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/branches/primop-branch-3/NOTES/HISTORY
ViewVC logotype

Annotation of /sml/branches/primop-branch-3/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 606 - (view) (download)
Original Path: sml/trunk/HISTORY

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 :     Date:
12 :     Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 591 ----------------------------------------------------------------------
15 : leunga 606 Name: Allen Leung
16 :     Date: 2000/04/09 19:09:00 EST
17 :     Tag: leunga-20000409-misc
18 :     Description:
19 :    
20 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
21 :    
22 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
23 :    
24 :     ----------------------------------------------------------------------
25 : monnier 605 Name: Stefan
26 :     Date: 2000/04/07 10:00:00 EDT
27 :     Tag: monnier-20000406-branch-handling
28 :     Description:
29 :    
30 :     Improved handling of branches (mostly those generated from
31 :     polymorphic equality), removed switchoff and changed the
32 :     default optimization settings (more cpsopt and less flintopt).
33 :    
34 :     ----------------------------------------------------------------------
35 : leunga 601 Name: Allen Leung
36 : leunga 602 Date: 2000/04/06 01:30:00 EST
37 :     Tag: leunga-20000406-peephole-x86-SSA-2
38 :     Description:
39 :    
40 :     Forgot a few files.
41 :    
42 :     ----------------------------------------------------------------------
43 :     Name: Allen Leung
44 : leunga 601 Date: 2000/04/06 00:36:00 EST
45 :     Tag: leunga-20000406-peephole-x86-SSA
46 :     Description:
47 :    
48 :     1. New Peephole code
49 :    
50 :     2. Minor improvement to X86 instruction selection
51 :    
52 :     3. Various fixes to SSA and machine description -> code translator
53 :    
54 :     ----------------------------------------------------------------------
55 : blume 595 Name: Matthias Blume
56 :     Date: 2000/04/05 12:30:00 JST
57 :     Tag: blume_main_v110p26p2_3
58 :     Description:
59 :    
60 :     This update just merges three minor cosmetic updates to CM's sources
61 :     to get ready for the 110.27 code freeze on Friday. No functionality
62 :     has changed.
63 :    
64 :     ----------------------------------------------------------------------
65 : leunga 593 Name: Allen Leung
66 : leunga 594 Date: 2000/04/04 19:39:00 EST
67 :     Tag: leunga-20000404-x86-asm
68 :     Description:
69 :    
70 :     1. Fixed a problem in X86 assembly.
71 :    
72 :     Things like
73 :    
74 :     jmp %eax
75 :     jmp (%eax)
76 :    
77 :     should be output as
78 :    
79 :     jmp *%eax
80 :     jmp *(%eax)
81 :    
82 :     2. Assembly output
83 :    
84 :     Added a new flag
85 :    
86 :     "asm-indent-copies" (default to false)
87 :    
88 :     When this flag is on, parallel copies will be indented an extra level.
89 :    
90 :     ----------------------------------------------------------------------
91 :     Name: Allen Leung
92 : leunga 593 Date: 2000/04/04 03:18:00 EST
93 :     Tag: leunga-20000404-C--Moby
94 :     Description:
95 :    
96 :     All of these fixes are related to C--, Moby, and my own optimization
97 :     stuff; so they shouldn't affect SML/NJ.
98 :    
99 :     1. X86
100 :    
101 :     Various fixes related floating point, and extensions.
102 :    
103 :     2. Alpha
104 :    
105 :     Some extra patterns related to loads with signed/zero extension
106 :     provided by Fermin.
107 :    
108 :     3. Assembly
109 :    
110 :     When generating assemby, resolve the value of client defined constants,
111 :     instead of generating symbolic values. This is controlled by the
112 :     new flag "asm-resolve-constants", which is default to true.
113 :    
114 :     4. Machine Descriptions
115 :    
116 :     a. The precedence parser was slightly broken when parsing infixr symbols.
117 :     b. The type generalizing code had the bound variables reversed, resulting
118 :     in a problem during arity raising.
119 :     c. Various fixes in machine descriptions.
120 :    
121 :     ----------------------------------------------------------------------
122 : blume 592 Name: Matthias Blume
123 :     Date: 2000/04/03 16:05:00 JST
124 :     Tag: blume_main_v110p26p2_2
125 :     Description:
126 :    
127 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
128 :     now done via the ordinary static environment that is context to each
129 :     compilation unit.
130 :    
131 :     To this end, I arranged that instead of "structure Core" as "structure
132 :     _Core" is bound in the pervasive environment. Core access is done via
133 :     _Core (which can never be accidentially rebound because _Core is not a
134 :     legal surface-syntax symbol).
135 :    
136 :     The current solution is much cleaner because the core environment is
137 :     now simply part of the pervasive environment which is part of every
138 :     compilation unit's context anyway. In particular, this eliminates all
139 :     special-case handling that was necessary until now in order to deal
140 :     with dynamic and symbolic parts of the core environment.
141 :    
142 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
143 :     compilation mananger's bootstrap compiler (actually: in the "init group"
144 :     handling). See the comments in src/system/smlnj/init/init.cmi for
145 :     more details.
146 :    
147 :     I also tried to track down all mentions of "Core" (as string argument
148 :     to Symbol.strSymbol) in the compiler and replaced them with a
149 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
150 :     name appears in one place only.
151 :    
152 :     Binfile and bootfile format have not changed, but the switchover from
153 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
154 :     bootfiles anyway.
155 :    
156 :     ----------------------------------------------------------------------
157 : leunga 591 Name: Allen Leung
158 :     Date: 2000/04/02 21:17:00 EST
159 :     Tag: leunga-20000402-mltree
160 :     Description:
161 : leunga 585
162 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
163 :     2. Added a bunch of files from my repository. These are currently
164 :     used by other non-SMLNJ backends.
165 :    
166 : leunga 576 ----------------------------------------------------------------------
167 : leunga 590 Name: Allen Leung
168 :     Date: 2000/03/31 21:15:00 EST
169 :     Tag: leunga-20000331-aliasing
170 :     Description:
171 :    
172 :     This update contains a rewritten (and hopefully more correct) module
173 :     for extracting aliasing information from CPS.
174 :    
175 :     To turn on this feature:
176 :    
177 :     Compiler.Control.CG.memDisambiguate := true
178 :    
179 :     To pretty print the region information with assembly
180 :    
181 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
182 :    
183 :     To control how many levels of aliasing information are printed, use:
184 :    
185 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
186 :    
187 :     The default of n is 3.
188 :    
189 :     ----------------------------------------------------------------------
190 : dbm 589 Name: David MacQueen
191 :     Date: 2000/03/31 11:15:00 EST
192 :     Tag: dbm-20000331-runtime_fix
193 :     Description:
194 :    
195 :     This update contains:
196 :    
197 :     1. runtime/c-lib/c-libraries.c
198 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
199 :    
200 :     2. fix for bug 1556
201 :     system/Basis/Implementation/NJ/internal-signals.sml
202 :    
203 :     ----------------------------------------------------------------------
204 : blume 587 Name: Matthias Blume
205 : blume 588 Date: 2000/03/31 18:00:00 JST
206 :     Tag: blume_main_v110p26p2_1
207 :     Description:
208 :    
209 :     This update contains:
210 :    
211 :     1. A small change to CM's handling of stable libraries:
212 :     CM now maintains one "global" modmap that is used for all stable
213 :     libraries. The use of such a global modmap maximizes sharing and
214 :     minimizes the need for re-traversing parts of environments during
215 :     modmap construction. (However, this has minor impact since modmap
216 :     construction seems to account for just one percent or less of total
217 :     compile time.)
218 :    
219 :     2. I added a "genmap" phase to the statistics. This is where I got the
220 :     "one percent" number (see above).
221 :    
222 :     3. CM's new tool parameter mechanism just became _even_ better. :)
223 :     - The parser understands named parameters and recursive options.
224 :     - The "make" and "shell" tools use these new features.
225 :     (This makes it a lot easier to cascade these tools.)
226 :     - There is a small syntax change: named parameters use a
227 :    
228 :     <name> : ( <option> ... ) or
229 :     <name> : <string>
230 :    
231 :     syntax. Previously, named parameters were implemented in an
232 :     ad-hoc fashion by each tool individually (by parsing strings)
233 :     and had the form
234 :    
235 :     <name>=<string>
236 :    
237 :     See the CM manual for a full description of these issues.
238 :    
239 :     ----------------------------------------------------------------------
240 :     Name: Matthias Blume
241 : blume 587 Date: 2000/03/30 18:00:00 JST
242 :     Tag: blume_main_v110p26p2_0
243 :     Description:
244 :    
245 :     !!!!! WARNING !!!!!!
246 :     !! New binfiles !!
247 :     !!!!!!!!!!!!!!!!!!!!
248 :    
249 :     This update contains:
250 :    
251 :     1. Moderate changes to CM:
252 :    
253 :     - Changes to CM's tools mechanism. In particular, it is now possible
254 :     to have tools that accept additional "command line" parameters
255 :     (specified in the .cm file at each instance where the tool's class is
256 :     used).
257 :    
258 :     This was done to accomodate the new "make" and "shell" tools which
259 :     facilitate fairly seemless hookup to portions of code managed using
260 :     Makefiles or Shell scripts.
261 :    
262 :     There are no classes "shared" or "private" anymore. Instead, the
263 :     sharing annotation is now a parameter to the "sml" class.
264 :    
265 :     There is a bit of generic machinery for implementing one's own
266 :     tools that accept command-line parameters. However, I am not yet fully
267 :     satisfied with that part, so expect changes here in the future.
268 :    
269 :     All existing tools are described in the CM manual.
270 :    
271 :     - Slightly better error handling. (CM now surpresses many followup
272 :     error messages that tended to be more annoying than helpful.)
273 :    
274 :     2. Major changes to the compiler's static environment data structures.
275 :    
276 :     - no CMStaticEnv anymore.
277 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
278 :     but it is called Environment), no conversions between different
279 :     kinds of static environments
280 :    
281 :     - There is still a notion of a "modmap", but such modmaps are generated
282 :     on demand at the time when they are needed. This sounds slow, but I
283 :     sped up the code that generates modmaps enough for this not to lead to
284 :     a slowdown of the compiler (at least I didn't detect any).
285 :    
286 :     - To facilitate rapid modmap generation, static environments now
287 :     contain an (optional) "modtree" structure. Modtree annotations are
288 :     constructed by the unpickler during unpickling. (This means that
289 :     the elaborator does not have to worry about modtrees at all.)
290 :     Modtrees have the advantage that they are compositional in the same
291 :     way as the environment data structure itself is compositional.
292 :     As a result, modtrees never hang on to parts of an environment that
293 :     has already been rendered "stale" by filtering or rebinding.
294 :    
295 :     - I went through many, many trials and errors before arriving at the
296 :     current solution. (The initial idea of "linkpaths" did not work.)
297 :     But the result of all this is that I have touched a lot of files that
298 :     depend on the "modules" and "types" data structures (most of the
299 :     elaborator). There were a lot of changes during my "linkpath" trials
300 :     that could have been reverted to their original state but weren't.
301 :     Please, don't be too harsh on me for messing with this code a bit more
302 :     than what was strictly necessary... (I _did_ resist the tempation
303 :     of doing any "global reformatting" to avoid an untimely death at
304 :     Dave's hands. :)
305 :    
306 :     - One positive aspect of the previous point: At least I made sure that
307 :     all files that I touched now compile without warnings (other than
308 :     "polyEqual").
309 :    
310 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
311 :     redundant modmaps)
312 :    
313 :     ----------------------------------------------------------------------
314 : leunga 580 Name: Allen Leung
315 : leunga 585 Date: 2000/03/29 18:00:00
316 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
317 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
318 : leunga 585 Description:
319 :    
320 :     This update contains *MAJOR* changes to the way code is generated from CPS
321 :     in the module mlriscGen, and in various backend modules.
322 :    
323 :     CHANGES
324 :     =======
325 :    
326 :     1. MLRiscGen: forward propagation fix.
327 :    
328 :     There was a bug in forward propagation introduced at about the same time
329 :     as the MLRISC x86 backend, which prohibits coalescing to be
330 :     performed effectively in loops.
331 :    
332 :     Effect: speed up of loops in RISC architectures.
333 :     By itself, this actually slowed down certain benchmarks on the x86.
334 :    
335 :     2. MLRiscGen: forward propagating addresses from consing.
336 :    
337 :     I've changed the way consing code is generated. Basically I separated
338 :     out the initialization part:
339 :    
340 :     store tag, offset(allocptr)
341 :     store elem1, offset+4(allocptr)
342 :     store elem2, offset+8(allocptr)
343 :     ...
344 :     store elemn, offset+4n(allocptr)
345 :    
346 :     and the address computation part:
347 :    
348 :     celladdr <- offset+4+alloctpr
349 :    
350 :     and move the address computation part
351 :    
352 :     Effect: register pressure is generally lower as a result. This
353 :     makes compilation of certain expressions much faster, such as
354 :     long lists with non-trivial elements.
355 :    
356 :     [(0,0), (0,0), .... (0,0)]
357 :    
358 :     3. MLRiscGen: base pointer elimination.
359 :    
360 :     As part of the linkage mechanism, we generate the sequence:
361 :    
362 :     L: ... <- start of the code fragment
363 :    
364 :     L1:
365 :     base pointer <- linkreg - L1 + L
366 :    
367 :     The base pointer was then used for computing relocatable addresses
368 :     in the code fragment. Frequently (such as in lots of continuations)
369 :     this is not needed. We now eliminate this sequence whenever possible.
370 :    
371 :     For compile time efficiency, I'm using a very stupid local heuristic.
372 :     But in general, this should be done as a control flow analysis.
373 :    
374 :     Effect: Smaller code size. Speed up of most programs.
375 :    
376 :     4. Hppa back end
377 :    
378 :     Long jumps in span dependence resolution used to depend on the existence
379 :     of the base pointer.
380 :    
381 :     A jump to a long label L was expanded into the following sequence:
382 :    
383 :     LDIL %hi(L-8192), %r29
384 :     LDO %lo(L-8192)(%r29), %r29
385 :     ADD %r29, baseptr, %r29
386 :     BV,n %r0(%r29)
387 :    
388 :     In the presence of change (3) above, this will not work. I've changed
389 :     it so that the following sequence of instructions are generated, which
390 :     doesn't mention the base pointer at all:
391 :    
392 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
393 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
394 :     BV,n %r0(%r29) /* Jump */
395 :    
396 :     5. Alpha back end
397 :    
398 :     New alpha instructions LDB/LDW have been added, as per Fermin's
399 :     suggestions. This is unrelated to all other changes.
400 :    
401 :     6. X86 back end
402 :    
403 :     I've changed andl to testl in the floating point test sequence
404 :     whenever appropriate. The Intel optimization guide states that
405 :     testl is perferable to andl.
406 :    
407 :     7. RA (x86 only)
408 :    
409 :     I've improved the spill propagation algorithm, using an approximation
410 :     of maximal weighted independent sets. This seems to be necessary to
411 :     alleviate the negative effect in light of the slow down in (1).
412 :    
413 :     I'll write down the algorithm one of these days.
414 :    
415 :     8. MLRiscGen: frequencies
416 :    
417 :     I've added an annotation that states that all call gc blocks have zero
418 :     execution frequencies. This improves register allocation on the x86.
419 :    
420 :     BENCHMARKS
421 :     ==========
422 :    
423 :     I've only perform the comparison on 110.25.
424 :    
425 :     The platforms are:
426 :    
427 :     HPPA A four processor HP machine (E9000) with 5G of memory.
428 :     X86 A 300Hhz Pentium II with 128M of memory, and
429 :     SPARC An Ultra sparc 2 with 512M of memory.
430 :    
431 :     I used the following parameters for the SML benchmarks:
432 :    
433 :     @SMLalloc
434 :     HPPA 256k
435 :     SPARC 512k
436 :     X86 256k
437 :    
438 :     COMPILATION TIME
439 :     ----------------
440 :     Here are the numbers comparing the compilation times of the compilers.
441 :     I've only compared 110.25 compiling the new sources versus
442 :     a fixpoint version of the new compiler compiling the same.
443 :    
444 :     110.25 New
445 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
446 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
447 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
448 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
449 :    
450 :     110.25 New
451 :     Code Size Code Size
452 :     HPPA 8596736 8561421
453 :     SPARC 8974299 8785143
454 :     X86 9029180 8716783
455 :    
456 :     So in summary, things are at least as good as before. Dramatic
457 :     reduction in compilation is obtained on the Sparc; I can't explain it,
458 :     but it is reproducible. Perhaps someone should try to reproduce this
459 :     on their own machines.
460 :    
461 :     SML BENCHMARKS
462 :     --------------
463 :    
464 :     On the average, all benchmarks perform at least as well as before.
465 :    
466 :     HPPA Compilation Time Spill+Reload Run Time
467 :     110.25 New 110.25 New 110.25 New
468 :    
469 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
470 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
471 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
472 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
473 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
474 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
475 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
476 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
477 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
478 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
479 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
480 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
481 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
482 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
483 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
484 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
485 :     --------------------------------------------------------------------------
486 :     Average 6.12% 4.09%
487 :    
488 :     SPARC Compilation Time Spill+Reload Run Time
489 :     110.25 New 110.25 New 110.25 New
490 :    
491 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
492 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
493 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
494 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
495 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
496 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
497 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
498 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
499 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
500 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
501 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
502 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
503 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
504 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
505 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
506 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
507 :     ----------------------------------------------------------------------------
508 :     Average 6.94% 2.64%
509 :    
510 :     X86 Compilation Time Spill+Reload Run Time
511 :     110.25 New 110.25 New 110.25 New
512 :    
513 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
514 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
515 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
516 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
517 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
518 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
519 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
520 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
521 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
522 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
523 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
524 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
525 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
526 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
527 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
528 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
529 :     ----------------------------------------------------------------------------
530 :     Average 1.22% 3.36%
531 :    
532 :     ----------------------------------------------------------------------
533 :     Name: Allen Leung
534 : leunga 583 Date: 2000/03/23 16:25:00
535 :     Tag: leunga-20000323-fix_x86_alpha
536 :     Description:
537 :    
538 :     1. X86 fixes/changes
539 :    
540 :     a. The old code generated for SETcc was completely wrong.
541 :     The Intel optimization guide is VERY misleading.
542 :    
543 :     2. ALPHA fixes/changes
544 :    
545 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
546 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
547 :     c. Added reassociation code for address computation.
548 :    
549 :     ----------------------------------------------------------------------
550 :     Name: Allen Leung
551 : leunga 580 Date: 2000/03/22 01:23:00
552 :     Tag: leunga-20000322-fix_x86_hppa_ra
553 :     Description:
554 :    
555 :     1. X86 fixes/changes
556 :    
557 :     a. x86Rewrite bug with MUL3 (found by Lal)
558 :     b. Added the instructions FSTS, FSTL
559 :    
560 :     2. PA-RISC fixes/changes
561 :    
562 :     a. B label should not be a delay slot candidate! Why did this work?
563 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
564 :     as it should be.
565 :     c. The assembly syntax for fstds and fstdd was wrong.
566 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
567 :     operand variant of COMCLR/LDO.
568 :    
569 :     3. Generic MLRISC
570 :    
571 :     a. shuffle.sml rewritten to be slightly more efficient
572 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
573 :    
574 :     4. Register Allocator
575 :    
576 :     a. I now release the interference graph earlier during spilling.
577 :     May improve memory usage.
578 :    
579 :     ----------------------------------------------------------------------
580 : blume 577 Name: Matthias Blume
581 : blume 578 Date: 2000/03/14 14:15:32
582 :     Tag: blume_main_v110p26p1_2
583 :     Description:
584 :    
585 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
586 :     additional argument called "template" which is an optional string that
587 :     specifiel the layout of the tool command line. See the CM manual for
588 :     explanation.
589 :    
590 :     2. A special-purpose tool can be "regisitered" by simply dropping the
591 :     corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
592 :     directory where the .cm file lives that uses this tool. (The
593 :     behavior/misfeature until now was to look for the tool description
594 :     files in the current working directory.) As before, tool description
595 :     files could also be anchored -- in which case they can live anywhere
596 :     they like. Following the recent e-mail discussion, this change should
597 :     make it easier to have special-purpose tools that are shipped together
598 :     with the sources of the program that uses them.
599 :    
600 :     ----------------------------------------------------------------------
601 :     Name: Matthias Blume
602 : blume 577 Date: 2000/03/10 07:48:34
603 :     Tag: blume_main_v110p26p1_1
604 :     Description:
605 :    
606 :     I added a re-written version of Dave's fixpt script to src/system.
607 :     Changes relative to the original version:
608 :     - sh-ified (not everybody has ksh)
609 :     - automatically figures out which architecture it runs on
610 :     - uses ./makeml a bit more cleverly
611 :     - never invokes ./installml (and, thus, does not clobber your
612 :     good and working installation of sml in case something goes wrong)
613 :     - accepts max iteration count using option "-iter <n>"
614 :     - accepts a "base" name using option "-base <base>"
615 :    
616 :     It does not build any extraneous heap images but directly rebuilds
617 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
618 :     it can incorporate existing bin- and boot- hierarchies. For example,
619 :     suppose the base is set to "sml" (which is the default). Then it
620 :     successively builds
621 :    
622 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
623 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
624 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
625 :     ...
626 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
627 :    
628 :     and so on. If any of these already exist, it will just use what's
629 :     there. In particular, many people will have the initial set of bin
630 :     and boot files around, so this saves time for at least one full
631 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
632 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
633 :     they might be the result of an earlier partial run of fixpt (which
634 :     perhaps got accidentially killed). In this case, fixpt will quickly
635 :     move through what exists before continuing where it left off earlier,
636 :     and, thus, saves a lot of time.
637 :    
638 :     ----------------------------------------------------------------------
639 : leunga 576 Name: Allen Leung
640 :     Date: 00/03/10 02:20:00
641 :     Tag: leunga-20000310-fix_x86_asm_ra
642 :     Description:
643 : dbm 570
644 : leunga 576 More assembly output problems involving the indexed addressing mode
645 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
646 :     fix.
647 :    
648 :     The interface and implementation of the register allocator have been changed
649 :     slightly to accommodate the possibility to skip the register allocation
650 :     phases completely and go directly to memory allocation. This is needed
651 :     for C-- use.
652 :    
653 : dbm 570 ----------------------------------------------------------------------
654 : blume 572 Name: Matthias Blume
655 : blume 575 Date: 00/03/09 10:23:53
656 :     Tag: blume_main_v110p26p1_0
657 :     Description:
658 :    
659 :     * Complete re-organization of library names. Many libraries have been
660 :     consolidated so that they share the same path anchor. For example,
661 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
662 :     are SML/NJ-specific are under "smlnj". Notice that names like
663 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
664 :     for a complete description of the new naming scheme. Quick reference:
665 :    
666 :     host-cmb.cm -> smlnj/cmb.cm
667 :     host-compiler.cm -> smlnj/compiler.cm
668 :     full-cm.cm -> smlnj/cm.cm
669 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
670 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
671 :    
672 :     * Bug fixes in CM.
673 :     - exceptions in user code are being passed through (i.e., reach top level)
674 :     - more bugs in paranoia mode fixed
675 :     - bug related to checking group owners fixed
676 :    
677 :     * New install.sh script that automagically fetches archive files:
678 :     The new file config/srcarchiveurl must contain the URL of the
679 :     (remote) directory that contains bin files (or other source archives).
680 :     If install.sh does not find the archive locally, it tries to get
681 :     it from that remote directory.
682 :     This should simplify installation further: For machines that have
683 :     access to the internet, just fetch <version>-config.tgz, unpack it,
684 :     edit config/targets, and go (run config/install.sh). The scipt will
685 :     fetch everything else that it might need all by itself.
686 :    
687 :     For CVS users, this mechanism is not relevant for source archives, but
688 :     it is convenient for getting new sets of binfiles.
689 :    
690 :     Archives should be tar files compressed with either gzip, compress, or
691 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
692 :    
693 :     ----------------------------------------------------------------------
694 :     Name: Matthias Blume
695 : blume 572 Date: 2000/03/07 04:01:04
696 :     Tag: blume_main_v110_26_2
697 : dbm 570 Description:
698 : blume 572 - size info in BOOTLIST
699 :     * no fixed upper limits for number of bootfiles or length of
700 :     bootfile names in runtime
701 :     * falling back to old behavior if no BOOTLIST size info found
702 :     - allocation size heuristics in .run-sml
703 :     * tries to read cache size from /proc/cpuinfo (this is important for
704 :     small-cache Celeron systems!)
705 :     - install.sh robustified
706 :     - CM manual updates
707 :     - paranoid mode
708 :     * no more CMB.deliver() (i.e., all done by CMB.make())
709 :     * can re-use existing sml.boot.* files
710 :     * init.cmi now treated as library
711 :     * library stamps for consistency checks
712 :     - sml.boot.<arch>-<os>/PIDMAP file
713 :     * This file is read by the CM startup code. This is used to minimize
714 :     the amount of dynamic state that needs to be stowed away for the
715 :     purpose of sharing between interactive system and user code.
716 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
717 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
718 :     like most other CM state variables.
719 :     - Compiler.CMSA eliminated
720 :     * No longer supported by CM anyway.
721 :     - fixed bugs in pickler that kept biting Stefan
722 :     * past refs to past refs (was caused by the possibility that
723 :     ad-hoc sharing is more discriminating than hash-cons sharing)
724 :     * integer overflow on LargeInt.minInt
725 :     - ml-{lex,yacc} build scripts now use new mechanism
726 :     for building standalone programs
727 :     - fixed several gcc -Wall warnings that were caused by missing header
728 :     files, missing initializations, etc., in runtime (not all warnings
729 :     eliminated, though)

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