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 666 - (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 :     Date:
12 :     Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 591 ----------------------------------------------------------------------
15 : blume 666 Name: Matthias Blume
16 :     Date: 2000/06/16 17:30:00 JST
17 :     Tag: blume-20000616-anchorenv
18 :     Description:
19 :    
20 :     This patch implements the long anticipated (just kidding :) "anchor
21 :     environment" mechanism. In the course of doing this, I also
22 :     re-implemented CM's internal "SrcPath" module from scratch. The new
23 :     one should be more robust in certain boundary cases. In any case, it
24 :     is a lot cleaner than its predecessor (IMHO).
25 :    
26 :     This time, although there is yet another boot file format change, I
27 :     kept the unpickler backward-compatible. As a result, no new bootfiles
28 :     are necessary and bootstrapping is straightforward. (You cannot read
29 :     new bootfiles into an old system, but the other way around is no
30 :     problem.)
31 :    
32 :     Visible changes:
33 :    
34 :     ** 1. The $<singlearc> abbreviation for $/<singlearc> for anchored
35 :     paths has finally vanished.
36 :    
37 :     John had critizised this as soon as I originally proposed and
38 :     implemented it, but at that time I did not really deeply believe
39 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
40 :     in another place where it cannot be seen as an abbreviation for
41 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
42 :     seems to mean (i.e., it "expands" into the corresponding anchor
43 :     value).
44 :    
45 :     However, when paths are used as members in CM description files, it
46 :     continues to be true that there must be at least another arc after the
47 :     anchor. This is now enforced separately during semantic analysis
48 :     (i.e., form a lexical/syntactical point of view, the notation is ok.)
49 :    
50 :     ** 2. The "cm" class now accepts an option "bind". The option's value
51 :     is a sub-option list of precisely two items -- one labeled "anchor"
52 :     and the other one labeled "value". As you might expect, "anchor" is
53 :     used to specify an anchor name to be bound, and "value" specifies what
54 :     the anchor is being bound to.
55 :    
56 :     The value must be a directory name and can be given in either standard
57 :     syntax (including the possibility that it is itself an anchored path)
58 :     or native syntax.
59 :    
60 :     Examples:
61 :    
62 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
63 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
64 :    
65 :     and so on.
66 :    
67 :     The meaning of this is that the .cm-file will be processed with an
68 :     augmented anchor environment where the given anchor(s) is/are bound to
69 :     the given values(s).
70 :    
71 :     The rationale for having this feature is this: Suppose you are trying
72 :     to use two different (already stable) libraries a.cm and b.cm (that
73 :     you perhaps didn't write yourself). Further, suppose each of these
74 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
75 :     Normally you would now have a problem because the anchor "lib" can not
76 :     be bound to more than one value globally. Therefore, the project that
77 :     uses both a.cm and b.cm must locally redirect the anchor to some other
78 :     place:
79 :    
80 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
81 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
82 :    
83 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
84 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
85 :    
86 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
87 :     you when you do so at the time of CM.stabilize). Therefore, you can
88 :     also use an anchored path as the value:
89 :    
90 :     a.cm (bind:(anchor:lib value:$a-lib))
91 :     b.cm (bind:(anchor:lib value:$b-lib))
92 :    
93 :     Now you can globally configure (using the usual CM.Anchor.anchor or
94 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
95 :     itself is always locally bound, setting it globally is no longer
96 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
97 :     can still be used as a global anchor for separate purposes. As a
98 :     matter of fact, one can locally define "lib" in terms of a global
99 :     "lib":
100 :    
101 :     a.cm (bind:(anchor:lib value:$lib/a))
102 :     b.cm (bind:(anchor:lib value:$lib/b))
103 :    
104 :     ** 3: The encoding of path names has changed. This affects the way
105 :     path names are shown in CM's progress report and also the internal
106 :     protocol encoding used for parallel make.
107 :    
108 :     The encoding now uses one or more ':'-separated segments. Each
109 :     segments corresponds to a file that has been specified relative to the
110 :     file given by its preceding segment. The first segment is either
111 :     relative to the CWD, absolute, or anchored. Each segment itself is
112 :     basically a Unix pathname; all segments but the first are relative.
113 :    
114 :     Example:
115 :    
116 :     $foo/bar/baz.cm:a/b/c.sml
117 :    
118 :     This path denotes the file bar/a/b/c.sml relative to the directory
119 :     denoted by anchor "foo". Notice that the encoding also includes
120 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
121 :     paths are resolved relative to the .cm-files directory, so baz.cm must
122 :     be ignored to get the "real" pathname.
123 :    
124 :     To make this fact more obvious, CM puts the names of such "virtual
125 :     arcs" into parentheses when they appear in progress reports. (No
126 :     parentheses will appear in the internal protocol encoding.) Thus,
127 :     what you really see is:
128 :    
129 :     $foo/bar/(baz.cm):a/b/c.sml
130 :    
131 :     I find this notation to be much more informative than before.
132 :    
133 :     Another new feature of the encoding is that special characters
134 :     including parentheses, colons, (back)slashes, and white space are
135 :     written as \ddd (where ddd is the decimal encoding of the character).
136 :    
137 :     ----------------------------------------------------------------------
138 : leunga 660 Name: Allen Leung
139 :     Date: 2000/06/15 00:38:00
140 :     Tag: leunga-20000615-x86-peephole
141 :    
142 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
143 :    
144 :     ----------------------------------------------------------------------
145 : blume 658 Name: Matthias Blume
146 :     Date: 2000/06/12 11:40:00
147 :     Tag: blume-20000612-parmakefix
148 :     Description:
149 :    
150 :     More cleanup after changing the file naming scheme: This time I
151 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
152 :    
153 :     ----------------------------------------------------------------------
154 : leunga 657 Name: Allen Leung
155 :     Date: 2000/06/09 01:25:00
156 :     Tag: leunga-20000609-various
157 :    
158 :     None of these things should affect normal SML/NJ operations
159 :    
160 :     1. Peephole improvements provided by Fermin (c--)
161 :     2. New annotation DEFUSE for adding extra dependence (moby)
162 :     3. New X86 LOCK instructions (moby)
163 :     4. New machine description language for reservation tables (scheduling)
164 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
165 :     trees etc.)
166 :     6. I've changed the CM files so that they can work with versions
167 :     110.0.6, 110.25 and 110.28
168 :    
169 :     ----------------------------------------------------------------------
170 : blume 652 Name: Matthias Blume
171 : blume 656 Date: 2000/06/09 12:40:00
172 :     Tag: blume-20000609-log
173 :     Description:
174 :    
175 :     - Removed all(?) remaining RCS Log entries from sources.
176 :    
177 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
178 :     anchored paths).
179 :    
180 :     ----------------------------------------------------------------------
181 :     Name: Matthias Blume
182 : blume 653 Date: 2000/06/07 17:00:00 JST
183 :     Tag: blume-20000607-no-implicit-anchors
184 :     Description:
185 :    
186 :     1. This update changes the default setting for
187 :     CM.Control.implicit_anchors from true to false. This means that
188 :     implicit anchors are no longer permitted by default. I also tried to
189 :     make sure that nothing else still relies on implicit anchors.
190 :     (This is the next step on the schedule towards a CM that does not even
191 :     have the notion of implicit anchors anymore.)
192 :    
193 :     2. More CM manual updates.
194 :    
195 :     3. I managed to track down and fix the pickling bug I mentioned last
196 :     time. Because of the previously existing workaround, this entails no
197 :     immediate practical changes.
198 :    
199 :     ----------------------------------------------------------------------
200 :     Name: Matthias Blume
201 : blume 652 Date: 2000/06/06 11:15:00 JST
202 :     Tag: blume-20000606-lazierpickle
203 :     Description:
204 :    
205 :     !!!! NEW BOOT FILES !!!!
206 :    
207 :     * The main purpose of this update is to make library pickles lazier in
208 :     order to reduce the initial space penalty for autoloading a library.
209 :     As a result, it is now possible to have $smlnj/compiler.cm
210 :     pre-registered. This should take care of the many complaints or
211 :     inquiries about missing structure Compiler. This required changes to
212 :     CM's internal data structures and small tweaks to some algorithms.
213 :    
214 :     As a neat additional effect, it is no longer necessary (for the sake
215 :     of lean heap image files) to distinguish between a "minimal" CM and a
216 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
217 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
218 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
219 :    
220 :     To make the life of compiler-hackers easier, "makeml" now also
221 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
222 :     words, after you bootstrap a new sml for the first time, you will not
223 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
224 :     around you will still have to do it, though.)
225 :    
226 :     * A second change consists of major updates to the CM manual. There
227 :     are now several appendices with summary information and also a full
228 :     specification of the CM description file syntax.
229 :    
230 :     * In directory src/system I added the script "allcross". This script
231 :     invokes sml and cross-compiles the compiler for all supported
232 :     architectures. (Useful when providing a new set of boot files.)
233 :    
234 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
235 :     added a small tweak to pickle-util.sml to work around this problem,
236 :     but it is not a proper fix yet. I will investigate further. (The
237 :     effect of the bug was an inflation of library pickle size.)
238 :    
239 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
240 :    
241 :     ----------------------------------------------------------------------
242 : leunga 646 Name: Allen Leung
243 : leunga 648 Date: 2000/05/25 17:28 EDT
244 :     Tag: leunga-20000525-ra
245 :     Description:
246 :    
247 :     Fixed a bug in freezing phase of the register allocator.
248 :    
249 :     ----------------------------------------------------------------------
250 :     Name: Allen Leung
251 : leunga 647 Date: 2000/05/15 22:53 EDT
252 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
253 :     Description:
254 :    
255 :     1. Alpha
256 :    
257 :     Slight cleanup. Removed the instruction SGNXL
258 :    
259 :     2. X86
260 :    
261 :     Added the following instructions to the instruction set:
262 :    
263 :     ROLx, RORx,
264 :     BTx, BTSx, BTLx, BTRx,
265 :     XCHGx, and variants with the LOCK prefix
266 :    
267 :     3. Register Allocation
268 :    
269 :     The module ra-rewrite-with-renaming has been improved.
270 :    
271 : leunga 647 These have no effect on SML/NJ.
272 :    
273 : leunga 646 ----------------------------------------------------------------------
274 : blume 642 Name: Matthias Blume
275 : blume 645 Date: 2000/05/15 16:20:00 JST
276 :     Tag: blume-20000515-lightrebuild
277 :     Description:
278 :    
279 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
280 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
281 :     defined. In effect, the command will not build any cross-compiler
282 :     backends and therefore finish more quickly.
283 :    
284 :     The "fixpt" script also takes a "-light" switch to be able to use
285 :     this new facility while compiling for a fixpoint.
286 :    
287 :     2. I replaced all mentions of anchored paths in group owner specifications
288 :     with simple relative paths (usually starting with "..").
289 :     The rationale is that a library's internal workings should not be
290 :     compromised by the lack of some anchor. (An anchor is necessary
291 :     for someone who wants to refer to the library by an anchored path,
292 :     but it should not be necessary to build the same library in the first
293 :     place.)
294 :    
295 :     3. I changed the way CM's tool mechanism determines the shell command
296 :     string used for things like ml-yacc etc. so that it does not break
297 :     when CM.Control.implicit_anchors is turned off.
298 :    
299 :     ----------------------------------------------------------------------
300 :     Name: Matthias Blume
301 : blume 644 Date: 2000/05/12 18:20:00 JST
302 :     Tag: blume-20000512-ml-build
303 :     Description:
304 :    
305 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
306 :     getting installed properly (by config/install.sh).
307 :    
308 :     ----------------------------------------------------------------------
309 :     Name: Matthias Blume
310 : blume 643 Date: 2000/05/12 17:30:00 JST
311 :     Tag: blume-20000512-anchors
312 :     Description:
313 :    
314 :     !!! NEW BOOT FILES !!!
315 :    
316 :     This change is in preparation of fading out support for "implicitly
317 :     anchored path names". I went through all sources and used the
318 :     explicit (and relatively new) $-notation. See system/README and the
319 :     CM manual for more info on this.
320 :    
321 :     I also modified the anchoring scheme for some things such as "smlnj",
322 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
323 :     anchors are more expressive: anchor name and first arc do not have to
324 :     coincide. This entails the following user-visible change:
325 :    
326 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
327 :     particular, when you fire up sml with a command-line argument, say,
328 :     e.g.:
329 :    
330 :     sml '$smlnj/cmb.cm'
331 :    
332 :     At the ML toplevel prompt:
333 :    
334 :     CM.autoload "$smlnj/cmb.cm";
335 :    
336 :     There is also a new controller in CM.Control that can be used to turn
337 :     off all remaining support for implicit anchors by saying:
338 :    
339 :     CM.autoload "$smlnj/
340 :     #set CM.Control.implicit_anchors false;
341 :    
342 :     This causes CM to reject implicitly anchored paths. This is (for the
343 :     time being) less permissive than the "final" version where there will
344 :     be no more such implicit anchors and relative paths will be just that:
345 :     relative.
346 :    
347 :     The next step (version after next version?) will be to make the
348 :     default for CM.Control.implicit_anchors false. After the dust has
349 :     settled, I can then produce the "final" version of this...
350 :    
351 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
352 :    
353 :     ----------------------------------------------------------------------
354 :     Name: Matthias Blume
355 : blume 642 Date: 2000/05/11 16:30:00 JST
356 :     Tag: blume-20000511-sources
357 :     Description:
358 :    
359 :     The main change is that I added function CM.sources as a generalized
360 :     version of the earlier CM.makedepend. This entails the following
361 :     additional changes:
362 :    
363 :     - CM.makedepend has been dropped.
364 :    
365 :     - CM manual has been updated.
366 :    
367 :     - TOOLS signature and API have been changed.
368 :    
369 :     ----------------------------------------------------------------------
370 : leunga 640 Name: Allen Leung
371 : leunga 641 Date: 2000/05/10 21:17 EDT
372 :     Tag: leunga-20000510-moby-c--ssa
373 :     Description:
374 :    
375 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
376 :     None of these affect SML/NJ.
377 :    
378 :     1. Register Allocation
379 :    
380 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
381 :     This module tries to remove local (i.e. basic block level) redundancies
382 :     during spilling.
383 :    
384 :     b. A new framework for performing region based register allocation.
385 :     Not yet entirely functional.
386 :    
387 :     2. X86
388 :    
389 :     a. DefUse for POP was missing the stack pointer [found by Lal]
390 :     b. Reload for CALL was incorrect in X86Spill [found by John]
391 :     c. Various fixes in X86Spill so that it can be used correctly for
392 :     the new spilling module.
393 :    
394 :     3. SSA/IR
395 :    
396 :     a. New module ir/dj-dataflow.sml implements elimination based
397 :     data flow analysis.
398 :    
399 :     4. MLRiscGen
400 :    
401 :     a. Fix for gc type annotation
402 :    
403 :     5. MDGen
404 :    
405 :     Various fixes for machine description -> ml code translation. For ssa
406 :     only.
407 :    
408 :     ----------------------------------------------------------------------
409 :     Name: Allen Leung
410 : leunga 640 Date: 2000/05/08 22:17 EDT
411 :     Tag: leunga-20000508-labexp
412 :     Description:
413 :    
414 :     Fermin has found a few assembly problems with constant expressions
415 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
416 :     which choke on dumb assemblers. This is his fix.
417 :    
418 :     ----------------------------------------------------------------------
419 : dbm 635 Name: Dave MacQueen
420 :     Date: 2000/04/09 14:00 EDT
421 :     Tag: dbm-20000502-Version_110_28
422 :     Description:
423 :    
424 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
425 :    
426 :     2. Updated config/version to 110.28
427 :    
428 :     3. Updated config/srcarchiveurl
429 :    
430 :     3. New boot files!
431 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
432 :    
433 :     ----------------------------------------------------------------------
434 : blume 634 Name: Matthias Blume
435 :     Date: 2000/05/01 19:05:00 JST
436 :     Tag: blume-20000501-noweb
437 :     Description:
438 :    
439 :     A new noweb tool has been added. The existing system is entirely
440 :     unaffected by this, but some CM users have asked for renewed noweb
441 :     support. Everything is documented in the CM manual.
442 :    
443 :     New (plugin) libraries:
444 :    
445 :     noweb-tool.cm
446 :     nw-ext.cm
447 :    
448 :     ----------------------------------------------------------------------
449 : dbm 633 Name: Dave MacQueen
450 :     Date: 2000/04/30 12:40PM EDT
451 :     Tag: dbm-20000430-bug_fixes
452 :     Description:
453 :    
454 :     1. Fix for bug 1498
455 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
456 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
457 :     added toRealArray function
458 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
459 :     added check for tag Obj.RealArray to array printing case in ppObj
460 :    
461 :     2. Fix for bug 1510
462 :     smlnj/src/compiler/Semant/types/typesutil.sml
463 :     fixed definition of dummyargs (used by equalTycon) so that
464 :     dummy args are distinct types
465 :    
466 :     ----------------------------------------------------------------------
467 : blume 629 Name: Matthias Blume
468 : blume 632 Date: 2000/04/30 01:00:00 JST
469 :     Tag: blume-20000430-versions
470 :     Description:
471 :    
472 :     1. CM version numbering added. This is an implementation of Lal's
473 :     proposal for adding version numbers and version checking to .cm
474 :     files. Lal said that his proposal was just that -- a proposal.
475 :     For the time being I went ahead and implemented it so that people
476 :     can comment on it. Everything is completely backward-compatible
477 :     (except for the stable library format, i.e., new bootfiles!).
478 :    
479 :     As usual, see the CM manual for details.
480 :    
481 :     2. An alternative syntax for anchored paths has been implemented.
482 :     Dave has recently voiced the same concerns that I had when I did
483 :     this, so there should be some support. My take is that eventually
484 :     I will let support for the current syntax (where anchors are
485 :     "implicit") fade out in favor of the new, explicit syntax.
486 :     In order to be backward-compatible, both old and new syntax are
487 :     currently supported.
488 :    
489 :     Again, see the CM manual for details.
490 :    
491 :     3. Parallel make is trying to be slightly smarter: When the master
492 :     process finds a "bottleneck", i.e., when there is only one
493 :     compilation unit that can be compiled and everybody else is
494 :     waiting on it, then it will simply compile it directly instead
495 :     of clumsily telling one of the slaves to do it.
496 :    
497 :     4. Support for "unsharing" added. This is necessary in order to be
498 :     able to have two different versions of the same library running
499 :     at the same time (e.g., for trying out a new MLRISC while still
500 :     having the old MLRISC linked into the current compiler, etc.)
501 :     See the CM manual.
502 :    
503 :     5. Simple "makedepend" functionality added for generating Makefile
504 :     dependency information. (This is rather crude at the moment.
505 :     Expect some changes here in the future.)
506 :    
507 :     6. ".fun" added as a recognized suffix for ML files. Also documented
508 :     explicitly in the manual that the fallback behavior (unknown suffix
509 :     -> ML file) is not an official feature!
510 :    
511 :     7. Small changes to the pickler for stable libraries.
512 :    
513 :     8. Several internal changes to CM (for cleanup/improvement).
514 :    
515 :    
516 :     !!!! NEW BINFILES !!!!
517 :    
518 :     ----------------------------------------------------------------------
519 :     Name: Matthias Blume
520 : blume 631 Date: 2000/04/28 17:30:00 JST
521 :     Tag: blume-20000428-pathconfig
522 :     Description:
523 :    
524 :     1. I changed config/install.sh to remove duplicate entries from the
525 :     lib/pathconfig file at the end. Moreover, the final version of
526 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
527 :     in src/system/installml.
528 :    
529 :     2. The config/install.sh script now consistently uses relative
530 :     pathnames in lib/pathconfig whenever the anchor is in the lib
531 :     directory. (So far this was true for the libraries that come
532 :     pre-compiled and bundled as part of the bootfiles but not for
533 :     libraries that are compiled by the script itself.)
534 :    
535 :     ----------------------------------------------------------------------
536 :     Name: Matthias Blume
537 : blume 629 Date: 2000/04/26 13:10:00 JST
538 :     Tag: blume-20000426-fun_suffix
539 :     Description:
540 :    
541 :     Added ".fun" as a recognized file name suffix (for ML code).
542 :    
543 :     ----------------------------------------------------------------------
544 : leunga 628 Name: Allen Leung
545 :     Date: 2000/04/25 17:00:00 EST
546 :     Tag: leunga-20000425-alpha-ra
547 :     Description:
548 :    
549 :     1. Alpha
550 :    
551 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
552 :     in C--.
553 :    
554 :     2. RA
555 :    
556 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
557 :     and the interference graph.
558 :    
559 :     ----------------------------------------------------------------------
560 : dbm 627 Name: Dave MacQueen
561 :     Date: 2000/04/25/
562 :     Tag: dbm-20000425-mlyacc_doc_examples
563 :     Description:
564 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
565 :     Updated all three examples in src/ml-yacc/examples to run
566 :     under 110.* using CM.make.
567 :    
568 :     ----------------------------------------------------------------------
569 : leunga 624 Name: Allen Leung
570 :     Date: 2000/04/20 23:04:00 EST
571 :     Tag: leunga-20000420-ssa-c---stuff
572 :     Description:
573 :    
574 :     This update synchronizes my repository with Yale's. Most of these
575 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
576 :    
577 :     1. Register Allocator
578 :    
579 :     a. An improvement in the interference graph construction:
580 :     Given a copy
581 :    
582 :     s <- t
583 :    
584 : leunga 628 no interference edge between s and t is added for this definition of s.
585 : leunga 624
586 :     b. I've added two new spill heuristic modules that Fermin and I developed
587 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
588 :     useful for others (Moby?)
589 :    
590 :     2. X86
591 :    
592 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
593 :    
594 :     3. Alpha
595 :    
596 :     a. Added the BSR instruction and code generation that goes with it [C--]
597 :     b. Other fixes too numerous to recount provided by Fermin [C--]
598 :    
599 :     4. Regmaps
600 :    
601 :     a. The regmaps are not initialized with the identity physical bindings
602 :     at creation time. This is unneeded.
603 :    
604 :     5. MLRISC Optimizations
605 :    
606 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
607 :     intersects with liveness incrementally in linear time! Woohoo!
608 :     This is now used in my new SSA construction algorithm.
609 :    
610 :     b. THe branch reorganization module is now smarter about linear chains of
611 :     basic blocks.
612 :    
613 :    
614 :     ----------------------------------------------------------------------
615 : blume 615 Name: Matthias Blume
616 :     Date: 2000/04/12 13:52:00 JST
617 :     Tag: blume_main_v110p27_1
618 :     Description:
619 :    
620 :     Changed install.sh script to handle archive files without version number
621 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
622 :     name of the boot file archive.
623 :    
624 :     ----------------------------------------------------------------------
625 : dbm 607 Name: Dave MacQueen
626 :     Date: 2000/04/09 14:00 EDT
627 :     Tag: dbm-20000410-Version_110_27
628 :     Description:
629 :    
630 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
631 :    
632 :     2. Updated src/config/version to 110.27
633 :    
634 :     3. New boot files!
635 :    
636 :     ----------------------------------------------------------------------
637 : leunga 606 Name: Allen Leung
638 :     Date: 2000/04/09 19:09:00 EST
639 :     Tag: leunga-20000409-misc
640 :     Description:
641 :    
642 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
643 :    
644 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
645 :    
646 :     ----------------------------------------------------------------------
647 : monnier 605 Name: Stefan
648 :     Date: 2000/04/07 10:00:00 EDT
649 :     Tag: monnier-20000406-branch-handling
650 :     Description:
651 :    
652 :     Improved handling of branches (mostly those generated from
653 :     polymorphic equality), removed switchoff and changed the
654 :     default optimization settings (more cpsopt and less flintopt).
655 :    
656 :     ----------------------------------------------------------------------
657 : leunga 601 Name: Allen Leung
658 : leunga 602 Date: 2000/04/06 01:30:00 EST
659 :     Tag: leunga-20000406-peephole-x86-SSA-2
660 :     Description:
661 :    
662 :     Forgot a few files.
663 :    
664 :     ----------------------------------------------------------------------
665 :     Name: Allen Leung
666 : leunga 601 Date: 2000/04/06 00:36:00 EST
667 :     Tag: leunga-20000406-peephole-x86-SSA
668 :     Description:
669 :    
670 :     1. New Peephole code
671 :    
672 :     2. Minor improvement to X86 instruction selection
673 :    
674 :     3. Various fixes to SSA and machine description -> code translator
675 :    
676 :     ----------------------------------------------------------------------
677 : blume 595 Name: Matthias Blume
678 :     Date: 2000/04/05 12:30:00 JST
679 :     Tag: blume_main_v110p26p2_3
680 :     Description:
681 :    
682 :     This update just merges three minor cosmetic updates to CM's sources
683 :     to get ready for the 110.27 code freeze on Friday. No functionality
684 :     has changed.
685 :    
686 :     ----------------------------------------------------------------------
687 : leunga 593 Name: Allen Leung
688 : leunga 594 Date: 2000/04/04 19:39:00 EST
689 :     Tag: leunga-20000404-x86-asm
690 :     Description:
691 :    
692 :     1. Fixed a problem in X86 assembly.
693 :    
694 :     Things like
695 :    
696 :     jmp %eax
697 :     jmp (%eax)
698 :    
699 :     should be output as
700 :    
701 :     jmp *%eax
702 :     jmp *(%eax)
703 :    
704 :     2. Assembly output
705 :    
706 :     Added a new flag
707 :    
708 :     "asm-indent-copies" (default to false)
709 :    
710 :     When this flag is on, parallel copies will be indented an extra level.
711 :    
712 :     ----------------------------------------------------------------------
713 :     Name: Allen Leung
714 : leunga 593 Date: 2000/04/04 03:18:00 EST
715 :     Tag: leunga-20000404-C--Moby
716 :     Description:
717 :    
718 :     All of these fixes are related to C--, Moby, and my own optimization
719 :     stuff; so they shouldn't affect SML/NJ.
720 :    
721 :     1. X86
722 :    
723 :     Various fixes related floating point, and extensions.
724 :    
725 :     2. Alpha
726 :    
727 :     Some extra patterns related to loads with signed/zero extension
728 :     provided by Fermin.
729 :    
730 :     3. Assembly
731 :    
732 : monnier 659 When generating assembly, resolve the value of client defined constants,
733 : leunga 593 instead of generating symbolic values. This is controlled by the
734 :     new flag "asm-resolve-constants", which is default to true.
735 :    
736 :     4. Machine Descriptions
737 :    
738 :     a. The precedence parser was slightly broken when parsing infixr symbols.
739 :     b. The type generalizing code had the bound variables reversed, resulting
740 :     in a problem during arity raising.
741 :     c. Various fixes in machine descriptions.
742 :    
743 :     ----------------------------------------------------------------------
744 : blume 592 Name: Matthias Blume
745 :     Date: 2000/04/03 16:05:00 JST
746 :     Tag: blume_main_v110p26p2_2
747 :     Description:
748 :    
749 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
750 :     now done via the ordinary static environment that is context to each
751 :     compilation unit.
752 :    
753 :     To this end, I arranged that instead of "structure Core" as "structure
754 :     _Core" is bound in the pervasive environment. Core access is done via
755 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
756 : blume 592 legal surface-syntax symbol).
757 :    
758 :     The current solution is much cleaner because the core environment is
759 :     now simply part of the pervasive environment which is part of every
760 :     compilation unit's context anyway. In particular, this eliminates all
761 :     special-case handling that was necessary until now in order to deal
762 :     with dynamic and symbolic parts of the core environment.
763 :    
764 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
765 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
766 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
767 :     more details.
768 :    
769 :     I also tried to track down all mentions of "Core" (as string argument
770 :     to Symbol.strSymbol) in the compiler and replaced them with a
771 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
772 :     name appears in one place only.
773 :    
774 :     Binfile and bootfile format have not changed, but the switchover from
775 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
776 :     bootfiles anyway.
777 :    
778 :     ----------------------------------------------------------------------
779 : leunga 591 Name: Allen Leung
780 :     Date: 2000/04/02 21:17:00 EST
781 :     Tag: leunga-20000402-mltree
782 :     Description:
783 : leunga 585
784 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
785 :     2. Added a bunch of files from my repository. These are currently
786 :     used by other non-SMLNJ backends.
787 :    
788 : leunga 576 ----------------------------------------------------------------------
789 : leunga 590 Name: Allen Leung
790 :     Date: 2000/03/31 21:15:00 EST
791 :     Tag: leunga-20000331-aliasing
792 :     Description:
793 :    
794 :     This update contains a rewritten (and hopefully more correct) module
795 :     for extracting aliasing information from CPS.
796 :    
797 :     To turn on this feature:
798 :    
799 :     Compiler.Control.CG.memDisambiguate := true
800 :    
801 :     To pretty print the region information with assembly
802 :    
803 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
804 :    
805 :     To control how many levels of aliasing information are printed, use:
806 :    
807 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
808 :    
809 :     The default of n is 3.
810 :    
811 :     ----------------------------------------------------------------------
812 : dbm 589 Name: David MacQueen
813 :     Date: 2000/03/31 11:15:00 EST
814 :     Tag: dbm-20000331-runtime_fix
815 :     Description:
816 :    
817 :     This update contains:
818 :    
819 :     1. runtime/c-lib/c-libraries.c
820 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
821 :    
822 :     2. fix for bug 1556
823 :     system/Basis/Implementation/NJ/internal-signals.sml
824 :    
825 :     ----------------------------------------------------------------------
826 : blume 587 Name: Matthias Blume
827 : blume 588 Date: 2000/03/31 18:00:00 JST
828 :     Tag: blume_main_v110p26p2_1
829 :     Description:
830 :    
831 :     This update contains:
832 :    
833 :     1. A small change to CM's handling of stable libraries:
834 :     CM now maintains one "global" modmap that is used for all stable
835 :     libraries. The use of such a global modmap maximizes sharing and
836 :     minimizes the need for re-traversing parts of environments during
837 :     modmap construction. (However, this has minor impact since modmap
838 :     construction seems to account for just one percent or less of total
839 :     compile time.)
840 :    
841 :     2. I added a "genmap" phase to the statistics. This is where I got the
842 :     "one percent" number (see above).
843 :    
844 :     3. CM's new tool parameter mechanism just became _even_ better. :)
845 :     - The parser understands named parameters and recursive options.
846 :     - The "make" and "shell" tools use these new features.
847 :     (This makes it a lot easier to cascade these tools.)
848 :     - There is a small syntax change: named parameters use a
849 :    
850 :     <name> : ( <option> ... ) or
851 :     <name> : <string>
852 :    
853 :     syntax. Previously, named parameters were implemented in an
854 :     ad-hoc fashion by each tool individually (by parsing strings)
855 :     and had the form
856 :    
857 :     <name>=<string>
858 :    
859 :     See the CM manual for a full description of these issues.
860 :    
861 :     ----------------------------------------------------------------------
862 :     Name: Matthias Blume
863 : blume 587 Date: 2000/03/30 18:00:00 JST
864 :     Tag: blume_main_v110p26p2_0
865 :     Description:
866 :    
867 :     !!!!! WARNING !!!!!!
868 :     !! New binfiles !!
869 :     !!!!!!!!!!!!!!!!!!!!
870 :    
871 :     This update contains:
872 :    
873 :     1. Moderate changes to CM:
874 :    
875 :     - Changes to CM's tools mechanism. In particular, it is now possible
876 :     to have tools that accept additional "command line" parameters
877 :     (specified in the .cm file at each instance where the tool's class is
878 :     used).
879 :    
880 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
881 :     facilitate fairly seamless hookup to portions of code managed using
882 : blume 587 Makefiles or Shell scripts.
883 :    
884 :     There are no classes "shared" or "private" anymore. Instead, the
885 :     sharing annotation is now a parameter to the "sml" class.
886 :    
887 :     There is a bit of generic machinery for implementing one's own
888 :     tools that accept command-line parameters. However, I am not yet fully
889 :     satisfied with that part, so expect changes here in the future.
890 :    
891 :     All existing tools are described in the CM manual.
892 :    
893 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
894 : blume 587 error messages that tended to be more annoying than helpful.)
895 :    
896 :     2. Major changes to the compiler's static environment data structures.
897 :    
898 :     - no CMStaticEnv anymore.
899 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
900 :     but it is called Environment), no conversions between different
901 :     kinds of static environments
902 :    
903 :     - There is still a notion of a "modmap", but such modmaps are generated
904 :     on demand at the time when they are needed. This sounds slow, but I
905 :     sped up the code that generates modmaps enough for this not to lead to
906 :     a slowdown of the compiler (at least I didn't detect any).
907 :    
908 :     - To facilitate rapid modmap generation, static environments now
909 :     contain an (optional) "modtree" structure. Modtree annotations are
910 :     constructed by the unpickler during unpickling. (This means that
911 :     the elaborator does not have to worry about modtrees at all.)
912 :     Modtrees have the advantage that they are compositional in the same
913 :     way as the environment data structure itself is compositional.
914 :     As a result, modtrees never hang on to parts of an environment that
915 :     has already been rendered "stale" by filtering or rebinding.
916 :    
917 :     - I went through many, many trials and errors before arriving at the
918 :     current solution. (The initial idea of "linkpaths" did not work.)
919 :     But the result of all this is that I have touched a lot of files that
920 :     depend on the "modules" and "types" data structures (most of the
921 :     elaborator). There were a lot of changes during my "linkpath" trials
922 :     that could have been reverted to their original state but weren't.
923 :     Please, don't be too harsh on me for messing with this code a bit more
924 :     than what was strictly necessary... (I _did_ resist the tempation
925 :     of doing any "global reformatting" to avoid an untimely death at
926 :     Dave's hands. :)
927 :    
928 :     - One positive aspect of the previous point: At least I made sure that
929 :     all files that I touched now compile without warnings (other than
930 :     "polyEqual").
931 :    
932 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
933 :     redundant modmaps)
934 :    
935 :     ----------------------------------------------------------------------
936 : leunga 580 Name: Allen Leung
937 : leunga 585 Date: 2000/03/29 18:00:00
938 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
939 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
940 : leunga 585 Description:
941 :    
942 :     This update contains *MAJOR* changes to the way code is generated from CPS
943 :     in the module mlriscGen, and in various backend modules.
944 :    
945 :     CHANGES
946 :     =======
947 :    
948 :     1. MLRiscGen: forward propagation fix.
949 :    
950 :     There was a bug in forward propagation introduced at about the same time
951 :     as the MLRISC x86 backend, which prohibits coalescing to be
952 :     performed effectively in loops.
953 :    
954 :     Effect: speed up of loops in RISC architectures.
955 :     By itself, this actually slowed down certain benchmarks on the x86.
956 :    
957 :     2. MLRiscGen: forward propagating addresses from consing.
958 :    
959 :     I've changed the way consing code is generated. Basically I separated
960 :     out the initialization part:
961 :    
962 :     store tag, offset(allocptr)
963 :     store elem1, offset+4(allocptr)
964 :     store elem2, offset+8(allocptr)
965 :     ...
966 :     store elemn, offset+4n(allocptr)
967 :    
968 :     and the address computation part:
969 :    
970 :     celladdr <- offset+4+alloctpr
971 :    
972 :     and move the address computation part
973 :    
974 :     Effect: register pressure is generally lower as a result. This
975 :     makes compilation of certain expressions much faster, such as
976 :     long lists with non-trivial elements.
977 :    
978 :     [(0,0), (0,0), .... (0,0)]
979 :    
980 :     3. MLRiscGen: base pointer elimination.
981 :    
982 :     As part of the linkage mechanism, we generate the sequence:
983 :    
984 :     L: ... <- start of the code fragment
985 :    
986 :     L1:
987 :     base pointer <- linkreg - L1 + L
988 :    
989 :     The base pointer was then used for computing relocatable addresses
990 :     in the code fragment. Frequently (such as in lots of continuations)
991 :     this is not needed. We now eliminate this sequence whenever possible.
992 :    
993 :     For compile time efficiency, I'm using a very stupid local heuristic.
994 :     But in general, this should be done as a control flow analysis.
995 :    
996 :     Effect: Smaller code size. Speed up of most programs.
997 :    
998 :     4. Hppa back end
999 :    
1000 :     Long jumps in span dependence resolution used to depend on the existence
1001 :     of the base pointer.
1002 :    
1003 :     A jump to a long label L was expanded into the following sequence:
1004 :    
1005 :     LDIL %hi(L-8192), %r29
1006 :     LDO %lo(L-8192)(%r29), %r29
1007 :     ADD %r29, baseptr, %r29
1008 :     BV,n %r0(%r29)
1009 :    
1010 :     In the presence of change (3) above, this will not work. I've changed
1011 :     it so that the following sequence of instructions are generated, which
1012 :     doesn't mention the base pointer at all:
1013 :    
1014 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
1015 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
1016 :     BV,n %r0(%r29) /* Jump */
1017 :    
1018 :     5. Alpha back end
1019 :    
1020 :     New alpha instructions LDB/LDW have been added, as per Fermin's
1021 :     suggestions. This is unrelated to all other changes.
1022 :    
1023 :     6. X86 back end
1024 :    
1025 :     I've changed andl to testl in the floating point test sequence
1026 :     whenever appropriate. The Intel optimization guide states that
1027 : monnier 659 testl is preferable to andl.
1028 : leunga 585
1029 :     7. RA (x86 only)
1030 :    
1031 :     I've improved the spill propagation algorithm, using an approximation
1032 :     of maximal weighted independent sets. This seems to be necessary to
1033 :     alleviate the negative effect in light of the slow down in (1).
1034 :    
1035 :     I'll write down the algorithm one of these days.
1036 :    
1037 :     8. MLRiscGen: frequencies
1038 :    
1039 :     I've added an annotation that states that all call gc blocks have zero
1040 :     execution frequencies. This improves register allocation on the x86.
1041 :    
1042 :     BENCHMARKS
1043 :     ==========
1044 :    
1045 :     I've only perform the comparison on 110.25.
1046 :    
1047 :     The platforms are:
1048 :    
1049 :     HPPA A four processor HP machine (E9000) with 5G of memory.
1050 :     X86 A 300Hhz Pentium II with 128M of memory, and
1051 :     SPARC An Ultra sparc 2 with 512M of memory.
1052 :    
1053 :     I used the following parameters for the SML benchmarks:
1054 :    
1055 :     @SMLalloc
1056 :     HPPA 256k
1057 :     SPARC 512k
1058 :     X86 256k
1059 :    
1060 :     COMPILATION TIME
1061 :     ----------------
1062 :     Here are the numbers comparing the compilation times of the compilers.
1063 :     I've only compared 110.25 compiling the new sources versus
1064 :     a fixpoint version of the new compiler compiling the same.
1065 :    
1066 :     110.25 New
1067 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
1068 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
1069 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
1070 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
1071 :    
1072 :     110.25 New
1073 :     Code Size Code Size
1074 :     HPPA 8596736 8561421
1075 :     SPARC 8974299 8785143
1076 :     X86 9029180 8716783
1077 :    
1078 :     So in summary, things are at least as good as before. Dramatic
1079 :     reduction in compilation is obtained on the Sparc; I can't explain it,
1080 :     but it is reproducible. Perhaps someone should try to reproduce this
1081 :     on their own machines.
1082 :    
1083 :     SML BENCHMARKS
1084 :     --------------
1085 :    
1086 :     On the average, all benchmarks perform at least as well as before.
1087 :    
1088 :     HPPA Compilation Time Spill+Reload Run Time
1089 :     110.25 New 110.25 New 110.25 New
1090 :    
1091 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
1092 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
1093 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
1094 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
1095 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
1096 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
1097 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
1098 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
1099 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
1100 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
1101 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
1102 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
1103 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
1104 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
1105 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
1106 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
1107 :     --------------------------------------------------------------------------
1108 :     Average 6.12% 4.09%
1109 :    
1110 :     SPARC Compilation Time Spill+Reload Run Time
1111 :     110.25 New 110.25 New 110.25 New
1112 :    
1113 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
1114 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
1115 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
1116 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
1117 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
1118 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
1119 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
1120 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
1121 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
1122 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
1123 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
1124 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
1125 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
1126 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
1127 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
1128 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
1129 :     ----------------------------------------------------------------------------
1130 :     Average 6.94% 2.64%
1131 :    
1132 :     X86 Compilation Time Spill+Reload Run Time
1133 :     110.25 New 110.25 New 110.25 New
1134 :    
1135 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
1136 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
1137 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
1138 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
1139 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
1140 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
1141 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
1142 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
1143 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
1144 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
1145 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
1146 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
1147 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
1148 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
1149 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
1150 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
1151 :     ----------------------------------------------------------------------------
1152 :     Average 1.22% 3.36%
1153 :    
1154 :     ----------------------------------------------------------------------
1155 :     Name: Allen Leung
1156 : leunga 583 Date: 2000/03/23 16:25:00
1157 :     Tag: leunga-20000323-fix_x86_alpha
1158 :     Description:
1159 :    
1160 :     1. X86 fixes/changes
1161 :    
1162 :     a. The old code generated for SETcc was completely wrong.
1163 :     The Intel optimization guide is VERY misleading.
1164 :    
1165 :     2. ALPHA fixes/changes
1166 :    
1167 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1168 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1169 :     c. Added reassociation code for address computation.
1170 :    
1171 :     ----------------------------------------------------------------------
1172 :     Name: Allen Leung
1173 : leunga 580 Date: 2000/03/22 01:23:00
1174 :     Tag: leunga-20000322-fix_x86_hppa_ra
1175 :     Description:
1176 :    
1177 :     1. X86 fixes/changes
1178 :    
1179 :     a. x86Rewrite bug with MUL3 (found by Lal)
1180 :     b. Added the instructions FSTS, FSTL
1181 :    
1182 :     2. PA-RISC fixes/changes
1183 :    
1184 :     a. B label should not be a delay slot candidate! Why did this work?
1185 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1186 :     as it should be.
1187 :     c. The assembly syntax for fstds and fstdd was wrong.
1188 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
1189 :     operand variant of COMCLR/LDO.
1190 :    
1191 :     3. Generic MLRISC
1192 :    
1193 :     a. shuffle.sml rewritten to be slightly more efficient
1194 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
1195 :    
1196 :     4. Register Allocator
1197 :    
1198 :     a. I now release the interference graph earlier during spilling.
1199 :     May improve memory usage.
1200 :    
1201 :     ----------------------------------------------------------------------
1202 : blume 577 Name: Matthias Blume
1203 : blume 578 Date: 2000/03/14 14:15:32
1204 :     Tag: blume_main_v110p26p1_2
1205 :     Description:
1206 :    
1207 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1208 :     additional argument called "template" which is an optional string that
1209 : monnier 659 specifies the layout of the tool command line. See the CM manual for
1210 : blume 578 explanation.
1211 :    
1212 : monnier 659 2. A special-purpose tool can be "registered" by simply dropping the
1213 : blume 578 corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1214 :     directory where the .cm file lives that uses this tool. (The
1215 :     behavior/misfeature until now was to look for the tool description
1216 :     files in the current working directory.) As before, tool description
1217 :     files could also be anchored -- in which case they can live anywhere
1218 :     they like. Following the recent e-mail discussion, this change should
1219 :     make it easier to have special-purpose tools that are shipped together
1220 :     with the sources of the program that uses them.
1221 :    
1222 :     ----------------------------------------------------------------------
1223 :     Name: Matthias Blume
1224 : blume 577 Date: 2000/03/10 07:48:34
1225 :     Tag: blume_main_v110p26p1_1
1226 :     Description:
1227 :    
1228 :     I added a re-written version of Dave's fixpt script to src/system.
1229 :     Changes relative to the original version:
1230 :     - sh-ified (not everybody has ksh)
1231 :     - automatically figures out which architecture it runs on
1232 :     - uses ./makeml a bit more cleverly
1233 :     - never invokes ./installml (and, thus, does not clobber your
1234 :     good and working installation of sml in case something goes wrong)
1235 :     - accepts max iteration count using option "-iter <n>"
1236 :     - accepts a "base" name using option "-base <base>"
1237 :    
1238 :     It does not build any extraneous heap images but directly rebuilds
1239 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
1240 :     it can incorporate existing bin- and boot- hierarchies. For example,
1241 :     suppose the base is set to "sml" (which is the default). Then it
1242 :     successively builds
1243 :    
1244 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
1245 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
1246 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
1247 :     ...
1248 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
1249 :    
1250 :     and so on. If any of these already exist, it will just use what's
1251 :     there. In particular, many people will have the initial set of bin
1252 :     and boot files around, so this saves time for at least one full
1253 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
1254 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
1255 :     they might be the result of an earlier partial run of fixpt (which
1256 : monnier 659 perhaps got accidentally killed). In this case, fixpt will quickly
1257 : blume 577 move through what exists before continuing where it left off earlier,
1258 :     and, thus, saves a lot of time.
1259 :    
1260 :     ----------------------------------------------------------------------
1261 : leunga 576 Name: Allen Leung
1262 :     Date: 00/03/10 02:20:00
1263 :     Tag: leunga-20000310-fix_x86_asm_ra
1264 :     Description:
1265 : dbm 570
1266 : leunga 576 More assembly output problems involving the indexed addressing mode
1267 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
1268 :     fix.
1269 :    
1270 :     The interface and implementation of the register allocator have been changed
1271 :     slightly to accommodate the possibility to skip the register allocation
1272 :     phases completely and go directly to memory allocation. This is needed
1273 :     for C-- use.
1274 :    
1275 : dbm 570 ----------------------------------------------------------------------
1276 : blume 572 Name: Matthias Blume
1277 : blume 575 Date: 00/03/09 10:23:53
1278 :     Tag: blume_main_v110p26p1_0
1279 :     Description:
1280 :    
1281 :     * Complete re-organization of library names. Many libraries have been
1282 :     consolidated so that they share the same path anchor. For example,
1283 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
1284 :     are SML/NJ-specific are under "smlnj". Notice that names like
1285 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
1286 :     for a complete description of the new naming scheme. Quick reference:
1287 :    
1288 :     host-cmb.cm -> smlnj/cmb.cm
1289 :     host-compiler.cm -> smlnj/compiler.cm
1290 :     full-cm.cm -> smlnj/cm.cm
1291 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
1292 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
1293 :    
1294 :     * Bug fixes in CM.
1295 :     - exceptions in user code are being passed through (i.e., reach top level)
1296 :     - more bugs in paranoia mode fixed
1297 :     - bug related to checking group owners fixed
1298 :    
1299 :     * New install.sh script that automagically fetches archive files:
1300 :     The new file config/srcarchiveurl must contain the URL of the
1301 :     (remote) directory that contains bin files (or other source archives).
1302 :     If install.sh does not find the archive locally, it tries to get
1303 :     it from that remote directory.
1304 :     This should simplify installation further: For machines that have
1305 :     access to the internet, just fetch <version>-config.tgz, unpack it,
1306 : monnier 659 edit config/targets, and go (run config/install.sh). The script will
1307 : blume 575 fetch everything else that it might need all by itself.
1308 :    
1309 :     For CVS users, this mechanism is not relevant for source archives, but
1310 :     it is convenient for getting new sets of binfiles.
1311 :    
1312 :     Archives should be tar files compressed with either gzip, compress, or
1313 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
1314 :    
1315 :     ----------------------------------------------------------------------
1316 :     Name: Matthias Blume
1317 : blume 572 Date: 2000/03/07 04:01:04
1318 :     Tag: blume_main_v110_26_2
1319 : dbm 570 Description:
1320 : blume 572 - size info in BOOTLIST
1321 :     * no fixed upper limits for number of bootfiles or length of
1322 :     bootfile names in runtime
1323 :     * falling back to old behavior if no BOOTLIST size info found
1324 :     - allocation size heuristics in .run-sml
1325 :     * tries to read cache size from /proc/cpuinfo (this is important for
1326 :     small-cache Celeron systems!)
1327 :     - install.sh robustified
1328 :     - CM manual updates
1329 :     - paranoid mode
1330 :     * no more CMB.deliver() (i.e., all done by CMB.make())
1331 :     * can re-use existing sml.boot.* files
1332 :     * init.cmi now treated as library
1333 :     * library stamps for consistency checks
1334 :     - sml.boot.<arch>-<os>/PIDMAP file
1335 :     * This file is read by the CM startup code. This is used to minimize
1336 :     the amount of dynamic state that needs to be stowed away for the
1337 :     purpose of sharing between interactive system and user code.
1338 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
1339 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
1340 :     like most other CM state variables.
1341 :     - Compiler.CMSA eliminated
1342 :     * No longer supported by CM anyway.
1343 :     - fixed bugs in pickler that kept biting Stefan
1344 :     * past refs to past refs (was caused by the possibility that
1345 :     ad-hoc sharing is more discriminating than hash-cons sharing)
1346 :     * integer overflow on LargeInt.minInt
1347 :     - ml-{lex,yacc} build scripts now use new mechanism
1348 :     for building standalone programs
1349 :     - fixed several gcc -Wall warnings that were caused by missing header
1350 :     files, missing initializations, etc., in runtime (not all warnings
1351 :     eliminated, though)

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0