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 876 - (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 : george 824
15 : leunga 815 ----------------------------------------------------------------------
16 : blume 826 Name: Matthias Blume
17 : blume 875 Date: 2001/07/10 17:30:00 EDT
18 :     Tag: Release_110_34
19 :     Description:
20 :    
21 : blume 876 Minor tweak to 110.34 (re-tagged):
22 :    
23 :     - README.html file added to CVS repository
24 :     - runtime compiles properly under FreeBSD 3.X and 4.X
25 :    
26 :     ----------------------------------------------------------------------
27 :     Name: Matthias Blume
28 :     Date: 2001/07/10 17:30:00 EDT
29 :     Tag: Release_110_34
30 :     Description:
31 :    
32 : blume 875 New version number (110.34). New bootfiles.
33 :    
34 :     ----------------------------------------------------------------------
35 :     Name: Matthias Blume
36 : blume 874 Date: 2001/07/09 16:00:00 EDT
37 :     Tag: blume-20010709-more-varargs
38 :     Description:
39 :    
40 :     I changed the handling of varargs in ml-nlffigen again:
41 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
42 :    
43 :     The immediate effect is that you can actually call a varargs function
44 :     from ML -- but you can't actually supply any arguments beyond the ones
45 :     specified explicitly. (For example, you can call printf with its format
46 :     string, but you cannot pass additional arguments.)
47 :    
48 :     This behavior is only marginally more useful than the one before, but
49 :     it has the advantage that a function or, more importantly, a function
50 :     type never gets dropped on the floor, thus avoiding follow-up problems with
51 :     other types that refer to the offending one.
52 :    
53 :     ----------------------------------------------------------------------
54 :     Name: Matthias Blume
55 : blume 873 Date: 2001/07/09 11:25:00 EDT
56 :     Tag: blume-20010709-varargs
57 :     Description:
58 :    
59 :     1. ckit-lib.cm now exports structure Error
60 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
61 :     with a warning accompanied by a source location. Moreover, it
62 :     merely skips the offending function or type and proceeds with the
63 :     rest of its work.u As a result, one can safely feed C code containing
64 :     "..." to ml-nlffigen.
65 :     3. There are some internal improvements to CM, providing slightly
66 :     more general string substitutions in the tools subsystem.
67 :    
68 :     ----------------------------------------------------------------------
69 :     Name: Matthias Blume
70 : blume 854 Date: 2001/06/27 15:10:00 EDT
71 :     Tag: blume-20010627-concur
72 :     Description:
73 :    
74 :     Fixed a small bug in CM's handling of parallel compilation.
75 :     (You could observe the bug by Control-C-interrupting an ordinary
76 :     CMB.make or CM.stabilize and then attaching some compile servers.
77 :     The result was that all of a sudden the previously interrupted
78 :     compilation would continue on its own. This was because of
79 :     an over-optimization: CM did not bother to clean out certain queues
80 :     when no servers were attached "anyway", resulting in the contents
81 :     of these queues to grab control when new servers did get attached.)
82 :    
83 :     There is also another minor update to the CM manual.
84 :    
85 :     ----------------------------------------------------------------------
86 :     Name: Matthias Blume
87 : blume 853 Date: 2001/06/26 16:15:00 EDT
88 :     Tag: blume-20010626-cmdoc
89 :     Description:
90 :    
91 :     Minor typo fixed in CM manual (syntax diagram for libraries).
92 :    
93 :     ----------------------------------------------------------------------
94 :     Name: Matthias Blume
95 : blume 852 Date: 2001/06/25 22:55:00 EDT
96 :     Tag: blume-20010625-x86pc
97 :     Description:
98 :    
99 :     Fixed a nasty bug in the X86 assembly code that caused signal
100 :     handlers to fail (crash) randomly.
101 :    
102 :     ----------------------------------------------------------------------
103 :     Name: Matthias Blume
104 : blume 846 Date: 2001/06/25 12:05:00 EDT
105 :     Tag: blume-20010625-nlffigen
106 :     Description:
107 :    
108 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
109 :     Nick Carter <nbc@andrew.cmu.edu>.
110 :    
111 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
112 :     2. Default names for generated files are now derived from the name of
113 :     the C file *without its directory*. In particular, this causes generated
114 :     files to be placed locally even if the C file is in some system directory.
115 :     3. Default names for generated signatures and structures are also derived
116 :     from the C file name without its directory. This avoids silly things
117 :     like "structure GL/GL".
118 :     (Other silly names are still possible because ml-nlffigen does not do
119 : blume 847 a thorough check of whether generated names are legal ML identifiers.
120 :     When in doubt, use command line arguments to force particular names.)
121 : blume 846
122 :     ----------------------------------------------------------------------
123 :     Name: Matthias Blume
124 : blume 845 Date: 2001/06/21 12:25:00 EDT
125 :     Tag: blume-20010621-eXene
126 :     Description:
127 :    
128 :     eXene now compiles and (sort of) works again.
129 :    
130 :     The library name (for version > 110.33) is $/eXene.cm.
131 :    
132 :     I also added an new example in src/eXene/examples/nbody. See the
133 :     README file there for details.
134 :    
135 :     ----------------------------------------------------------------------
136 :     Name: Matthias Blume
137 : blume 844 Date: 2001/06/20 16:40:00 EDT
138 :     Tag: blume-20010620-cml
139 :     Description:
140 :    
141 :     CML now compiles and works again.
142 :    
143 :     Libraries (for version > 110.33):
144 :    
145 :     $cml/cml.cm Main CML library.
146 :     $cml/basis.cm CML's version of $/basis.cm.
147 :     $cml/cml-internal.cm Internal helper library.
148 :     $cml/core-cml.cm Internal helper library.
149 :     $cml-lib/trace-cml.cm Tracing facility.
150 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
151 :    
152 :     The installer (config/install.sh) has been taught how to properly
153 :     install this stuff.
154 :    
155 :     ----------------------------------------------------------------------
156 :     Name: Matthias Blume
157 : blume 843 Date: 2001/06/19 17:55:00 EDT
158 :     Tag: blume-20010619-instantiate
159 :     Description:
160 :    
161 :     This un-breaks the fix for bug 1432.
162 :     (The bug was originally fixed in 110.9 but I broke it again some
163 :     time after that.)
164 :    
165 :     ----------------------------------------------------------------------
166 :     Name: Matthias Blume
167 : blume 842 Date: 2001/06/19 17:25:00 EDT
168 :     Tag: blume-20010619-signals
169 :     Description:
170 :    
171 :     This should (hopefully) fix the long-standing signal handling bug.
172 :     (The runtime system was constructing a continuation record with an
173 :     incorrect descriptor which would cause the GC to drop data on the floor...)
174 :    
175 :     ----------------------------------------------------------------------
176 :     Name: Matthias Blume
177 : blume 840 Date: 2001/06/15 15:05:00 EDT
178 : blume 841 Tag: blume-20010615-moresparc
179 :     Description:
180 :    
181 :     Here is a short late-hour update related to Sparc c-calls:
182 :    
183 :     -- made handling of double-word arguments a bit smarter
184 :    
185 :     -- instruction selection phase tries to collapse certain clumsily
186 :     constructed ML-Trees; typical example:
187 :    
188 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
189 :    
190 :     This currently has no further impact on SML/NJ since mlriscGen does
191 :     not seem to generate such patterns in the first place, and c-calls
192 :     (which did generate them in the beginning) has meanwhile been fixed
193 :     so as to avoid them as well.
194 :    
195 :     ----------------------------------------------------------------------
196 :     Name: Matthias Blume
197 :     Date: 2001/06/15 15:05:00 EDT
198 : blume 840 Tag: blume-20010615-sparc
199 :     Description:
200 :    
201 :     The purpose of this update is to provide an implementation of NLFFI
202 :     on Sparc machines.
203 :    
204 :     Here are the changes in detail:
205 :    
206 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
207 :     the Sparc implementation of the c-calls API.
208 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
209 :     for accessing the ML frame. Thus, we have a real frame pointer and
210 :     can freely modify %sp without need for an omit-frame-ptr phase.
211 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
212 :     this case.
213 :     * ml-nlffigen has been taught to produce code for different architectures
214 :     and calling conventions.
215 :     * In a way similar to what was done in the x86 case, the Sparc
216 :     backend uses its own specific extension to mltree. (For example,
217 :     it needs to be able to generate UNIMP instructions which are part
218 :     of the calling convention.)
219 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
220 :     to make it easier to plug in new machine- and os-dependent parts).
221 :    
222 :     There are some other fairly unrelated bug fixes and cleanups as well:
223 :    
224 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
225 :     that they properly share their libraries with existing SML/NJ libraries.
226 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
227 :     follow-up error messages.
228 :     * Updates to CM/CMB documentation.
229 :    
230 :     TODO items:
231 :    
232 :     * MLRISC should use a different register as its asmTemp on the Sparc.
233 :     (The current %o2 is a really bad choice because it is part of the
234 :     calling conventions, so things might interfere in unexpected ways.)
235 :    
236 :     ----------------------------------------------------------------------
237 :     Name: Matthias Blume
238 : blume 839 Date: 2001/06/07
239 :     Tag: blume-20010607-calls
240 :     Description:
241 :    
242 :     A number of internal changes related to C calls and calling conventions:
243 :    
244 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
245 :     number of bytes popped implicitly (by the callee). In most cases
246 :     this field is 0 but on x86/win32 it is some non-zero value. This
247 :     is information provided for the benefit of the "omit-frameptr" pass.
248 :     2. The CALL instruction on the x86 carries a similar "pops" field.
249 :     The instruction selection phase copies its value from the ML-Tree
250 :     CALL statement.
251 :     3. On all other architectures, the instruction selection phase checks
252 :     whether "pops=0" and complains if not.
253 :     4. The c-calls implementation for x86 now accepts two calling conventions:
254 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
255 :     up after the call and pops is set to 0. For "stdcall", the caller
256 :     does nothing, leaving the cleanup to the callee; pops is set to
257 :     the number of bytes that were pushed onto the stack.
258 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
259 :     distinguish between "ccall" and "stdcall".
260 :     6. The UNIMP instruction has been added to the supported Sparc instruction
261 :     set. (This is needed for implementing the official C calling convention
262 :     on this architecture.)
263 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
264 :     work with the latest CM.
265 :    
266 :     ----------------------------------------------------------------------
267 :     Name: Matthias Blume
268 : blume 838 Date: 2001/06/05 15:10:00 EDT
269 :     Tag: blume-20010605-cm-index
270 :     Description:
271 :    
272 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
273 :    
274 :     1. CM can now generate "index files". These are human-readable files
275 :     that list on a per-.cm-file basis each toplevel symbol defined or
276 :     imported. The location of the index file for
277 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
278 :     To enable index-file generation, set CM.Control.generate_index to true
279 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
280 :    
281 :     The CM manual has been updated accordingly.
282 :    
283 :     2. I made some slight modifications to the c-calls API in MLRISC.
284 :    
285 :     a) There is now a callback to support saving/restoring of
286 :     dedicated but caller-save registers around the actual call
287 :     instruction.
288 :     b) One can optionally specify a comment-annotation for the
289 :     call instruction.
290 :    
291 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
292 :     (For example, the comment annotation shows the C prototype of
293 :     the function being called.)
294 :    
295 :     ----------------------------------------------------------------------
296 :     Name: Matthias Blume
297 : blume 837 Date: 2001/06/01 13:30:00 EDT
298 :     Tag: blume-20010601-nlffi-cleanup
299 :     Description:
300 :    
301 :     This is mostly a cleanup of MLFFI stuff:
302 :    
303 :     - some signature files have been put into a more exposed place
304 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
305 :     - ml-nlffigen changed accordingly
306 :     - tutorial updated
307 :    
308 :     Other changes:
309 :    
310 :     - author's affiliation in CM manual(s) updated
311 :     - some more recognized keywords added to Allen's sml.sty
312 :    
313 :     ----------------------------------------------------------------------
314 :     Name: Matthias Blume
315 : blume 836 Date: 2001/05/25 15:30:00 EDT
316 :     Tag: blume-20010525-iptr
317 :     Description:
318 :    
319 :     - put the official 110.33-README (as it appears on the ftp server) under
320 :     CVS
321 :     - fixed a small bug related to incomplete pointer types in
322 :     ml-nlffigen
323 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
324 :     (it does not need the 'f type parameter)
325 :    
326 :     ----------------------------------------------------------------------
327 :     Name: Matthias Blume
328 : blume 834 Date: 2001/05/23 14:30:00 EDT
329 :     Tag: Release_110_33
330 :     Description:
331 :    
332 :     New version number (110.33). New bootfiles.
333 :    
334 :     ----------------------------------------------------------------------
335 :     Name: Matthias Blume
336 : blume 833 Date: 2001/05/22 18:06:00 EDT
337 :     Tag: blume-20010522-targets
338 :     Description:
339 :    
340 :     Made install.sh use file config/targets.customized if it exists, falling
341 :     back to config/targets if it doesn't. This way one can have a customized
342 :     version of the targets file without touching the "real thing", thus
343 :     eliminating the constant fear of accidentally checking something bogus
344 :     back into the CVS repository... (File config/targets.customized must
345 :     not be added to the repository!)
346 :    
347 :     ----------------------------------------------------------------------
348 :     Name: Matthias Blume
349 : blume 832 Date: 2001/05/22 16:30:00 EDT
350 :     Tag: blume-20010522-minitut
351 :     Description:
352 :    
353 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
354 :     struct returns.
355 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
356 :     incomplete, preliminary documentation for NLFFI.
357 :    
358 :     ----------------------------------------------------------------------
359 :     Name: Matthias Blume
360 : blume 829 Date: 2001/05/14 11:30:00 EDT
361 :     Tag: blume-20010514-script
362 :     Description:
363 :    
364 :     Some bugs in install script fixed.
365 :    
366 :     In addition to that I also made a slight change to the NLFFI API:
367 :     Functors generated by ml-nlffigen now take the dynamic library as a
368 :     straight functor argument, not as a suspended one. (The original
369 :     functor code used to force the suspension right away anyway, so there
370 :     was nothing gained by this complication of the interface.)
371 :    
372 :     ----------------------------------------------------------------------
373 :     Name: Matthias Blume
374 : blume 828 Date: 2001/05/11 14:35:00 EDT
375 :     Tag: blume-20010511-ml-nlffi
376 :     Description:
377 :    
378 :     I finally took the plunge and added my new FFI code to the main
379 :     repository. For x86-linux it is now ready for prime-time.
380 :    
381 :     There are two new subdirectories of "src":
382 :    
383 :     - ml-nlffi-lib:
384 :     The utility library for programs using the FFI interface.
385 :     Here is the implementation of $/c.cm and its associated low-level
386 :     partners $/c-int.cm and $/memory.cm.
387 :     - ml-nlffigen:
388 :     A stand-alone program for generating ML glue code from C source
389 :     code.
390 :    
391 :     Building ml-nlffigen requires $/ckit-lib.cm.
392 :    
393 :     The config/install.sh script has been updates to do the Right Thing
394 :     (hopefully).
395 :    
396 :     Notice that the source tree for the C-Kit will not be put under "src"
397 :     but directly under the installation root directory. (This is the
398 :     structure that currently exists on the CVS server when you check out
399 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
400 :    
401 :     Bugs: No documentation yet.
402 :    
403 :     ----------------------------------------------------------------------
404 :     Name: Matthias Blume
405 : blume 826 Date: 2001/05/09 16:35:00 EDT
406 :     Tag: blume-20010509-cpscontract
407 :     Description:
408 :    
409 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
410 :     wrapper/unwrapper elimination did not decrement usage counts and some
411 :     dead variables got overlooked by the dead-up logic.)
412 :    
413 :     ----------------------------------------------------------------------
414 : george 824 Name: Lal George
415 :     Date: 2001/05/08 17:26:09 EDT
416 :     Tag: george-20010508-omit-frameptr
417 :     Description:
418 :    
419 :     Changes to implement the omit-frame-pointer optimization to support
420 :     raw C calls. For now, there is only support on the Intel x86, but
421 :     other architectures will follow as more experience is gained with this.
422 :    
423 :    
424 :     ----------------------------------------------------------------------
425 : blume 818 Name: Matthias Blume
426 : blume 819 Date: 2001/05/07 14:40:00 EDT
427 :     Tag: blume-20010507-proxies
428 :     Description:
429 :    
430 :     I made into "proxy libraries" all libraries that qualify for such a
431 :     change. (A qualifying library is a library that has another library or
432 :     groups as its sole member and repeats that member's export list
433 :     verbatim. A proxy library avoids this repetition by omitting its export
434 :     list, effectively inheriting the list that its (only) member exports.
435 :     See the CM manual for more explanation.)
436 :     The main effect is that explicit export lists for these libraries
437 :     do not have to be kepts in sync, making maintenance a bit easier.
438 :    
439 :     I also added copyright notices to many .cm-files.
440 :    
441 :     Last but not least, I made a new set of bootfiles.
442 :    
443 :     ----------------------------------------------------------------------
444 :     Name: Matthias Blume
445 : blume 818 Date: 2001/05/04 17:00:00 EDT
446 :     Tag: blume-20010504-cm-lsplit
447 :     Description:
448 :    
449 :     0. John merged pending changes to $/smlnj-lib.cm
450 :    
451 :     1. Allen's previous change accidentally backed out of one of Lal's
452 :     earlier changes. I undid this mistake (re-introducing Lal's change).
453 :    
454 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
455 :     within the compiler where applicable. There is some code simplification
456 :     because of that.
457 :    
458 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
459 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
460 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
461 :     it can now also be controlled on a per-source basis: CM has been taught
462 :     a new tool parameter applicable to ML source files.
463 :    
464 :     - To turn lambda-splitting off completely:
465 :     local open Compiler.Control.LambdaSplitting in
466 :     val _ = set Off
467 :     end
468 :     - To make "no lambda-splitting" the global default (but allow per-source
469 :     overriding); this is the initial setting:
470 :     local open Compiler.Control.LambdaSplitting in
471 :     val _ = set (Default NONE)
472 :     end
473 :     - To make "lambda-splitting with aggressiveness a" the global default
474 :     (and allow per-source overriding):
475 :     local open Compiler.Control.LambdaSplitting in
476 :     val _ = set (Default (SOME a))
477 :     end
478 :    
479 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
480 :     write (in the respective .cm-file):
481 :     a.sml (lambdasplitting:off)
482 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
483 :     a.sml (lambdasplitting:on)
484 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
485 :     <a> is a decimal non-negative integer):
486 :     a.sml (lambdasplitting:<a>)
487 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
488 :     a.sml (lambdasplitting:infinity)
489 :     - To use the global default for a.sml:
490 :     a.sml (lambdasplitting:default)
491 :     or simply
492 :     a.sml
493 :    
494 :     ----------------------------------------------------------------------
495 : leunga 815 Name: Allen Leung
496 :     Date: 2001/05/04 01:57:00 EDT
497 :     Tag: leunga-20010504-sync
498 :     Description:
499 : leunga 788
500 : leunga 815 MLRISC features.
501 :    
502 :     1. Fix to CMPXCHG instructions.
503 :     2. Changed RA interface to allow annotations in callbacks.
504 :     3. Added a new method to the stream interface to allow annotations updates.
505 :    
506 : george 749 ----------------------------------------------------------------------
507 : blume 812 Name: Matthias Blume
508 : blume 814 Date: 2001/05/01 11:45:00 EDT
509 :     Tag: blume-20010501-pcedittmp
510 :     Description:
511 :    
512 :     Changed install.sh to use the current working directory instead of
513 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
514 :     of /usr/tmp caused trouble with MacOS X because of file premission
515 :     problems.
516 :    
517 :     ----------------------------------------------------------------------
518 :     Name: Matthias Blume
519 : blume 812 Date: 2001/04/20 11:10:00 EDT
520 :     Tag: blume-20010420-inMLflag
521 :     Description:
522 :    
523 :     - added vp_limitPtrMask to vproc-state.h
524 :     (for use by the raw-C-calls mechanism to implement proper interrupt
525 :     handling)
526 :     - made the ML compiler aware of various data-structure offsets so it
527 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
528 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
529 :     raw C-calls
530 :    
531 :     ----------------------------------------------------------------------
532 : george 810 Name: Lal George
533 :     Date: 2001/04/20 09:15:28 EDT
534 :     Tag: george-20010420-macosX
535 :     Description:
536 :    
537 :     - Changes to port to Mac OS X; Darwin.
538 :    
539 :     - In the process I found that sqrt was broken on the PPC, because the
540 :     fsqrt instruction is not implemented.
541 :    
542 :     ----------------------------------------------------------------------
543 : blume 799 Name: Matthias Blume
544 : blume 808 Date: 2001/04/18 12:45:00 EDT
545 :     Tag: blume-20010418-ccalls
546 :     Description:
547 :    
548 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
549 :     (this bug prevented structure arguments containing pointers from being
550 :     passed correctly)
551 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
552 :     structure arguments are represented as a pointer to the beginning
553 :     of the structure (instead of having a series of synthesized arguments,
554 :     one for each structure member)
555 :    
556 :     - made makeml script's verbosity level configurable via environment
557 :     variable (MAKEML_VERBOSITY)
558 :    
559 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
560 : blume 808 in rawmem-x86.sml; we are now using the real thing
561 :    
562 :     ----------------------------------------------------------------------
563 :     Name: Matthias Blume
564 : blume 806 Date: 2001/03/22 16:25:00 EST
565 :     Tag: blume-20010322-bootfiles
566 :     Description:
567 :    
568 :     Created a new set of bootfiles (for your automatic installation convenience).
569 :    
570 :     ----------------------------------------------------------------------
571 :     Name: Matthias Blume
572 : blume 805 Date: 2001/03/22 15:10:00 EST
573 :     Tag: blume-20010322-rawmem-parcm
574 :     Description:
575 :    
576 :     1. All "raw memory access" primitives for the new FFI are implemented now
577 :     (at least on the x86).
578 :     2. Some further cleanup of CM's parallel make mechanism.
579 :    
580 :     ----------------------------------------------------------------------
581 :     Name: Matthias Blume
582 : blume 801 Date: 2001/03/19 17:53:00 EST
583 :     Tag: blume-20010319-parallel
584 :     Description:
585 :    
586 :     Parallel make (using compile servers) now works again.
587 :    
588 :     To this end, CM.stabilize and CMB.make have been modified to work in
589 :     two passes when compile servers are attached:
590 :     1. Compile everything, do not perform stabilization; this pass
591 :     uses compile servers
592 :     2. Stabilize everything; this pass does not use compile servers
593 :     If there are no compile servers, the two passes are combined into one
594 :     (as before). Splitting the passes increases the inherent parallelism
595 :     in the dependency graph because the entire graph including all
596 :     libraries is available at the same time. This, in turn, improves
597 :     server utilization. The downside is that the master process will
598 :     have to do some extra work after compilation is done (because for
599 :     technical reasons it must re-read all the binfiles during stabilization).
600 :    
601 :     ----------------------------------------------------------------------
602 :     Name: Matthias Blume
603 : blume 800 Date: 2001/03/16 12:22:00 EST
604 :     Tag: blume-20010316-bootfiles
605 :     Description:
606 :    
607 :     Created a new set of bootfiles (for your automatic installation convenience).
608 :    
609 :     ----------------------------------------------------------------------
610 :     Name: Matthias Blume
611 : blume 799 Date: 2001/03/16 11:00:00 EST
612 :     Tag: blume-20010316-MLTREE-fixup
613 :     Description:
614 :    
615 :     This is a minor fixup for an (untagged) earlier commit by Allen.
616 :     (A file was missing).
617 :    
618 :     ----------------------------------------------------------------------
619 : leunga 788 Name: Allen Leung
620 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
621 :     Tag: leunga-20010305-cut-support
622 :    
623 :     1. New support for alternative control-flow in MLTREE.
624 :     Currently we support
625 :    
626 :     FLOW_TO(CALL ...., [k1,...,kn])
627 :    
628 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
629 :     in Moby
630 :    
631 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
632 :    
633 :     2. Register Allocator
634 :    
635 :     Changes in interface [from Fermin, John]
636 :    
637 :     3. Alpha 8-bit SLL support [Fermin]
638 :    
639 :     4. All architectures
640 :    
641 :     A new module (ClusterExpandCopies) for expanding parallel copies.
642 :    
643 :     ----------------------------------------------------------------------
644 :     Name: Allen Leung
645 : leunga 788 Date: 2001/02/27 23:07:00 EST
646 :     Tag: leunga-20010227-minor-stuff
647 :    
648 :     1. Alpha bug fix for CMOVNE
649 :     2. Handle mltree COND(..,FCMP ...,...)
650 :     3. Bug fix in simplifier
651 :    
652 :     ----------------------------------------------------------------------
653 : blume 777 Name: Matthias Blume
654 : blume 781 Date: 2001/01/30 17:50:00 EST
655 :     Tag: blume-20010130-sync
656 :     Description:
657 :    
658 :     This is just a minor update to sync my devel branch with the main brach.
659 :     The only visible change is the addition of some README files.
660 :    
661 :     ----------------------------------------------------------------------
662 :     Name: Matthias Blume
663 : blume 778 Date: 2001/01/12 23:30:00 JST
664 :     Tag: blume-20010112-bootfiles
665 :     Description:
666 :    
667 :     Made a new set of bootfiles that goes with the current state of the
668 :     repository.
669 :    
670 :     ----------------------------------------------------------------------
671 :     Name: Matthias Blume
672 : blume 777 Date: 2001/01/12 21:20:00 JST
673 :     Tag: blume-20010112-sync
674 :     Description:
675 :    
676 :     I am just flushing out some minor changes that had accumulated in
677 :     my private branch in order to sync with the main tree. (This is
678 :     mainly because I had CVS trouble when trying to merge _into_ my
679 :     private branch.)
680 :    
681 :     Most people should be completely unaffected by this.
682 :    
683 :     ----------------------------------------------------------------------
684 : leunga 775 Name: Allen Leung
685 :     Date: Thu Jan 11 21:03:00 EST 2001
686 :     Tag: leunga-20010111-labexp=mltree
687 :     Description:
688 :    
689 :     1. Removed the type LabelExp and replace it by MLTree.
690 :     2. Rewritten mltree-simplify with the pattern matcher tool.
691 :     3. There were some bugs in alpha code generator which would break
692 :     64-bit code generation.
693 :     4. Redo the tools to generate code with the
694 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
695 :     are now generated by perl scripts.
696 :    
697 :     ----------------------------------------------------------------------
698 : blume 771 Name: Matthias Blume
699 : blume 774 Date: 2001/01/10 21:55:00 JST
700 :     Tag: blume-20010110-rcc
701 :     Description:
702 :    
703 :     The RCC stuff now seems to work (but only on the x86).
704 :     This required hacking of the c-calls interface (and -implementation) in
705 :     MLRISC.
706 :    
707 :     Normal compiler users should be unaffected.
708 :    
709 :     ----------------------------------------------------------------------
710 :     Name: Matthias Blume
711 : blume 773 Date: 2001/01/09 01:20:00 JST
712 :     Tag: blume-20010109-rcc
713 :     Description:
714 :    
715 :     This is a fairly big patch, flushing out a large number of pending
716 :     changes that I made to my development copy over the last couple of days.
717 :    
718 :     Of practical relevance at this moment is a workaround for a pickling
719 :     bug that Allen ran into the other day. The cause of the bug itself is
720 :     still unknown and it might be hard to fix it properly, but the
721 :     workaround has some merits of its own (namely somewhat reducing pickling
722 :     overhead for certain libraries). Therefore, I think this solution should
723 :     be satisfactory at this time.
724 :    
725 :     The rest of the changes (i.e., the vast majority) has to do with my
726 :     ongoing efforts of providing direct support for C function calls from
727 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
728 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
729 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
730 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
731 :     handle RCC.
732 :    
733 :     The last part (the changes to mlriscGen.sml) are still known to be
734 :     wrong on the x86 and not implemented on all other architectures. But
735 :     the infrastructure is in place. I had to change a few functor
736 :     signatures in the backend to be able to route the CCalls interface
737 :     from MLRISC there, and I had to specialize the mltree type (on the
738 :     x86) to include the necessary extensions. (The extensions themselves
739 :     were already there and redy to go in MLRISC/x86).
740 :    
741 :     Everything should be very happy as soon as someone helps me with
742 :     mlriscGen.sml...
743 :    
744 :     In any case, nothing of this should matter to anyone as long as the
745 :     new primop is not being used (which is going to be the case unless you
746 :     find it where I hid it :). The rest of the compiler is completely
747 :     unaffected.
748 :    
749 :     ----------------------------------------------------------------------
750 :     Name: Matthias Blume
751 : blume 772 Date: 2001/01/05 00:30:00 JST
752 :     Tag: blume-20010105-primops
753 :     Description:
754 :    
755 :     Added some experimental support for work that I am doing right now.
756 :     These changes mostly concern added primops, but there is also a new
757 :     experimental C library in the runtime system (but currently not enabled
758 :     anywhere except on Linux/X86).
759 :    
760 :     In the course of adding primops (and playing with them), I discovered that
761 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
762 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
763 :     To recover, I made type information in INL_PRIM mandatory and changed
764 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
765 :     has complete, correct type information (i.e., no bottom types).
766 :    
767 :     Since all these changes mean that we need new binfiles, I also bumped the
768 :     version number to 110.32.1.
769 :    
770 :     ----------------------------------------------------------------------
771 :     Name: Matthias Blume
772 : blume 771 Date: 2000/12/30 22:10:00 JST
773 :     Tag: blume-20001230-various
774 :     Description:
775 :    
776 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
777 :     to each other using path anchors. (See CM manual for explanation.)
778 :    
779 :     Updated CM documentation.
780 :    
781 :     Fixed some bugs in CM.
782 :    
783 :     Implemented "proxy" libraries (= syntactic sugar for CM).
784 :    
785 :     Added "-quiet" option to makeml and changed runtime system accordingly.
786 :    
787 :     Added cleanup handler for exportML to reset timers and compiler stats.
788 :    
789 :     ----------------------------------------------------------------------
790 : george 761 Name: Lal George
791 :     Date: 2000/12/22 22:22:58 EST 2000
792 :     Tag: Release_110_32
793 :     Description:
794 :    
795 :     Infinite precision used throughout MLRISC.
796 :     see MLRISC/mltree/machine-int.sig
797 :    
798 :     ----------------------------------------------------------------------
799 : blume 756 Name: Matthias Blume
800 : blume 760 Date: 2000/12/22 23:16:00 JST
801 :     Tag: blume-20001222-warn
802 :     Description:
803 : blume 759
804 : blume 760 Corrected wording and formatting of some CM warning message which I
805 :     broke in my previous patch.
806 : blume 759
807 :     ----------------------------------------------------------------------
808 :     Name: Matthias Blume
809 : blume 763 Date: 2000/12/22 21:20:00 JST
810 :     Tag: blume-20001222-anchorenv
811 :     Description:
812 :    
813 :     Fixed CM's handling of anchor environments in connection with CMB.make.
814 :    
815 :     ----------------------------------------------------------------------
816 :     Name: Matthias Blume
817 : blume 758 Date: 2000/12/22 13:15:00 JST
818 :     Tag: blume-20001222-cleanup
819 :     Description:
820 :    
821 :     Removed src/cm/ffi which does not (and did not) belong here.
822 :    
823 :     ----------------------------------------------------------------------
824 :     Name: Matthias Blume
825 : blume 757 Date: 2000/12/21 23:55:00 JST
826 :     Tag: blume-20001221-exn
827 :     Description:
828 :    
829 :     Probably most important: CM no longer silently swallows all exceptions
830 :     in the compiler.
831 :     Plus: some other minor CM changes. For example, CM now reports some
832 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
833 :    
834 :     ----------------------------------------------------------------------
835 :     Name: Matthias Blume
836 : blume 756 Date: 2000/12/15 00:01:05 JST
837 :     Tag: blume-20001215-dirtool
838 :     Description:
839 :    
840 :     - "dir" tool added.
841 :     - improvements and cleanup to Tools structure
842 :     - documentation updates
843 :    
844 :     ----------------------------------------------------------------------
845 : leunga 752 Name: Allen Leung
846 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
847 :     Description:
848 :     Tag: leunga-20001214-int-inf
849 :     Description:
850 :    
851 :     In IntInf, added these standard functions, which are missing from our
852 :     implementation:
853 :    
854 :     andb : int * int -> int
855 :     xorb : int * int -> int
856 :     orb : int * int -> int
857 :     notb : int -> int
858 :     << : int * word -> int
859 :     ~>> : int * word -> int
860 :    
861 :     Not tested, I hope they are correct.
862 :    
863 :     ----------------------------------------------------------------------
864 :     Name: Allen Leung
865 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
866 :     Description:
867 :     Tag: leunga-20001208-nowhere
868 :     Description:
869 :    
870 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
871 :     to generate better error messages etc. Plus a brief manual.
872 :    
873 :     ----------------------------------------------------------------------
874 : george 749 Name: Lal George
875 :     Date: 2000/12/08 09:54:02 EST 2000
876 :     Tag: Release_110_31
877 :     Description:
878 : george 705
879 : george 749 - Version 110.31
880 : leunga 591 ----------------------------------------------------------------------
881 : leunga 744 Name: Allen Leung
882 :     Date: Thu Dec 7 22:01:04 EST 2000
883 :     Tag: leunga-20001207-cell-monster-hack
884 :     Description:
885 :    
886 :     Major MLRISC internal changes. Affect all clients.
887 :     Summary:
888 :    
889 :     1. Type CELLS.cell = int is now replaced by a datatype.
890 :     As a result, the old regmap is now gone. Almost all interfaces
891 :     in MLRISC change as a consequence.
892 :    
893 :     2. A new brand version of machine description tool (v3.0) that generates
894 :     modules expecting the new interface. The old version is removed.
895 :    
896 :     3. The RA interface has been further abstracted into two new functors.
897 :     RISC_RA and X86RA. These functors have much simpler interfaces.
898 :     [See also directory MLRISC/demo.]
899 :    
900 :     4. Some other new source->source code generation tools are available:
901 :    
902 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
903 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
904 :     I use this tool to generate the peephole optimizers---with the new
905 :     cell type changes, peephole rules are becoming difficult to write
906 :     without conditional pattern matching.
907 :    
908 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
909 :     cover the entire MLRISC source tree so many things broke.
910 :    
911 :     6. CM files have been moved to the subdirectory MLRISC/cm.
912 :     They are moved because there are a lot of them and they clutter up the
913 :     root dir.
914 :    
915 :     7. More detailed documentation to come...
916 :    
917 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
918 :     a makeml -rebuild first. This is because of other other
919 :     changes that Matthias has made (see below).
920 :    
921 :    
922 :     ----------------------------------------------------------------------
923 : blume 733 Name: Matthias Blume
924 : blume 742 Date: 2000/11/30 23:12:00 JST
925 :     Tag: blume-20001130-filereorg
926 :     Description:
927 :    
928 :     Some manual updates and some file reorganizations in CM.
929 :    
930 :     ----------------------------------------------------------------------
931 :     Name: Matthias Blume
932 : blume 737 Date: 2000/11/24 17:45:00 JST
933 :     Tag: blume-20001124-link
934 :     Description:
935 :    
936 :     Drastically improved link traversal code for the case that the dynamic
937 :     value was already loaded at bootstrap time. As a result, CM and CMB
938 :     now both load blazingly fast -- even on a very slow machine. Also,
939 :     memory consumption has been further reduced by this.
940 :    
941 :     Warning: The format of the PIDMAP file has changed. THerefore, to
942 :     bootstrap you have to do this:
943 :    
944 :     1. Run CMB.make
945 :     2. Make a symbolic link for the boot directory:
946 :     ln -s sml.boot.ARCH-OS xxx
947 :     3. "Rebuild" the boot directory:
948 :     ./makeml -boot xxx -rebuild sml ; rm xxx
949 :     4. Boot normally:
950 :     ./makeml
951 :    
952 :     ----------------------------------------------------------------------
953 :     Name: Matthias Blume
954 : blume 735 Date: 2000/11/21 21:20:00 JST
955 :     Tag: blume-20001121-tools
956 :     Description:
957 :    
958 :     Continued hacking on autoloading problem -- with success this time.
959 :     Also changed tool-plugin mechanism. See new CM manual.
960 :    
961 :     ----------------------------------------------------------------------
962 :     Name: Matthias Blume
963 : blume 734 Date: 2000/11/19 14:30:00 JST
964 :     Tag: blume-20001119-autoload
965 :     Description:
966 :    
967 :     Some hacking to make autoloading faster. Success for CMB, no success
968 :     so far for CM. There is a reduced structure CM' that autoloads faster.
969 :     (This is a temporary, non-documented hack to be eliminated again when
970 :     the general problem is solved.)
971 :    
972 :     ----------------------------------------------------------------------
973 :     Name: Matthias Blume
974 : blume 733 Date: 2000/11/17 14:10:00 JST
975 :     Tag: blume-20001117-pickle-lib
976 :     Description:
977 :    
978 :     1. Eliminated comp-lib.cm
979 :     2. Made pickle-lib.cm
980 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
981 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
982 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
983 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
984 :     Duplicated intmap modules thrown out.
985 :     6. Hunted down all duplicated SCC code and replaced it with
986 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
987 :     7. Rewrote Feedback module.
988 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
989 :     should be thrown out and equivalent modules from smlnj-lib.cm
990 :     should be used (IntRedBlackSet, IntListSet, ...).
991 :    
992 :     Confirmed that compiler compiles to fixpoint.
993 :    
994 :     ----------------------------------------------------------------------
995 : leunga 731 Name: Allen Leung
996 :     Date: 2000/11/10 18:00:00
997 :     Tag: leunga-20001110-new-x86-fp
998 :    
999 :     A new x86 floating point code generator has been added.
1000 :     By default this is turned off. To turn this on, do:
1001 :    
1002 :     CM.autoload "$smlnj/compiler.cm";
1003 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1004 :    
1005 :     Changes:
1006 :    
1007 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
1008 :     2. Changed the extension callback for FTANGENT to generate:
1009 :    
1010 :     fptan
1011 :     fstp %st(0)
1012 :     instead of
1013 :     fptan
1014 :     fstpl ftempmem
1015 :    
1016 :     3. Numerous assembly fixes for x86.
1017 :    
1018 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
1019 :     support for a whole bunch of instructions and addressing modes:
1020 :    
1021 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
1022 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
1023 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
1024 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1025 :     fxch %st(n)
1026 :     fld %st(n)
1027 :     fst %st(n)
1028 :     fst mem
1029 :     fstp %st(n)
1030 :     fucom %st(n)
1031 :     fucomp %st(n)
1032 :    
1033 :     All these are now generated when the fast fp mode is turned on.
1034 :    
1035 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1036 :    
1037 :     ----------------------------------------------------------------------
1038 : blume 729 Name: Matthias Blume
1039 :     Date: 2000/11/09 11:20:00 JST
1040 :     Tag: blume-20001109-scc
1041 :     Description:
1042 :    
1043 :     Eliminated some code duplication:
1044 :    
1045 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1046 :     (Otherwise the functor is useless.)
1047 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
1048 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
1049 :    
1050 :     ----------------------------------------------------------------------
1051 : george 721 Name: Lal George
1052 :     Date: 2000/11/06 09:02:21 EST 2000
1053 :     Tag: Release_110_30
1054 :     Description:
1055 :    
1056 :     - Version 110.30
1057 :     ----------------------------------------------------------------------
1058 : blume 715 Name: Matthias Blume
1059 : blume 716 Date: 2000/11/04 14:45:00
1060 :     Tag: blume-20001104-mlbuild
1061 :     Description:
1062 :    
1063 :     - Made ml-build faster on startup.
1064 :     - Documentation fixes.
1065 :    
1066 :     ----------------------------------------------------------------------
1067 :     Name: Matthias Blume
1068 : blume 715 Date: 2000/11/02 17:00:00 JST
1069 :     Tag: blume-20001102-condcomp
1070 :     Description:
1071 :    
1072 :     - Small tweaks to pickler -- new BOOTFILES!
1073 :     - Version bumped to 110.29.2.
1074 :     - Added conditional compilation facility to init.cmi (see comment there).
1075 :     ----------------------------------------------------------------------
1076 : leunga 713 Name: Allen Leung
1077 :     Date: 2000/10/23 19:31:00
1078 :     Tag: leunga-20001023-demo-ra
1079 :    
1080 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1081 :     2. Test programs for the graph library updated
1082 :     3. Some new MLRISC demo programs added
1083 :    
1084 :     ----------------------------------------------------------------------
1085 : blume 710 Name: Matthias Blume
1086 : blume 711 Date: 2000/08/31 22:15:00 JST
1087 :     Tag: blume-20001017-errmsg
1088 :     Description:
1089 :    
1090 :     More error message grief: Where there used to be no messages, there
1091 :     now were some that had bogus error regions. Fixed.
1092 :    
1093 :     ----------------------------------------------------------------------
1094 :     Name: Matthias Blume
1095 : blume 710 Date: 2000/08/31 17:30:00 JST
1096 :     Tag: blume-20001017-v110p29p1
1097 :     Description:
1098 :    
1099 :     I made a version 110.29.1 with new bootfiles.
1100 :    
1101 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
1102 :     CM documentation changes and a small bugfix in CM's error reporting.
1103 :    
1104 :     ----------------------------------------------------------------------
1105 : george 705 Name: Lal George
1106 :     Date: 2000/09/27 14:42:35 EDT
1107 :     Tag: george-20000927-nodestatus
1108 :     Description:
1109 :    
1110 :     Changed the type of the nodestatus, so that:
1111 :    
1112 :     SPILLED(~1) is now SPILLED
1113 :     SPILLED(m) where m>=0 is now MEMREG(m)
1114 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
1115 :    
1116 :     ----------------------------------------------------------------------
1117 : blume 697 Name: Matthias Blume
1118 : blume 703 Date: 2000/09/07 14:45:00 JST
1119 :     Tag: blume-20000907-cmerrmsg
1120 :     Description:
1121 :    
1122 :     Small tweak to CM to avoid getting ML syntax error messages twice.
1123 :    
1124 :     ----------------------------------------------------------------------
1125 :     Name: Matthias Blume
1126 : blume 701 Date: 2000/08/31 18:00:00 JST
1127 :     Tag: blume-20000831-cvsbootfiles
1128 :     Description:
1129 :    
1130 :     New URL for boot files (because the 110.29 files on the BL server do
1131 :     now work correctly with my updated install scripts for yacc and lex).
1132 :    
1133 :     ----------------------------------------------------------------------
1134 :     Name: Matthias Blume
1135 : blume 697 Date: 2000/08/08 12:33:00 JST
1136 :     Tag: blume-20000808-manual
1137 :     Description:
1138 :    
1139 :     Tiny update to CM manual.
1140 :    
1141 :     ----------------------------------------------------------------------
1142 : leunga 695 Name: Allen Leung
1143 :     Date: 2000/08/7 19:31:00
1144 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
1145 :    
1146 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
1147 :     and Mobdy.
1148 :    
1149 :     1. x86
1150 :    
1151 :     a. Fixes to peephole module by John and Dan.
1152 :     b. Assembly fix to SETcc by Allen.
1153 :     c. Fix to c-call by John.
1154 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
1155 :     e. Instruction selection optimization to SETcc as suggested by John.
1156 :    
1157 :     For example,
1158 :    
1159 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1160 :    
1161 :     should generate:
1162 :    
1163 :     MOVL a, x
1164 :     SUBL b, x
1165 :     SHRL 31, x
1166 :    
1167 :     2. IR stuff
1168 :    
1169 :     A bunch of new DJ-graph related algorithms added. These
1170 :     speed up SSA construction.
1171 :    
1172 :     3. SSA + Scheduling
1173 :    
1174 :     Added code for SSA and scheduling to the repository
1175 :    
1176 :     ----------------------------------------------------------------------
1177 : george 694 Name: Lal George
1178 :     Date: 2000/07/27 11:53:14 EDT
1179 :    
1180 :     Tag: lal-20000727-linux-ppc
1181 :     Description:
1182 :    
1183 :     Made changes to support Linux PPC.
1184 :     p.s. I have confirmation that the 110.29 boot files work fine.
1185 :    
1186 :     ----------------------------------------------------------------------
1187 : blume 684 Name: Matthias Blume
1188 : blume 692 Date: 2000/07/27 17:40:00 JST
1189 :     Tag: blume-20000727-scripts
1190 :     Description:
1191 :    
1192 :     !!!! WARNING !!!!
1193 :     You must recompile the runtime system!
1194 :     !!!! WARNING !!!!
1195 :    
1196 :     This is basically another round of script-enhancements:
1197 :    
1198 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1199 :     and undefine CM preprocessor symbols.
1200 :    
1201 :     2. ml-build avoids generating a new heap image if it finds that the
1202 :     existing one is still ok. (The condition is that no ML file had to
1203 :     be recompiled and all ML files are found to be older that the heap
1204 :     file.)
1205 :    
1206 :     To make this work smoothly, I also hacked the runtime system as
1207 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
1208 :     (.sparc-solaris, ...) that is currently being used.
1209 :    
1210 :     Moreover, the signature of CM.mk_standalone has changed. See the
1211 :     CM manual.
1212 :    
1213 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
1214 :     CM manual for details.)
1215 :    
1216 :     4. More CM manual updates:
1217 :     - all of the above has been documented.
1218 :     - there is now a section describing the (CM-related) command line
1219 :     arguments that are accepted by the "sml" command
1220 :    
1221 :     ----------------------------------------------------------------------
1222 :     Name: Matthias Blume
1223 : blume 691 Date: 2000/07/25 16:20:00 JST
1224 :     Tag: blume-20000725-makedepend
1225 :     Description:
1226 :    
1227 :     Added a script called ml-makedepend. This can be used in makefiles
1228 :     for Unix' make in a way very similar to the "makedepend" command for
1229 :     C.
1230 :    
1231 :     The script internally uses function CM.sources.
1232 :    
1233 :     Synopsis:
1234 :    
1235 :     ml-makedepend [-f makefile] cmfile targetname
1236 :    
1237 :     The default for the makefile is "makefile" (or "Makefile" should
1238 :     "makefile" not exist).
1239 :    
1240 :     ml-makedepend adds a cmfile/targetname-specific section to this
1241 :     makefile (after removing the previous version of this section). The
1242 :     section contains a single dependency specification with targetname on
1243 :     the LHS (targetname is an arbitrary name), and a list of files derived
1244 :     from the cmfile on the RHS. Some of the files on the RHS are
1245 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
1246 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1247 :     path names. The makefile writer is responsible for making sure that
1248 :     these variables have correct at the time "make" is invoked.
1249 :    
1250 :     ----------------------------------------------------------------------
1251 :     Name: Matthias Blume
1252 : blume 690 Date: 2000/07/22 23:30:00 JST
1253 :     Tag: blume-20000722-urlupdate
1254 :     Description:
1255 :    
1256 :     Changed BOOT and config/srcarchiveurl to point to BL server:
1257 :    
1258 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1259 :    
1260 :     ----------------------------------------------------------------------
1261 :     Name: Matthias Blume
1262 : blume 686 Date: 2000/07/18 18:00:00 JST
1263 :     Tag: blume-20000718-Version_110_29
1264 :     Description:
1265 :    
1266 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1267 :    
1268 :     2. Updated config/version to 110.29
1269 :    
1270 :     3. Updated config/srcarchiveurl
1271 :    
1272 :     3. New boot files!
1273 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1274 :    
1275 :     ----------------------------------------------------------------------
1276 :     Name: Matthias Blume
1277 : blume 684 Date: 2000/07/11 13:58:00 JST
1278 :     Tag: blume-20000711-doctypo
1279 :     Description:
1280 :    
1281 :     Fixed a few typos in CM manual.
1282 :    
1283 :     ----------------------------------------------------------------------
1284 : leunga 683 Name: Allen Leung
1285 :     Date: 2000/06/15 00:38:00
1286 :     Tag: leunga-20000704-sparc-x86
1287 :    
1288 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
1289 :     2. fix to x86 RET bug [found by Dan Grossman]
1290 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
1291 :    
1292 :     Affects c-- and moby only
1293 :    
1294 :     ----------------------------------------------------------------------
1295 : blume 680 Name: Matthias Blume
1296 : blume 682 Date: 2000/07/04 15:26:00
1297 :     Tag: blume-20000704-trigger
1298 :     Description:
1299 :    
1300 :     1. Improvements to CM manual.
1301 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1302 :     of getting a back-trace. The function, when called, raises an
1303 :     internal exception which explicitly carries the full back-trace history,
1304 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
1305 :     or not). The interactive loop will print that history once it arrives
1306 :     at top level.
1307 :     Short of having all exceptions implicitly carry the full history, the
1308 :     recommended way of using this facility is:
1309 :     - compile your program with instrumentation "on"
1310 :     - run it, when it raises an exception, look at the history
1311 :     - if the history is "cut off" because of some handler, go and modify
1312 :     your program so that it explicitly calls BTrace.trigger
1313 :     - recompile (still instrumented), and rerun; look at the full history
1314 :    
1315 :     ----------------------------------------------------------------------
1316 :     Name: Matthias Blume
1317 :     Date: 2000/07/03 15:36:00 JST
1318 : blume 680 Tag: blume-20000702-manual
1319 :     Description:
1320 :    
1321 :     Small corrections and updates to CM manual.
1322 :    
1323 :     ----------------------------------------------------------------------
1324 : blume 682 Name: Matthias Blume
1325 : blume 679 Date: 2000/06/29 16:04:00 JST
1326 :     Tag: blume-20000629-yacctool
1327 :     Description:
1328 :    
1329 :     Changes:
1330 :    
1331 :     1. Class "mlyacc" now takes separate arguments to pass options to
1332 :     generated .sml- and .sig-files independently.
1333 :     2. Corresponding CM manual updates.
1334 :     3. BTrace module now also reports call sites. (However, for loop clusters
1335 :     it only shows from where the cluster was entered.) There are associated
1336 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1337 :    
1338 :     ----------------------------------------------------------------------
1339 : blume 675 Name: Matthias Blume
1340 : blume 678 Date: 2000/06/27 16:51:00 JST
1341 :     Tag: blume-20000627-noweb
1342 :     Description:
1343 :    
1344 :     Changes:
1345 :    
1346 :     1. Implemented "subdir" and "witness" options for noweb tool.
1347 :     This caused some slight internal changes in CM's tool implementation.
1348 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
1349 :     some remaining issues from earlier path anchor changes.
1350 :     3. Updated CM manual accordingly.
1351 :    
1352 :     4. Changed implementation of back-tracing so that I now consider it
1353 :     ready for prime-time.
1354 :    
1355 :     In particular, you don't have to explicitly trigger the back-trace
1356 :     anymore. Instead, if you are running BTrace-instrumented code and
1357 :     there is an uncaught exception (regardless of whether or not it was
1358 :     raised in instrumented code), the top-level evalloop will print
1359 :     the back-trace.
1360 :    
1361 :     Features:
1362 :    
1363 :     - Instrumented and uninstrumented code work together seemlessly.
1364 :     (Of course, uninstrumented code is never mentioned in actual
1365 :     back-traces.)
1366 :    
1367 :     - Asymptotic time- and space-complexity of instrumented code is
1368 :     equal to that of uninstrumented code. (This means that
1369 :     tail-recursion is preserved by the instrumentation phase.)
1370 :    
1371 :     - Modules whose code has been instrumented in different sessions
1372 :     work together without problem.
1373 :    
1374 :     - There is no penalty whatsoever on uninstrumented code.
1375 :    
1376 :     - There is no penalty on "raise" expressions, even in
1377 :     instrumented code.
1378 :    
1379 :     A potential bug (or perhaps it is a feature, too):
1380 :    
1381 :     A back-trace reaches no further than the outermost instrumented
1382 :     non-trivial "raise". Here, a "trivial" raise is one that is the
1383 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
1384 :    
1385 :     <exp> handle e => raise e
1386 :    
1387 :     and even
1388 :    
1389 :     <exp> handle Foo => raise Bar
1390 :    
1391 :     and, of course, through
1392 :    
1393 :     <exp> handle Foo => ...
1394 :    
1395 :     if the exception was not Foo.
1396 :    
1397 :     Back-traces always reach right through any un-instrumented code
1398 :     including any of its "handle" expressions, trivial or not.
1399 :    
1400 :     To try this out, do the following:
1401 :    
1402 :     - Erase all existing binfiles for your program.
1403 :     (You may keep binfiles for those modules where you think you
1404 :     definitely don't need back-tracing.)
1405 :     - Turn on back-trace instrumentation:
1406 :     SMLofNJ.Internals.BTrace.mode (SOME true);
1407 :     - Recompile your program. (I.e., run "CM.make" or "use".)
1408 :     - You may now turn instrumentation off again (if you want):
1409 :     SMLofNJ.Internals.BTrace.mode (SOME false);
1410 :     - Run your program as usual. If it raises an exception that
1411 :     reaches the interactive toplevel, then a back-trace will
1412 :     automatically be printed. After that, the toplevel loop
1413 :     will print the exception history as usual.
1414 :    
1415 :     ----------------------------------------------------------------------
1416 :     Name: Matthias Blume
1417 : blume 677 Date: 2000/06/26 09:56:46 JST
1418 :     Tag: blume-20000626-setup
1419 :     Description:
1420 :    
1421 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1422 :     ML code before and after compiling a file (e.g., to set compiler
1423 :     flags)
1424 :    
1425 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1426 :     - associated changes to btrace.sml (BTrace instrumentation pass)
1427 :     - cleaner implementation of btimp.sml (BTrace tracing and report
1428 :     module)
1429 :    
1430 :     CM manual: * new path encoding documented
1431 :     * description of setup-parameter to "sml" added
1432 :    
1433 :     The biggest user-visible change to back-tracing is that it is no
1434 :     longer necessary to compile all traced modules within the same
1435 :     session. (This was a real limitation.)
1436 :    
1437 :     ----------------------------------------------------------------------
1438 :     Name: Matthias Blume
1439 : blume 676 Date: 2000/06/24 12:40:00 JST
1440 :     Tag: blume-20000624-startup
1441 :     Description:
1442 :    
1443 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
1444 :     bit too often -- to put it mildly. :)
1445 :    
1446 :     ----------------------------------------------------------------------
1447 :     Name: Matthias Blume
1448 : blume 675 Date: 2000/06/23 18:20:00 JST
1449 :     Tag: blume-20000623-btrace
1450 :     Description:
1451 :    
1452 :     This updates adds a backtrace facility to aid programmers in debugging
1453 :     their programs. This involves the following changes:
1454 :    
1455 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1456 :     keeping track of the current call stack. When programs are compiled
1457 :     in a special mode, the compiler will insert calls to these hooks
1458 :     into the user program.
1459 :     "Hook" means that it is possible for different implementations of
1460 :     back-tracing to register themselves (at different times).
1461 :    
1462 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1463 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
1464 :     It can be turned on using this call:
1465 :     SMLofNJ.Internals.BTrace.mode (SOME true);
1466 :     Turning it off again:
1467 :     SMLofNJ.Internals.BTrace.mode (SOME false);
1468 :     Querying the current status:
1469 :     SMLofNJ.Internals.BTrace.mode NONE;
1470 :     Annotated programs are about twice as big as normal ones, and they
1471 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1472 :     where all hooks do nothing). The slowdown with a plugin that is
1473 :     actually useful (such as the one supplied by default) is even greater,
1474 :     but in the case of the default plugin it is still only an constant
1475 :     factor (amortized).
1476 :    
1477 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1478 :     with a sub-structure BTrace for controlling back-tracing. In particular,
1479 :     the above-mentioned function "mode" controls whether the annotation
1480 :     phase is invoked by the compiler. Another important function is
1481 :     "trigger": when called it aborts the current execution and causes
1482 :     the top-level loop to print a full back-trace.
1483 :    
1484 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1485 :     for back-tracing. It keeps track of the dynamic call stack and in
1486 :     addition to that it keeps a partial history at each "level" of that
1487 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
1488 :     d and b (at separate times, dynamically), then the report will show:
1489 :    
1490 :     GOTO d
1491 :     /c
1492 :     GOTO \b
1493 :     CALL a
1494 :    
1495 :     This shows that there was an initial non-tail call of a, then a
1496 :     tail-call to b or c, looping behavior in a cluster of functions that
1497 :     consist of b and c, and then a goto from that cluster (i.e., either from
1498 :     b or from c) to d.
1499 :    
1500 :     Note that (depending on the user program) the amount of information
1501 :     that the back-trace module has to keep track of at each level is bounded
1502 :     by a constant. Thus, the whole implementation has the same asymptotical
1503 :     complexity as the original program (both in space and in time).
1504 :    
1505 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
1506 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
1507 :     which is raised by the "trigger" function mentioned above.
1508 :    
1509 :     Notes on usage:
1510 :    
1511 :     - Annotated code works well together with unannotated code:
1512 :     Unannotated calls simply do not show up at all in the backtrace.
1513 :    
1514 :     - It is not a good idea to let modules that were annotated during
1515 :     different sessions run at the same time. This is because the compiler
1516 :     chooses small integers to identify individual functions, and there
1517 :     will be clashes if different modules were compiled in separate sessions.
1518 :     (Nothing will crash, and you will even be told about the clashes, but
1519 :     back-trace information will in general not be useful.)
1520 :    
1521 :     - Back-tracing can be confused by callcc and capture.
1522 :    
1523 :     - The only way of getting a back-trace right now is to explicitly
1524 :     invoke the "trigger" function from your user program. Eventually, we
1525 :     should make every exception carry back-trace information (if
1526 :     available). But since this creates more overhead at "raise"-time
1527 :     (similar to the current exnHistory overhead), I have not yet
1528 :     implemented this. (The implementation will be rather easy.) With
1529 :     exceptions carrying back-trace information, this facility will be even
1530 :     more useful because users don't need to modify their programs...
1531 :    
1532 :     - While it is possible to compile the compiler with back-trace
1533 :     annotations turned on (I did it to get some confidence in
1534 :     correctness), you must make absolutely sure that core.sml and
1535 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
1536 :     be compiled with annotation because there is no core access yet, but
1537 :     if you compile btimp.sml with annotation, then the system will go into
1538 :     an infinite recursion and crash.)
1539 :     Since CM currently does not know about BTrace, the only way to turn
1540 :     annotations on and off for different modules of the compiler is to
1541 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
1542 :     this is awkward and clumsy.
1543 :    
1544 :     Sample sessions:
1545 :    
1546 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1547 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
1548 :     [autoloading]
1549 :     [autoloading done]
1550 :     val it = false : bool
1551 :     - structure X = struct
1552 :     - fun main n = let
1553 :     - fun a (x, 0) = d x
1554 :     - | a (x, n) = b (x, n - 1)
1555 :     - and b (x, n) = c (x, n)
1556 :     - and c (x, n) = a (x, n)
1557 :     - and d x = e (x, 3)
1558 :     - and e (x, 0) = f x
1559 :     - | e (x, n) = e (x, n - 1)
1560 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1561 :     - | f n = n * g (n - 1)
1562 :     - and g n = a (n, 3)
1563 :     - in
1564 :     - f n
1565 :     - end
1566 :     - end;
1567 :     structure X : sig val main : int -> int end
1568 :     - X.main 3;
1569 :     *** BACK-TRACE ***
1570 :     GOTO stdIn:4.2-13.20: X.main[2].f
1571 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1572 :     GOTO stdIn:4.2-13.20: X.main[2].d
1573 :     / stdIn:4.2-13.20: X.main[2].a
1574 :     | stdIn:4.2-13.20: X.main[2].b
1575 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1576 :     CALL stdIn:4.2-13.20: X.main[2].g
1577 :     GOTO stdIn:4.2-13.20: X.main[2].f
1578 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1579 :     GOTO stdIn:4.2-13.20: X.main[2].d
1580 :     / stdIn:4.2-13.20: X.main[2].a
1581 :     | stdIn:4.2-13.20: X.main[2].b
1582 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1583 :     CALL stdIn:4.2-13.20: X.main[2].g
1584 :     GOTO stdIn:4.2-13.20: X.main[2].f
1585 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1586 :     GOTO stdIn:4.2-13.20: X.main[2].d
1587 :     / stdIn:4.2-13.20: X.main[2].a
1588 :     | stdIn:4.2-13.20: X.main[2].b
1589 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1590 :     CALL stdIn:4.2-13.20: X.main[2].g
1591 :     GOTO stdIn:4.2-13.20: X.main[2].f
1592 :     CALL stdIn:2.15-17.4: X.main[2]
1593 :     -
1594 :    
1595 :     (Note that because of a FLINt bug the above code currently does not
1596 :     compile without BTrace turned on.)
1597 :    
1598 :     Here is another example, using my modified Tiger compiler:
1599 :    
1600 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1601 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
1602 :     [autoloading]
1603 :     [autoloading done]
1604 :     val it = false : bool
1605 :     - CM.make "sources.cm";
1606 :     [autoloading]
1607 :     ...
1608 :     [autoloading done]
1609 :     [scanning sources.cm]
1610 :     [parsing (sources.cm):parse.sml]
1611 :     [creating directory CM/SKEL ...]
1612 :     [parsing (sources.cm):tiger.lex.sml]
1613 :     ...
1614 :     [wrote CM/sparc-unix/semant.sml]
1615 :     [compiling (sources.cm):main.sml]
1616 :     [wrote CM/sparc-unix/main.sml]
1617 :     [New bindings added.]
1618 :     val it = true : bool
1619 :     - Main.compile ("../testcases/merge.tig", "foo.out");
1620 :     *** BACK-TRACE ***
1621 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1622 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1623 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1624 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1625 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1626 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1627 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1628 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1629 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1630 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1631 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1632 :     CALL main.sml:1.18-118.4: Main.compile[2]
1633 :     -
1634 :    
1635 :     ----------------------------------------------------------------------
1636 : blume 673 Name: Matthias Blumen
1637 :     Date: 2000/06/21 18:00:00 JST
1638 :     Tag: blume-20000621-manual
1639 :     Description:
1640 :    
1641 :     CM manual update: Path environments documented.
1642 :    
1643 :     ----------------------------------------------------------------------
1644 : blume 666 Name: Matthias Blume
1645 : blume 672 Date: 2000/06/19 13:40:00
1646 :     Tag: blume-20000619-manual
1647 :     Description:
1648 :    
1649 :     CM manual and system/README update. This only covers the fact that
1650 :     there are no more implicit anchors. (Path environments and the "bind"
1651 :     option to "cm" have yet to be documented.)
1652 :    
1653 :     ----------------------------------------------------------------------
1654 :     Name: Matthias Blume
1655 : blume 671 Date: 2000/06/19 11:05:00 JST
1656 :     Tag: blume-20000619-chdir-bugfix
1657 :     Description:
1658 :    
1659 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1660 :    
1661 :     ----------------------------------------------------------------------
1662 :     Name: Matthias Blume
1663 : blume 670 Date: 2000/06/18 22:00:10 JST
1664 :     Tag: blume-20000618-implicit-anchors-really-gone
1665 :     Description:
1666 :    
1667 :     I updates the previous HISTORY entry where I forgot to mention that
1668 :     implicit anchors are no longer with us.
1669 :    
1670 :     The current update also gets rid of the (now useless) controller
1671 :     CM.Control.implicit_anchors.
1672 :    
1673 :     ----------------------------------------------------------------------
1674 :     Name: Matthias Blume
1675 : blume 666 Date: 2000/06/16 17:30:00 JST
1676 :     Tag: blume-20000616-anchorenv
1677 :     Description:
1678 :    
1679 :     This patch implements the long anticipated (just kidding :) "anchor
1680 :     environment" mechanism. In the course of doing this, I also
1681 :     re-implemented CM's internal "SrcPath" module from scratch. The new
1682 :     one should be more robust in certain boundary cases. In any case, it
1683 :     is a lot cleaner than its predecessor (IMHO).
1684 :    
1685 :     This time, although there is yet another boot file format change, I
1686 :     kept the unpickler backward-compatible. As a result, no new bootfiles
1687 :     are necessary and bootstrapping is straightforward. (You cannot read
1688 :     new bootfiles into an old system, but the other way around is no
1689 :     problem.)
1690 :    
1691 :     Visible changes:
1692 :    
1693 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
1694 :     longer recognized at all. This means that such path names are not
1695 :     illegal either. For example, the name basis.cm simply refers to a
1696 :     local file called "basis.cm" (i.e, the name is an ordinary path
1697 :     relative to .cm-files directory). Or, to put it differently, only
1698 :     names that start with $ are anchored paths.
1699 : blume 666
1700 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1701 :     vanished.
1702 :    
1703 :     John (Reppy) had critizised this as soon as I originally proposed and
1704 : blume 666 implemented it, but at that time I did not really deeply believe
1705 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
1706 :     in another place where it cannot be seen as an abbreviation for
1707 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
1708 :     seems to mean (i.e., it "expands" into the corresponding anchor
1709 :     value).
1710 :    
1711 :     However, when paths are used as members in CM description files, it
1712 :     continues to be true that there must be at least another arc after the
1713 :     anchor. This is now enforced separately during semantic analysis
1714 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
1715 : blume 666
1716 :     ** 2. The "cm" class now accepts an option "bind". The option's value
1717 :     is a sub-option list of precisely two items -- one labeled "anchor"
1718 :     and the other one labeled "value". As you might expect, "anchor" is
1719 :     used to specify an anchor name to be bound, and "value" specifies what
1720 :     the anchor is being bound to.
1721 :    
1722 :     The value must be a directory name and can be given in either standard
1723 :     syntax (including the possibility that it is itself an anchored path)
1724 :     or native syntax.
1725 :    
1726 :     Examples:
1727 :    
1728 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
1729 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
1730 :    
1731 :     and so on.
1732 :    
1733 :     The meaning of this is that the .cm-file will be processed with an
1734 :     augmented anchor environment where the given anchor(s) is/are bound to
1735 :     the given values(s).
1736 :    
1737 :     The rationale for having this feature is this: Suppose you are trying
1738 :     to use two different (already stable) libraries a.cm and b.cm (that
1739 :     you perhaps didn't write yourself). Further, suppose each of these
1740 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
1741 :     Normally you would now have a problem because the anchor "lib" can not
1742 :     be bound to more than one value globally. Therefore, the project that
1743 :     uses both a.cm and b.cm must locally redirect the anchor to some other
1744 :     place:
1745 :    
1746 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1747 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1748 :    
1749 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1750 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1751 :    
1752 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1753 :     you when you do so at the time of CM.stabilize). Therefore, you can
1754 :     also use an anchored path as the value:
1755 :    
1756 :     a.cm (bind:(anchor:lib value:$a-lib))
1757 :     b.cm (bind:(anchor:lib value:$b-lib))
1758 :    
1759 :     Now you can globally configure (using the usual CM.Anchor.anchor or
1760 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
1761 :     itself is always locally bound, setting it globally is no longer
1762 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
1763 :     can still be used as a global anchor for separate purposes. As a
1764 :     matter of fact, one can locally define "lib" in terms of a global
1765 :     "lib":
1766 :    
1767 :     a.cm (bind:(anchor:lib value:$lib/a))
1768 :     b.cm (bind:(anchor:lib value:$lib/b))
1769 :    
1770 :     ** 3: The encoding of path names has changed. This affects the way
1771 :     path names are shown in CM's progress report and also the internal
1772 :     protocol encoding used for parallel make.
1773 :    
1774 :     The encoding now uses one or more ':'-separated segments. Each
1775 :     segments corresponds to a file that has been specified relative to the
1776 :     file given by its preceding segment. The first segment is either
1777 :     relative to the CWD, absolute, or anchored. Each segment itself is
1778 :     basically a Unix pathname; all segments but the first are relative.
1779 :    
1780 :     Example:
1781 :    
1782 :     $foo/bar/baz.cm:a/b/c.sml
1783 :    
1784 :     This path denotes the file bar/a/b/c.sml relative to the directory
1785 :     denoted by anchor "foo". Notice that the encoding also includes
1786 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
1787 :     paths are resolved relative to the .cm-files directory, so baz.cm must
1788 :     be ignored to get the "real" pathname.
1789 :    
1790 :     To make this fact more obvious, CM puts the names of such "virtual
1791 :     arcs" into parentheses when they appear in progress reports. (No
1792 :     parentheses will appear in the internal protocol encoding.) Thus,
1793 :     what you really see is:
1794 :    
1795 :     $foo/bar/(baz.cm):a/b/c.sml
1796 :    
1797 :     I find this notation to be much more informative than before.
1798 :    
1799 :     Another new feature of the encoding is that special characters
1800 :     including parentheses, colons, (back)slashes, and white space are
1801 :     written as \ddd (where ddd is the decimal encoding of the character).
1802 :    
1803 : blume 670 *** The CM manual still needs to be updated.
1804 :    
1805 : blume 666 ----------------------------------------------------------------------
1806 : leunga 660 Name: Allen Leung
1807 :     Date: 2000/06/15 00:38:00
1808 :     Tag: leunga-20000615-x86-peephole
1809 :    
1810 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
1811 :    
1812 :     ----------------------------------------------------------------------
1813 : blume 658 Name: Matthias Blume
1814 :     Date: 2000/06/12 11:40:00
1815 :     Tag: blume-20000612-parmakefix
1816 :     Description:
1817 :    
1818 :     More cleanup after changing the file naming scheme: This time I
1819 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
1820 :    
1821 :     ----------------------------------------------------------------------
1822 : leunga 657 Name: Allen Leung
1823 :     Date: 2000/06/09 01:25:00
1824 :     Tag: leunga-20000609-various
1825 :    
1826 :     None of these things should affect normal SML/NJ operations
1827 :    
1828 :     1. Peephole improvements provided by Fermin (c--)
1829 :     2. New annotation DEFUSE for adding extra dependence (moby)
1830 :     3. New X86 LOCK instructions (moby)
1831 :     4. New machine description language for reservation tables (scheduling)
1832 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
1833 :     trees etc.)
1834 :     6. I've changed the CM files so that they can work with versions
1835 :     110.0.6, 110.25 and 110.28
1836 :    
1837 :     ----------------------------------------------------------------------
1838 : blume 652 Name: Matthias Blume
1839 : blume 656 Date: 2000/06/09 12:40:00
1840 :     Tag: blume-20000609-log
1841 :     Description:
1842 :    
1843 :     - Removed all(?) remaining RCS Log entries from sources.
1844 :    
1845 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1846 :     anchored paths).
1847 :    
1848 :     ----------------------------------------------------------------------
1849 :     Name: Matthias Blume
1850 : blume 653 Date: 2000/06/07 17:00:00 JST
1851 :     Tag: blume-20000607-no-implicit-anchors
1852 :     Description:
1853 :    
1854 :     1. This update changes the default setting for
1855 :     CM.Control.implicit_anchors from true to false. This means that
1856 :     implicit anchors are no longer permitted by default. I also tried to
1857 :     make sure that nothing else still relies on implicit anchors.
1858 :     (This is the next step on the schedule towards a CM that does not even
1859 :     have the notion of implicit anchors anymore.)
1860 :    
1861 :     2. More CM manual updates.
1862 :    
1863 :     3. I managed to track down and fix the pickling bug I mentioned last
1864 :     time. Because of the previously existing workaround, this entails no
1865 :     immediate practical changes.
1866 :    
1867 :     ----------------------------------------------------------------------
1868 :     Name: Matthias Blume
1869 : blume 652 Date: 2000/06/06 11:15:00 JST
1870 :     Tag: blume-20000606-lazierpickle
1871 :     Description:
1872 :    
1873 :     !!!! NEW BOOT FILES !!!!
1874 :    
1875 :     * The main purpose of this update is to make library pickles lazier in
1876 :     order to reduce the initial space penalty for autoloading a library.
1877 :     As a result, it is now possible to have $smlnj/compiler.cm
1878 :     pre-registered. This should take care of the many complaints or
1879 :     inquiries about missing structure Compiler. This required changes to
1880 :     CM's internal data structures and small tweaks to some algorithms.
1881 :    
1882 :     As a neat additional effect, it is no longer necessary (for the sake
1883 :     of lean heap image files) to distinguish between a "minimal" CM and a
1884 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
1885 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1886 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
1887 :    
1888 :     To make the life of compiler-hackers easier, "makeml" now also
1889 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
1890 :     words, after you bootstrap a new sml for the first time, you will not
1891 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
1892 :     around you will still have to do it, though.)
1893 :    
1894 :     * A second change consists of major updates to the CM manual. There
1895 :     are now several appendices with summary information and also a full
1896 :     specification of the CM description file syntax.
1897 :    
1898 :     * In directory src/system I added the script "allcross". This script
1899 :     invokes sml and cross-compiles the compiler for all supported
1900 :     architectures. (Useful when providing a new set of boot files.)
1901 :    
1902 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
1903 :     added a small tweak to pickle-util.sml to work around this problem,
1904 :     but it is not a proper fix yet. I will investigate further. (The
1905 :     effect of the bug was an inflation of library pickle size.)
1906 :    
1907 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
1908 :    
1909 :     ----------------------------------------------------------------------
1910 : leunga 646 Name: Allen Leung
1911 : leunga 648 Date: 2000/05/25 17:28 EDT
1912 :     Tag: leunga-20000525-ra
1913 :     Description:
1914 :    
1915 :     Fixed a bug in freezing phase of the register allocator.
1916 :    
1917 :     ----------------------------------------------------------------------
1918 :     Name: Allen Leung
1919 : leunga 647 Date: 2000/05/15 22:53 EDT
1920 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
1921 :     Description:
1922 :    
1923 :     1. Alpha
1924 :    
1925 :     Slight cleanup. Removed the instruction SGNXL
1926 :    
1927 :     2. X86
1928 :    
1929 :     Added the following instructions to the instruction set:
1930 :    
1931 :     ROLx, RORx,
1932 :     BTx, BTSx, BTLx, BTRx,
1933 :     XCHGx, and variants with the LOCK prefix
1934 :    
1935 :     3. Register Allocation
1936 :    
1937 :     The module ra-rewrite-with-renaming has been improved.
1938 :    
1939 : leunga 647 These have no effect on SML/NJ.
1940 :    
1941 : leunga 646 ----------------------------------------------------------------------
1942 : blume 642 Name: Matthias Blume
1943 : blume 645 Date: 2000/05/15 16:20:00 JST
1944 :     Tag: blume-20000515-lightrebuild
1945 :     Description:
1946 :    
1947 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
1948 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
1949 :     defined. In effect, the command will not build any cross-compiler
1950 :     backends and therefore finish more quickly.
1951 :    
1952 :     The "fixpt" script also takes a "-light" switch to be able to use
1953 :     this new facility while compiling for a fixpoint.
1954 :    
1955 :     2. I replaced all mentions of anchored paths in group owner specifications
1956 :     with simple relative paths (usually starting with "..").
1957 :     The rationale is that a library's internal workings should not be
1958 :     compromised by the lack of some anchor. (An anchor is necessary
1959 :     for someone who wants to refer to the library by an anchored path,
1960 :     but it should not be necessary to build the same library in the first
1961 :     place.)
1962 :    
1963 :     3. I changed the way CM's tool mechanism determines the shell command
1964 :     string used for things like ml-yacc etc. so that it does not break
1965 :     when CM.Control.implicit_anchors is turned off.
1966 :    
1967 :     ----------------------------------------------------------------------
1968 :     Name: Matthias Blume
1969 : blume 644 Date: 2000/05/12 18:20:00 JST
1970 :     Tag: blume-20000512-ml-build
1971 :     Description:
1972 :    
1973 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1974 :     getting installed properly (by config/install.sh).
1975 :    
1976 :     ----------------------------------------------------------------------
1977 :     Name: Matthias Blume
1978 : blume 643 Date: 2000/05/12 17:30:00 JST
1979 :     Tag: blume-20000512-anchors
1980 :     Description:
1981 :    
1982 :     !!! NEW BOOT FILES !!!
1983 :    
1984 :     This change is in preparation of fading out support for "implicitly
1985 :     anchored path names". I went through all sources and used the
1986 :     explicit (and relatively new) $-notation. See system/README and the
1987 :     CM manual for more info on this.
1988 :    
1989 :     I also modified the anchoring scheme for some things such as "smlnj",
1990 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
1991 :     anchors are more expressive: anchor name and first arc do not have to
1992 :     coincide. This entails the following user-visible change:
1993 :    
1994 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
1995 :     particular, when you fire up sml with a command-line argument, say,
1996 :     e.g.:
1997 :    
1998 :     sml '$smlnj/cmb.cm'
1999 :    
2000 :     At the ML toplevel prompt:
2001 :    
2002 :     CM.autoload "$smlnj/cmb.cm";
2003 :    
2004 :     There is also a new controller in CM.Control that can be used to turn
2005 :     off all remaining support for implicit anchors by saying:
2006 :    
2007 :     CM.autoload "$smlnj/
2008 :     #set CM.Control.implicit_anchors false;
2009 :    
2010 :     This causes CM to reject implicitly anchored paths. This is (for the
2011 :     time being) less permissive than the "final" version where there will
2012 :     be no more such implicit anchors and relative paths will be just that:
2013 :     relative.
2014 :    
2015 :     The next step (version after next version?) will be to make the
2016 :     default for CM.Control.implicit_anchors false. After the dust has
2017 :     settled, I can then produce the "final" version of this...
2018 :    
2019 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
2020 :    
2021 :     ----------------------------------------------------------------------
2022 :     Name: Matthias Blume
2023 : blume 642 Date: 2000/05/11 16:30:00 JST
2024 :     Tag: blume-20000511-sources
2025 :     Description:
2026 :    
2027 :     The main change is that I added function CM.sources as a generalized
2028 :     version of the earlier CM.makedepend. This entails the following
2029 :     additional changes:
2030 :    
2031 :     - CM.makedepend has been dropped.
2032 :    
2033 :     - CM manual has been updated.
2034 :    
2035 :     - TOOLS signature and API have been changed.
2036 :    
2037 :     ----------------------------------------------------------------------
2038 : leunga 640 Name: Allen Leung
2039 : leunga 641 Date: 2000/05/10 21:17 EDT
2040 :     Tag: leunga-20000510-moby-c--ssa
2041 :     Description:
2042 :    
2043 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2044 :     None of these affect SML/NJ.
2045 :    
2046 :     1. Register Allocation
2047 :    
2048 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2049 :     This module tries to remove local (i.e. basic block level) redundancies
2050 :     during spilling.
2051 :    
2052 :     b. A new framework for performing region based register allocation.
2053 :     Not yet entirely functional.
2054 :    
2055 :     2. X86
2056 :    
2057 :     a. DefUse for POP was missing the stack pointer [found by Lal]
2058 :     b. Reload for CALL was incorrect in X86Spill [found by John]
2059 :     c. Various fixes in X86Spill so that it can be used correctly for
2060 :     the new spilling module.
2061 :    
2062 :     3. SSA/IR
2063 :    
2064 :     a. New module ir/dj-dataflow.sml implements elimination based
2065 :     data flow analysis.
2066 :    
2067 :     4. MLRiscGen
2068 :    
2069 :     a. Fix for gc type annotation
2070 :    
2071 :     5. MDGen
2072 :    
2073 :     Various fixes for machine description -> ml code translation. For ssa
2074 :     only.
2075 :    
2076 :     ----------------------------------------------------------------------
2077 :     Name: Allen Leung
2078 : leunga 640 Date: 2000/05/08 22:17 EDT
2079 :     Tag: leunga-20000508-labexp
2080 :     Description:
2081 :    
2082 :     Fermin has found a few assembly problems with constant expressions
2083 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
2084 :     which choke on dumb assemblers. This is his fix.
2085 :    
2086 :     ----------------------------------------------------------------------
2087 : dbm 635 Name: Dave MacQueen
2088 :     Date: 2000/04/09 14:00 EDT
2089 :     Tag: dbm-20000502-Version_110_28
2090 :     Description:
2091 :    
2092 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2093 :    
2094 :     2. Updated config/version to 110.28
2095 :    
2096 :     3. Updated config/srcarchiveurl
2097 :    
2098 :     3. New boot files!
2099 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2100 :    
2101 :     ----------------------------------------------------------------------
2102 : blume 634 Name: Matthias Blume
2103 :     Date: 2000/05/01 19:05:00 JST
2104 :     Tag: blume-20000501-noweb
2105 :     Description:
2106 :    
2107 :     A new noweb tool has been added. The existing system is entirely
2108 :     unaffected by this, but some CM users have asked for renewed noweb
2109 :     support. Everything is documented in the CM manual.
2110 :    
2111 :     New (plugin) libraries:
2112 :    
2113 :     noweb-tool.cm
2114 :     nw-ext.cm
2115 :    
2116 :     ----------------------------------------------------------------------
2117 : dbm 633 Name: Dave MacQueen
2118 :     Date: 2000/04/30 12:40PM EDT
2119 :     Tag: dbm-20000430-bug_fixes
2120 :     Description:
2121 :    
2122 :     1. Fix for bug 1498
2123 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2124 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2125 :     added toRealArray function
2126 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
2127 :     added check for tag Obj.RealArray to array printing case in ppObj
2128 :    
2129 :     2. Fix for bug 1510
2130 :     smlnj/src/compiler/Semant/types/typesutil.sml
2131 :     fixed definition of dummyargs (used by equalTycon) so that
2132 :     dummy args are distinct types
2133 :    
2134 :     ----------------------------------------------------------------------
2135 : blume 629 Name: Matthias Blume
2136 : blume 632 Date: 2000/04/30 01:00:00 JST
2137 :     Tag: blume-20000430-versions
2138 :     Description:
2139 :    
2140 :     1. CM version numbering added. This is an implementation of Lal's
2141 :     proposal for adding version numbers and version checking to .cm
2142 :     files. Lal said that his proposal was just that -- a proposal.
2143 :     For the time being I went ahead and implemented it so that people
2144 :     can comment on it. Everything is completely backward-compatible
2145 :     (except for the stable library format, i.e., new bootfiles!).
2146 :    
2147 :     As usual, see the CM manual for details.
2148 :    
2149 :     2. An alternative syntax for anchored paths has been implemented.
2150 :     Dave has recently voiced the same concerns that I had when I did
2151 :     this, so there should be some support. My take is that eventually
2152 :     I will let support for the current syntax (where anchors are
2153 :     "implicit") fade out in favor of the new, explicit syntax.
2154 :     In order to be backward-compatible, both old and new syntax are
2155 :     currently supported.
2156 :    
2157 :     Again, see the CM manual for details.
2158 :    
2159 :     3. Parallel make is trying to be slightly smarter: When the master
2160 :     process finds a "bottleneck", i.e., when there is only one
2161 :     compilation unit that can be compiled and everybody else is
2162 :     waiting on it, then it will simply compile it directly instead
2163 :     of clumsily telling one of the slaves to do it.
2164 :    
2165 :     4. Support for "unsharing" added. This is necessary in order to be
2166 :     able to have two different versions of the same library running
2167 :     at the same time (e.g., for trying out a new MLRISC while still
2168 :     having the old MLRISC linked into the current compiler, etc.)
2169 :     See the CM manual.
2170 :    
2171 :     5. Simple "makedepend" functionality added for generating Makefile
2172 :     dependency information. (This is rather crude at the moment.
2173 :     Expect some changes here in the future.)
2174 :    
2175 :     6. ".fun" added as a recognized suffix for ML files. Also documented
2176 :     explicitly in the manual that the fallback behavior (unknown suffix
2177 :     -> ML file) is not an official feature!
2178 :    
2179 :     7. Small changes to the pickler for stable libraries.
2180 :    
2181 :     8. Several internal changes to CM (for cleanup/improvement).
2182 :    
2183 :    
2184 :     !!!! NEW BINFILES !!!!
2185 :    
2186 :     ----------------------------------------------------------------------
2187 :     Name: Matthias Blume
2188 : blume 631 Date: 2000/04/28 17:30:00 JST
2189 :     Tag: blume-20000428-pathconfig
2190 :     Description:
2191 :    
2192 :     1. I changed config/install.sh to remove duplicate entries from the
2193 :     lib/pathconfig file at the end. Moreover, the final version of
2194 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
2195 :     in src/system/installml.
2196 :    
2197 :     2. The config/install.sh script now consistently uses relative
2198 :     pathnames in lib/pathconfig whenever the anchor is in the lib
2199 :     directory. (So far this was true for the libraries that come
2200 :     pre-compiled and bundled as part of the bootfiles but not for
2201 :     libraries that are compiled by the script itself.)
2202 :    
2203 :     ----------------------------------------------------------------------
2204 :     Name: Matthias Blume
2205 : blume 629 Date: 2000/04/26 13:10:00 JST
2206 :     Tag: blume-20000426-fun_suffix
2207 :     Description:
2208 :    
2209 :     Added ".fun" as a recognized file name suffix (for ML code).
2210 :    
2211 :     ----------------------------------------------------------------------
2212 : leunga 628 Name: Allen Leung
2213 :     Date: 2000/04/25 17:00:00 EST
2214 :     Tag: leunga-20000425-alpha-ra
2215 :     Description:
2216 :    
2217 :     1. Alpha
2218 :    
2219 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
2220 :     in C--.
2221 :    
2222 :     2. RA
2223 :    
2224 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
2225 :     and the interference graph.
2226 :    
2227 :     ----------------------------------------------------------------------
2228 : dbm 627 Name: Dave MacQueen
2229 :     Date: 2000/04/25/
2230 :     Tag: dbm-20000425-mlyacc_doc_examples
2231 :     Description:
2232 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2233 :     Updated all three examples in src/ml-yacc/examples to run
2234 :     under 110.* using CM.make.
2235 :    
2236 :     ----------------------------------------------------------------------
2237 : leunga 624 Name: Allen Leung
2238 :     Date: 2000/04/20 23:04:00 EST
2239 :     Tag: leunga-20000420-ssa-c---stuff
2240 :     Description:
2241 :    
2242 :     This update synchronizes my repository with Yale's. Most of these
2243 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
2244 :    
2245 :     1. Register Allocator
2246 :    
2247 :     a. An improvement in the interference graph construction:
2248 :     Given a copy
2249 :    
2250 :     s <- t
2251 :    
2252 : leunga 628 no interference edge between s and t is added for this definition of s.
2253 : leunga 624
2254 :     b. I've added two new spill heuristic modules that Fermin and I developed
2255 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
2256 :     useful for others (Moby?)
2257 :    
2258 :     2. X86
2259 :    
2260 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
2261 :    
2262 :     3. Alpha
2263 :    
2264 :     a. Added the BSR instruction and code generation that goes with it [C--]
2265 :     b. Other fixes too numerous to recount provided by Fermin [C--]
2266 :    
2267 :     4. Regmaps
2268 :    
2269 :     a. The regmaps are not initialized with the identity physical bindings
2270 :     at creation time. This is unneeded.
2271 :    
2272 :     5. MLRISC Optimizations
2273 :    
2274 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
2275 :     intersects with liveness incrementally in linear time! Woohoo!
2276 :     This is now used in my new SSA construction algorithm.
2277 :    
2278 :     b. THe branch reorganization module is now smarter about linear chains of
2279 :     basic blocks.
2280 :    
2281 :    
2282 :     ----------------------------------------------------------------------
2283 : blume 615 Name: Matthias Blume
2284 :     Date: 2000/04/12 13:52:00 JST
2285 :     Tag: blume_main_v110p27_1
2286 :     Description:
2287 :    
2288 :     Changed install.sh script to handle archive files without version number
2289 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2290 :     name of the boot file archive.
2291 :    
2292 :     ----------------------------------------------------------------------
2293 : dbm 607 Name: Dave MacQueen
2294 :     Date: 2000/04/09 14:00 EDT
2295 :     Tag: dbm-20000410-Version_110_27
2296 :     Description:
2297 :    
2298 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2299 :    
2300 :     2. Updated src/config/version to 110.27
2301 :    
2302 :     3. New boot files!
2303 :    
2304 :     ----------------------------------------------------------------------
2305 : leunga 606 Name: Allen Leung
2306 :     Date: 2000/04/09 19:09:00 EST
2307 :     Tag: leunga-20000409-misc
2308 :     Description:
2309 :    
2310 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
2311 :    
2312 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
2313 :    
2314 :     ----------------------------------------------------------------------
2315 : monnier 605 Name: Stefan
2316 :     Date: 2000/04/07 10:00:00 EDT
2317 :     Tag: monnier-20000406-branch-handling
2318 :     Description:
2319 :    
2320 :     Improved handling of branches (mostly those generated from
2321 :     polymorphic equality), removed switchoff and changed the
2322 :     default optimization settings (more cpsopt and less flintopt).
2323 :    
2324 :     ----------------------------------------------------------------------
2325 : leunga 601 Name: Allen Leung
2326 : leunga 602 Date: 2000/04/06 01:30:00 EST
2327 :     Tag: leunga-20000406-peephole-x86-SSA-2
2328 :     Description:
2329 :    
2330 :     Forgot a few files.
2331 :    
2332 :     ----------------------------------------------------------------------
2333 :     Name: Allen Leung
2334 : leunga 601 Date: 2000/04/06 00:36:00 EST
2335 :     Tag: leunga-20000406-peephole-x86-SSA
2336 :     Description:
2337 :    
2338 :     1. New Peephole code
2339 :    
2340 :     2. Minor improvement to X86 instruction selection
2341 :    
2342 :     3. Various fixes to SSA and machine description -> code translator
2343 :    
2344 :     ----------------------------------------------------------------------
2345 : blume 595 Name: Matthias Blume
2346 :     Date: 2000/04/05 12:30:00 JST
2347 :     Tag: blume_main_v110p26p2_3
2348 :     Description:
2349 :    
2350 :     This update just merges three minor cosmetic updates to CM's sources
2351 :     to get ready for the 110.27 code freeze on Friday. No functionality
2352 :     has changed.
2353 :    
2354 :     ----------------------------------------------------------------------
2355 : leunga 593 Name: Allen Leung
2356 : leunga 594 Date: 2000/04/04 19:39:00 EST
2357 :     Tag: leunga-20000404-x86-asm
2358 :     Description:
2359 :    
2360 :     1. Fixed a problem in X86 assembly.
2361 :    
2362 :     Things like
2363 :    
2364 :     jmp %eax
2365 :     jmp (%eax)
2366 :    
2367 :     should be output as
2368 :    
2369 :     jmp *%eax
2370 :     jmp *(%eax)
2371 :    
2372 :     2. Assembly output
2373 :    
2374 :     Added a new flag
2375 :    
2376 :     "asm-indent-copies" (default to false)
2377 :    
2378 :     When this flag is on, parallel copies will be indented an extra level.
2379 :    
2380 :     ----------------------------------------------------------------------
2381 :     Name: Allen Leung
2382 : leunga 593 Date: 2000/04/04 03:18:00 EST
2383 :     Tag: leunga-20000404-C--Moby
2384 :     Description:
2385 :    
2386 :     All of these fixes are related to C--, Moby, and my own optimization
2387 :     stuff; so they shouldn't affect SML/NJ.
2388 :    
2389 :     1. X86
2390 :    
2391 :     Various fixes related floating point, and extensions.
2392 :    
2393 :     2. Alpha
2394 :    
2395 :     Some extra patterns related to loads with signed/zero extension
2396 :     provided by Fermin.
2397 :    
2398 :     3. Assembly
2399 :    
2400 : monnier 659 When generating assembly, resolve the value of client defined constants,
2401 : leunga 593 instead of generating symbolic values. This is controlled by the
2402 :     new flag "asm-resolve-constants", which is default to true.
2403 :    
2404 :     4. Machine Descriptions
2405 :    
2406 :     a. The precedence parser was slightly broken when parsing infixr symbols.
2407 :     b. The type generalizing code had the bound variables reversed, resulting
2408 :     in a problem during arity raising.
2409 :     c. Various fixes in machine descriptions.
2410 :    
2411 :     ----------------------------------------------------------------------
2412 : blume 592 Name: Matthias Blume
2413 :     Date: 2000/04/03 16:05:00 JST
2414 :     Tag: blume_main_v110p26p2_2
2415 :     Description:
2416 :    
2417 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
2418 :     now done via the ordinary static environment that is context to each
2419 :     compilation unit.
2420 :    
2421 :     To this end, I arranged that instead of "structure Core" as "structure
2422 :     _Core" is bound in the pervasive environment. Core access is done via
2423 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
2424 : blume 592 legal surface-syntax symbol).
2425 :    
2426 :     The current solution is much cleaner because the core environment is
2427 :     now simply part of the pervasive environment which is part of every
2428 :     compilation unit's context anyway. In particular, this eliminates all
2429 :     special-case handling that was necessary until now in order to deal
2430 :     with dynamic and symbolic parts of the core environment.
2431 :    
2432 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2433 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
2434 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
2435 :     more details.
2436 :    
2437 :     I also tried to track down all mentions of "Core" (as string argument
2438 :     to Symbol.strSymbol) in the compiler and replaced them with a
2439 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
2440 :     name appears in one place only.
2441 :    
2442 :     Binfile and bootfile format have not changed, but the switchover from
2443 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
2444 :     bootfiles anyway.
2445 :    
2446 :     ----------------------------------------------------------------------
2447 : leunga 591 Name: Allen Leung
2448 :     Date: 2000/04/02 21:17:00 EST
2449 :     Tag: leunga-20000402-mltree
2450 :     Description:
2451 : leunga 585
2452 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
2453 :     2. Added a bunch of files from my repository. These are currently
2454 :     used by other non-SMLNJ backends.
2455 :    
2456 : leunga 576 ----------------------------------------------------------------------
2457 : leunga 590 Name: Allen Leung
2458 :     Date: 2000/03/31 21:15:00 EST
2459 :     Tag: leunga-20000331-aliasing
2460 :     Description:
2461 :    
2462 :     This update contains a rewritten (and hopefully more correct) module
2463 :     for extracting aliasing information from CPS.
2464 :    
2465 :     To turn on this feature:
2466 :    
2467 :     Compiler.Control.CG.memDisambiguate := true
2468 :    
2469 :     To pretty print the region information with assembly
2470 :    
2471 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
2472 :    
2473 :     To control how many levels of aliasing information are printed, use:
2474 :    
2475 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
2476 :    
2477 :     The default of n is 3.
2478 :    
2479 :     ----------------------------------------------------------------------
2480 : dbm 589 Name: David MacQueen
2481 :     Date: 2000/03/31 11:15:00 EST
2482 :     Tag: dbm-20000331-runtime_fix
2483 :     Description:
2484 :    
2485 :     This update contains:
2486 :    
2487 :     1. runtime/c-lib/c-libraries.c
2488 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
2489 :    
2490 :     2. fix for bug 1556
2491 :     system/Basis/Implementation/NJ/internal-signals.sml
2492 :    
2493 :     ----------------------------------------------------------------------
2494 : blume 587 Name: Matthias Blume
2495 : blume 588 Date: 2000/03/31 18:00:00 JST
2496 :     Tag: blume_main_v110p26p2_1
2497 :     Description:
2498 :    
2499 :     This update contains:
2500 :    
2501 :     1. A small change to CM's handling of stable libraries:
2502 :     CM now maintains one "global" modmap that is used for all stable
2503 :     libraries. The use of such a global modmap maximizes sharing and
2504 :     minimizes the need for re-traversing parts of environments during
2505 :     modmap construction. (However, this has minor impact since modmap
2506 :     construction seems to account for just one percent or less of total
2507 :     compile time.)
2508 :    
2509 :     2. I added a "genmap" phase to the statistics. This is where I got the
2510 :     "one percent" number (see above).
2511 :    
2512 :     3. CM's new tool parameter mechanism just became _even_ better. :)
2513 :     - The parser understands named parameters and recursive options.
2514 :     - The "make" and "shell" tools use these new features.
2515 :     (This makes it a lot easier to cascade these tools.)
2516 :     - There is a small syntax change: named parameters use a
2517 :    
2518 :     <name> : ( <option> ... ) or
2519 :     <name> : <string>
2520 :    
2521 :     syntax. Previously, named parameters were implemented in an
2522 :     ad-hoc fashion by each tool individually (by parsing strings)
2523 :     and had the form
2524 :    
2525 :     <name>=<string>
2526 :    
2527 :     See the CM manual for a full description of these issues.
2528 :    
2529 :     ----------------------------------------------------------------------
2530 :     Name: Matthias Blume
2531 : blume 587 Date: 2000/03/30 18:00:00 JST
2532 :     Tag: blume_main_v110p26p2_0
2533 :     Description:
2534 :    
2535 :     !!!!! WARNING !!!!!!
2536 :     !! New binfiles !!
2537 :     !!!!!!!!!!!!!!!!!!!!
2538 :    
2539 :     This update contains:
2540 :    
2541 :     1. Moderate changes to CM:
2542 :    
2543 :     - Changes to CM's tools mechanism. In particular, it is now possible
2544 :     to have tools that accept additional "command line" parameters
2545 :     (specified in the .cm file at each instance where the tool's class is
2546 :     used).
2547 :    
2548 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
2549 :     facilitate fairly seamless hookup to portions of code managed using
2550 : blume 587 Makefiles or Shell scripts.
2551 :    
2552 :     There are no classes "shared" or "private" anymore. Instead, the
2553 :     sharing annotation is now a parameter to the "sml" class.
2554 :    
2555 :     There is a bit of generic machinery for implementing one's own
2556 :     tools that accept command-line parameters. However, I am not yet fully
2557 :     satisfied with that part, so expect changes here in the future.
2558 :    
2559 :     All existing tools are described in the CM manual.
2560 :    
2561 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
2562 : blume 587 error messages that tended to be more annoying than helpful.)
2563 :    
2564 :     2. Major changes to the compiler's static environment data structures.
2565 :    
2566 :     - no CMStaticEnv anymore.
2567 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
2568 :     but it is called Environment), no conversions between different
2569 :     kinds of static environments
2570 :    
2571 :     - There is still a notion of a "modmap", but such modmaps are generated
2572 :     on demand at the time when they are needed. This sounds slow, but I
2573 :     sped up the code that generates modmaps enough for this not to lead to
2574 :     a slowdown of the compiler (at least I didn't detect any).
2575 :    
2576 :     - To facilitate rapid modmap generation, static environments now
2577 :     contain an (optional) "modtree" structure. Modtree annotations are
2578 :     constructed by the unpickler during unpickling. (This means that
2579 :     the elaborator does not have to worry about modtrees at all.)
2580 :     Modtrees have the advantage that they are compositional in the same
2581 :     way as the environment data structure itself is compositional.
2582 :     As a result, modtrees never hang on to parts of an environment that
2583 :     has already been rendered "stale" by filtering or rebinding.
2584 :    
2585 :     - I went through many, many trials and errors before arriving at the
2586 :     current solution. (The initial idea of "linkpaths" did not work.)
2587 :     But the result of all this is that I have touched a lot of files that
2588 :     depend on the "modules" and "types" data structures (most of the
2589 :     elaborator). There were a lot of changes during my "linkpath" trials
2590 :     that could have been reverted to their original state but weren't.
2591 :     Please, don't be too harsh on me for messing with this code a bit more
2592 :     than what was strictly necessary... (I _did_ resist the tempation
2593 :     of doing any "global reformatting" to avoid an untimely death at
2594 :     Dave's hands. :)
2595 :    
2596 :     - One positive aspect of the previous point: At least I made sure that
2597 :     all files that I touched now compile without warnings (other than
2598 :     "polyEqual").
2599 :    
2600 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
2601 :     redundant modmaps)
2602 :    
2603 :     ----------------------------------------------------------------------
2604 : leunga 580 Name: Allen Leung
2605 : leunga 585 Date: 2000/03/29 18:00:00
2606 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
2607 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
2608 : leunga 585 Description:
2609 :    
2610 :     This update contains *MAJOR* changes to the way code is generated from CPS
2611 :     in the module mlriscGen, and in various backend modules.
2612 :    
2613 :     CHANGES
2614 :     =======
2615 :    
2616 :     1. MLRiscGen: forward propagation fix.
2617 :    
2618 :     There was a bug in forward propagation introduced at about the same time
2619 :     as the MLRISC x86 backend, which prohibits coalescing to be
2620 :     performed effectively in loops.
2621 :    
2622 :     Effect: speed up of loops in RISC architectures.
2623 :     By itself, this actually slowed down certain benchmarks on the x86.
2624 :    
2625 :     2. MLRiscGen: forward propagating addresses from consing.
2626 :    
2627 :     I've changed the way consing code is generated. Basically I separated
2628 :     out the initialization part:
2629 :    
2630 :     store tag, offset(allocptr)
2631 :     store elem1, offset+4(allocptr)
2632 :     store elem2, offset+8(allocptr)
2633 :     ...
2634 :     store elemn, offset+4n(allocptr)
2635 :    
2636 :     and the address computation part:
2637 :    
2638 :     celladdr <- offset+4+alloctpr
2639 :    
2640 :     and move the address computation part
2641 :    
2642 :     Effect: register pressure is generally lower as a result. This
2643 :     makes compilation of certain expressions much faster, such as
2644 :     long lists with non-trivial elements.
2645 :    
2646 :     [(0,0), (0,0), .... (0,0)]
2647 :    
2648 :     3. MLRiscGen: base pointer elimination.
2649 :    
2650 :     As part of the linkage mechanism, we generate the sequence:
2651 :    
2652 :     L: ... <- start of the code fragment
2653 :    
2654 :     L1:
2655 :     base pointer <- linkreg - L1 + L
2656 :    
2657 :     The base pointer was then used for computing relocatable addresses
2658 :     in the code fragment. Frequently (such as in lots of continuations)
2659 :     this is not needed. We now eliminate this sequence whenever possible.
2660 :    
2661 :     For compile time efficiency, I'm using a very stupid local heuristic.
2662 :     But in general, this should be done as a control flow analysis.
2663 :    
2664 :     Effect: Smaller code size. Speed up of most programs.
2665 :    
2666 :     4. Hppa back end
2667 :    
2668 :     Long jumps in span dependence resolution used to depend on the existence
2669 :     of the base pointer.
2670 :    
2671 :     A jump to a long label L was expanded into the following sequence:
2672 :    
2673 :     LDIL %hi(L-8192), %r29
2674 :     LDO %lo(L-8192)(%r29), %r29
2675 :     ADD %r29, baseptr, %r29
2676 :     BV,n %r0(%r29)
2677 :    
2678 :     In the presence of change (3) above, this will not work. I've changed
2679 :     it so that the following sequence of instructions are generated, which
2680 :     doesn't mention the base pointer at all:
2681 :    
2682 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
2683 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
2684 :     BV,n %r0(%r29) /* Jump */
2685 :    
2686 :     5. Alpha back end
2687 :    
2688 :     New alpha instructions LDB/LDW have been added, as per Fermin's
2689 :     suggestions. This is unrelated to all other changes.
2690 :    
2691 :     6. X86 back end
2692 :    
2693 :     I've changed andl to testl in the floating point test sequence
2694 :     whenever appropriate. The Intel optimization guide states that
2695 : monnier 659 testl is preferable to andl.
2696 : leunga 585
2697 :     7. RA (x86 only)
2698 :    
2699 :     I've improved the spill propagation algorithm, using an approximation
2700 :     of maximal weighted independent sets. This seems to be necessary to
2701 :     alleviate the negative effect in light of the slow down in (1).
2702 :    
2703 :     I'll write down the algorithm one of these days.
2704 :    
2705 :     8. MLRiscGen: frequencies
2706 :    
2707 :     I've added an annotation that states that all call gc blocks have zero
2708 :     execution frequencies. This improves register allocation on the x86.
2709 :    
2710 :     BENCHMARKS
2711 :     ==========
2712 :    
2713 :     I've only perform the comparison on 110.25.
2714 :    
2715 :     The platforms are:
2716 :    
2717 :     HPPA A four processor HP machine (E9000) with 5G of memory.
2718 :     X86 A 300Hhz Pentium II with 128M of memory, and
2719 :     SPARC An Ultra sparc 2 with 512M of memory.
2720 :    
2721 :     I used the following parameters for the SML benchmarks:
2722 :    
2723 :     @SMLalloc
2724 :     HPPA 256k
2725 :     SPARC 512k
2726 :     X86 256k
2727 :    
2728 :     COMPILATION TIME
2729 :     ----------------
2730 :     Here are the numbers comparing the compilation times of the compilers.
2731 :     I've only compared 110.25 compiling the new sources versus
2732 :     a fixpoint version of the new compiler compiling the same.
2733 :    
2734 :     110.25 New
2735 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
2736 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
2737 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
2738 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
2739 :    
2740 :     110.25 New
2741 :     Code Size Code Size
2742 :     HPPA 8596736 8561421
2743 :     SPARC 8974299 8785143
2744 :     X86 9029180 8716783
2745 :    
2746 :     So in summary, things are at least as good as before. Dramatic
2747 :     reduction in compilation is obtained on the Sparc; I can't explain it,
2748 :     but it is reproducible. Perhaps someone should try to reproduce this
2749 :     on their own machines.
2750 :    
2751 :     SML BENCHMARKS
2752 :     --------------
2753 :    
2754 :     On the average, all benchmarks perform at least as well as before.
2755 :    
2756 :     HPPA Compilation Time Spill+Reload Run Time
2757 :     110.25 New 110.25 New 110.25 New
2758 :    
2759 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
2760 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
2761 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
2762 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
2763 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
2764 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
2765 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
2766 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
2767 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
2768 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
2769 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
2770 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
2771 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
2772 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
2773 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
2774 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
2775 :     --------------------------------------------------------------------------
2776 :     Average 6.12% 4.09%
2777 :    
2778 :     SPARC Compilation Time Spill+Reload Run Time
2779 :     110.25 New 110.25 New 110.25 New
2780 :    
2781 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
2782 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
2783 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
2784 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
2785 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
2786 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
2787 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
2788 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
2789 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
2790 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
2791 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
2792 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
2793 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
2794 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
2795 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
2796 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
2797 :     ----------------------------------------------------------------------------
2798 :     Average 6.94% 2.64%
2799 :    
2800 :     X86 Compilation Time Spill+Reload Run Time
2801 :     110.25 New 110.25 New 110.25 New
2802 :    
2803 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
2804 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
2805 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
2806 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
2807 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
2808 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
2809 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
2810 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
2811 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
2812 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
2813 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
2814 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
2815 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
2816 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
2817 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
2818 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
2819 :     ----------------------------------------------------------------------------
2820 :     Average 1.22% 3.36%
2821 :    
2822 :     ----------------------------------------------------------------------
2823 :     Name: Allen Leung
2824 : leunga 583 Date: 2000/03/23 16:25:00
2825 :     Tag: leunga-20000323-fix_x86_alpha
2826 :     Description:
2827 :    
2828 :     1. X86 fixes/changes
2829 :    
2830 :     a. The old code generated for SETcc was completely wrong.
2831 :     The Intel optimization guide is VERY misleading.
2832 :    
2833 :     2. ALPHA fixes/changes
2834 :    
2835 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2836 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2837 :     c. Added reassociation code for address computation.
2838 :    
2839 :     ----------------------------------------------------------------------
2840 :     Name: Allen Leung
2841 : leunga 580 Date: 2000/03/22 01:23:00
2842 :     Tag: leunga-20000322-fix_x86_hppa_ra
2843 :     Description:
2844 :    
2845 :     1. X86 fixes/changes
2846 :    
2847 :     a. x86Rewrite bug with MUL3 (found by Lal)
2848 :     b. Added the instructions FSTS, FSTL
2849 :    
2850 :     2. PA-RISC fixes/changes
2851 :    
2852 :     a. B label should not be a delay slot candidate! Why did this work?
2853 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
2854 :     as it should be.
2855 :     c. The assembly syntax for fstds and fstdd was wrong.
2856 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
2857 :     operand variant of COMCLR/LDO.
2858 :    
2859 :     3. Generic MLRISC
2860 :    
2861 :     a. shuffle.sml rewritten to be slightly more efficient
2862 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
2863 :    
2864 :     4. Register Allocator
2865 :    
2866 :     a. I now release the interference graph earlier during spilling.
2867 :     May improve memory usage.
2868 :    
2869 :     ----------------------------------------------------------------------
2870 : blume 577 Name: Matthias Blume
2871 : blume 578 Date: 2000/03/14 14:15:32
2872 :     Tag: blume_main_v110p26p1_2
2873 :     Description:
2874 :    
2875 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2876 :     additional argument called "template" which is an optional string that
2877 : monnier 659 specifies the layout of the tool command line. See the CM manual for
2878 : blume 578 explanation.
2879 :    
2880 : monnier 659 2. A special-purpose tool can be "registered" by simply dropping the
2881 : blume 578 corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2882 :     directory where the .cm file lives that uses this tool. (The
2883 :     behavior/misfeature until now was to look for the tool description
2884 :     files in the current working directory.) As before, tool description
2885 :     files could also be anchored -- in which case they can live anywhere
2886 :     they like. Following the recent e-mail discussion, this change should
2887 :     make it easier to have special-purpose tools that are shipped together
2888 :     with the sources of the program that uses them.
2889 :    
2890 :     ----------------------------------------------------------------------
2891 :     Name: Matthias Blume
2892 : blume 577 Date: 2000/03/10 07:48:34
2893 :     Tag: blume_main_v110p26p1_1
2894 :     Description:
2895 :    
2896 :     I added a re-written version of Dave's fixpt script to src/system.
2897 :     Changes relative to the original version:
2898 :     - sh-ified (not everybody has ksh)
2899 :     - automatically figures out which architecture it runs on
2900 :     - uses ./makeml a bit more cleverly
2901 :     - never invokes ./installml (and, thus, does not clobber your
2902 :     good and working installation of sml in case something goes wrong)
2903 :     - accepts max iteration count using option "-iter <n>"
2904 :     - accepts a "base" name using option "-base <base>"
2905 :    
2906 :     It does not build any extraneous heap images but directly rebuilds
2907 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
2908 :     it can incorporate existing bin- and boot- hierarchies. For example,
2909 :     suppose the base is set to "sml" (which is the default). Then it
2910 :     successively builds
2911 :    
2912 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
2913 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
2914 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
2915 :     ...
2916 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
2917 :    
2918 :     and so on. If any of these already exist, it will just use what's
2919 :     there. In particular, many people will have the initial set of bin
2920 :     and boot files around, so this saves time for at least one full
2921 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2922 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
2923 :     they might be the result of an earlier partial run of fixpt (which
2924 : monnier 659 perhaps got accidentally killed). In this case, fixpt will quickly
2925 : blume 577 move through what exists before continuing where it left off earlier,
2926 :     and, thus, saves a lot of time.
2927 :    
2928 :     ----------------------------------------------------------------------
2929 : leunga 576 Name: Allen Leung
2930 :     Date: 00/03/10 02:20:00
2931 :     Tag: leunga-20000310-fix_x86_asm_ra
2932 :     Description:
2933 : dbm 570
2934 : leunga 576 More assembly output problems involving the indexed addressing mode
2935 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
2936 :     fix.
2937 :    
2938 :     The interface and implementation of the register allocator have been changed
2939 :     slightly to accommodate the possibility to skip the register allocation
2940 :     phases completely and go directly to memory allocation. This is needed
2941 :     for C-- use.
2942 :    
2943 : dbm 570 ----------------------------------------------------------------------
2944 : blume 572 Name: Matthias Blume
2945 : blume 575 Date: 00/03/09 10:23:53
2946 :     Tag: blume_main_v110p26p1_0
2947 :     Description:
2948 :    
2949 :     * Complete re-organization of library names. Many libraries have been
2950 :     consolidated so that they share the same path anchor. For example,
2951 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
2952 :     are SML/NJ-specific are under "smlnj". Notice that names like
2953 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
2954 :     for a complete description of the new naming scheme. Quick reference:
2955 :    
2956 :     host-cmb.cm -> smlnj/cmb.cm
2957 :     host-compiler.cm -> smlnj/compiler.cm
2958 :     full-cm.cm -> smlnj/cm.cm
2959 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
2960 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
2961 :    
2962 :     * Bug fixes in CM.
2963 :     - exceptions in user code are being passed through (i.e., reach top level)
2964 :     - more bugs in paranoia mode fixed
2965 :     - bug related to checking group owners fixed
2966 :    
2967 :     * New install.sh script that automagically fetches archive files:
2968 :     The new file config/srcarchiveurl must contain the URL of the
2969 :     (remote) directory that contains bin files (or other source archives).
2970 :     If install.sh does not find the archive locally, it tries to get
2971 :     it from that remote directory.
2972 :     This should simplify installation further: For machines that have
2973 :     access to the internet, just fetch <version>-config.tgz, unpack it,
2974 : monnier 659 edit config/targets, and go (run config/install.sh). The script will
2975 : blume 575 fetch everything else that it might need all by itself.
2976 :    
2977 :     For CVS users, this mechanism is not relevant for source archives, but
2978 :     it is convenient for getting new sets of binfiles.
2979 :    
2980 :     Archives should be tar files compressed with either gzip, compress, or
2981 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
2982 :    
2983 :     ----------------------------------------------------------------------
2984 :     Name: Matthias Blume
2985 : blume 572 Date: 2000/03/07 04:01:04
2986 :     Tag: blume_main_v110_26_2
2987 : dbm 570 Description:
2988 : blume 572 - size info in BOOTLIST
2989 :     * no fixed upper limits for number of bootfiles or length of
2990 :     bootfile names in runtime
2991 :     * falling back to old behavior if no BOOTLIST size info found
2992 :     - allocation size heuristics in .run-sml
2993 :     * tries to read cache size from /proc/cpuinfo (this is important for
2994 :     small-cache Celeron systems!)
2995 :     - install.sh robustified
2996 :     - CM manual updates
2997 :     - paranoid mode
2998 :     * no more CMB.deliver() (i.e., all done by CMB.make())
2999 :     * can re-use existing sml.boot.* files
3000 :     * init.cmi now treated as library
3001 :     * library stamps for consistency checks
3002 :     - sml.boot.<arch>-<os>/PIDMAP file
3003 :     * This file is read by the CM startup code. This is used to minimize
3004 :     the amount of dynamic state that needs to be stowed away for the
3005 :     purpose of sharing between interactive system and user code.
3006 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
3007 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
3008 :     like most other CM state variables.
3009 :     - Compiler.CMSA eliminated
3010 :     * No longer supported by CM anyway.
3011 :     - fixed bugs in pickler that kept biting Stefan
3012 :     * past refs to past refs (was caused by the possibility that
3013 :     ad-hoc sharing is more discriminating than hash-cons sharing)
3014 :     * integer overflow on LargeInt.minInt
3015 :     - ml-{lex,yacc} build scripts now use new mechanism
3016 :     for building standalone programs
3017 :     - fixed several gcc -Wall warnings that were caused by missing header
3018 :     files, missing initializations, etc., in runtime (not all warnings
3019 :     eliminated, though)

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