes and cleanups as well:
1770 :    
1771 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
1772 :     that they properly share their libraries with existing SML/NJ libraries.
1773 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
1774 :     follow-up error messages.
1775 :     * Updates to CM/CMB documentation.
1776 :    
1777 :     TODO items:
1778 :    
1779 :     * MLRISC should use a different register as its asmTemp on the Sparc.
1780 :     (The current %o2 is a really bad choice because it is part of the
1781 :     calling conventions, so things might interfere in unexpected ways.)
1782 :    
1783 :     ----------------------------------------------------------------------
1784 :     Name: Matthias Blume
1785 : blume 839 Date: 2001/06/07
1786 :     Tag: blume-20010607-calls
1787 :     Description:
1788 :    
1789 :     A number of internal changes related to C calls and calling conventions:
1790 :    
1791 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
1792 :     number of bytes popped implicitly (by the callee). In most cases
1793 :     this field is 0 but on x86/win32 it is some non-zero value. This
1794 :     is information provided for the benefit of the "omit-frameptr" pass.
1795 :     2. The CALL instruction on the x86 carries a similar "pops" field.
1796 :     The instruction selection phase copies its value from the ML-Tree
1797 :     CALL statement.
1798 :     3. On all other architectures, the instruction selection phase checks
1799 :     whether "pops=0" and complains if not.
1800 :     4. The c-calls implementation for x86 now accepts two calling conventions:
1801 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
1802 :     up after the call and pops is set to 0. For "stdcall", the caller
1803 :     does nothing, leaving the cleanup to the callee; pops is set to
1804 :     the number of bytes that were pushed onto the stack.
1805 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1806 :     distinguish between "ccall" and "stdcall".
1807 :     6. The UNIMP instruction has been added to the supported Sparc instruction
1808 :     set. (This is needed for implementing the official C calling convention
1809 :     on this architecture.)
1810 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1811 :     work with the latest CM.
1812 :    
1813 :     ----------------------------------------------------------------------
1814 :     Name: Matthias Blume
1815 : blume 838 Date: 2001/06/05 15:10:00 EDT
1816 :     Tag: blume-20010605-cm-index
1817 :     Description:
1818 :    
1819 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1820 :    
1821 :     1. CM can now generate "index files". These are human-readable files
1822 :     that list on a per-.cm-file basis each toplevel symbol defined or
1823 :     imported. The location of the index file for
1824 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1825 :     To enable index-file generation, set CM.Control.generate_index to true
1826 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
1827 :    
1828 :     The CM manual has been updated accordingly.
1829 :    
1830 :     2. I made some slight modifications to the c-calls API in MLRISC.
1831 :    
1832 :     a) There is now a callback to support saving/restoring of
1833 :     dedicated but caller-save registers around the actual call
1834 :     instruction.
1835 :     b) One can optionally specify a comment-annotation for the
1836 :     call instruction.
1837 :    
1838 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1839 :     (For example, the comment annotation shows the C prototype of
1840 :     the function being called.)
1841 :    
1842 :     ----------------------------------------------------------------------
1843 :     Name: Matthias Blume
1844 : blume 837 Date: 2001/06/01 13:30:00 EDT
1845 :     Tag: blume-20010601-nlffi-cleanup
1846 :     Description:
1847 :    
1848 :     This is mostly a cleanup of MLFFI stuff:
1849 :    
1850 :     - some signature files have been put into a more exposed place
1851 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
1852 :     - ml-nlffigen changed accordingly
1853 :     - tutorial updated
1854 :    
1855 :     Other changes:
1856 :    
1857 :     - author's affiliation in CM manual(s) updated
1858 :     - some more recognized keywords added to Allen's sml.sty
1859 :    
1860 :     ----------------------------------------------------------------------
1861 :     Name: Matthias Blume
1862 : blume 836 Date: 2001/05/25 15:30:00 EDT
1863 :     Tag: blume-20010525-iptr
1864 :     Description:
1865 :    
1866 :     - put the official 110.33-README (as it appears on the ftp server) under
1867 :     CVS
1868 :     - fixed a small bug related to incomplete pointer types in
1869 :     ml-nlffigen
1870 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1871 :     (it does not need the 'f type parameter)
1872 :    
1873 :     ----------------------------------------------------------------------
1874 :     Name: Matthias Blume
1875 : blume 834 Date: 2001/05/23 14:30:00 EDT
1876 :     Tag: Release_110_33
1877 :     Description:
1878 :    
1879 :     New version number (110.33). New bootfiles.
1880 :    
1881 :     ----------------------------------------------------------------------
1882 :     Name: Matthias Blume
1883 : blume 833 Date: 2001/05/22 18:06:00 EDT
1884 :     Tag: blume-20010522-targets
1885 :     Description:
1886 :    
1887 :     Made install.sh use file config/targets.customized if it exists, falling
1888 :     back to config/targets if it doesn't. This way one can have a customized
1889 :     version of the targets file without touching the "real thing", thus
1890 :     eliminating the constant fear of accidentally checking something bogus
1891 :     back into the CVS repository... (File config/targets.customized must
1892 :     not be added to the repository!)
1893 :    
1894 :     ----------------------------------------------------------------------
1895 :     Name: Matthias Blume
1896 : blume 832 Date: 2001/05/22 16:30:00 EDT
1897 :     Tag: blume-20010522-minitut
1898 :     Description:
1899 :    
1900 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1901 :     struct returns.
1902 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
1903 :     incomplete, preliminary documentation for NLFFI.
1904 :    
1905 :     ----------------------------------------------------------------------
1906 :     Name: Matthias Blume
1907 : blume 829 Date: 2001/05/14 11:30:00 EDT
1908 :     Tag: blume-20010514-script
1909 :     Description:
1910 :    
1911 :     Some bugs in install script fixed.
1912 :    
1913 :     In addition to that I also made a slight change to the NLFFI API:
1914 :     Functors generated by ml-nlffigen now take the dynamic library as a
1915 :     straight functor argument, not as a suspended one. (The original
1916 :     functor code used to force the suspension right away anyway, so there
1917 :     was nothing gained by this complication of the interface.)
1918 :    
1919 :     ----------------------------------------------------------------------
1920 :     Name: Matthias Blume
1921 : blume 828 Date: 2001/05/11 14:35:00 EDT
1922 :     Tag: blume-20010511-ml-nlffi
1923 :     Description:
1924 :    
1925 :     I finally took the plunge and added my new FFI code to the main
1926 :     repository. For x86-linux it is now ready for prime-time.
1927 :    
1928 :     There are two new subdirectories of "src":
1929 :    
1930 :     - ml-nlffi-lib:
1931 :     The utility library for programs using the FFI interface.
1932 :     Here is the implementation of $/c.cm and its associated low-level
1933 :     partners $/c-int.cm and $/memory.cm.
1934 :     - ml-nlffigen:
1935 :     A stand-alone program for generating ML glue code from C source
1936 :     code.
1937 :    
1938 :     Building ml-nlffigen requires $/ckit-lib.cm.
1939 :    
1940 :     The config/install.sh script has been updates to do the Right Thing
1941 :     (hopefully).
1942 :    
1943 :     Notice that the source tree for the C-Kit will not be put under "src"
1944 :     but directly under the installation root directory. (This is the
1945 :     structure that currently exists on the CVS server when you check out
1946 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
1947 :    
1948 :     Bugs: No documentation yet.
1949 :    
1950 :     ----------------------------------------------------------------------
1951 :     Name: Matthias Blume
1952 : blume 826 Date: 2001/05/09 16:35:00 EDT
1953 :     Tag: blume-20010509-cpscontract
1954 :     Description:
1955 :    
1956 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
1957 :     wrapper/unwrapper elimination did not decrement usage counts and some
1958 :     dead variables got overlooked by the dead-up logic.)
1959 :    
1960 :     ----------------------------------------------------------------------
1961 : george 824 Name: Lal George
1962 :     Date: 2001/05/08 17:26:09 EDT
1963 :     Tag: george-20010508-omit-frameptr
1964 :     Description:
1965 :    
1966 :     Changes to implement the omit-frame-pointer optimization to support
1967 :     raw C calls. For now, there is only support on the Intel x86, but
1968 :     other architectures will follow as more experience is gained with this.
1969 :    
1970 :    
1971 :     ----------------------------------------------------------------------
1972 : blume 818 Name: Matthias Blume
1973 : blume 819 Date: 2001/05/07 14:40:00 EDT
1974 :     Tag: blume-20010507-proxies
1975 :     Description:
1976 :    
1977 :     I made into "proxy libraries" all libraries that qualify for such a
1978 :     change. (A qualifying library is a library that has another library or
1979 :     groups as its sole member and repeats that member's export list
1980 :     verbatim. A proxy library avoids this repetition by omitting its export
1981 :     list, effectively inheriting the list that its (only) member exports.
1982 :     See the CM manual for more explanation.)
1983 :     The main effect is that explicit export lists for these libraries
1984 :     do not have to be kepts in sync, making maintenance a bit easier.
1985 :    
1986 :     I also added copyright notices to many .cm-files.
1987 :    
1988 :     Last but not least, I made a new set of bootfiles.
1989 :    
1990 :     ----------------------------------------------------------------------
1991 :     Name: Matthias Blume
1992 : blume 818 Date: 2001/05/04 17:00:00 EDT
1993 :     Tag: blume-20010504-cm-lsplit
1994 :     Description:
1995 :    
1996 :     0. John merged pending changes to $/smlnj-lib.cm
1997 :    
1998 :     1. Allen's previous change accidentally backed out of one of Lal's
1999 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2000 :    
2001 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2002 :     within the compiler where applicable. There is some code simplification
2003 :     because of that.
2004 :    
2005 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2006 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2007 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2008 :     it can now also be controlled on a per-source basis: CM has been taught
2009 :     a new tool parameter applicable to ML source files.
2010 :    
2011 :     - To turn lambda-splitting off completely:
2012 :     local open Compiler.Control.LambdaSplitting in
2013 :     val _ = set Off
2014 :     end
2015 :     - To make "no lambda-splitting" the global default (but allow per-source
2016 :     overriding); this is the initial setting:
2017 :     local open Compiler.Control.LambdaSplitting in
2018 :     val _ = set (Default NONE)
2019 :     end
2020 :     - To make "lambda-splitting with aggressiveness a" the global default
2021 :     (and allow per-source overriding):
2022 :     local open Compiler.Control.LambdaSplitting in
2023 :     val _ = set (Default (SOME a))
2024 :     end
2025 :    
2026 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2027 :     write (in the respective .cm-file):
2028 :     a.sml (lambdasplitting:off)
2029 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2030 :     a.sml (lambdasplitting:on)
2031 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2032 :     <a> is a decimal non-negative integer):
2033 :     a.sml (lambdasplitting:<a>)
2034 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2035 :     a.sml (lambdasplitting:infinity)
2036 :     - To use the global default for a.sml:
2037 :     a.sml (lambdasplitting:default)
2038 :     or simply
2039 :     a.sml
2040 :    
2041 :     ----------------------------------------------------------------------
2042 : leunga 815 Name: Allen Leung
2043 :     Date: 2001/05/04 01:57:00 EDT
2044 :     Tag: leunga-20010504-sync
2045 :     Description:
2046 : leunga 788
2047 : leunga 815 MLRISC features.
2048 :    
2049 :     1. Fix to CMPXCHG instructions.
2050 :     2. Changed RA interface to allow annotations in callbacks.
2051 :     3. Added a new method to the stream interface to allow annotations updates.
2052 :    
2053 : george 749 ----------------------------------------------------------------------
2054 : blume 812 Name: Matthias Blume
2055 : blume 814 Date: 2001/05/01 11:45:00 EDT
2056 :     Tag: blume-20010501-pcedittmp
2057 :     Description:
2058 :    
2059 :     Changed install.sh to use the current working directory instead of
2060 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2061 :     of /usr/tmp caused trouble with MacOS X because of file premission
2062 :     problems.
2063 :    
2064 :     ----------------------------------------------------------------------
2065 :     Name: Matthias Blume
2066 : blume 812 Date: 2001/04/20 11:10:00 EDT
2067 :     Tag: blume-20010420-inMLflag
2068 :     Description:
2069 :    
2070 :     - added vp_limitPtrMask to vproc-state.h
2071 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2072 :     handling)
2073 :     - made the ML compiler aware of various data-structure offsets so it
2074 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2075 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2076 :     raw C-calls
2077 :    
2078 :     ----------------------------------------------------------------------
2079 : george 810 Name: Lal George
2080 :     Date: 2001/04/20 09:15:28 EDT
2081 :     Tag: george-20010420-macosX
2082 :     Description:
2083 :    
2084 :     - Changes to port to Mac OS X; Darwin.
2085 :    
2086 :     - In the process I found that sqrt was broken on the PPC, because the
2087 :     fsqrt instruction is not implemented.
2088 :    
2089 :     ----------------------------------------------------------------------
2090 : blume 799 Name: Matthias Blume
2091 : blume 808 Date: 2001/04/18 12:45:00 EDT
2092 :     Tag: blume-20010418-ccalls
2093 :     Description:
2094 :    
2095 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2096 :     (this bug prevented structure arguments containing pointers from being
2097 :     passed correctly)
2098 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
2099 :     structure arguments are represented as a pointer to the beginning
2100 :     of the structure (instead of having a series of synthesized arguments,
2101 :     one for each structure member)
2102 :    
2103 :     - made makeml script's verbosity level configurable via environment
2104 :     variable (MAKEML_VERBOSITY)
2105 :    
2106 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2107 : blume 808 in rawmem-x86.sml; we are now using the real thing
2108 :    
2109 :     ----------------------------------------------------------------------
2110 :     Name: Matthias Blume
2111 : blume 806 Date: 2001/03/22 16:25:00 EST
2112 :     Tag: blume-20010322-bootfiles
2113 :     Description:
2114 :    
2115 :     Created a new set of bootfiles (for your automatic installation convenience).
2116 :    
2117 :     ----------------------------------------------------------------------
2118 :     Name: Matthias Blume
2119 : blume 805 Date: 2001/03/22 15:10:00 EST
2120 :     Tag: blume-20010322-rawmem-parcm
2121 :     Description:
2122 :    
2123 :     1. All "raw memory access" primitives for the new FFI are implemented now
2124 :     (at least on the x86).
2125 :     2. Some further cleanup of CM's parallel make mechanism.
2126 :    
2127 :     ----------------------------------------------------------------------
2128 :     Name: Matthias Blume
2129 : blume 801 Date: 2001/03/19 17:53:00 EST
2130 :     Tag: blume-20010319-parallel
2131 :     Description:
2132 :    
2133 :     Parallel make (using compile servers) now works again.
2134 :    
2135 :     To this end, CM.stabilize and CMB.make have been modified to work in
2136 :     two passes when compile servers are attached:
2137 :     1. Compile everything, do not perform stabilization; this pass
2138 :     uses compile servers
2139 :     2. Stabilize everything; this pass does not use compile servers
2140 :     If there are no compile servers, the two passes are combined into one
2141 :     (as before). Splitting the passes increases the inherent parallelism
2142 :     in the dependency graph because the entire graph including all
2143 :     libraries is available at the same time. This, in turn, improves
2144 :     server utilization. The downside is that the master process will
2145 :     have to do some extra work after compilation is done (because for
2146 :     technical reasons it must re-read all the binfiles during stabilization).
2147 :    
2148 :     ----------------------------------------------------------------------
2149 :     Name: Matthias Blume
2150 : blume 800 Date: 2001/03/16 12:22:00 EST
2151 :     Tag: blume-20010316-bootfiles
2152 :     Description:
2153 :    
2154 :     Created a new set of bootfiles (for your automatic installation convenience).
2155 :    
2156 :     ----------------------------------------------------------------------
2157 :     Name: Matthias Blume
2158 : blume 799 Date: 2001/03/16 11:00:00 EST
2159 :     Tag: blume-20010316-MLTREE-fixup
2160 :     Description:
2161 :    
2162 :     This is a minor fixup for an (untagged) earlier commit by Allen.
2163 :     (A file was missing).
2164 :    
2165 :     ----------------------------------------------------------------------
2166 : leunga 788 Name: Allen Leung
2167 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
2168 :     Tag: leunga-20010305-cut-support
2169 :    
2170 :     1. New support for alternative control-flow in MLTREE.
2171 :     Currently we support
2172 :    
2173 :     FLOW_TO(CALL ...., [k1,...,kn])
2174 :    
2175 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
2176 :     in Moby
2177 :    
2178 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2179 :    
2180 :     2. Register Allocator
2181 :    
2182 :     Changes in interface [from Fermin, John]
2183 :    
2184 :     3. Alpha 8-bit SLL support [Fermin]
2185 :    
2186 :     4. All architectures
2187 :    
2188 :     A new module (ClusterExpandCopies) for expanding parallel copies.
2189 :    
2190 :     ----------------------------------------------------------------------
2191 :     Name: Allen Leung
2192 : leunga 788 Date: 2001/02/27 23:07:00 EST
2193 :     Tag: leunga-20010227-minor-stuff
2194 :    
2195 :     1. Alpha bug fix for CMOVNE
2196 :     2. Handle mltree COND(..,FCMP ...,...)
2197 :     3. Bug fix in simplifier
2198 :    
2199 :     ----------------------------------------------------------------------
2200 : blume 777 Name: Matthias Blume
2201 : blume 781 Date: 2001/01/30 17:50:00 EST
2202 :     Tag: blume-20010130-sync
2203 :     Description:
2204 :    
2205 :     This is just a minor update to sync my devel branch with the main brach.
2206 :     The only visible change is the addition of some README files.
2207 :    
2208 :     ----------------------------------------------------------------------
2209 :     Name: Matthias Blume
2210 : blume 778 Date: 2001/01/12 23:30:00 JST
2211 :     Tag: blume-20010112-bootfiles
2212 :     Description:
2213 :    
2214 :     Made a new set of bootfiles that goes with the current state of the
2215 :     repository.
2216 :    
2217 :     ----------------------------------------------------------------------
2218 :     Name: Matthias Blume
2219 : blume 777 Date: 2001/01/12 21:20:00 JST
2220 :     Tag: blume-20010112-sync
2221 :     Description:
2222 :    
2223 :     I am just flushing out some minor changes that had accumulated in
2224 :     my private branch in order to sync with the main tree. (This is
2225 :     mainly because I had CVS trouble when trying to merge _into_ my
2226 :     private branch.)
2227 :    
2228 :     Most people should be completely unaffected by this.
2229 :    
2230 :     ----------------------------------------------------------------------
2231 : leunga 775 Name: Allen Leung
2232 :     Date: Thu Jan 11 21:03:00 EST 2001
2233 :     Tag: leunga-20010111-labexp=mltree
2234 :     Description:
2235 :    
2236 :     1. Removed the type LabelExp and replace it by MLTree.
2237 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2238 :     3. There were some bugs in alpha code generator which would break
2239 :     64-bit code generation.
2240 :     4. Redo the tools to generate code with the
2241 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2242 :     are now generated by perl scripts.
2243 :    
2244 :     ----------------------------------------------------------------------
2245 : blume 771 Name: Matthias Blume
2246 : blume 774 Date: 2001/01/10 21:55:00 JST
2247 :     Tag: blume-20010110-rcc
2248 :     Description:
2249 :    
2250 :     The RCC stuff now seems to work (but only on the x86).
2251 :     This required hacking of the c-calls interface (and -implementation) in
2252 :     MLRISC.
2253 :    
2254 :     Normal compiler users should be unaffected.
2255 :    
2256 :     ----------------------------------------------------------------------
2257 :     Name: Matthias Blume
2258 : blume 773 Date: 2001/01/09 01:20:00 JST
2259 :     Tag: blume-20010109-rcc
2260 :     Description:
2261 :    
2262 :     This is a fairly big patch, flushing out a large number of pending
2263 :     changes that I made to my development copy over the last couple of days.
2264 :    
2265 :     Of practical relevance at this moment is a workaround for a pickling
2266 :     bug that Allen ran into the other day. The cause of the bug itself is
2267 :     still unknown and it might be hard to fix it properly, but the
2268 :     workaround has some merits of its own (namely somewhat reducing pickling
2269 :     overhead for certain libraries). Therefore, I think this solution should
2270 :     be satisfactory at this time.
2271 :    
2272 :     The rest of the changes (i.e., the vast majority) has to do with my
2273 :     ongoing efforts of providing direct support for C function calls from
2274 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2275 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2276 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2277 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2278 :     handle RCC.
2279 :    
2280 :     The last part (the changes to mlriscGen.sml) are still known to be
2281 :     wrong on the x86 and not implemented on all other architectures. But
2282 :     the infrastructure is in place. I had to change a few functor
2283 :     signatures in the backend to be able to route the CCalls interface
2284 :     from MLRISC there, and I had to specialize the mltree type (on the
2285 :     x86) to include the necessary extensions. (The extensions themselves
2286 :     were already there and redy to go in MLRISC/x86).
2287 :    
2288 :     Everything should be very happy as soon as someone helps me with
2289 :     mlriscGen.sml...
2290 :    
2291 :     In any case, nothing of this should matter to anyone as long as the
2292 :     new primop is not being used (which is going to be the case unless you
2293 :     find it where I hid it :). The rest of the compiler is completely
2294 :     unaffected.
2295 :    
2296 :     ----------------------------------------------------------------------
2297 :     Name: Matthias Blume
2298 : blume 772 Date: 2001/01/05 00:30:00 JST
2299 :     Tag: blume-20010105-primops
2300 :     Description:
2301 :    
2302 :     Added some experimental support for work that I am doing right now.
2303 :     These changes mostly concern added primops, but there is also a new
2304 :     experimental C library in the runtime system (but currently not enabled
2305 :     anywhere except on Linux/X86).
2306 :    
2307 :     In the course of adding primops (and playing with them), I discovered that
2308 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2309 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2310 :     To recover, I made type information in INL_PRIM mandatory and changed
2311 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2312 :     has complete, correct type information (i.e., no bottom types).
2313 :    
2314 :     Since all these changes mean that we need new binfiles, I also bumped the
2315 :     version number to 110.32.1.
2316 :    
2317 :     ----------------------------------------------------------------------
2318 :     Name: Matthias Blume
2319 : blume 771 Date: 2000/12/30 22:10:00 JST
2320 :     Tag: blume-20001230-various
2321 :     Description:
2322 :    
2323 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2324 :     to each other using path anchors. (See CM manual for explanation.)
2325 :    
2326 :     Updated CM documentation.
2327 :    
2328 :     Fixed some bugs in CM.
2329 :    
2330 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2331 :    
2332 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2333 :    
2334 :     Added cleanup handler for exportML to reset timers and compiler stats.
2335 :    
2336 :     ----------------------------------------------------------------------
2337 : george 761 Name: Lal George
2338 :     Date: 2000/12/22 22:22:58 EST 2000
2339 :     Tag: Release_110_32
2340 :     Description:
2341 :    
2342 :     Infinite precision used throughout MLRISC.
2343 :     see MLRISC/mltree/machine-int.sig
2344 :    
2345 :     ----------------------------------------------------------------------
2346 : blume 756 Name: Matthias Blume
2347 : blume 760 Date: 2000/12/22 23:16:00 JST
2348 :     Tag: blume-20001222-warn
2349 :     Description:
2350 : blume 759
2351 : blume 760 Corrected wording and formatting of some CM warning message which I
2352 :     broke in my previous patch.
2353 : blume 759
2354 :     ----------------------------------------------------------------------
2355 :     Name: Matthias Blume
2356 : blume 763 Date: 2000/12/22 21:20:00 JST
2357 :     Tag: blume-20001222-anchorenv
2358 :     Description:
2359 :    
2360 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2361 :    
2362 :     ----------------------------------------------------------------------
2363 :     Name: Matthias Blume
2364 : blume 758 Date: 2000/12/22 13:15:00 JST
2365 :     Tag: blume-20001222-cleanup
2366 :     Description:
2367 :    
2368 :     Removed src/cm/ffi which does not (and did not) belong here.
2369 :    
2370 :     ----------------------------------------------------------------------
2371 :     Name: Matthias Blume
2372 : blume 757 Date: 2000/12/21 23:55:00 JST
2373 :     Tag: blume-20001221-exn
2374 :     Description:
2375 :    
2376 :     Probably most important: CM no longer silently swallows all exceptions
2377 :     in the compiler.
2378 :     Plus: some other minor CM changes. For example, CM now reports some
2379 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2380 :    
2381 :     ----------------------------------------------------------------------
2382 :     Name: Matthias Blume
2383 : blume 756 Date: 2000/12/15 00:01:05 JST
2384 :     Tag: blume-20001215-dirtool
2385 :     Description:
2386 :    
2387 :     - "dir" tool added.
2388 :     - improvements and cleanup to Tools structure
2389 :     - documentation updates
2390 :    
2391 :     ----------------------------------------------------------------------
2392 : leunga 752 Name: Allen Leung
2393 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2394 :     Description:
2395 :     Tag: leunga-20001214-int-inf
2396 :     Description:
2397 :    
2398 :     In IntInf, added these standard functions, which are missing from our
2399 :     implementation:
2400 :    
2401 :     andb : int * int -> int
2402 :     xorb : int * int -> int
2403 :     orb : int * int -> int
2404 :     notb : int -> int
2405 :     << : int * word -> int
2406 :     ~>> : int * word -> int
2407 :    
2408 :     Not tested, I hope they are correct.
2409 :    
2410 :     ----------------------------------------------------------------------
2411 :     Name: Allen Leung
2412 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2413 :     Description:
2414 :     Tag: leunga-20001208-nowhere
2415 :     Description:
2416 :    
2417 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2418 :     to generate better error messages etc. Plus a brief manual.
2419 :    
2420 :     ----------------------------------------------------------------------
2421 : george 749 Name: Lal George
2422 :     Date: 2000/12/08 09:54:02 EST 2000
2423 :     Tag: Release_110_31
2424 :     Description:
2425 : george 705
2426 : george 749 - Version 110.31
2427 : leunga 591 ----------------------------------------------------------------------
2428 : leunga 744 Name: Allen Leung
2429 :     Date: Thu Dec 7 22:01:04 EST 2000
2430 :     Tag: leunga-20001207-cell-monster-hack
2431 :     Description:
2432 :    
2433 :     Major MLRISC internal changes. Affect all clients.
2434 :     Summary:
2435 :    
2436 :     1. Type CELLS.cell = int is now replaced by a datatype.
2437 :     As a result, the old regmap is now gone. Almost all interfaces
2438 :     in MLRISC change as a consequence.
2439 :    
2440 :     2. A new brand version of machine description tool (v3.0) that generates
2441 :     modules expecting the new interface. The old version is removed.
2442 :    
2443 :     3. The RA interface has been further abstracted into two new functors.
2444 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2445 :     [See also directory MLRISC/demo.]
2446 :    
2447 :     4. Some other new source->source code generation tools are available:
2448 :    
2449 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2450 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2451 :     I use this tool to generate the peephole optimizers---with the new
2452 :     cell type changes, peephole rules are becoming difficult to write
2453 :     without conditional pattern matching.
2454 :    
2455 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2456 :     cover the entire MLRISC source tree so many things broke.
2457 :    
2458 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2459 :     They are moved because there are a lot of them and they clutter up the
2460 :     root dir.
2461 :    
2462 :     7. More detailed documentation to come...
2463 :    
2464 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2465 :     a makeml -rebuild first. This is because of other other
2466 :     changes that Matthias has made (see below).
2467 :    
2468 :    
2469 :     ----------------------------------------------------------------------
2470 : blume 733 Name: Matthias Blume
2471 : blume 742 Date: 2000/11/30 23:12:00 JST
2472 :     Tag: blume-20001130-filereorg
2473 :     Description:
2474 :    
2475 :     Some manual updates and some file reorganizations in CM.
2476 :    
2477 :     ----------------------------------------------------------------------
2478 :     Name: Matthias Blume
2479 : blume 737 Date: 2000/11/24 17:45:00 JST
2480 :     Tag: blume-20001124-link
2481 :     Description:
2482 :    
2483 :     Drastically improved link traversal code for the case that the dynamic
2484 :     value was already loaded at bootstrap time. As a result, CM and CMB
2485 :     now both load blazingly fast -- even on a very slow machine. Also,
2486 :     memory consumption has been further reduced by this.
2487 :    
2488 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2489 :     bootstrap you have to do this:
2490 :    
2491 :     1. Run CMB.make
2492 :     2. Make a symbolic link for the boot directory:
2493 :     ln -s sml.boot.ARCH-OS xxx
2494 :     3. "Rebuild" the boot directory:
2495 :     ./makeml -boot xxx -rebuild sml ; rm xxx
2496 :     4. Boot normally:
2497 :     ./makeml
2498 :    
2499 :     ----------------------------------------------------------------------
2500 :     Name: Matthias Blume
2501 : blume 735 Date: 2000/11/21 21:20:00 JST
2502 :     Tag: blume-20001121-tools
2503 :     Description:
2504 :    
2505 :     Continued hacking on autoloading problem -- with success this time.
2506 :     Also changed tool-plugin mechanism. See new CM manual.
2507 :    
2508 :     ----------------------------------------------------------------------
2509 :     Name: Matthias Blume
2510 : blume 734 Date: 2000/11/19 14:30:00 JST
2511 :     Tag: blume-20001119-autoload
2512 :     Description:
2513 :    
2514 :     Some hacking to make autoloading faster. Success for CMB, no success
2515 :     so far for CM. There is a reduced structure CM' that autoloads faster.
2516 :     (This is a temporary, non-documented hack to be eliminated again when
2517 :     the general problem is solved.)
2518 :    
2519 :     ----------------------------------------------------------------------
2520 :     Name: Matthias Blume
2521 : blume 733 Date: 2000/11/17 14:10:00 JST
2522 :     Tag: blume-20001117-pickle-lib
2523 :     Description:
2524 :    
2525 :     1. Eliminated comp-lib.cm
2526 :     2. Made pickle-lib.cm
2527 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
2528 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
2529 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
2530 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
2531 :     Duplicated intmap modules thrown out.
2532 :     6. Hunted down all duplicated SCC code and replaced it with
2533 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
2534 :     7. Rewrote Feedback module.
2535 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
2536 :     should be thrown out and equivalent modules from smlnj-lib.cm
2537 :     should be used (IntRedBlackSet, IntListSet, ...).
2538 :    
2539 :     Confirmed that compiler compiles to fixpoint.
2540 :    
2541 :     ----------------------------------------------------------------------
2542 : leunga 731 Name: Allen Leung
2543 :     Date: 2000/11/10 18:00:00
2544 :     Tag: leunga-20001110-new-x86-fp
2545 :    
2546 :     A new x86 floating point code generator has been added.
2547 :     By default this is turned off. To turn this on, do:
2548 :    
2549 :     CM.autoload "$smlnj/compiler.cm";
2550 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
2551 :    
2552 :     Changes:
2553 :    
2554 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
2555 :     2. Changed the extension callback for FTANGENT to generate:
2556 :    
2557 :     fptan
2558 :     fstp %st(0)
2559 :     instead of
2560 :     fptan
2561 :     fstpl ftempmem
2562 :    
2563 :     3. Numerous assembly fixes for x86.
2564 :    
2565 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
2566 :     support for a whole bunch of instructions and addressing modes:
2567 :    
2568 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
2569 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
2570 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
2571 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
2572 :     fxch %st(n)
2573 :     fld %st(n)
2574 :     fst %st(n)
2575 :     fst mem
2576 :     fstp %st(n)
2577 :     fucom %st(n)
2578 :     fucomp %st(n)
2579 :    
2580 :     All these are now generated when the fast fp mode is turned on.
2581 :    
2582 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
2583 :    
2584 :     ----------------------------------------------------------------------
2585 : blume 729 Name: Matthias Blume
2586 :     Date: 2000/11/09 11:20:00 JST
2587 :     Tag: blume-20001109-scc
2588 :     Description:
2589 :    
2590 :     Eliminated some code duplication:
2591 :    
2592 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
2593 :     (Otherwise the functor is useless.)
2594 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
2595 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
2596 :    
2597 :     ----------------------------------------------------------------------
2598 : george 721 Name: Lal George
2599 :     Date: 2000/11/06 09:02:21 EST 2000
2600 :     Tag: Release_110_30
2601 :     Description:
2602 :    
2603 :     - Version 110.30
2604 :     ----------------------------------------------------------------------
2605 : blume 715 Name: Matthias Blume
2606 : blume 716 Date: 2000/11/04 14:45:00
2607 :     Tag: blume-20001104-mlbuild
2608 :     Description:
2609 :    
2610 :     - Made ml-build faster on startup.
2611 :     - Documentation fixes.
2612 :    
2613 :     ----------------------------------------------------------------------
2614 :     Name: Matthias Blume
2615 : blume 715 Date: 2000/11/02 17:00:00 JST
2616 :     Tag: blume-20001102-condcomp
2617 :     Description:
2618 :    
2619 :     - Small tweaks to pickler -- new BOOTFILES!
2620 :     - Version bumped to 110.29.2.
2621 :     - Added conditional compilation facility to init.cmi (see comment there).
2622 :     ----------------------------------------------------------------------
2623 : leunga 713 Name: Allen Leung
2624 :     Date: 2000/10/23 19:31:00
2625 :     Tag: leunga-20001023-demo-ra
2626 :    
2627 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
2628 :     2. Test programs for the graph library updated
2629 :     3. Some new MLRISC demo programs added
2630 :    
2631 :     ----------------------------------------------------------------------
2632 : blume 710 Name: Matthias Blume
2633 : blume 711 Date: 2000/08/31 22:15:00 JST
2634 :     Tag: blume-20001017-errmsg
2635 :     Description:
2636 :    
2637 :     More error message grief: Where there used to be no messages, there
2638 :     now were some that had bogus error regions. Fixed.
2639 :    
2640 :     ----------------------------------------------------------------------
2641 :     Name: Matthias Blume
2642 : blume 710 Date: 2000/08/31 17:30:00 JST
2643 :     Tag: blume-20001017-v110p29p1
2644 :     Description:
2645 :    
2646 :     I made a version 110.29.1 with new bootfiles.
2647 :    
2648 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
2649 :     CM documentation changes and a small bugfix in CM's error reporting.
2650 :    
2651 :     ----------------------------------------------------------------------
2652 : george 705 Name: Lal George
2653 :     Date: 2000/09/27 14:42:35 EDT
2654 :     Tag: george-20000927-nodestatus
2655 :     Description:
2656 :    
2657 :     Changed the type of the nodestatus, so that:
2658 :    
2659 :     SPILLED(~1) is now SPILLED
2660 :     SPILLED(m) where m>=0 is now MEMREG(m)
2661 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
2662 :    
2663 :     ----------------------------------------------------------------------
2664 : blume 697 Name: Matthias Blume
2665 : blume 703 Date: 2000/09/07 14:45:00 JST
2666 :     Tag: blume-20000907-cmerrmsg
2667 :     Description:
2668 :    
2669 :     Small tweak to CM to avoid getting ML syntax error messages twice.
2670 :    
2671 :     ----------------------------------------------------------------------
2672 :     Name: Matthias Blume
2673 : blume 701 Date: 2000/08/31 18:00:00 JST
2674 :     Tag: blume-20000831-cvsbootfiles
2675 :     Description:
2676 :    
2677 :     New URL for boot files (because the 110.29 files on the BL server do
2678 :     now work correctly with my updated install scripts for yacc and lex).
2679 :    
2680 :     ----------------------------------------------------------------------
2681 :     Name: Matthias Blume
2682 : blume 697 Date: 2000/08/08 12:33:00 JST
2683 :     Tag: blume-20000808-manual
2684 :     Description:
2685 :    
2686 :     Tiny update to CM manual.
2687 :    
2688 :     ----------------------------------------------------------------------
2689 : leunga 695 Name: Allen Leung
2690 :     Date: 2000/08/7 19:31:00
2691 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
2692 :    
2693 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
2694 :     and Mobdy.
2695 :    
2696 :     1. x86
2697 :    
2698 :     a. Fixes to peephole module by John and Dan.
2699 :     b. Assembly fix to SETcc by Allen.
2700 :     c. Fix to c-call by John.
2701 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
2702 :     e. Instruction selection optimization to SETcc as suggested by John.
2703 :    
2704 :     For example,
2705 :    
2706 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
2707 :    
2708 :     should generate:
2709 :    
2710 :     MOVL a, x
2711 :     SUBL b, x
2712 :     SHRL 31, x
2713 :    
2714 :     2. IR stuff
2715 :    
2716 :     A bunch of new DJ-graph related algorithms added. These
2717 :     speed up SSA construction.
2718 :    
2719 :     3. SSA + Scheduling
2720 :    
2721 :     Added code for SSA and scheduling to the repository
2722 :    
2723 :     ----------------------------------------------------------------------
2724 : george 694 Name: Lal George
2725 :     Date: 2000/07/27 11:53:14 EDT
2726 :    
2727 :     Tag: lal-20000727-linux-ppc
2728 :     Description:
2729 :    
2730 :     Made changes to support Linux PPC.
2731 :     p.s. I have confirmation that the 110.29 boot files work fine.
2732 :    
2733 :     ----------------------------------------------------------------------
2734 : blume 684 Name: Matthias Blume
2735 : blume 692 Date: 2000/07/27 17:40:00 JST
2736 :     Tag: blume-20000727-scripts
2737 :     Description:
2738 :    
2739 :     !!!! WARNING !!!!
2740 :     You must recompile the runtime system!
2741 :     !!!! WARNING !!!!
2742 :    
2743 :     This is basically another round of script-enhancements:
2744 :    
2745 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
2746 :     and undefine CM preprocessor symbols.
2747 :    
2748 :     2. ml-build avoids generating a new heap image if it finds that the
2749 :     existing one is still ok. (The condition is that no ML file had to
2750 :     be recompiled and all ML files are found to be older that the heap
2751 :     file.)
2752 :    
2753 :     To make this work smoothly, I also hacked the runtime system as
2754 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
2755 :     (.sparc-solaris, ...) that is currently being used.
2756 :    
2757 :     Moreover, the signature of CM.mk_standalone has changed. See the
2758 :     CM manual.
2759 :    
2760 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
2761 :     CM manual for details.)
2762 :    
2763 :     4. More CM manual updates:
2764 :     - all of the above has been documented.
2765 :     - there is now a section describing the (CM-related) command line
2766 :     arguments that are accepted by the "sml" command
2767 :    
2768 :     ----------------------------------------------------------------------
2769 :     Name: Matthias Blume
2770 : blume 691 Date: 2000/07/25 16:20:00 JST
2771 :     Tag: blume-20000725-makedepend
2772 :     Description:
2773 :    
2774 :     Added a script called ml-makedepend. This can be used in makefiles
2775 :     for Unix' make in a way very similar to the "makedepend" command for
2776 :     C.
2777 :    
2778 :     The script internally uses function CM.sources.
2779 :    
2780 :     Synopsis:
2781 :    
2782 :     ml-makedepend [-f makefile] cmfile targetname
2783 :    
2784 :     The default for the makefile is "makefile" (or "Makefile" should
2785 :     "makefile" not exist).
2786 :    
2787 :     ml-makedepend adds a cmfile/targetname-specific section to this
2788 :     makefile (after removing the previous version of this section). The
2789 :     section contains a single dependency specification with targetname on
2790 :     the LHS (targetname is an arbitrary name), and a list of files derived
2791 :     from the cmfile on the RHS. Some of the files on the RHS are
2792 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
2793 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
2794 :     path names. The makefile writer is responsible for making sure that
2795 :     these variables have correct at the time "make" is invoked.
2796 :    
2797 :     ----------------------------------------------------------------------
2798 :     Name: Matthias Blume
2799 : blume 690 Date: 2000/07/22 23:30:00 JST
2800 :     Tag: blume-20000722-urlupdate
2801 :     Description:
2802 :    
2803 :     Changed BOOT and config/srcarchiveurl to point to BL server:
2804 :    
2805 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
2806 :    
2807 :     ----------------------------------------------------------------------
2808 :     Name: Matthias Blume
2809 : blume 686 Date: 2000/07/18 18:00:00 JST
2810 :     Tag: blume-20000718-Version_110_29
2811 :     Description:
2812 :    
2813 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
2814 :    
2815 :     2. Updated config/version to 110.29
2816 :    
2817 :     3. Updated config/srcarchiveurl
2818 :    
2819 :     3. New boot files!
2820 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
2821 :    
2822 :     ----------------------------------------------------------------------
2823 :     Name: Matthias Blume
2824 : blume 684 Date: 2000/07/11 13:58:00 JST
2825 :     Tag: blume-20000711-doctypo
2826 :     Description:
2827 :    
2828 :     Fixed a few typos in CM manual.
2829 :    
2830 :     ----------------------------------------------------------------------
2831 : leunga 683 Name: Allen Leung
2832 :     Date: 2000/06/15 00:38:00
2833 :     Tag: leunga-20000704-sparc-x86
2834 :    
2835 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
2836 :     2. fix to x86 RET bug [found by Dan Grossman]
2837 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
2838 :    
2839 :     Affects c-- and moby only
2840 :    
2841 :     ----------------------------------------------------------------------
2842 : blume 680 Name: Matthias Blume
2843 : blume 682 Date: 2000/07/04 15:26:00
2844 :     Tag: blume-20000704-trigger
2845 :     Description:
2846 :    
2847 :     1. Improvements to CM manual.
2848 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
2849 :     of getting a back-trace. The function, when called, raises an
2850 :     internal exception which explicitly carries the full back-trace history,
2851 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
2852 :     or not). The interactive loop will print that history once it arrives
2853 :     at top level.
2854 :     Short of having all exceptions implicitly carry the full history, the
2855 :     recommended way of using this facility is:
2856 :     - compile your program with instrumentation "on"
2857 :     - run it, when it raises an exception, look at the history
2858 :     - if the history is "cut off" because of some handler, go and modify
2859 :     your program so that it explicitly calls BTrace.trigger
2860 :     - recompile (still instrumented), and rerun; look at the full history
2861 :    
2862 :     ----------------------------------------------------------------------
2863 :     Name: Matthias Blume
2864 :     Date: 2000/07/03 15:36:00 JST
2865 : blume 680 Tag: blume-20000702-manual
2866 :     Description:
2867 :    
2868 :     Small corrections and updates to CM manual.
2869 :    
2870 :     ----------------------------------------------------------------------
2871 : blume 682 Name: Matthias Blume
2872 : blume 679 Date: 2000/06/29 16:04:00 JST
2873 :     Tag: blume-20000629-yacctool
2874 :     Description:
2875 :    
2876 :     Changes:
2877 :    
2878 :     1. Class "mlyacc" now takes separate arguments to pass options to
2879 :     generated .sml- and .sig-files independently.
2880 :     2. Corresponding CM manual updates.
2881 :     3. BTrace module now also reports call sites. (However, for loop clusters
2882 :     it only shows from where the cluster was entered.) There are associated
2883 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
2884 :    
2885 :     ----------------------------------------------------------------------
2886 : blume 675 Name: Matthias Blume
2887 : blume 678 Date: 2000/06/27 16:51:00 JST
2888 :     Tag: blume-20000627-noweb
2889 :     Description:
2890 :    
2891 :     Changes:
2892 :    
2893 :     1. Implemented "subdir" and "witness" options for noweb tool.
2894 :     This caused some slight internal changes in CM's tool implementation.
2895 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
2896 :     some remaining issues from earlier path anchor changes.
2897 :     3. Updated CM manual accordingly.
2898 :    
2899 :     4. Changed implementation of back-tracing so that I now consider it
2900 :     ready for prime-time.
2901 :    
2902 :     In particular, you don't have to explicitly trigger the back-trace
2903 :     anymore. Instead, if you are running BTrace-instrumented code and
2904 :     there is an uncaught exception (regardless of whether or not it was
2905 :     raised in instrumented code), the top-level evalloop will print
2906 :     the back-trace.
2907 :    
2908 :     Features:
2909 :    
2910 :     - Instrumented and uninstrumented code work together seemlessly.
2911 :     (Of course, uninstrumented code is never mentioned in actual
2912 :     back-traces.)
2913 :    
2914 :     - Asymptotic time- and space-complexity of instrumented code is
2915 :     equal to that of uninstrumented code. (This means that
2916 :     tail-recursion is preserved by the instrumentation phase.)
2917 :    
2918 :     - Modules whose code has been instrumented in different sessions
2919 :     work together without problem.
2920 :    
2921 :     - There is no penalty whatsoever on uninstrumented code.
2922 :    
2923 :     - There is no penalty on "raise" expressions, even in
2924 :     instrumented code.
2925 :    
2926 :     A potential bug (or perhaps it is a feature, too):
2927 :    
2928 :     A back-trace reaches no further than the outermost instrumented
2929 :     non-trivial "raise". Here, a "trivial" raise is one that is the
2930 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
2931 :    
2932 :     <exp> handle e => raise e
2933 :    
2934 :     and even
2935 :    
2936 :     <exp> handle Foo => raise Bar
2937 :    
2938 :     and, of course, through
2939 :    
2940 :     <exp> handle Foo => ...
2941 :    
2942 :     if the exception was not Foo.
2943 :    
2944 :     Back-traces always reach right through any un-instrumented code
2945 :     including any of its "handle" expressions, trivial or not.
2946 :    
2947 :     To try this out, do the following:
2948 :    
2949 :     - Erase all existing binfiles for your program.
2950 :     (You may keep binfiles for those modules where you think you
2951 :     definitely don't need back-tracing.)
2952 :     - Turn on back-trace instrumentation:
2953 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2954 :     - Recompile your program. (I.e., run "CM.make" or "use".)
2955 :     - You may now turn instrumentation off again (if you want):
2956 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2957 :     - Run your program as usual. If it raises an exception that
2958 :     reaches the interactive toplevel, then a back-trace will
2959 :     automatically be printed. After that, the toplevel loop
2960 :     will print the exception history as usual.
2961 :    
2962 :     ----------------------------------------------------------------------
2963 :     Name: Matthias Blume
2964 : blume 677 Date: 2000/06/26 09:56:46 JST
2965 :     Tag: blume-20000626-setup
2966 :     Description:
2967 :    
2968 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2969 :     ML code before and after compiling a file (e.g., to set compiler
2970 :     flags)
2971 :    
2972 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2973 :     - associated changes to btrace.sml (BTrace instrumentation pass)
2974 :     - cleaner implementation of btimp.sml (BTrace tracing and report
2975 :     module)
2976 :    
2977 :     CM manual: * new path encoding documented
2978 :     * description of setup-parameter to "sml" added
2979 :    
2980 :     The biggest user-visible change to back-tracing is that it is no
2981 :     longer necessary to compile all traced modules within the same
2982 :     session. (This was a real limitation.)
2983 :    
2984 :     ----------------------------------------------------------------------
2985 :     Name: Matthias Blume
2986 : blume 676 Date: 2000/06/24 12:40:00 JST
2987 :     Tag: blume-20000624-startup
2988 :     Description:
2989 :    
2990 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
2991 :     bit too often -- to put it mildly. :)
2992 :    
2993 :     ----------------------------------------------------------------------
2994 :     Name: Matthias Blume
2995 : blume 675 Date: 2000/06/23 18:20:00 JST
2996 :     Tag: blume-20000623-btrace
2997 :     Description:
2998 :    
2999 :     This updates adds a backtrace facility to aid programmers in debugging
3000 :     their programs. This involves the following changes:
3001 :    
3002 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3003 :     keeping track of the current call stack. When programs are compiled
3004 :     in a special mode, the compiler will insert calls to these hooks
3005 :     into the user program.
3006 :     "Hook" means that it is possible for different implementations of
3007 :     back-tracing to register themselves (at different times).
3008 :    
3009 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3010 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3011 :     It can be turned on using this call:
3012 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3013 :     Turning it off again:
3014 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3015 :     Querying the current status:
3016 :     SMLofNJ.Internals.BTrace.mode NONE;
3017 :     Annotated programs are about twice as big as normal ones, and they
3018 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3019 :     where all hooks do nothing). The slowdown with a plugin that is
3020 :     actually useful (such as the one supplied by default) is even greater,
3021 :     but in the case of the default plugin it is still only an constant
3022 :     factor (amortized).
3023 :    
3024 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3025 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3026 :     the above-mentioned function "mode" controls whether the annotation
3027 :     phase is invoked by the compiler. Another important function is
3028 :     "trigger": when called it aborts the current execution and causes
3029 :     the top-level loop to print a full back-trace.
3030 :    
3031 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3032 :     for back-tracing. It keeps track of the dynamic call stack and in
3033 :     addition to that it keeps a partial history at each "level" of that
3034 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3035 :     d and b (at separate times, dynamically), then the report will show:
3036 :    
3037 :     GOTO d
3038 :     /c
3039 :     GOTO \b
3040 :     CALL a
3041 :    
3042 :     This shows that there was an initial non-tail call of a, then a
3043 :     tail-call to b or c, looping behavior in a cluster of functions that
3044 :     consist of b and c, and then a goto from that cluster (i.e., either from
3045 :     b or from c) to d.
3046 :    
3047 :     Note that (depending on the user program) the amount of information
3048 :     that the back-trace module has to keep track of at each level is bounded
3049 :     by a constant. Thus, the whole implementation has the same asymptotical
3050 :     complexity as the original program (both in space and in time).
3051 :    
3052 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3053 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3054 :     which is raised by the "trigger" function mentioned above.
3055 :    
3056 :     Notes on usage:
3057 :    
3058 :     - Annotated code works well together with unannotated code:
3059 :     Unannotated calls simply do not show up at all in the backtrace.
3060 :    
3061 :     - It is not a good idea to let modules that were annotated during
3062 :     different sessions run at the same time. This is because the compiler
3063 :     chooses small integers to identify individual functions, and there
3064 :     will be clashes if different modules were compiled in separate sessions.
3065 :     (Nothing will crash, and you will even be told about the clashes, but
3066 :     back-trace information will in general not be useful.)
3067 :    
3068 :     - Back-tracing can be confused by callcc and capture.
3069 :    
3070 :     - The only way of getting a back-trace right now is to explicitly
3071 :     invoke the "trigger" function from your user program. Eventually, we
3072 :     should make every exception carry back-trace information (if
3073 :     available). But since this creates more overhead at "raise"-time
3074 :     (similar to the current exnHistory overhead), I have not yet
3075 :     implemented this. (The implementation will be rather easy.) With
3076 :     exceptions carrying back-trace information, this facility will be even
3077 :     more useful because users don't need to modify their programs...
3078 :    
3079 :     - While it is possible to compile the compiler with back-trace
3080 :     annotations turned on (I did it to get some confidence in
3081 :     correctness), you must make absolutely sure that core.sml and
3082 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3083 :     be compiled with annotation because there is no core access yet, but
3084 :     if you compile btimp.sml with annotation, then the system will go into
3085 :     an infinite recursion and crash.)
3086 :     Since CM currently does not know about BTrace, the only way to turn
3087 :     annotations on and off for different modules of the compiler is to
3088 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3089 :     this is awkward and clumsy.
3090 :    
3091 :     Sample sessions:
3092 :    
3093 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3094 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3095 :     [autoloading]
3096 :     [autoloading done]
3097 :     val it = false : bool
3098 :     - structure X = struct
3099 :     - fun main n = let
3100 :     - fun a (x, 0) = d x
3101 :     - | a (x, n) = b (x, n - 1)
3102 :     - and b (x, n) = c (x, n)
3103 :     - and c (x, n) = a (x, n)
3104 :     - and d x = e (x, 3)
3105 :     - and e (x, 0) = f x
3106 :     - | e (x, n) = e (x, n - 1)
3107 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
3108 :     - | f n = n * g (n - 1)
3109 :     - and g n = a (n, 3)
3110 :     - in
3111 :     - f n
3112 :     - end
3113 :     - end;
3114 :     structure X : sig val main : int -> int end
3115 :     - X.main 3;
3116 :     *** BACK-TRACE ***
3117 :     GOTO stdIn:4.2-13.20: X.main[2].f
3118 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3119 :     GOTO stdIn:4.2-13.20: X.main[2].d
3120 :     / stdIn:4.2-13.20: X.main[2].a
3121 :     | stdIn:4.2-13.20: X.main[2].b
3122 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3123 :     CALL stdIn:4.2-13.20: X.main[2].g
3124 :     GOTO stdIn:4.2-13.20: X.main[2].f
3125 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3126 :     GOTO stdIn:4.2-13.20: X.main[2].d
3127 :     / stdIn:4.2-13.20: X.main[2].a
3128 :     | stdIn:4.2-13.20: X.main[2].b
3129 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3130 :     CALL stdIn:4.2-13.20: X.main[2].g
3131 :     GOTO stdIn:4.2-13.20: X.main[2].f
3132 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3133 :     GOTO stdIn:4.2-13.20: X.main[2].d
3134 :     / stdIn:4.2-13.20: X.main[2].a
3135 :     | stdIn:4.2-13.20: X.main[2].b
3136 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3137 :     CALL stdIn:4.2-13.20: X.main[2].g
3138 :     GOTO stdIn:4.2-13.20: X.main[2].f
3139 :     CALL stdIn:2.15-17.4: X.main[2]
3140 :     -
3141 :    
3142 :     (Note that because of a FLINt bug the above code currently does not
3143 :     compile without BTrace turned on.)
3144 :    
3145 :     Here is another example, using my modified Tiger compiler:
3146 :    
3147 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3148 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3149 :     [autoloading]
3150 :     [autoloading done]
3151 :     val it = false : bool
3152 :     - CM.make "sources.cm";
3153 :     [autoloading]
3154 :     ...
3155 :     [autoloading done]
3156 :     [scanning sources.cm]
3157 :     [parsing (sources.cm):parse.sml]
3158 :     [creating directory CM/SKEL ...]
3159 :     [parsing (sources.cm):tiger.lex.sml]
3160 :     ...
3161 :     [wrote CM/sparc-unix/semant.sml]
3162 :     [compiling (sources.cm):main.sml]
3163 :     [wrote CM/sparc-unix/main.sml]
3164 :     [New bindings added.]
3165 :     val it = true : bool
3166 :     - Main.compile ("../testcases/merge.tig", "foo.out");
3167 :     *** BACK-TRACE ***
3168 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
3169 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3170 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3171 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3172 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3173 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3174 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
3175 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
3176 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
3177 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3178 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
3179 :     CALL main.sml:1.18-118.4: Main.compile[2]
3180 :     -
3181 :    
3182 :     ----------------------------------------------------------------------
3183 : blume 673 Name: Matthias Blumen
3184 :     Date: 2000/06/21 18:00:00 JST
3185 :     Tag: blume-20000621-manual
3186 :     Description:
3187 :    
3188 :     CM manual update: Path environments documented.
3189 :    
3190 :     ----------------------------------------------------------------------
3191 : blume 666 Name: Matthias Blume
3192 : blume 672 Date: 2000/06/19 13:40:00
3193 :     Tag: blume-20000619-manual
3194 :     Description:
3195 :    
3196 :     CM manual and system/README update. This only covers the fact that
3197 :     there are no more implicit anchors. (Path environments and the "bind"
3198 :     option to "cm" have yet to be documented.)
3199 :    
3200 :     ----------------------------------------------------------------------
3201 :     Name: Matthias Blume
3202 : blume 671 Date: 2000/06/19 11:05:00 JST
3203 :     Tag: blume-20000619-chdir-bugfix
3204 :     Description:
3205 :    
3206 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
3207 :    
3208 :     ----------------------------------------------------------------------
3209 :     Name: Matthias Blume
3210 : blume 670 Date: 2000/06/18 22:00:10 JST
3211 :     Tag: blume-20000618-implicit-anchors-really-gone
3212 :     Description:
3213 :    
3214 :     I updates the previous HISTORY entry where I forgot to mention that
3215 :     implicit anchors are no longer with us.
3216 :    
3217 :     The current update also gets rid of the (now useless) controller
3218 :     CM.Control.implicit_anchors.
3219 :    
3220 :     ----------------------------------------------------------------------
3221 :     Name: Matthias Blume
3222 : blume 666 Date: 2000/06/16 17:30:00 JST
3223 :     Tag: blume-20000616-anchorenv
3224 :     Description:
3225 :    
3226 :     This patch implements the long anticipated (just kidding :) "anchor
3227 :     environment" mechanism. In the course of doing this, I also
3228 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3229 :     one should be more robust in certain boundary cases. In any case, it
3230 :     is a lot cleaner than its predecessor (IMHO).
3231 :    
3232 :     This time, although there is yet another boot file format change, I
3233 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3234 :     are necessary and bootstrapping is straightforward. (You cannot read
3235 :     new bootfiles into an old system, but the other way around is no
3236 :     problem.)
3237 :    
3238 :     Visible changes:
3239 :    
3240 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3241 :     longer recognized at all. This means that such path names are not
3242 :     illegal either. For example, the name basis.cm simply refers to a
3243 :     local file called "basis.cm" (i.e, the name is an ordinary path
3244 :     relative to .cm-files directory). Or, to put it differently, only
3245 :     names that start with $ are anchored paths.
3246 : blume 666
3247 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3248 :     vanished.
3249 :    
3250 :     John (Reppy) had critizised this as soon as I originally proposed and
3251 : blume 666 implemented it, but at that time I did not really deeply believe
3252 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3253 :     in another place where it cannot be seen as an abbreviation for
3254 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3255 :     seems to mean (i.e., it "expands" into the corresponding anchor
3256 :     value).
3257 :    
3258 :     However, when paths are used as members in CM description files, it
3259 :     continues to be true that there must be at least another arc after the
3260 :     anchor. This is now enforced separately during semantic analysis
3261 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3262 : blume 666
3263 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3264 :     is a sub-option list of precisely two items -- one labeled "anchor"
3265 :     and the other one labeled "value". As you might expect, "anchor" is
3266 :     used to specify an anchor name to be bound, and "value" specifies what
3267 :     the anchor is being bound to.
3268 :    
3269 :     The value must be a directory name and can be given in either standard
3270 :     syntax (including the possibility that it is itself an anchored path)
3271 :     or native syntax.
3272 :    
3273 :     Examples:
3274 :    
3275 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3276 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3277 :    
3278 :     and so on.
3279 :    
3280 :     The meaning of this is that the .cm-file will be processed with an
3281 :     augmented anchor environment where the given anchor(s) is/are bound to
3282 :     the given values(s).
3283 :    
3284 :     The rationale for having this feature is this: Suppose you are trying
3285 :     to use two different (already stable) libraries a.cm and b.cm (that
3286 :     you perhaps didn't write yourself). Further, suppose each of these
3287 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3288 :     Normally you would now have a problem because the anchor "lib" can not
3289 :     be bound to more than one value globally. Therefore, the project that
3290 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3291 :     place:
3292 :    
3293 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3294 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3295 :    
3296 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3297 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3298 :    
3299 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3300 :     you when you do so at the time of CM.stabilize). Therefore, you can
3301 :     also use an anchored path as the value:
3302 :    
3303 :     a.cm (bind:(anchor:lib value:$a-lib))
3304 :     b.cm (bind:(anchor:lib value:$b-lib))
3305 :    
3306 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3307 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3308 :     itself is always locally bound, setting it globally is no longer
3309 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3310 :     can still be used as a global anchor for separate purposes. As a
3311 :     matter of fact, one can locally define "lib" in terms of a global
3312 :     "lib":
3313 :    
3314 :     a.cm (bind:(anchor:lib value:$lib/a))
3315 :     b.cm (bind:(anchor:lib value:$lib/b))
3316 :    
3317 :     ** 3: The encoding of path names has changed. This affects the way
3318 :     path names are shown in CM's progress report and also the internal
3319 :     protocol encoding used for parallel make.
3320 :    
3321 :     The encoding now uses one or more ':'-separated segments. Each
3322 :     segments corresponds to a file that has been specified relative to the
3323 :     file given by its preceding segment. The first segment is either
3324 :     relative to the CWD, absolute, or anchored. Each segment itself is
3325 :     basically a Unix pathname; all segments but the first are relative.
3326 :    
3327 :     Example:
3328 :    
3329 :     $foo/bar/baz.cm:a/b/c.sml
3330 :    
3331 :     This path denotes the file bar/a/b/c.sml relative to the directory
3332 :     denoted by anchor "foo". Notice that the encoding also includes
3333 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3334 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3335 :     be ignored to get the "real" pathname.
3336 :    
3337 :     To make this fact more obvious, CM puts the names of such "virtual
3338 :     arcs" into parentheses when they appear in progress reports. (No
3339 :     parentheses will appear in the internal protocol encoding.) Thus,
3340 :     what you really see is:
3341 :    
3342 :     $foo/bar/(baz.cm):a/b/c.sml
3343 :    
3344 :     I find this notation to be much more informative than before.
3345 :    
3346 :     Another new feature of the encoding is that special characters
3347 :     including parentheses, colons, (back)slashes, and white space are
3348 :     written as \ddd (where ddd is the decimal encoding of the character).
3349 :    
3350 : blume 670 *** The CM manual still needs to be updated.
3351 :    
3352 : blume 666 ----------------------------------------------------------------------
3353 : leunga 660 Name: Allen Leung
3354 :     Date: 2000/06/15 00:38:00
3355 :     Tag: leunga-20000615-x86-peephole
3356 :    
3357 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3358 :    
3359 :     ----------------------------------------------------------------------
3360 : blume 658 Name: Matthias Blume
3361 :     Date: 2000/06/12 11:40:00
3362 :     Tag: blume-20000612-parmakefix
3363 :     Description:
3364 :    
3365 :     More cleanup after changing the file naming scheme: This time I
3366 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3367 :    
3368 :     ----------------------------------------------------------------------
3369 : leunga 657 Name: Allen Leung
3370 :     Date: 2000/06/09 01:25:00
3371 :     Tag: leunga-20000609-various
3372 :    
3373 :     None of these things should affect normal SML/NJ operations
3374 :    
3375 :     1. Peephole improvements provided by Fermin (c--)
3376 :     2. New annotation DEFUSE for adding extra dependence (moby)
3377 :     3. New X86 LOCK instructions (moby)
3378 :     4. New machine description language for reservation tables (scheduling)
3379 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3380 :     trees etc.)
3381 :     6. I've changed the CM files so that they can work with versions
3382 :     110.0.6, 110.25 and 110.28
3383 :    
3384 :     ----------------------------------------------------------------------
3385 : blume 652 Name: Matthias Blume
3386 : blume 656 Date: 2000/06/09 12:40:00
3387 :     Tag: blume-20000609-log
3388 :     Description:
3389 :    
3390 :     - Removed all(?) remaining RCS Log entries from sources.
3391 :    
3392 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3393 :     anchored paths).
3394 :    
3395 :     ----------------------------------------------------------------------
3396 :     Name: Matthias Blume
3397 : blume 653 Date: 2000/06/07 17:00:00 JST
3398 :     Tag: blume-20000607-no-implicit-anchors
3399 :     Description:
3400 :    
3401 :     1. This update changes the default setting for
3402 :     CM.Control.implicit_anchors from true to false. This means that
3403 :     implicit anchors are no longer permitted by default. I also tried to
3404 :     make sure that nothing else still relies on implicit anchors.
3405 :     (This is the next step on the schedule towards a CM that does not even
3406 :     have the notion of implicit anchors anymore.)
3407 :    
3408 :     2. More CM manual updates.
3409 :    
3410 :     3. I managed to track down and fix the pickling bug I mentioned last
3411 :     time. Because of the previously existing workaround, this entails no
3412 :     immediate practical changes.
3413 :    
3414 :     ----------------------------------------------------------------------
3415 :     Name: Matthias Blume
3416 : blume 652 Date: 2000/06/06 11:15:00 JST
3417 :     Tag: blume-20000606-lazierpickle
3418 :     Description:
3419 :    
3420 :     !!!! NEW BOOT FILES !!!!
3421 :    
3422 :     * The main purpose of this update is to make library pickles lazier in
3423 :     order to reduce the initial space penalty for autoloading a library.
3424 :     As a result, it is now possible to have $smlnj/compiler.cm
3425 :     pre-registered. This should take care of the many complaints or
3426 :     inquiries about missing structure Compiler. This required changes to
3427 :     CM's internal data structures and small tweaks to some algorithms.
3428 :    
3429 :     As a neat additional effect, it is no longer necessary (for the sake
3430 :     of lean heap image files) to distinguish between a "minimal" CM and a
3431 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3432 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3433 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3434 :    
3435 :     To make the life of compiler-hackers easier, "makeml" now also
3436 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3437 :     words, after you bootstrap a new sml for the first time, you will not
3438 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3439 :     around you will still have to do it, though.)
3440 :    
3441 :     * A second change consists of major updates to the CM manual. There
3442 :     are now several appendices with summary information and also a full
3443 :     specification of the CM description file syntax.
3444 :    
3445 :     * In directory src/system I added the script "allcross". This script
3446 :     invokes sml and cross-compiles the compiler for all supported
3447 :     architectures. (Useful when providing a new set of boot files.)
3448 :    
3449 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3450 :     added a small tweak to pickle-util.sml to work around this problem,
3451 :     but it is not a proper fix yet. I will investigate further. (The
3452 :     effect of the bug was an inflation of library pickle size.)
3453 :    
3454 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3455 :    
3456 :     ----------------------------------------------------------------------
3457 : leunga 646 Name: Allen Leung
3458 : leunga 648 Date: 2000/05/25 17:28 EDT
3459 :     Tag: leunga-20000525-ra
3460 :     Description:
3461 :    
3462 :     Fixed a bug in freezing phase of the register allocator.
3463 :    
3464 :     ----------------------------------------------------------------------
3465 :     Name: Allen Leung
3466 : leunga 647 Date: 2000/05/15 22:53 EDT
3467 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3468 :     Description:
3469 :    
3470 :     1. Alpha
3471 :    
3472 :     Slight cleanup. Removed the instruction SGNXL
3473 :    
3474 :     2. X86
3475 :    
3476 :     Added the following instructions to the instruction set:
3477 :    
3478 :     ROLx, RORx,
3479 :     BTx, BTSx, BTLx, BTRx,
3480 :     XCHGx, and variants with the LOCK prefix
3481 :    
3482 :     3. Register Allocation
3483 :    
3484 :     The module ra-rewrite-with-renaming has been improved.
3485 :    
3486 : leunga 647 These have no effect on SML/NJ.
3487 :    
3488 : leunga 646 ----------------------------------------------------------------------
3489 : blume 642 Name: Matthias Blume
3490 : blume 645 Date: 2000/05/15 16:20:00 JST
3491 :     Tag: blume-20000515-lightrebuild
3492 :     Description:
3493 :    
3494 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
3495 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
3496 :     defined. In effect, the command will not build any cross-compiler
3497 :     backends and therefore finish more quickly.
3498 :    
3499 :     The "fixpt" script also takes a "-light" switch to be able to use
3500 :     this new facility while compiling for a fixpoint.
3501 :    
3502 :     2. I replaced all mentions of anchored paths in group owner specifications
3503 :     with simple relative paths (usually starting with "..").
3504 :     The rationale is that a library's internal workings should not be
3505 :     compromised by the lack of some anchor. (An anchor is necessary
3506 :     for someone who wants to refer to the library by an anchored path,
3507 :     but it should not be necessary to build the same library in the first
3508 :     place.)
3509 :    
3510 :     3. I changed the way CM's tool mechanism determines the shell command
3511 :     string used for things like ml-yacc etc. so that it does not break
3512 :     when CM.Control.implicit_anchors is turned off.
3513 :    
3514 :     ----------------------------------------------------------------------
3515 :     Name: Matthias Blume
3516 : blume 644 Date: 2000/05/12 18:20:00 JST
3517 :     Tag: blume-20000512-ml-build
3518 :     Description:
3519 :    
3520 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
3521 :     getting installed properly (by config/install.sh).
3522 :    
3523 :     ----------------------------------------------------------------------
3524 :     Name: Matthias Blume
3525 : blume 643 Date: 2000/05/12 17:30:00 JST
3526 :     Tag: blume-20000512-anchors
3527 :     Description:
3528 :    
3529 :     !!! NEW BOOT FILES !!!
3530 :    
3531 :     This change is in preparation of fading out support for "implicitly
3532 :     anchored path names". I went through all sources and used the
3533 :     explicit (and relatively new) $-notation. See system/README and the
3534 :     CM manual for more info on this.
3535 :    
3536 :     I also modified the anchoring scheme for some things such as "smlnj",
3537 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
3538 :     anchors are more expressive: anchor name and first arc do not have to
3539 :     coincide. This entails the following user-visible change:
3540 :    
3541 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
3542 :     particular, when you fire up sml with a command-line argument, say,
3543 :     e.g.:
3544 :    
3545 :     sml '$smlnj/cmb.cm'
3546 :    
3547 :     At the ML toplevel prompt:
3548 :    
3549 :     CM.autoload "$smlnj/cmb.cm";
3550 :    
3551 :     There is also a new controller in CM.Control that can be used to turn
3552 :     off all remaining support for implicit anchors by saying:
3553 :    
3554 :     CM.autoload "$smlnj/
3555 :     #set CM.Control.implicit_anchors false;
3556 :    
3557 :     This causes CM to reject implicitly anchored paths. This is (for the
3558 :     time being) less permissive than the "final" version where there will
3559 :     be no more such implicit anchors and relative paths will be just that:
3560 :     relative.
3561 :    
3562 :     The next step (version after next version?) will be to make the
3563 :     default for CM.Control.implicit_anchors false. After the dust has
3564 :     settled, I can then produce the "final" version of this...
3565 :    
3566 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
3567 :    
3568 :     ----------------------------------------------------------------------
3569 :     Name: Matthias Blume
3570 : blume 642 Date: 2000/05/11 16:30:00 JST
3571 :     Tag: blume-20000511-sources
3572 :     Description:
3573 :    
3574 :     The main change is that I added function CM.sources as a generalized
3575 :     version of the earlier CM.makedepend. This entails the following
3576 :     additional changes:
3577 :    
3578 :     - CM.makedepend has been dropped.
3579 :    
3580 :     - CM manual has been updated.
3581 :    
3582 :     - TOOLS signature and API have been changed.
3583 :    
3584 :     ----------------------------------------------------------------------
3585 : leunga 640 Name: Allen Leung
3586 : leunga 641 Date: 2000/05/10 21:17 EDT
3587 :     Tag: leunga-20000510-moby-c--ssa
3588 :     Description:
3589 :    
3590 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
3591 :     None of these affect SML/NJ.
3592 :    
3593 :     1. Register Allocation
3594 :    
3595 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
3596 :     This module tries to remove local (i.e. basic block level) redundancies
3597 :     during spilling.
3598 :    
3599 :     b. A new framework for performing region based register allocation.
3600 :     Not yet entirely functional.
3601 :    
3602 :     2. X86
3603 :    
3604 :     a. DefUse for POP was missing the stack pointer [found by Lal]
3605 :     b. Reload for CALL was incorrect in X86Spill [found by John]
3606 :     c. Various fixes in X86Spill so that it can be used correctly for
3607 :     the new spilling module.
3608 :    
3609 :     3. SSA/IR
3610 :    
3611 :     a. New module ir/dj-dataflow.sml implements elimination based
3612 :     data flow analysis.
3613 :    
3614 :     4. MLRiscGen
3615 :    
3616 :     a. Fix for gc type annotation
3617 :    
3618 :     5. MDGen
3619 :    
3620 :     Various fixes for machine description -> ml code translation. For ssa
3621 :     only.
3622 :    
3623 :     ----------------------------------------------------------------------
3624 :     Name: Allen Leung
3625 : leunga 640 Date: 2000/05/08 22:17 EDT
3626 :     Tag: leunga-20000508-labexp
3627 :     Description:
3628 :    
3629 :     Fermin has found a few assembly problems with constant expressions
3630 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
3631 :     which choke on dumb assemblers. This is his fix.
3632 :    
3633 :     ----------------------------------------------------------------------
3634 : dbm 635 Name: Dave MacQueen
3635 :     Date: 2000/04/09 14:00 EDT
3636 :     Tag: dbm-20000502-Version_110_28
3637 :     Description:
3638 :    
3639 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
3640 :    
3641 :     2. Updated config/version to 110.28
3642 :    
3643 :     3. Updated config/srcarchiveurl
3644 :    
3645 :     3. New boot files!
3646 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
3647 :    
3648 :     ----------------------------------------------------------------------
3649 : blume 634 Name: Matthias Blume
3650 :     Date: 2000/05/01 19:05:00 JST
3651 :     Tag: blume-20000501-noweb
3652 :     Description:
3653 :    
3654 :     A new noweb tool has been added. The existing system is entirely
3655 :     unaffected by this, but some CM users have asked for renewed noweb
3656 :     support. Everything is documented in the CM manual.
3657 :    
3658 :     New (plugin) libraries:
3659 :    
3660 :     noweb-tool.cm
3661 :     nw-ext.cm
3662 :    
3663 :     ----------------------------------------------------------------------
3664 : dbm 633 Name: Dave MacQueen
3665 :     Date: 2000/04/30 12:40PM EDT
3666 :     Tag: dbm-20000430-bug_fixes
3667 :     Description:
3668 :    
3669 :     1. Fix for bug 1498
3670 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
3671 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
3672 :     added toRealArray function
3673 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
3674 :     added check for tag Obj.RealArray to array printing case in ppObj
3675 :    
3676 :     2. Fix for bug 1510
3677 :     smlnj/src/compiler/Semant/types/typesutil.sml
3678 :     fixed definition of dummyargs (used by equalTycon) so that
3679 :     dummy args are distinct types
3680 :    
3681 :     ----------------------------------------------------------------------
3682 : blume 629 Name: Matthias Blume
3683 : blume 632 Date: 2000/04/30 01:00:00 JST
3684 :     Tag: blume-20000430-versions
3685 :     Description:
3686 :    
3687 :     1. CM version numbering added. This is an implementation of Lal's
3688 :     proposal for adding version numbers and version checking to .cm
3689 :     files. Lal said that his proposal was just that -- a proposal.
3690 :     For the time being I went ahead and implemented it so that people
3691 :     can comment on it. Everything is completely backward-compatible
3692 :     (except for the stable library format, i.e., new bootfiles!).
3693 :    
3694 :     As usual, see the CM manual for details.
3695 :    
3696 :     2. An alternative syntax for anchored paths has been implemented.
3697 :     Dave has recently voiced the same concerns that I had when I did
3698 :     this, so there should be some support. My take is that eventually
3699 :     I will let support for the current syntax (where anchors are
3700 :     "implicit") fade out in favor of the new, explicit syntax.
3701 :     In order to be backward-compatible, both old and new syntax are
3702 :     currently supported.
3703 :    
3704 :     Again, see the CM manual for details.
3705 :    
3706 :     3. Parallel make is trying to be slightly smarter: When the master
3707 :     process finds a "bottleneck", i.e., when there is only one
3708 :     compilation unit that can be compiled and everybody else is
3709 :     waiting on it, then it will simply compile it directly instead
3710 :     of clumsily telling one of the slaves to do it.
3711 :    
3712 :     4. Support for "unsharing" added. This is necessary in order to be
3713 :     able to have two different versions of the same library running
3714 :     at the same time (e.g., for trying out a new MLRISC while still
3715 :     having the old MLRISC linked into the current compiler, etc.)
3716 :     See the CM manual.
3717 :    
3718 :     5. Simple "makedepend" functionality added for generating Makefile
3719 :     dependency information. (This is rather crude at the moment.
3720 :     Expect some changes here in the future.)
3721 :    
3722 :     6. ".fun" added as a recognized suffix for ML files. Also documented
3723 :     explicitly in the manual that the fallback behavior (unknown suffix
3724 :     -> ML file) is not an official feature!
3725 :    
3726 :     7. Small changes to the pickler for stable libraries.
3727 :    
3728 :     8. Several internal changes to CM (for cleanup/improvement).
3729 :    
3730 :    
3731 :     !!!! NEW BINFILES !!!!
3732 :    
3733 :     ----------------------------------------------------------------------
3734 :     Name: Matthias Blume
3735 : blume 631 Date: 2000/04/28 17:30:00 JST
3736 :     Tag: blume-20000428-pathconfig
3737 :     Description:
3738 :    
3739 :     1. I changed config/install.sh to remove duplicate entries from the
3740 :     lib/pathconfig file at the end. Moreover, the final version of
3741 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
3742 :     in src/system/installml.
3743 :    
3744 :     2. The config/install.sh script now consistently uses relative
3745 :     pathnames in lib/pathconfig whenever the anchor is in the lib
3746 :     directory. (So far this was true for the libraries that come
3747 :     pre-compiled and bundled as part of the bootfiles but not for
3748 :     libraries that are compiled by the script itself.)
3749 :    
3750 :     ----------------------------------------------------------------------
3751 :     Name: Matthias Blume
3752 : blume 629 Date: 2000/04/26 13:10:00 JST
3753 :     Tag: blume-20000426-fun_suffix
3754 :     Description:
3755 :    
3756 :     Added ".fun" as a recognized file name suffix (for ML code).
3757 :    
3758 :     ----------------------------------------------------------------------
3759 : leunga 628 Name: Allen Leung
3760 :     Date: 2000/04/25 17:00:00 EST
3761 :     Tag: leunga-20000425-alpha-ra
3762 :     Description:
3763 :    
3764 :     1. Alpha
3765 :    
3766 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
3767 :     in C--.
3768 :    
3769 :     2. RA
3770 :    
3771 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
3772 :     and the interference graph.
3773 :    
3774 :     ----------------------------------------------------------------------
3775 : dbm 627 Name: Dave MacQueen
3776 :     Date: 2000/04/25/
3777 :     Tag: dbm-20000425-mlyacc_doc_examples
3778 :     Description:
3779 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
3780 :     Updated all three examples in src/ml-yacc/examples to run
3781 :     under 110.* using CM.make.
3782 :    
3783 :     ----------------------------------------------------------------------
3784 : leunga 624 Name: Allen Leung
3785 :     Date: 2000/04/20 23:04:00 EST
3786 :     Tag: leunga-20000420-ssa-c---stuff
3787 :     Description:
3788 :    
3789 :     This update synchronizes my repository with Yale's. Most of these
3790 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
3791 :    
3792 :     1. Register Allocator
3793 :    
3794 :     a. An improvement in the interference graph construction:
3795 :     Given a copy
3796 :    
3797 :     s <- t
3798 :    
3799 : leunga 628 no interference edge between s and t is added for this definition of s.
3800 : leunga 624
3801 :     b. I've added two new spill heuristic modules that Fermin and I developed
3802 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
3803 :     useful for others (Moby?)
3804 :    
3805 :     2. X86
3806 :    
3807 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
3808 :    
3809 :     3. Alpha
3810 :    
3811 :     a. Added the BSR instruction and code generation that goes with it [C--]
3812 :     b. Other fixes too numerous to recount provided by Fermin [C--]
3813 :    
3814 :     4. Regmaps
3815 :    
3816 :     a. The regmaps are not initialized with the identity physical bindings
3817 :     at creation time. This is unneeded.
3818 :    
3819 :     5. MLRISC Optimizations
3820 :    
3821 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
3822 :     intersects with liveness incrementally in linear time! Woohoo!
3823 :     This is now used in my new SSA construction algorithm.
3824 :    
3825 :     b. THe branch reorganization module is now smarter about linear chains of
3826 :     basic blocks.
3827 :    
3828 :    
3829 :     ----------------------------------------------------------------------
3830 : blume 615 Name: Matthias Blume
3831 :     Date: 2000/04/12 13:52:00 JST
3832 :     Tag: blume_main_v110p27_1
3833 :     Description:
3834 :    
3835 :     Changed install.sh script to handle archive files without version number
3836 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
3837 :     name of the boot file archive.
3838 :    
3839 :     ----------------------------------------------------------------------
3840 : dbm 607 Name: Dave MacQueen
3841 :     Date: 2000/04/09 14:00 EDT
3842 :     Tag: dbm-20000410-Version_110_27
3843 :     Description:
3844 :    
3845 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
3846 :    
3847 :     2. Updated src/config/version to 110.27
3848 :    
3849 :     3. New boot files!
3850 :    
3851 :     ----------------------------------------------------------------------
3852 : leunga 606 Name: Allen Leung
3853 :     Date: 2000/04/09 19:09:00 EST
3854 :     Tag: leunga-20000409-misc
3855 :     Description:
3856 :    
3857 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
3858 :    
3859 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
3860 :    
3861 :     ----------------------------------------------------------------------
3862 : monnier 605 Name: Stefan
3863 :     Date: 2000/04/07 10:00:00 EDT
3864 :     Tag: monnier-20000406-branch-handling
3865 :     Description:
3866 :    
3867 :     Improved handling of branches (mostly those generated from
3868 :     polymorphic equality), removed switchoff and changed the
3869 :     default optimization settings (more cpsopt and less flintopt).
3870 :    
3871 :     ----------------------------------------------------------------------
3872 : leunga 601 Name: Allen Leung
3873 : leunga 602 Date: 2000/04/06 01:30:00 EST
3874 :     Tag: leunga-20000406-peephole-x86-SSA-2
3875 :     Description:
3876 :    
3877 :     Forgot a few files.
3878 :    
3879 :     ----------------------------------------------------------------------
3880 :     Name: Allen Leung
3881 : leunga 601 Date: 2000/04/06 00:36:00 EST
3882 :     Tag: leunga-20000406-peephole-x86-SSA
3883 :     Description:
3884 :    
3885 :     1. New Peephole code
3886 :    
3887 :     2. Minor improvement to X86 instruction selection
3888 :    
3889 :     3. Various fixes to SSA and machine description -> code translator
3890 :    
3891 :     ----------------------------------------------------------------------
3892 : blume 595 Name: Matthias Blume
3893 :     Date: 2000/04/05 12:30:00 JST
3894 :     Tag: blume_main_v110p26p2_3
3895 :     Description:
3896 :    
3897 :     This update just merges three minor cosmetic updates to CM's sources
3898 :     to get ready for the 110.27 code freeze on Friday. No functionality
3899 :     has changed.
3900 :    
3901 :     ----------------------------------------------------------------------
3902 : leunga 593 Name: Allen Leung
3903 : leunga 594 Date: 2000/04/04 19:39:00 EST
3904 :     Tag: leunga-20000404-x86-asm
3905 :     Description:
3906 :    
3907 :     1. Fixed a problem in X86 assembly.
3908 :    
3909 :     Things like
3910 :    
3911 :     jmp %eax
3912 :     jmp (%eax)
3913 :    
3914 :     should be output as
3915 :    
3916 :     jmp *%eax
3917 :     jmp *(%eax)
3918 :    
3919 :     2. Assembly output
3920 :    
3921 :     Added a new flag
3922 :    
3923 :     "asm-indent-copies" (default to false)
3924 :    
3925 :     When this flag is on, parallel copies will be indented an extra level.
3926 :    
3927 :     ----------------------------------------------------------------------
3928 :     Name: Allen Leung
3929 : leunga 593 Date: 2000/04/04 03:18:00 EST
3930 :     Tag: leunga-20000404-C--Moby
3931 :     Description:
3932 :    
3933 :     All of these fixes are related to C--, Moby, and my own optimization
3934 :     stuff; so they shouldn't affect SML/NJ.
3935 :    
3936 :     1. X86
3937 :    
3938 :     Various fixes related floating point, and extensions.
3939 :    
3940 :     2. Alpha
3941 :    
3942 :     Some extra patterns related to loads with signed/zero extension
3943 :     provided by Fermin.
3944 :    
3945 :     3. Assembly
3946 :    
3947 : monnier 659 When generating assembly, resolve the value of client defined constants,
3948 : leunga 593 instead of generating symbolic values. This is controlled by the
3949 :     new flag "asm-resolve-constants", which is default to true.
3950 :    
3951 :     4. Machine Descriptions
3952 :    
3953 :     a. The precedence parser was slightly broken when parsing infixr symbols.
3954 :     b. The type generalizing code had the bound variables reversed, resulting
3955 :     in a problem during arity raising.
3956 :     c. Various fixes in machine descriptions.
3957 :    
3958 :     ----------------------------------------------------------------------
3959 : blume 592 Name: Matthias Blume
3960 :     Date: 2000/04/03 16:05:00 JST
3961 :     Tag: blume_main_v110p26p2_2
3962 :     Description:
3963 :    
3964 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
3965 :     now done via the ordinary static environment that is context to each
3966 :     compilation unit.
3967 :    
3968 :     To this end, I arranged that instead of "structure Core" as "structure
3969 :     _Core" is bound in the pervasive environment. Core access is done via
3970 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
3971 : blume 592 legal surface-syntax symbol).
3972 :    
3973 :     The current solution is much cleaner because the core environment is
3974 :     now simply part of the pervasive environment which is part of every
3975 :     compilation unit's context anyway. In particular, this eliminates all
3976 :     special-case handling that was necessary until now in order to deal
3977 :     with dynamic and symbolic parts of the core environment.
3978 :    
3979 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3980 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
3981 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
3982 :     more details.
3983 :    
3984 :     I also tried to track down all mentions of "Core" (as string argument
3985 :     to Symbol.strSymbol) in the compiler and replaced them with a
3986 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
3987 :     name appears in one place only.
3988 :    
3989 :     Binfile and bootfile format have not changed, but the switchover from
3990 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
3991 :     bootfiles anyway.
3992 :    
3993 :     ----------------------------------------------------------------------
3994 : leunga 591 Name: Allen Leung
3995 :     Date: 2000/04/02 21:17:00 EST
3996 :     Tag: leunga-20000402-mltree
3997 :     Description:
3998 : leunga 585
3999 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
4000 :     2. Added a bunch of files from my repository. These are currently
4001 :     used by other non-SMLNJ backends.
4002 :    
4003 : leunga 576 ----------------------------------------------------------------------
4004 : leunga 590 Name: Allen Leung
4005 :     Date: 2000/03/31 21:15:00 EST
4006 :     Tag: leunga-20000331-aliasing
4007 :     Description:
4008 :    
4009 :     This update contains a rewritten (and hopefully more correct) module
4010 :     for extracting aliasing information from CPS.
4011 :    
4012 :     To turn on this feature:
4013 :    
4014 :     Compiler.Control.CG.memDisambiguate := true
4015 :    
4016 :     To pretty print the region information with assembly
4017 :    
4018 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
4019 :    
4020 :     To control how many levels of aliasing information are printed, use:
4021 :    
4022 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
4023 :    
4024 :     The default of n is 3.
4025 :    
4026 :     ----------------------------------------------------------------------
4027 : dbm 589 Name: David MacQueen
4028 :     Date: 2000/03/31 11:15:00 EST
4029 :     Tag: dbm-20000331-runtime_fix
4030 :     Description:
4031 :    
4032 :     This update contains:
4033 :    
4034 :     1. runtime/c-lib/c-libraries.c
4035 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
4036 :    
4037 :     2. fix for bug 1556
4038 :     system/Basis/Implementation/NJ/internal-signals.sml
4039 :    
4040 :     ----------------------------------------------------------------------
4041 : blume 587 Name: Matthias Blume
4042 : blume 588 Date: 2000/03/31 18:00:00 JST
4043 :     Tag: blume_main_v110p26p2_1
4044 :     Description:
4045 :    
4046 :     This update contains:
4047 :    
4048 :     1. A small change to CM's handling of stable libraries:
4049 :     CM now maintains one "global" modmap that is used for all stable
4050 :     libraries. The use of such a global modmap maximizes sharing and
4051 :     minimizes the need for re-traversing parts of environments during
4052 :     modmap construction. (However, this has minor impact since modmap
4053 :     construction seems to account for just one percent or less of total
4054 :     compile time.)
4055 :    
4056 :     2. I added a "genmap" phase to the statistics. This is where I got the
4057 :     "one percent" number (see above).
4058 :    
4059 :     3. CM's new tool parameter mechanism just became _even_ better. :)
4060 :     - The parser understands named parameters and recursive options.
4061 :     - The "make" and "shell" tools use these new features.
4062 :     (This makes it a lot easier to cascade these tools.)
4063 :     - There is a small syntax change: named parameters use a
4064 :    
4065 :     <name> : ( <option> ... ) or
4066 :     <name> : <string>
4067 :    
4068 :     syntax. Previously, named parameters were implemented in an
4069 :     ad-hoc fashion by each tool individually (by parsing strings)
4070 :     and had the form
4071 :    
4072 :     <name>=<string>
4073 :    
4074 :     See the CM manual for a full description of these issues.
4075 :    
4076 :     ----------------------------------------------------------------------
4077 :     Name: Matthias Blume
4078 : blume 587 Date: 2000/03/30 18:00:00 JST
4079 :     Tag: blume_main_v110p26p2_0
4080 :     Description:
4081 :    
4082 :     !!!!! WARNING !!!!!!
4083 :     !! New binfiles !!
4084 :     !!!!!!!!!!!!!!!!!!!!
4085 :    
4086 :     This update contains:
4087 :    
4088 :     1. Moderate changes to CM:
4089 :    
4090 :     - Changes to CM's tools mechanism. In particular, it is now possible
4091 :     to have tools that accept additional "command line" parameters
4092 :     (specified in the .cm file at each instance where the tool's class is
4093 :     used).
4094 :    
4095 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
4096 :     facilitate fairly seamless hookup to portions of code managed using
4097 : blume 587 Makefiles or Shell scripts.
4098 :    
4099 :     There are no classes "shared" or "private" anymore. Instead, the
4100 :     sharing annotation is now a parameter to the "sml" class.
4101 :    
4102 :     There is a bit of generic machinery for implementing one's own
4103 :     tools that accept command-line parameters. However, I am not yet fully
4104 :     satisfied with that part, so expect changes here in the future.
4105 :    
4106 :     All existing tools are described in the CM manual.
4107 :    
4108 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
4109 : blume 587 error messages that tended to be more annoying than helpful.)
4110 :    
4111 :     2. Major changes to the compiler's static environment data structures.
4112 :    
4113 :     - no CMStaticEnv anymore.
4114 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
4115 :     but it is called Environment), no conversions between different
4116 :     kinds of static environments
4117 :    
4118 :     - There is still a notion of a "modmap", but such modmaps are generated
4119 :     on demand at the time when they are needed. This sounds slow, but I
4120 :     sped up the code that generates modmaps enough for this not to lead to
4121 :     a slowdown of the compiler (at least I didn't detect any).
4122 :    
4123 :     - To facilitate rapid modmap generation, static environments now
4124 :     contain an (optional) "modtree" structure. Modtree annotations are
4125 :     constructed by the unpickler during unpickling. (This means that
4126 :     the elaborator does not have to worry about modtrees at all.)
4127 :     Modtrees have the advantage that they are compositional in the same
4128 :     way as the environment data structure itself is compositional.
4129 :     As a result, modtrees never hang on to parts of an environment that
4130 :     has already been rendered "stale" by filtering or rebinding.
4131 :    
4132 :     - I went through many, many trials and errors before arriving at the
4133 :     current solution. (The initial idea of "linkpaths" did not work.)
4134 :     But the result of all this is that I have touched a lot of files that
4135 :     depend on the "modules" and "types" data structures (most of the
4136 :     elaborator). There were a lot of changes during my "linkpath" trials
4137 :     that could have been reverted to their original state but weren't.
4138 :     Please, don't be too harsh on me for messing with this code a bit more
4139 : monnier 921 than what was strictly necessary... (I _did_ resist the temptation
4140 : blume 587 of doing any "global reformatting" to avoid an untimely death at
4141 :     Dave's hands. :)
4142 :    
4143 :     - One positive aspect of the previous point: At least I made sure that
4144 :     all files that I touched now compile without warnings (other than
4145 :     "polyEqual").
4146 :    
4147 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
4148 :     redundant modmaps)
4149 :    
4150 :     ----------------------------------------------------------------------
4151 : leunga 580 Name: Allen Leung
4152 : leunga 585 Date: 2000/03/29 18:00:00
4153 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
4154 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
4155 : leunga 585 Description:
4156 :    
4157 :     This update contains *MAJOR* changes to the way code is generated from CPS
4158 :     in the module mlriscGen, and in various backend modules.
4159 :    
4160 :     CHANGES
4161 :     =======
4162 :    
4163 :     1. MLRiscGen: forward propagation fix.
4164 :    
4165 :     There was a bug in forward propagation introduced at about the same time
4166 :     as the MLRISC x86 backend, which prohibits coalescing to be
4167 :     performed effectively in loops.
4168 :    
4169 :     Effect: speed up of loops in RISC architectures.
4170 :     By itself, this actually slowed down certain benchmarks on the x86.
4171 :    
4172 :     2. MLRiscGen: forward propagating addresses from consing.
4173 :    
4174 :     I've changed the way consing code is generated. Basically I separated
4175 :     out the initialization part:
4176 :    
4177 :     store tag, offset(allocptr)
4178 :     store elem1, offset+4(allocptr)
4179 :     store elem2, offset+8(allocptr)
4180 :     ...
4181 :     store elemn, offset+4n(allocptr)
4182 :    
4183 :     and the address computation part:
4184 :    
4185 :     celladdr <- offset+4+alloctpr
4186 :    
4187 :     and move the address computation part
4188 :    
4189 :     Effect: register pressure is generally lower as a result. This
4190 :     makes compilation of certain expressions much faster, such as
4191 :     long lists with non-trivial elements.
4192 :    
4193 :     [(0,0), (0,0), .... (0,0)]
4194 :    
4195 :     3. MLRiscGen: base pointer elimination.
4196 :    
4197 :     As part of the linkage mechanism, we generate the sequence:
4198 :    
4199 :     L: ... <- start of the code fragment
4200 :    
4201 :     L1:
4202 :     base pointer <- linkreg - L1 + L
4203 :    
4204 :     The base pointer was then used for computing relocatable addresses
4205 :     in the code fragment. Frequently (such as in lots of continuations)
4206 :     this is not needed. We now eliminate this sequence whenever possible.
4207 :    
4208 :     For compile time efficiency, I'm using a very stupid local heuristic.
4209 :     But in general, this should be done as a control flow analysis.
4210 :    
4211 :     Effect: Smaller code size. Speed up of most programs.
4212 :    
4213 :     4. Hppa back end
4214 :    
4215 :     Long jumps in span dependence resolution used to depend on the existence
4216 :     of the base pointer.
4217 :    
4218 :     A jump to a long label L was expanded into the following sequence:
4219 :    
4220 :     LDIL %hi(L-8192), %r29
4221 :     LDO %lo(L-8192)(%r29), %r29
4222 :     ADD %r29, baseptr, %r29
4223 :     BV,n %r0(%r29)
4224 :    
4225 :     In the presence of change (3) above, this will not work. I've changed
4226 :     it so that the following sequence of instructions are generated, which
4227 :     doesn't mention the base pointer at all:
4228 :    
4229 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
4230 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
4231 :     BV,n %r0(%r29) /* Jump */
4232 :    
4233 :     5. Alpha back end
4234 :    
4235 :     New alpha instructions LDB/LDW have been added, as per Fermin's
4236 :     suggestions. This is unrelated to all other changes.
4237 :    
4238 :     6. X86 back end
4239 :    
4240 :     I've changed andl to testl in the floating point test sequence
4241 :     whenever appropriate. The Intel optimization guide states that
4242 : monnier 659 testl is preferable to andl.
4243 : leunga 585
4244 :     7. RA (x86 only)
4245 :    
4246 :     I've improved the spill propagation algorithm, using an approximation
4247 :     of maximal weighted independent sets. This seems to be necessary to
4248 :     alleviate the negative effect in light of the slow down in (1).
4249 :    
4250 :     I'll write down the algorithm one of these days.
4251 :    
4252 :     8. MLRiscGen: frequencies
4253 :    
4254 :     I've added an annotation that states that all call gc blocks have zero
4255 :     execution frequencies. This improves register allocation on the x86.
4256 :    
4257 :     BENCHMARKS
4258 :     ==========
4259 :    
4260 :     I've only perform the comparison on 110.25.
4261 :    
4262 :     The platforms are:
4263 :    
4264 :     HPPA A four processor HP machine (E9000) with 5G of memory.
4265 :     X86 A 300Hhz Pentium II with 128M of memory, and
4266 :     SPARC An Ultra sparc 2 with 512M of memory.
4267 :    
4268 :     I used the following parameters for the SML benchmarks:
4269 :    
4270 :     @SMLalloc
4271 :     HPPA 256k
4272 :     SPARC 512k
4273 :     X86 256k
4274 :    
4275 :     COMPILATION TIME
4276 :     ----------------
4277 :     Here are the numbers comparing the compilation times of the compilers.
4278 :     I've only compared 110.25 compiling the new sources versus
4279 :     a fixpoint version of the new compiler compiling the same.
4280 :    
4281 :     110.25 New
4282 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
4283 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
4284 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
4285 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
4286 :    
4287 :     110.25 New
4288 :     Code Size Code Size
4289 :     HPPA 8596736 8561421
4290 :     SPARC 8974299 8785143
4291 :     X86 9029180 8716783
4292 :    
4293 :     So in summary, things are at least as good as before. Dramatic
4294 :     reduction in compilation is obtained on the Sparc; I can't explain it,
4295 :     but it is reproducible. Perhaps someone should try to reproduce this
4296 :     on their own machines.
4297 :    
4298 :     SML BENCHMARKS
4299 :     --------------
4300 :    
4301 :     On the average, all benchmarks perform at least as well as before.
4302 :    
4303 :     HPPA Compilation Time Spill+Reload Run Time
4304 :     110.25 New 110.25 New 110.25 New
4305 :    
4306 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
4307 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
4308 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
4309 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
4310 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
4311 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
4312 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
4313 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
4314 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
4315 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
4316 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
4317 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
4318 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
4319 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
4320 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
4321 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
4322 :     --------------------------------------------------------------------------
4323 :     Average 6.12% 4.09%
4324 :    
4325 :     SPARC Compilation Time Spill+Reload Run Time
4326 :     110.25 New 110.25 New 110.25 New
4327 :    
4328 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
4329 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
4330 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
4331 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
4332 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
4333 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
4334 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
4335 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
4336 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
4337 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
4338 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
4339 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
4340 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
4341 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
4342 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
4343 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
4344 :     ----------------------------------------------------------------------------
4345 :     Average 6.94% 2.64%
4346 :    
4347 :     X86 Compilation Time Spill+Reload Run Time
4348 :     110.25 New 110.25 New 110.25 New
4349 :    
4350 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
4351 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
4352 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
4353 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
4354 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
4355 :