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 896 - (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 : george 890 Name: Lal George
17 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
18 :     Tag: george-20010731-x86-fmalloc
19 :     Description: Fixed bug in x86 calls
20 :    
21 :     There was a bug where call instructions would mysteriously
22 :     vanish. The call instruction had to be one that returned
23 :     a floating point value.
24 :    
25 :     ----------------------------------------------------------------------
26 :     Name: Lal George
27 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
28 :     Tag: george-20010719-simple-cells
29 :     Description:
30 :    
31 :     I have dramatically simplified the interface for CELLS in MLRISC.
32 :    
33 :     In summary, the cells interface is broken up into three parts:
34 :    
35 :     1. CellsBasis : CELLS_BASIS
36 :    
37 :     CellsBasis is a top level structure and common for all
38 :     architectures. it contains the definitions of basic datatypes
39 :     and utility functions over these types.
40 :    
41 :     2. functor Cells() : CELLS
42 :    
43 :     Cells generates an interface for CELLS that incorporates the
44 :     specific resources on the target architecture, such as the
45 :     presence of special register classes, their number and size,
46 :     and various useful substructures.
47 :    
48 :     3. <ARCH>CELLS
49 :    
50 :     e.g. SparcCells: SPARCCELLS
51 :    
52 :     <ARCH>CELLS usually contains additional bindings for special
53 :     registers on the architecture, such as:
54 :    
55 :     val r0 : cell (* register zero *)
56 :     val y : cell (* Y register *)
57 :     val psr : cell (* processor status register *)
58 :     ...
59 :    
60 :     The structure returned by applying the Cells functor is opened
61 :     in this interface.
62 :    
63 :     The main implication of all this is that the datatypes for cells is
64 :     split between CellsBasis and CELLS -- a fairly simple change for user
65 :     code.
66 :    
67 :     In the old scheme the CELLS interface had a definitional binding of
68 :     the form:
69 :    
70 :     signature CELLS = sig
71 :    
72 :     structure CellsBasis = CellsBasis
73 :    
74 :     ...
75 :    
76 :     end
77 :    
78 :     With all the sharing constraints that goes on in MLRISC, this old
79 :     design quickly leads to errors such as:
80 :    
81 :     "structure definition spec inside of sharing ... "
82 :    
83 :    
84 :     and appears to require an unacceptable amount of sharing and where
85 :     constraint hackery.
86 :    
87 :     I think this error message (the interaction of definitional specs and
88 :     sharing) requires more explanation on our web page.
89 :    
90 :     ----------------------------------------------------------------------
91 : blume 826 Name: Matthias Blume
92 : blume 879 Date: 2001/07/19 15:00:00 EDT
93 :     Tag: blume-20010719-libreorg
94 :     Description:
95 :    
96 :     This update puts together a fairly extensive but straightforward change
97 :     to the way the libraries that implement the interactive system are
98 :     organized:
99 :    
100 :     The biggest change is the elimination of structure Compiler. As a
101 :     replacement for this structure, there is now a CM library
102 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
103 :     that exports all the substructures of the original structure Compiler
104 :     directly. So instead of saying Compiler.Foo.bar one now simply
105 :     says Foo.bar. (The CM libraries actually export a collection of
106 :     structures that is richer than the collection of substructures of
107 :     structure Compiler.)
108 :    
109 :     To make the transition smooth, there is a separate library called
110 :     $smlnj/compiler/compiler.cm which puts together and exports the
111 :     original structure Compiler (or at least something very close to it).
112 :    
113 :     There are five members of the original structure Compiler
114 :     that are not exported directly but which instead became members
115 :     of a new structure Backend (described by signature BACKEND). These are:
116 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
117 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
118 :     architecture (: string).
119 :    
120 :     Structure Compiler.Version has become structure CompilerVersion.
121 :    
122 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
123 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
124 :     or x86, respectively.
125 :     Each of these exports the same frontend structures that
126 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
127 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
128 :     Hppa, PPC, Sparc, or X86, respectively.
129 :    
130 :     Library $smlnj/compiler/all.cm exports the union of the exports of
131 :     $smlnj/compiler/<arch>.cm
132 :    
133 :     There are no structures <Arch>Compiler anymore, use
134 :     $smlnj/compiler/<arch>.cm instead.
135 :    
136 :     Library host-compiler-0.cm is gone. Instead, the internal library
137 :     that instantiates CM is now called cm0.cm. Selection of the host
138 :     compiler (backend) is no longer done here but. (Responsibility for it
139 :     now lies with $smlnj/compiler/current.cm. This seems to be more
140 :     logical.)
141 :    
142 :     Many individual files have been moved or renamed. Some files have
143 :     been split into multiple files, and some "dead" files have been deleted.
144 :    
145 :     Aside from these changes to library organization, there are also changes
146 :     to the way the code itself is organized:
147 :    
148 :     Structure Binfile has been re-implemented in such a way that it no
149 :     longer needs any knowledge of the compiler. It exclusively deals
150 :     with the details of binfile layout. It no longer invokes the
151 :     compiler (for the purpose of creating new prospective binfile
152 :     content), and it no longer has any knowledge of how to interpret
153 :     pickles.
154 :    
155 :     Structure Compile (: COMPILE) has been stripped down to the bare
156 :     essentials of compilation. It no longer deals with linking/execution.
157 :     The interface has been cleaned up considerably.
158 :    
159 :     Utility routines for dealing with linking and execution have been
160 :     moved into their own substructures.
161 :    
162 :     (The ultimate goal of these changes is to provide a light-weight
163 :     binfile loader/linker (at least for, e.g., stable libraries) that
164 :     does not require CM or the compiler to be present.)
165 :    
166 :     CM documentation has been updated to reflect the changes to library
167 :     organization.
168 :    
169 :     ----------------------------------------------------------------------
170 :     Name: Matthias Blume
171 : blume 875 Date: 2001/07/10 17:30:00 EDT
172 :     Tag: Release_110_34
173 :     Description:
174 :    
175 : blume 876 Minor tweak to 110.34 (re-tagged):
176 :    
177 :     - README.html file added to CVS repository
178 :     - runtime compiles properly under FreeBSD 3.X and 4.X
179 :    
180 :     ----------------------------------------------------------------------
181 :     Name: Matthias Blume
182 :     Date: 2001/07/10 17:30:00 EDT
183 :     Tag: Release_110_34
184 :     Description:
185 :    
186 : blume 875 New version number (110.34). New bootfiles.
187 :    
188 :     ----------------------------------------------------------------------
189 :     Name: Matthias Blume
190 : blume 874 Date: 2001/07/09 16:00:00 EDT
191 :     Tag: blume-20010709-more-varargs
192 :     Description:
193 :    
194 :     I changed the handling of varargs in ml-nlffigen again:
195 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
196 :    
197 :     The immediate effect is that you can actually call a varargs function
198 :     from ML -- but you can't actually supply any arguments beyond the ones
199 :     specified explicitly. (For example, you can call printf with its format
200 :     string, but you cannot pass additional arguments.)
201 :    
202 :     This behavior is only marginally more useful than the one before, but
203 :     it has the advantage that a function or, more importantly, a function
204 :     type never gets dropped on the floor, thus avoiding follow-up problems with
205 :     other types that refer to the offending one.
206 :    
207 :     ----------------------------------------------------------------------
208 :     Name: Matthias Blume
209 : blume 873 Date: 2001/07/09 11:25:00 EDT
210 :     Tag: blume-20010709-varargs
211 :     Description:
212 :    
213 :     1. ckit-lib.cm now exports structure Error
214 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
215 :     with a warning accompanied by a source location. Moreover, it
216 :     merely skips the offending function or type and proceeds with the
217 :     rest of its work.u As a result, one can safely feed C code containing
218 :     "..." to ml-nlffigen.
219 :     3. There are some internal improvements to CM, providing slightly
220 :     more general string substitutions in the tools subsystem.
221 :    
222 :     ----------------------------------------------------------------------
223 :     Name: Matthias Blume
224 : blume 854 Date: 2001/06/27 15:10:00 EDT
225 :     Tag: blume-20010627-concur
226 :     Description:
227 :    
228 :     Fixed a small bug in CM's handling of parallel compilation.
229 :     (You could observe the bug by Control-C-interrupting an ordinary
230 :     CMB.make or CM.stabilize and then attaching some compile servers.
231 :     The result was that all of a sudden the previously interrupted
232 :     compilation would continue on its own. This was because of
233 :     an over-optimization: CM did not bother to clean out certain queues
234 :     when no servers were attached "anyway", resulting in the contents
235 :     of these queues to grab control when new servers did get attached.)
236 :    
237 :     There is also another minor update to the CM manual.
238 :    
239 :     ----------------------------------------------------------------------
240 :     Name: Matthias Blume
241 : blume 853 Date: 2001/06/26 16:15:00 EDT
242 :     Tag: blume-20010626-cmdoc
243 :     Description:
244 :    
245 :     Minor typo fixed in CM manual (syntax diagram for libraries).
246 :    
247 :     ----------------------------------------------------------------------
248 :     Name: Matthias Blume
249 : blume 852 Date: 2001/06/25 22:55:00 EDT
250 :     Tag: blume-20010625-x86pc
251 :     Description:
252 :    
253 :     Fixed a nasty bug in the X86 assembly code that caused signal
254 :     handlers to fail (crash) randomly.
255 :    
256 :     ----------------------------------------------------------------------
257 :     Name: Matthias Blume
258 : blume 846 Date: 2001/06/25 12:05:00 EDT
259 :     Tag: blume-20010625-nlffigen
260 :     Description:
261 :    
262 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
263 :     Nick Carter <nbc@andrew.cmu.edu>.
264 :    
265 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
266 :     2. Default names for generated files are now derived from the name of
267 :     the C file *without its directory*. In particular, this causes generated
268 :     files to be placed locally even if the C file is in some system directory.
269 :     3. Default names for generated signatures and structures are also derived
270 :     from the C file name without its directory. This avoids silly things
271 :     like "structure GL/GL".
272 :     (Other silly names are still possible because ml-nlffigen does not do
273 : blume 847 a thorough check of whether generated names are legal ML identifiers.
274 :     When in doubt, use command line arguments to force particular names.)
275 : blume 846
276 :     ----------------------------------------------------------------------
277 :     Name: Matthias Blume
278 : blume 845 Date: 2001/06/21 12:25:00 EDT
279 :     Tag: blume-20010621-eXene
280 :     Description:
281 :    
282 :     eXene now compiles and (sort of) works again.
283 :    
284 :     The library name (for version > 110.33) is $/eXene.cm.
285 :    
286 :     I also added an new example in src/eXene/examples/nbody. See the
287 :     README file there for details.
288 :    
289 :     ----------------------------------------------------------------------
290 :     Name: Matthias Blume
291 : blume 844 Date: 2001/06/20 16:40:00 EDT
292 :     Tag: blume-20010620-cml
293 :     Description:
294 :    
295 :     CML now compiles and works again.
296 :    
297 :     Libraries (for version > 110.33):
298 :    
299 :     $cml/cml.cm Main CML library.
300 :     $cml/basis.cm CML's version of $/basis.cm.
301 :     $cml/cml-internal.cm Internal helper library.
302 :     $cml/core-cml.cm Internal helper library.
303 :     $cml-lib/trace-cml.cm Tracing facility.
304 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
305 :    
306 :     The installer (config/install.sh) has been taught how to properly
307 :     install this stuff.
308 :    
309 :     ----------------------------------------------------------------------
310 :     Name: Matthias Blume
311 : blume 843 Date: 2001/06/19 17:55:00 EDT
312 :     Tag: blume-20010619-instantiate
313 :     Description:
314 :    
315 :     This un-breaks the fix for bug 1432.
316 :     (The bug was originally fixed in 110.9 but I broke it again some
317 :     time after that.)
318 :    
319 :     ----------------------------------------------------------------------
320 :     Name: Matthias Blume
321 : blume 842 Date: 2001/06/19 17:25:00 EDT
322 :     Tag: blume-20010619-signals
323 :     Description:
324 :    
325 :     This should (hopefully) fix the long-standing signal handling bug.
326 :     (The runtime system was constructing a continuation record with an
327 :     incorrect descriptor which would cause the GC to drop data on the floor...)
328 :    
329 :     ----------------------------------------------------------------------
330 :     Name: Matthias Blume
331 : blume 840 Date: 2001/06/15 15:05:00 EDT
332 : blume 841 Tag: blume-20010615-moresparc
333 :     Description:
334 :    
335 :     Here is a short late-hour update related to Sparc c-calls:
336 :    
337 :     -- made handling of double-word arguments a bit smarter
338 :    
339 :     -- instruction selection phase tries to collapse certain clumsily
340 :     constructed ML-Trees; typical example:
341 :    
342 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
343 :    
344 :     This currently has no further impact on SML/NJ since mlriscGen does
345 :     not seem to generate such patterns in the first place, and c-calls
346 :     (which did generate them in the beginning) has meanwhile been fixed
347 :     so as to avoid them as well.
348 :    
349 :     ----------------------------------------------------------------------
350 :     Name: Matthias Blume
351 :     Date: 2001/06/15 15:05:00 EDT
352 : blume 840 Tag: blume-20010615-sparc
353 :     Description:
354 :    
355 :     The purpose of this update is to provide an implementation of NLFFI
356 :     on Sparc machines.
357 :    
358 :     Here are the changes in detail:
359 :    
360 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
361 :     the Sparc implementation of the c-calls API.
362 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
363 :     for accessing the ML frame. Thus, we have a real frame pointer and
364 :     can freely modify %sp without need for an omit-frame-ptr phase.
365 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
366 :     this case.
367 :     * ml-nlffigen has been taught to produce code for different architectures
368 :     and calling conventions.
369 :     * In a way similar to what was done in the x86 case, the Sparc
370 :     backend uses its own specific extension to mltree. (For example,
371 :     it needs to be able to generate UNIMP instructions which are part
372 :     of the calling convention.)
373 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
374 :     to make it easier to plug in new machine- and os-dependent parts).
375 :    
376 :     There are some other fairly unrelated bug fixes and cleanups as well:
377 :    
378 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
379 :     that they properly share their libraries with existing SML/NJ libraries.
380 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
381 :     follow-up error messages.
382 :     * Updates to CM/CMB documentation.
383 :    
384 :     TODO items:
385 :    
386 :     * MLRISC should use a different register as its asmTemp on the Sparc.
387 :     (The current %o2 is a really bad choice because it is part of the
388 :     calling conventions, so things might interfere in unexpected ways.)
389 :    
390 :     ----------------------------------------------------------------------
391 :     Name: Matthias Blume
392 : blume 839 Date: 2001/06/07
393 :     Tag: blume-20010607-calls
394 :     Description:
395 :    
396 :     A number of internal changes related to C calls and calling conventions:
397 :    
398 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
399 :     number of bytes popped implicitly (by the callee). In most cases
400 :     this field is 0 but on x86/win32 it is some non-zero value. This
401 :     is information provided for the benefit of the "omit-frameptr" pass.
402 :     2. The CALL instruction on the x86 carries a similar "pops" field.
403 :     The instruction selection phase copies its value from the ML-Tree
404 :     CALL statement.
405 :     3. On all other architectures, the instruction selection phase checks
406 :     whether "pops=0" and complains if not.
407 :     4. The c-calls implementation for x86 now accepts two calling conventions:
408 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
409 :     up after the call and pops is set to 0. For "stdcall", the caller
410 :     does nothing, leaving the cleanup to the callee; pops is set to
411 :     the number of bytes that were pushed onto the stack.
412 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
413 :     distinguish between "ccall" and "stdcall".
414 :     6. The UNIMP instruction has been added to the supported Sparc instruction
415 :     set. (This is needed for implementing the official C calling convention
416 :     on this architecture.)
417 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
418 :     work with the latest CM.
419 :    
420 :     ----------------------------------------------------------------------
421 :     Name: Matthias Blume
422 : blume 838 Date: 2001/06/05 15:10:00 EDT
423 :     Tag: blume-20010605-cm-index
424 :     Description:
425 :    
426 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
427 :    
428 :     1. CM can now generate "index files". These are human-readable files
429 :     that list on a per-.cm-file basis each toplevel symbol defined or
430 :     imported. The location of the index file for
431 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
432 :     To enable index-file generation, set CM.Control.generate_index to true
433 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
434 :    
435 :     The CM manual has been updated accordingly.
436 :    
437 :     2. I made some slight modifications to the c-calls API in MLRISC.
438 :    
439 :     a) There is now a callback to support saving/restoring of
440 :     dedicated but caller-save registers around the actual call
441 :     instruction.
442 :     b) One can optionally specify a comment-annotation for the
443 :     call instruction.
444 :    
445 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
446 :     (For example, the comment annotation shows the C prototype of
447 :     the function being called.)
448 :    
449 :     ----------------------------------------------------------------------
450 :     Name: Matthias Blume
451 : blume 837 Date: 2001/06/01 13:30:00 EDT
452 :     Tag: blume-20010601-nlffi-cleanup
453 :     Description:
454 :    
455 :     This is mostly a cleanup of MLFFI stuff:
456 :    
457 :     - some signature files have been put into a more exposed place
458 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
459 :     - ml-nlffigen changed accordingly
460 :     - tutorial updated
461 :    
462 :     Other changes:
463 :    
464 :     - author's affiliation in CM manual(s) updated
465 :     - some more recognized keywords added to Allen's sml.sty
466 :    
467 :     ----------------------------------------------------------------------
468 :     Name: Matthias Blume
469 : blume 836 Date: 2001/05/25 15:30:00 EDT
470 :     Tag: blume-20010525-iptr
471 :     Description:
472 :    
473 :     - put the official 110.33-README (as it appears on the ftp server) under
474 :     CVS
475 :     - fixed a small bug related to incomplete pointer types in
476 :     ml-nlffigen
477 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
478 :     (it does not need the 'f type parameter)
479 :    
480 :     ----------------------------------------------------------------------
481 :     Name: Matthias Blume
482 : blume 834 Date: 2001/05/23 14:30:00 EDT
483 :     Tag: Release_110_33
484 :     Description:
485 :    
486 :     New version number (110.33). New bootfiles.
487 :    
488 :     ----------------------------------------------------------------------
489 :     Name: Matthias Blume
490 : blume 833 Date: 2001/05/22 18:06:00 EDT
491 :     Tag: blume-20010522-targets
492 :     Description:
493 :    
494 :     Made install.sh use file config/targets.customized if it exists, falling
495 :     back to config/targets if it doesn't. This way one can have a customized
496 :     version of the targets file without touching the "real thing", thus
497 :     eliminating the constant fear of accidentally checking something bogus
498 :     back into the CVS repository... (File config/targets.customized must
499 :     not be added to the repository!)
500 :    
501 :     ----------------------------------------------------------------------
502 :     Name: Matthias Blume
503 : blume 832 Date: 2001/05/22 16:30:00 EDT
504 :     Tag: blume-20010522-minitut
505 :     Description:
506 :    
507 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
508 :     struct returns.
509 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
510 :     incomplete, preliminary documentation for NLFFI.
511 :    
512 :     ----------------------------------------------------------------------
513 :     Name: Matthias Blume
514 : blume 829 Date: 2001/05/14 11:30:00 EDT
515 :     Tag: blume-20010514-script
516 :     Description:
517 :    
518 :     Some bugs in install script fixed.
519 :    
520 :     In addition to that I also made a slight change to the NLFFI API:
521 :     Functors generated by ml-nlffigen now take the dynamic library as a
522 :     straight functor argument, not as a suspended one. (The original
523 :     functor code used to force the suspension right away anyway, so there
524 :     was nothing gained by this complication of the interface.)
525 :    
526 :     ----------------------------------------------------------------------
527 :     Name: Matthias Blume
528 : blume 828 Date: 2001/05/11 14:35:00 EDT
529 :     Tag: blume-20010511-ml-nlffi
530 :     Description:
531 :    
532 :     I finally took the plunge and added my new FFI code to the main
533 :     repository. For x86-linux it is now ready for prime-time.
534 :    
535 :     There are two new subdirectories of "src":
536 :    
537 :     - ml-nlffi-lib:
538 :     The utility library for programs using the FFI interface.
539 :     Here is the implementation of $/c.cm and its associated low-level
540 :     partners $/c-int.cm and $/memory.cm.
541 :     - ml-nlffigen:
542 :     A stand-alone program for generating ML glue code from C source
543 :     code.
544 :    
545 :     Building ml-nlffigen requires $/ckit-lib.cm.
546 :    
547 :     The config/install.sh script has been updates to do the Right Thing
548 :     (hopefully).
549 :    
550 :     Notice that the source tree for the C-Kit will not be put under "src"
551 :     but directly under the installation root directory. (This is the
552 :     structure that currently exists on the CVS server when you check out
553 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
554 :    
555 :     Bugs: No documentation yet.
556 :    
557 :     ----------------------------------------------------------------------
558 :     Name: Matthias Blume
559 : blume 826 Date: 2001/05/09 16:35:00 EDT
560 :     Tag: blume-20010509-cpscontract
561 :     Description:
562 :    
563 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
564 :     wrapper/unwrapper elimination did not decrement usage counts and some
565 :     dead variables got overlooked by the dead-up logic.)
566 :    
567 :     ----------------------------------------------------------------------
568 : george 824 Name: Lal George
569 :     Date: 2001/05/08 17:26:09 EDT
570 :     Tag: george-20010508-omit-frameptr
571 :     Description:
572 :    
573 :     Changes to implement the omit-frame-pointer optimization to support
574 :     raw C calls. For now, there is only support on the Intel x86, but
575 :     other architectures will follow as more experience is gained with this.
576 :    
577 :    
578 :     ----------------------------------------------------------------------
579 : blume 818 Name: Matthias Blume
580 : blume 819 Date: 2001/05/07 14:40:00 EDT
581 :     Tag: blume-20010507-proxies
582 :     Description:
583 :    
584 :     I made into "proxy libraries" all libraries that qualify for such a
585 :     change. (A qualifying library is a library that has another library or
586 :     groups as its sole member and repeats that member's export list
587 :     verbatim. A proxy library avoids this repetition by omitting its export
588 :     list, effectively inheriting the list that its (only) member exports.
589 :     See the CM manual for more explanation.)
590 :     The main effect is that explicit export lists for these libraries
591 :     do not have to be kepts in sync, making maintenance a bit easier.
592 :    
593 :     I also added copyright notices to many .cm-files.
594 :    
595 :     Last but not least, I made a new set of bootfiles.
596 :    
597 :     ----------------------------------------------------------------------
598 :     Name: Matthias Blume
599 : blume 818 Date: 2001/05/04 17:00:00 EDT
600 :     Tag: blume-20010504-cm-lsplit
601 :     Description:
602 :    
603 :     0. John merged pending changes to $/smlnj-lib.cm
604 :    
605 :     1. Allen's previous change accidentally backed out of one of Lal's
606 :     earlier changes. I undid this mistake (re-introducing Lal's change).
607 :    
608 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
609 :     within the compiler where applicable. There is some code simplification
610 :     because of that.
611 :    
612 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
613 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
614 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
615 :     it can now also be controlled on a per-source basis: CM has been taught
616 :     a new tool parameter applicable to ML source files.
617 :    
618 :     - To turn lambda-splitting off completely:
619 :     local open Compiler.Control.LambdaSplitting in
620 :     val _ = set Off
621 :     end
622 :     - To make "no lambda-splitting" the global default (but allow per-source
623 :     overriding); this is the initial setting:
624 :     local open Compiler.Control.LambdaSplitting in
625 :     val _ = set (Default NONE)
626 :     end
627 :     - To make "lambda-splitting with aggressiveness a" the global default
628 :     (and allow per-source overriding):
629 :     local open Compiler.Control.LambdaSplitting in
630 :     val _ = set (Default (SOME a))
631 :     end
632 :    
633 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
634 :     write (in the respective .cm-file):
635 :     a.sml (lambdasplitting:off)
636 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
637 :     a.sml (lambdasplitting:on)
638 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
639 :     <a> is a decimal non-negative integer):
640 :     a.sml (lambdasplitting:<a>)
641 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
642 :     a.sml (lambdasplitting:infinity)
643 :     - To use the global default for a.sml:
644 :     a.sml (lambdasplitting:default)
645 :     or simply
646 :     a.sml
647 :    
648 :     ----------------------------------------------------------------------
649 : leunga 815 Name: Allen Leung
650 :     Date: 2001/05/04 01:57:00 EDT
651 :     Tag: leunga-20010504-sync
652 :     Description:
653 : leunga 788
654 : leunga 815 MLRISC features.
655 :    
656 :     1. Fix to CMPXCHG instructions.
657 :     2. Changed RA interface to allow annotations in callbacks.
658 :     3. Added a new method to the stream interface to allow annotations updates.
659 :    
660 : george 749 ----------------------------------------------------------------------
661 : blume 812 Name: Matthias Blume
662 : blume 814 Date: 2001/05/01 11:45:00 EDT
663 :     Tag: blume-20010501-pcedittmp
664 :     Description:
665 :    
666 :     Changed install.sh to use the current working directory instead of
667 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
668 :     of /usr/tmp caused trouble with MacOS X because of file premission
669 :     problems.
670 :    
671 :     ----------------------------------------------------------------------
672 :     Name: Matthias Blume
673 : blume 812 Date: 2001/04/20 11:10:00 EDT
674 :     Tag: blume-20010420-inMLflag
675 :     Description:
676 :    
677 :     - added vp_limitPtrMask to vproc-state.h
678 :     (for use by the raw-C-calls mechanism to implement proper interrupt
679 :     handling)
680 :     - made the ML compiler aware of various data-structure offsets so it
681 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
682 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
683 :     raw C-calls
684 :    
685 :     ----------------------------------------------------------------------
686 : george 810 Name: Lal George
687 :     Date: 2001/04/20 09:15:28 EDT
688 :     Tag: george-20010420-macosX
689 :     Description:
690 :    
691 :     - Changes to port to Mac OS X; Darwin.
692 :    
693 :     - In the process I found that sqrt was broken on the PPC, because the
694 :     fsqrt instruction is not implemented.
695 :    
696 :     ----------------------------------------------------------------------
697 : blume 799 Name: Matthias Blume
698 : blume 808 Date: 2001/04/18 12:45:00 EDT
699 :     Tag: blume-20010418-ccalls
700 :     Description:
701 :    
702 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
703 :     (this bug prevented structure arguments containing pointers from being
704 :     passed correctly)
705 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
706 :     structure arguments are represented as a pointer to the beginning
707 :     of the structure (instead of having a series of synthesized arguments,
708 :     one for each structure member)
709 :    
710 :     - made makeml script's verbosity level configurable via environment
711 :     variable (MAKEML_VERBOSITY)
712 :    
713 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
714 : blume 808 in rawmem-x86.sml; we are now using the real thing
715 :    
716 :     ----------------------------------------------------------------------
717 :     Name: Matthias Blume
718 : blume 806 Date: 2001/03/22 16:25:00 EST
719 :     Tag: blume-20010322-bootfiles
720 :     Description:
721 :    
722 :     Created a new set of bootfiles (for your automatic installation convenience).
723 :    
724 :     ----------------------------------------------------------------------
725 :     Name: Matthias Blume
726 : blume 805 Date: 2001/03/22 15:10:00 EST
727 :     Tag: blume-20010322-rawmem-parcm
728 :     Description:
729 :    
730 :     1. All "raw memory access" primitives for the new FFI are implemented now
731 :     (at least on the x86).
732 :     2. Some further cleanup of CM's parallel make mechanism.
733 :    
734 :     ----------------------------------------------------------------------
735 :     Name: Matthias Blume
736 : blume 801 Date: 2001/03/19 17:53:00 EST
737 :     Tag: blume-20010319-parallel
738 :     Description:
739 :    
740 :     Parallel make (using compile servers) now works again.
741 :    
742 :     To this end, CM.stabilize and CMB.make have been modified to work in
743 :     two passes when compile servers are attached:
744 :     1. Compile everything, do not perform stabilization; this pass
745 :     uses compile servers
746 :     2. Stabilize everything; this pass does not use compile servers
747 :     If there are no compile servers, the two passes are combined into one
748 :     (as before). Splitting the passes increases the inherent parallelism
749 :     in the dependency graph because the entire graph including all
750 :     libraries is available at the same time. This, in turn, improves
751 :     server utilization. The downside is that the master process will
752 :     have to do some extra work after compilation is done (because for
753 :     technical reasons it must re-read all the binfiles during stabilization).
754 :    
755 :     ----------------------------------------------------------------------
756 :     Name: Matthias Blume
757 : blume 800 Date: 2001/03/16 12:22:00 EST
758 :     Tag: blume-20010316-bootfiles
759 :     Description:
760 :    
761 :     Created a new set of bootfiles (for your automatic installation convenience).
762 :    
763 :     ----------------------------------------------------------------------
764 :     Name: Matthias Blume
765 : blume 799 Date: 2001/03/16 11:00:00 EST
766 :     Tag: blume-20010316-MLTREE-fixup
767 :     Description:
768 :    
769 :     This is a minor fixup for an (untagged) earlier commit by Allen.
770 :     (A file was missing).
771 :    
772 :     ----------------------------------------------------------------------
773 : leunga 788 Name: Allen Leung
774 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
775 :     Tag: leunga-20010305-cut-support
776 :    
777 :     1. New support for alternative control-flow in MLTREE.
778 :     Currently we support
779 :    
780 :     FLOW_TO(CALL ...., [k1,...,kn])
781 :    
782 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
783 :     in Moby
784 :    
785 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
786 :    
787 :     2. Register Allocator
788 :    
789 :     Changes in interface [from Fermin, John]
790 :    
791 :     3. Alpha 8-bit SLL support [Fermin]
792 :    
793 :     4. All architectures
794 :    
795 :     A new module (ClusterExpandCopies) for expanding parallel copies.
796 :    
797 :     ----------------------------------------------------------------------
798 :     Name: Allen Leung
799 : leunga 788 Date: 2001/02/27 23:07:00 EST
800 :     Tag: leunga-20010227-minor-stuff
801 :    
802 :     1. Alpha bug fix for CMOVNE
803 :     2. Handle mltree COND(..,FCMP ...,...)
804 :     3. Bug fix in simplifier
805 :    
806 :     ----------------------------------------------------------------------
807 : blume 777 Name: Matthias Blume
808 : blume 781 Date: 2001/01/30 17:50:00 EST
809 :     Tag: blume-20010130-sync
810 :     Description:
811 :    
812 :     This is just a minor update to sync my devel branch with the main brach.
813 :     The only visible change is the addition of some README files.
814 :    
815 :     ----------------------------------------------------------------------
816 :     Name: Matthias Blume
817 : blume 778 Date: 2001/01/12 23:30:00 JST
818 :     Tag: blume-20010112-bootfiles
819 :     Description:
820 :    
821 :     Made a new set of bootfiles that goes with the current state of the
822 :     repository.
823 :    
824 :     ----------------------------------------------------------------------
825 :     Name: Matthias Blume
826 : blume 777 Date: 2001/01/12 21:20:00 JST
827 :     Tag: blume-20010112-sync
828 :     Description:
829 :    
830 :     I am just flushing out some minor changes that had accumulated in
831 :     my private branch in order to sync with the main tree. (This is
832 :     mainly because I had CVS trouble when trying to merge _into_ my
833 :     private branch.)
834 :    
835 :     Most people should be completely unaffected by this.
836 :    
837 :     ----------------------------------------------------------------------
838 : leunga 775 Name: Allen Leung
839 :     Date: Thu Jan 11 21:03:00 EST 2001
840 :     Tag: leunga-20010111-labexp=mltree
841 :     Description:
842 :    
843 :     1. Removed the type LabelExp and replace it by MLTree.
844 :     2. Rewritten mltree-simplify with the pattern matcher tool.
845 :     3. There were some bugs in alpha code generator which would break
846 :     64-bit code generation.
847 :     4. Redo the tools to generate code with the
848 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
849 :     are now generated by perl scripts.
850 :    
851 :     ----------------------------------------------------------------------
852 : blume 771 Name: Matthias Blume
853 : blume 774 Date: 2001/01/10 21:55:00 JST
854 :     Tag: blume-20010110-rcc
855 :     Description:
856 :    
857 :     The RCC stuff now seems to work (but only on the x86).
858 :     This required hacking of the c-calls interface (and -implementation) in
859 :     MLRISC.
860 :    
861 :     Normal compiler users should be unaffected.
862 :    
863 :     ----------------------------------------------------------------------
864 :     Name: Matthias Blume
865 : blume 773 Date: 2001/01/09 01:20:00 JST
866 :     Tag: blume-20010109-rcc
867 :     Description:
868 :    
869 :     This is a fairly big patch, flushing out a large number of pending
870 :     changes that I made to my development copy over the last couple of days.
871 :    
872 :     Of practical relevance at this moment is a workaround for a pickling
873 :     bug that Allen ran into the other day. The cause of the bug itself is
874 :     still unknown and it might be hard to fix it properly, but the
875 :     workaround has some merits of its own (namely somewhat reducing pickling
876 :     overhead for certain libraries). Therefore, I think this solution should
877 :     be satisfactory at this time.
878 :    
879 :     The rest of the changes (i.e., the vast majority) has to do with my
880 :     ongoing efforts of providing direct support for C function calls from
881 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
882 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
883 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
884 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
885 :     handle RCC.
886 :    
887 :     The last part (the changes to mlriscGen.sml) are still known to be
888 :     wrong on the x86 and not implemented on all other architectures. But
889 :     the infrastructure is in place. I had to change a few functor
890 :     signatures in the backend to be able to route the CCalls interface
891 :     from MLRISC there, and I had to specialize the mltree type (on the
892 :     x86) to include the necessary extensions. (The extensions themselves
893 :     were already there and redy to go in MLRISC/x86).
894 :    
895 :     Everything should be very happy as soon as someone helps me with
896 :     mlriscGen.sml...
897 :    
898 :     In any case, nothing of this should matter to anyone as long as the
899 :     new primop is not being used (which is going to be the case unless you
900 :     find it where I hid it :). The rest of the compiler is completely
901 :     unaffected.
902 :    
903 :     ----------------------------------------------------------------------
904 :     Name: Matthias Blume
905 : blume 772 Date: 2001/01/05 00:30:00 JST
906 :     Tag: blume-20010105-primops
907 :     Description:
908 :    
909 :     Added some experimental support for work that I am doing right now.
910 :     These changes mostly concern added primops, but there is also a new
911 :     experimental C library in the runtime system (but currently not enabled
912 :     anywhere except on Linux/X86).
913 :    
914 :     In the course of adding primops (and playing with them), I discovered that
915 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
916 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
917 :     To recover, I made type information in INL_PRIM mandatory and changed
918 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
919 :     has complete, correct type information (i.e., no bottom types).
920 :    
921 :     Since all these changes mean that we need new binfiles, I also bumped the
922 :     version number to 110.32.1.
923 :    
924 :     ----------------------------------------------------------------------
925 :     Name: Matthias Blume
926 : blume 771 Date: 2000/12/30 22:10:00 JST
927 :     Tag: blume-20001230-various
928 :     Description:
929 :    
930 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
931 :     to each other using path anchors. (See CM manual for explanation.)
932 :    
933 :     Updated CM documentation.
934 :    
935 :     Fixed some bugs in CM.
936 :    
937 :     Implemented "proxy" libraries (= syntactic sugar for CM).
938 :    
939 :     Added "-quiet" option to makeml and changed runtime system accordingly.
940 :    
941 :     Added cleanup handler for exportML to reset timers and compiler stats.
942 :    
943 :     ----------------------------------------------------------------------
944 : george 761 Name: Lal George
945 :     Date: 2000/12/22 22:22:58 EST 2000
946 :     Tag: Release_110_32
947 :     Description:
948 :    
949 :     Infinite precision used throughout MLRISC.
950 :     see MLRISC/mltree/machine-int.sig
951 :    
952 :     ----------------------------------------------------------------------
953 : blume 756 Name: Matthias Blume
954 : blume 760 Date: 2000/12/22 23:16:00 JST
955 :     Tag: blume-20001222-warn
956 :     Description:
957 : blume 759
958 : blume 760 Corrected wording and formatting of some CM warning message which I
959 :     broke in my previous patch.
960 : blume 759
961 :     ----------------------------------------------------------------------
962 :     Name: Matthias Blume
963 : blume 763 Date: 2000/12/22 21:20:00 JST
964 :     Tag: blume-20001222-anchorenv
965 :     Description:
966 :    
967 :     Fixed CM's handling of anchor environments in connection with CMB.make.
968 :    
969 :     ----------------------------------------------------------------------
970 :     Name: Matthias Blume
971 : blume 758 Date: 2000/12/22 13:15:00 JST
972 :     Tag: blume-20001222-cleanup
973 :     Description:
974 :    
975 :     Removed src/cm/ffi which does not (and did not) belong here.
976 :    
977 :     ----------------------------------------------------------------------
978 :     Name: Matthias Blume
979 : blume 757 Date: 2000/12/21 23:55:00 JST
980 :     Tag: blume-20001221-exn
981 :     Description:
982 :    
983 :     Probably most important: CM no longer silently swallows all exceptions
984 :     in the compiler.
985 :     Plus: some other minor CM changes. For example, CM now reports some
986 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
987 :    
988 :     ----------------------------------------------------------------------
989 :     Name: Matthias Blume
990 : blume 756 Date: 2000/12/15 00:01:05 JST
991 :     Tag: blume-20001215-dirtool
992 :     Description:
993 :    
994 :     - "dir" tool added.
995 :     - improvements and cleanup to Tools structure
996 :     - documentation updates
997 :    
998 :     ----------------------------------------------------------------------
999 : leunga 752 Name: Allen Leung
1000 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
1001 :     Description:
1002 :     Tag: leunga-20001214-int-inf
1003 :     Description:
1004 :    
1005 :     In IntInf, added these standard functions, which are missing from our
1006 :     implementation:
1007 :    
1008 :     andb : int * int -> int
1009 :     xorb : int * int -> int
1010 :     orb : int * int -> int
1011 :     notb : int -> int
1012 :     << : int * word -> int
1013 :     ~>> : int * word -> int
1014 :    
1015 :     Not tested, I hope they are correct.
1016 :    
1017 :     ----------------------------------------------------------------------
1018 :     Name: Allen Leung
1019 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
1020 :     Description:
1021 :     Tag: leunga-20001208-nowhere
1022 :     Description:
1023 :    
1024 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
1025 :     to generate better error messages etc. Plus a brief manual.
1026 :    
1027 :     ----------------------------------------------------------------------
1028 : george 749 Name: Lal George
1029 :     Date: 2000/12/08 09:54:02 EST 2000
1030 :     Tag: Release_110_31
1031 :     Description:
1032 : george 705
1033 : george 749 - Version 110.31
1034 : leunga 591 ----------------------------------------------------------------------
1035 : leunga 744 Name: Allen Leung
1036 :     Date: Thu Dec 7 22:01:04 EST 2000
1037 :     Tag: leunga-20001207-cell-monster-hack
1038 :     Description:
1039 :    
1040 :     Major MLRISC internal changes. Affect all clients.
1041 :     Summary:
1042 :    
1043 :     1. Type CELLS.cell = int is now replaced by a datatype.
1044 :     As a result, the old regmap is now gone. Almost all interfaces
1045 :     in MLRISC change as a consequence.
1046 :    
1047 :     2. A new brand version of machine description tool (v3.0) that generates
1048 :     modules expecting the new interface. The old version is removed.
1049 :    
1050 :     3. The RA interface has been further abstracted into two new functors.
1051 :     RISC_RA and X86RA. These functors have much simpler interfaces.
1052 :     [See also directory MLRISC/demo.]
1053 :    
1054 :     4. Some other new source->source code generation tools are available:
1055 :    
1056 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1057 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1058 :     I use this tool to generate the peephole optimizers---with the new
1059 :     cell type changes, peephole rules are becoming difficult to write
1060 :     without conditional pattern matching.
1061 :    
1062 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
1063 :     cover the entire MLRISC source tree so many things broke.
1064 :    
1065 :     6. CM files have been moved to the subdirectory MLRISC/cm.
1066 :     They are moved because there are a lot of them and they clutter up the
1067 :     root dir.
1068 :    
1069 :     7. More detailed documentation to come...
1070 :    
1071 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1072 :     a makeml -rebuild first. This is because of other other
1073 :     changes that Matthias has made (see below).
1074 :    
1075 :    
1076 :     ----------------------------------------------------------------------
1077 : blume 733 Name: Matthias Blume
1078 : blume 742 Date: 2000/11/30 23:12:00 JST
1079 :     Tag: blume-20001130-filereorg
1080 :     Description:
1081 :    
1082 :     Some manual updates and some file reorganizations in CM.
1083 :    
1084 :     ----------------------------------------------------------------------
1085 :     Name: Matthias Blume
1086 : blume 737 Date: 2000/11/24 17:45:00 JST
1087 :     Tag: blume-20001124-link
1088 :     Description:
1089 :    
1090 :     Drastically improved link traversal code for the case that the dynamic
1091 :     value was already loaded at bootstrap time. As a result, CM and CMB
1092 :     now both load blazingly fast -- even on a very slow machine. Also,
1093 :     memory consumption has been further reduced by this.
1094 :    
1095 :     Warning: The format of the PIDMAP file has changed. THerefore, to
1096 :     bootstrap you have to do this:
1097 :    
1098 :     1. Run CMB.make
1099 :     2. Make a symbolic link for the boot directory:
1100 :     ln -s sml.boot.ARCH-OS xxx
1101 :     3. "Rebuild" the boot directory:
1102 :     ./makeml -boot xxx -rebuild sml ; rm xxx
1103 :     4. Boot normally:
1104 :     ./makeml
1105 :    
1106 :     ----------------------------------------------------------------------
1107 :     Name: Matthias Blume
1108 : blume 735 Date: 2000/11/21 21:20:00 JST
1109 :     Tag: blume-20001121-tools
1110 :     Description:
1111 :    
1112 :     Continued hacking on autoloading problem -- with success this time.
1113 :     Also changed tool-plugin mechanism. See new CM manual.
1114 :    
1115 :     ----------------------------------------------------------------------
1116 :     Name: Matthias Blume
1117 : blume 734 Date: 2000/11/19 14:30:00 JST
1118 :     Tag: blume-20001119-autoload
1119 :     Description:
1120 :    
1121 :     Some hacking to make autoloading faster. Success for CMB, no success
1122 :     so far for CM. There is a reduced structure CM' that autoloads faster.
1123 :     (This is a temporary, non-documented hack to be eliminated again when
1124 :     the general problem is solved.)
1125 :    
1126 :     ----------------------------------------------------------------------
1127 :     Name: Matthias Blume
1128 : blume 733 Date: 2000/11/17 14:10:00 JST
1129 :     Tag: blume-20001117-pickle-lib
1130 :     Description:
1131 :    
1132 :     1. Eliminated comp-lib.cm
1133 :     2. Made pickle-lib.cm
1134 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
1135 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1136 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
1137 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
1138 :     Duplicated intmap modules thrown out.
1139 :     6. Hunted down all duplicated SCC code and replaced it with
1140 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1141 :     7. Rewrote Feedback module.
1142 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
1143 :     should be thrown out and equivalent modules from smlnj-lib.cm
1144 :     should be used (IntRedBlackSet, IntListSet, ...).
1145 :    
1146 :     Confirmed that compiler compiles to fixpoint.
1147 :    
1148 :     ----------------------------------------------------------------------
1149 : leunga 731 Name: Allen Leung
1150 :     Date: 2000/11/10 18:00:00
1151 :     Tag: leunga-20001110-new-x86-fp
1152 :    
1153 :     A new x86 floating point code generator has been added.
1154 :     By default this is turned off. To turn this on, do:
1155 :    
1156 :     CM.autoload "$smlnj/compiler.cm";
1157 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1158 :    
1159 :     Changes:
1160 :    
1161 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
1162 :     2. Changed the extension callback for FTANGENT to generate:
1163 :    
1164 :     fptan
1165 :     fstp %st(0)
1166 :     instead of
1167 :     fptan
1168 :     fstpl ftempmem
1169 :    
1170 :     3. Numerous assembly fixes for x86.
1171 :    
1172 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
1173 :     support for a whole bunch of instructions and addressing modes:
1174 :    
1175 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
1176 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
1177 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
1178 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1179 :     fxch %st(n)
1180 :     fld %st(n)
1181 :     fst %st(n)
1182 :     fst mem
1183 :     fstp %st(n)
1184 :     fucom %st(n)
1185 :     fucomp %st(n)
1186 :    
1187 :     All these are now generated when the fast fp mode is turned on.
1188 :    
1189 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1190 :    
1191 :     ----------------------------------------------------------------------
1192 : blume 729 Name: Matthias Blume
1193 :     Date: 2000/11/09 11:20:00 JST
1194 :     Tag: blume-20001109-scc
1195 :     Description:
1196 :    
1197 :     Eliminated some code duplication:
1198 :    
1199 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1200 :     (Otherwise the functor is useless.)
1201 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
1202 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
1203 :    
1204 :     ----------------------------------------------------------------------
1205 : george 721 Name: Lal George
1206 :     Date: 2000/11/06 09:02:21 EST 2000
1207 :     Tag: Release_110_30
1208 :     Description:
1209 :    
1210 :     - Version 110.30
1211 :     ----------------------------------------------------------------------
1212 : blume 715 Name: Matthias Blume
1213 : blume 716 Date: 2000/11/04 14:45:00
1214 :     Tag: blume-20001104-mlbuild
1215 :     Description:
1216 :    
1217 :     - Made ml-build faster on startup.
1218 :     - Documentation fixes.
1219 :    
1220 :     ----------------------------------------------------------------------
1221 :     Name: Matthias Blume
1222 : blume 715 Date: 2000/11/02 17:00:00 JST
1223 :     Tag: blume-20001102-condcomp
1224 :     Description:
1225 :    
1226 :     - Small tweaks to pickler -- new BOOTFILES!
1227 :     - Version bumped to 110.29.2.
1228 :     - Added conditional compilation facility to init.cmi (see comment there).
1229 :     ----------------------------------------------------------------------
1230 : leunga 713 Name: Allen Leung
1231 :     Date: 2000/10/23 19:31:00
1232 :     Tag: leunga-20001023-demo-ra
1233 :    
1234 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1235 :     2. Test programs for the graph library updated
1236 :     3. Some new MLRISC demo programs added
1237 :    
1238 :     ----------------------------------------------------------------------
1239 : blume 710 Name: Matthias Blume
1240 : blume 711 Date: 2000/08/31 22:15:00 JST
1241 :     Tag: blume-20001017-errmsg
1242 :     Description:
1243 :    
1244 :     More error message grief: Where there used to be no messages, there
1245 :     now were some that had bogus error regions. Fixed.
1246 :    
1247 :     ----------------------------------------------------------------------
1248 :     Name: Matthias Blume
1249 : blume 710 Date: 2000/08/31 17:30:00 JST
1250 :     Tag: blume-20001017-v110p29p1
1251 :     Description:
1252 :    
1253 :     I made a version 110.29.1 with new bootfiles.
1254 :    
1255 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
1256 :     CM documentation changes and a small bugfix in CM's error reporting.
1257 :    
1258 :     ----------------------------------------------------------------------
1259 : george 705 Name: Lal George
1260 :     Date: 2000/09/27 14:42:35 EDT
1261 :     Tag: george-20000927-nodestatus
1262 :     Description:
1263 :    
1264 :     Changed the type of the nodestatus, so that:
1265 :    
1266 :     SPILLED(~1) is now SPILLED
1267 :     SPILLED(m) where m>=0 is now MEMREG(m)
1268 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
1269 :    
1270 :     ----------------------------------------------------------------------
1271 : blume 697 Name: Matthias Blume
1272 : blume 703 Date: 2000/09/07 14:45:00 JST
1273 :     Tag: blume-20000907-cmerrmsg
1274 :     Description:
1275 :    
1276 :     Small tweak to CM to avoid getting ML syntax error messages twice.
1277 :    
1278 :     ----------------------------------------------------------------------
1279 :     Name: Matthias Blume
1280 : blume 701 Date: 2000/08/31 18:00:00 JST
1281 :     Tag: blume-20000831-cvsbootfiles
1282 :     Description:
1283 :    
1284 :     New URL for boot files (because the 110.29 files on the BL server do
1285 :     now work correctly with my updated install scripts for yacc and lex).
1286 :    
1287 :     ----------------------------------------------------------------------
1288 :     Name: Matthias Blume
1289 : blume 697 Date: 2000/08/08 12:33:00 JST
1290 :     Tag: blume-20000808-manual
1291 :     Description:
1292 :    
1293 :     Tiny update to CM manual.
1294 :    
1295 :     ----------------------------------------------------------------------
1296 : leunga 695 Name: Allen Leung
1297 :     Date: 2000/08/7 19:31:00
1298 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
1299 :    
1300 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
1301 :     and Mobdy.
1302 :    
1303 :     1. x86
1304 :    
1305 :     a. Fixes to peephole module by John and Dan.
1306 :     b. Assembly fix to SETcc by Allen.
1307 :     c. Fix to c-call by John.
1308 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
1309 :     e. Instruction selection optimization to SETcc as suggested by John.
1310 :    
1311 :     For example,
1312 :    
1313 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1314 :    
1315 :     should generate:
1316 :    
1317 :     MOVL a, x
1318 :     SUBL b, x
1319 :     SHRL 31, x
1320 :    
1321 :     2. IR stuff
1322 :    
1323 :     A bunch of new DJ-graph related algorithms added. These
1324 :     speed up SSA construction.
1325 :    
1326 :     3. SSA + Scheduling
1327 :    
1328 :     Added code for SSA and scheduling to the repository
1329 :    
1330 :     ----------------------------------------------------------------------
1331 : george 694 Name: Lal George
1332 :     Date: 2000/07/27 11:53:14 EDT
1333 :    
1334 :     Tag: lal-20000727-linux-ppc
1335 :     Description:
1336 :    
1337 :     Made changes to support Linux PPC.
1338 :     p.s. I have confirmation that the 110.29 boot files work fine.
1339 :    
1340 :     ----------------------------------------------------------------------
1341 : blume 684 Name: Matthias Blume
1342 : blume 692 Date: 2000/07/27 17:40:00 JST
1343 :     Tag: blume-20000727-scripts
1344 :     Description:
1345 :    
1346 :     !!!! WARNING !!!!
1347 :     You must recompile the runtime system!
1348 :     !!!! WARNING !!!!
1349 :    
1350 :     This is basically another round of script-enhancements:
1351 :    
1352 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1353 :     and undefine CM preprocessor symbols.
1354 :    
1355 :     2. ml-build avoids generating a new heap image if it finds that the
1356 :     existing one is still ok. (The condition is that no ML file had to
1357 :     be recompiled and all ML files are found to be older that the heap
1358 :     file.)
1359 :    
1360 :     To make this work smoothly, I also hacked the runtime system as
1361 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
1362 :     (.sparc-solaris, ...) that is currently being used.
1363 :    
1364 :     Moreover, the signature of CM.mk_standalone has changed. See the
1365 :     CM manual.
1366 :    
1367 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
1368 :     CM manual for details.)
1369 :    
1370 :     4. More CM manual updates:
1371 :     - all of the above has been documented.
1372 :     - there is now a section describing the (CM-related) command line
1373 :     arguments that are accepted by the "sml" command
1374 :    
1375 :     ----------------------------------------------------------------------
1376 :     Name: Matthias Blume
1377 : blume 691 Date: 2000/07/25 16:20:00 JST
1378 :     Tag: blume-20000725-makedepend
1379 :     Description:
1380 :    
1381 :     Added a script called ml-makedepend. This can be used in makefiles
1382 :     for Unix' make in a way very similar to the "makedepend" command for
1383 :     C.
1384 :    
1385 :     The script internally uses function CM.sources.
1386 :    
1387 :     Synopsis:
1388 :    
1389 :     ml-makedepend [-f makefile] cmfile targetname
1390 :    
1391 :     The default for the makefile is "makefile" (or "Makefile" should
1392 :     "makefile" not exist).
1393 :    
1394 :     ml-makedepend adds a cmfile/targetname-specific section to this
1395 :     makefile (after removing the previous version of this section). The
1396 :     section contains a single dependency specification with targetname on
1397 :     the LHS (targetname is an arbitrary name), and a list of files derived
1398 :     from the cmfile on the RHS. Some of the files on the RHS are
1399 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
1400 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1401 :     path names. The makefile writer is responsible for making sure that
1402 :     these variables have correct at the time "make" is invoked.
1403 :    
1404 :     ----------------------------------------------------------------------
1405 :     Name: Matthias Blume
1406 : blume 690 Date: 2000/07/22 23:30:00 JST
1407 :     Tag: blume-20000722-urlupdate
1408 :     Description:
1409 :    
1410 :     Changed BOOT and config/srcarchiveurl to point to BL server:
1411 :    
1412 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1413 :    
1414 :     ----------------------------------------------------------------------
1415 :     Name: Matthias Blume
1416 : blume 686 Date: 2000/07/18 18:00:00 JST
1417 :     Tag: blume-20000718-Version_110_29
1418 :     Description:
1419 :    
1420 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1421 :    
1422 :     2. Updated config/version to 110.29
1423 :    
1424 :     3. Updated config/srcarchiveurl
1425 :    
1426 :     3. New boot files!
1427 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1428 :    
1429 :     ----------------------------------------------------------------------
1430 :     Name: Matthias Blume
1431 : blume 684 Date: 2000/07/11 13:58:00 JST
1432 :     Tag: blume-20000711-doctypo
1433 :     Description:
1434 :    
1435 :     Fixed a few typos in CM manual.
1436 :    
1437 :     ----------------------------------------------------------------------
1438 : leunga 683 Name: Allen Leung
1439 :     Date: 2000/06/15 00:38:00
1440 :     Tag: leunga-20000704-sparc-x86
1441 :    
1442 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
1443 :     2. fix to x86 RET bug [found by Dan Grossman]
1444 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
1445 :    
1446 :     Affects c-- and moby only
1447 :    
1448 :     ----------------------------------------------------------------------
1449 : blume 680 Name: Matthias Blume
1450 : blume 682 Date: 2000/07/04 15:26:00
1451 :     Tag: blume-20000704-trigger
1452 :     Description:
1453 :    
1454 :     1. Improvements to CM manual.
1455 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1456 :     of getting a back-trace. The function, when called, raises an
1457 :     internal exception which explicitly carries the full back-trace history,
1458 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
1459 :     or not). The interactive loop will print that history once it arrives
1460 :     at top level.
1461 :     Short of having all exceptions implicitly carry the full history, the
1462 :     recommended way of using this facility is:
1463 :     - compile your program with instrumentation "on"
1464 :     - run it, when it raises an exception, look at the history
1465 :     - if the history is "cut off" because of some handler, go and modify
1466 :     your program so that it explicitly calls BTrace.trigger
1467 :     - recompile (still instrumented), and rerun; look at the full history
1468 :    
1469 :     ----------------------------------------------------------------------
1470 :     Name: Matthias Blume
1471 :     Date: 2000/07/03 15:36:00 JST
1472 : blume 680 Tag: blume-20000702-manual
1473 :     Description:
1474 :    
1475 :     Small corrections and updates to CM manual.
1476 :    
1477 :     ----------------------------------------------------------------------
1478 : blume 682 Name: Matthias Blume
1479 : blume 679 Date: 2000/06/29 16:04:00 JST
1480 :     Tag: blume-20000629-yacctool
1481 :     Description:
1482 :    
1483 :     Changes:
1484 :    
1485 :     1. Class "mlyacc" now takes separate arguments to pass options to
1486 :     generated .sml- and .sig-files independently.
1487 :     2. Corresponding CM manual updates.
1488 :     3. BTrace module now also reports call sites. (However, for loop clusters
1489 :     it only shows from where the cluster was entered.) There are associated
1490 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1491 :    
1492 :     ----------------------------------------------------------------------
1493 : blume 675 Name: Matthias Blume
1494 : blume 678 Date: 2000/06/27 16:51:00 JST
1495 :     Tag: blume-20000627-noweb
1496 :     Description:
1497 :    
1498 :     Changes:
1499 :    
1500 :     1. Implemented "subdir" and "witness" options for noweb tool.
1501 :     This caused some slight internal changes in CM's tool implementation.
1502 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
1503 :     some remaining issues from earlier path anchor changes.
1504 :     3. Updated CM manual accordingly.
1505 :    
1506 :     4. Changed implementation of back-tracing so that I now consider it
1507 :     ready for prime-time.
1508 :    
1509 :     In particular, you don't have to explicitly trigger the back-trace
1510 :     anymore. Instead, if you are running BTrace-instrumented code and
1511 :     there is an uncaught exception (regardless of whether or not it was
1512 :     raised in instrumented code), the top-level evalloop will print
1513 :     the back-trace.
1514 :    
1515 :     Features:
1516 :    
1517 :     - Instrumented and uninstrumented code work together seemlessly.
1518 :     (Of course, uninstrumented code is never mentioned in actual
1519 :     back-traces.)
1520 :    
1521 :     - Asymptotic time- and space-complexity of instrumented code is
1522 :     equal to that of uninstrumented code. (This means that
1523 :     tail-recursion is preserved by the instrumentation phase.)
1524 :    
1525 :     - Modules whose code has been instrumented in different sessions
1526 :     work together without problem.
1527 :    
1528 :     - There is no penalty whatsoever on uninstrumented code.
1529 :    
1530 :     - There is no penalty on "raise" expressions, even in
1531 :     instrumented code.
1532 :    
1533 :     A potential bug (or perhaps it is a feature, too):
1534 :    
1535 :     A back-trace reaches no further than the outermost instrumented
1536 :     non-trivial "raise". Here, a "trivial" raise is one that is the
1537 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
1538 :    
1539 :     <exp> handle e => raise e
1540 :    
1541 :     and even
1542 :    
1543 :     <exp> handle Foo => raise Bar
1544 :    
1545 :     and, of course, through
1546 :    
1547 :     <exp> handle Foo => ...
1548 :    
1549 :     if the exception was not Foo.
1550 :    
1551 :     Back-traces always reach right through any un-instrumented code
1552 :     including any of its "handle" expressions, trivial or not.
1553 :    
1554 :     To try this out, do the following:
1555 :    
1556 :     - Erase all existing binfiles for your program.
1557 :     (You may keep binfiles for those modules where you think you
1558 :     definitely don't need back-tracing.)
1559 :     - Turn on back-trace instrumentation:
1560 :     SMLofNJ.Internals.BTrace.mode (SOME true);
1561 :     - Recompile your program. (I.e., run "CM.make" or "use".)
1562 :     - You may now turn instrumentation off again (if you want):
1563 :     SMLofNJ.Internals.BTrace.mode (SOME false);
1564 :     - Run your program as usual. If it raises an exception that
1565 :     reaches the interactive toplevel, then a back-trace will
1566 :     automatically be printed. After that, the toplevel loop
1567 :     will print the exception history as usual.
1568 :    
1569 :     ----------------------------------------------------------------------
1570 :     Name: Matthias Blume
1571 : blume 677 Date: 2000/06/26 09:56:46 JST
1572 :     Tag: blume-20000626-setup
1573 :     Description:
1574 :    
1575 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1576 :     ML code before and after compiling a file (e.g., to set compiler
1577 :     flags)
1578 :    
1579 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1580 :     - associated changes to btrace.sml (BTrace instrumentation pass)
1581 :     - cleaner implementation of btimp.sml (BTrace tracing and report
1582 :     module)
1583 :    
1584 :     CM manual: * new path encoding documented
1585 :     * description of setup-parameter to "sml" added
1586 :    
1587 :     The biggest user-visible change to back-tracing is that it is no
1588 :     longer necessary to compile all traced modules within the same
1589 :     session. (This was a real limitation.)
1590 :    
1591 :     ----------------------------------------------------------------------
1592 :     Name: Matthias Blume
1593 : blume 676 Date: 2000/06/24 12:40:00 JST
1594 :     Tag: blume-20000624-startup
1595 :     Description:
1596 :    
1597 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
1598 :     bit too often -- to put it mildly. :)
1599 :    
1600 :     ----------------------------------------------------------------------
1601 :     Name: Matthias Blume
1602 : blume 675 Date: 2000/06/23 18:20:00 JST
1603 :     Tag: blume-20000623-btrace
1604 :     Description:
1605 :    
1606 :     This updates adds a backtrace facility to aid programmers in debugging
1607 :     their programs. This involves the following changes:
1608 :    
1609 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1610 :     keeping track of the current call stack. When programs are compiled
1611 :     in a special mode, the compiler will insert calls to these hooks
1612 :     into the user program.
1613 :     "Hook" means that it is possible for different implementations of
1614 :     back-tracing to register themselves (at different times).
1615 :    
1616 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1617 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
1618 :     It can be turned on using this call:
1619 :     SMLofNJ.Internals.BTrace.mode (SOME true);
1620 :     Turning it off again:
1621 :     SMLofNJ.Internals.BTrace.mode (SOME false);
1622 :     Querying the current status:
1623 :     SMLofNJ.Internals.BTrace.mode NONE;
1624 :     Annotated programs are about twice as big as normal ones, and they
1625 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1626 :     where all hooks do nothing). The slowdown with a plugin that is
1627 :     actually useful (such as the one supplied by default) is even greater,
1628 :     but in the case of the default plugin it is still only an constant
1629 :     factor (amortized).
1630 :    
1631 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1632 :     with a sub-structure BTrace for controlling back-tracing. In particular,
1633 :     the above-mentioned function "mode" controls whether the annotation
1634 :     phase is invoked by the compiler. Another important function is
1635 :     "trigger": when called it aborts the current execution and causes
1636 :     the top-level loop to print a full back-trace.
1637 :    
1638 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1639 :     for back-tracing. It keeps track of the dynamic call stack and in
1640 :     addition to that it keeps a partial history at each "level" of that
1641 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
1642 :     d and b (at separate times, dynamically), then the report will show:
1643 :    
1644 :     GOTO d
1645 :     /c
1646 :     GOTO \b
1647 :     CALL a
1648 :    
1649 :     This shows that there was an initial non-tail call of a, then a
1650 :     tail-call to b or c, looping behavior in a cluster of functions that
1651 :     consist of b and c, and then a goto from that cluster (i.e., either from
1652 :     b or from c) to d.
1653 :    
1654 :     Note that (depending on the user program) the amount of information
1655 :     that the back-trace module has to keep track of at each level is bounded
1656 :     by a constant. Thus, the whole implementation has the same asymptotical
1657 :     complexity as the original program (both in space and in time).
1658 :    
1659 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
1660 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
1661 :     which is raised by the "trigger" function mentioned above.
1662 :    
1663 :     Notes on usage:
1664 :    
1665 :     - Annotated code works well together with unannotated code:
1666 :     Unannotated calls simply do not show up at all in the backtrace.
1667 :    
1668 :     - It is not a good idea to let modules that were annotated during
1669 :     different sessions run at the same time. This is because the compiler
1670 :     chooses small integers to identify individual functions, and there
1671 :     will be clashes if different modules were compiled in separate sessions.
1672 :     (Nothing will crash, and you will even be told about the clashes, but
1673 :     back-trace information will in general not be useful.)
1674 :    
1675 :     - Back-tracing can be confused by callcc and capture.
1676 :    
1677 :     - The only way of getting a back-trace right now is to explicitly
1678 :     invoke the "trigger" function from your user program. Eventually, we
1679 :     should make every exception carry back-trace information (if
1680 :     available). But since this creates more overhead at "raise"-time
1681 :     (similar to the current exnHistory overhead), I have not yet
1682 :     implemented this. (The implementation will be rather easy.) With
1683 :     exceptions carrying back-trace information, this facility will be even
1684 :     more useful because users don't need to modify their programs...
1685 :    
1686 :     - While it is possible to compile the compiler with back-trace
1687 :     annotations turned on (I did it to get some confidence in
1688 :     correctness), you must make absolutely sure that core.sml and
1689 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
1690 :     be compiled with annotation because there is no core access yet, but
1691 :     if you compile btimp.sml with annotation, then the system will go into
1692 :     an infinite recursion and crash.)
1693 :     Since CM currently does not know about BTrace, the only way to turn
1694 :     annotations on and off for different modules of the compiler is to
1695 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
1696 :     this is awkward and clumsy.
1697 :    
1698 :     Sample sessions:
1699 :    
1700 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1701 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
1702 :     [autoloading]
1703 :     [autoloading done]
1704 :     val it = false : bool
1705 :     - structure X = struct
1706 :     - fun main n = let
1707 :     - fun a (x, 0) = d x
1708 :     - | a (x, n) = b (x, n - 1)
1709 :     - and b (x, n) = c (x, n)
1710 :     - and c (x, n) = a (x, n)
1711 :     - and d x = e (x, 3)
1712 :     - and e (x, 0) = f x
1713 :     - | e (x, n) = e (x, n - 1)
1714 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1715 :     - | f n = n * g (n - 1)
1716 :     - and g n = a (n, 3)
1717 :     - in
1718 :     - f n
1719 :     - end
1720 :     - end;
1721 :     structure X : sig val main : int -> int end
1722 :     - X.main 3;
1723 :     *** BACK-TRACE ***
1724 :     GOTO stdIn:4.2-13.20: X.main[2].f
1725 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1726 :     GOTO stdIn:4.2-13.20: X.main[2].d
1727 :     / stdIn:4.2-13.20: X.main[2].a
1728 :     | stdIn:4.2-13.20: X.main[2].b
1729 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1730 :     CALL stdIn:4.2-13.20: X.main[2].g
1731 :     GOTO stdIn:4.2-13.20: X.main[2].f
1732 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1733 :     GOTO stdIn:4.2-13.20: X.main[2].d
1734 :     / stdIn:4.2-13.20: X.main[2].a
1735 :     | stdIn:4.2-13.20: X.main[2].b
1736 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1737 :     CALL stdIn:4.2-13.20: X.main[2].g
1738 :     GOTO stdIn:4.2-13.20: X.main[2].f
1739 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
1740 :     GOTO stdIn:4.2-13.20: X.main[2].d
1741 :     / stdIn:4.2-13.20: X.main[2].a
1742 :     | stdIn:4.2-13.20: X.main[2].b
1743 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
1744 :     CALL stdIn:4.2-13.20: X.main[2].g
1745 :     GOTO stdIn:4.2-13.20: X.main[2].f
1746 :     CALL stdIn:2.15-17.4: X.main[2]
1747 :     -
1748 :    
1749 :     (Note that because of a FLINt bug the above code currently does not
1750 :     compile without BTrace turned on.)
1751 :    
1752 :     Here is another example, using my modified Tiger compiler:
1753 :    
1754 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1755 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
1756 :     [autoloading]
1757 :     [autoloading done]
1758 :     val it = false : bool
1759 :     - CM.make "sources.cm";
1760 :     [autoloading]
1761 :     ...
1762 :     [autoloading done]
1763 :     [scanning sources.cm]
1764 :     [parsing (sources.cm):parse.sml]
1765 :     [creating directory CM/SKEL ...]
1766 :     [parsing (sources.cm):tiger.lex.sml]
1767 :     ...
1768 :     [wrote CM/sparc-unix/semant.sml]
1769 :     [compiling (sources.cm):main.sml]
1770 :     [wrote CM/sparc-unix/main.sml]
1771 :     [New bindings added.]
1772 :     val it = true : bool
1773 :     - Main.compile ("../testcases/merge.tig", "foo.out");
1774 :     *** BACK-TRACE ***
1775 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1776 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1777 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1778 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1779 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1780 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1781 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1782 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1783 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1784 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1785 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1786 :     CALL main.sml:1.18-118.4: Main.compile[2]
1787 :     -
1788 :    
1789 :     ----------------------------------------------------------------------
1790 : blume 673 Name: Matthias Blumen
1791 :     Date: 2000/06/21 18:00:00 JST
1792 :     Tag: blume-20000621-manual
1793 :     Description:
1794 :    
1795 :     CM manual update: Path environments documented.
1796 :    
1797 :     ----------------------------------------------------------------------
1798 : blume 666 Name: Matthias Blume
1799 : blume 672 Date: 2000/06/19 13:40:00
1800 :     Tag: blume-20000619-manual
1801 :     Description:
1802 :    
1803 :     CM manual and system/README update. This only covers the fact that
1804 :     there are no more implicit anchors. (Path environments and the "bind"
1805 :     option to "cm" have yet to be documented.)
1806 :    
1807 :     ----------------------------------------------------------------------
1808 :     Name: Matthias Blume
1809 : blume 671 Date: 2000/06/19 11:05:00 JST
1810 :     Tag: blume-20000619-chdir-bugfix
1811 :     Description:
1812 :    
1813 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1814 :    
1815 :     ----------------------------------------------------------------------
1816 :     Name: Matthias Blume
1817 : blume 670 Date: 2000/06/18 22:00:10 JST
1818 :     Tag: blume-20000618-implicit-anchors-really-gone
1819 :     Description:
1820 :    
1821 :     I updates the previous HISTORY entry where I forgot to mention that
1822 :     implicit anchors are no longer with us.
1823 :    
1824 :     The current update also gets rid of the (now useless) controller
1825 :     CM.Control.implicit_anchors.
1826 :    
1827 :     ----------------------------------------------------------------------
1828 :     Name: Matthias Blume
1829 : blume 666 Date: 2000/06/16 17:30:00 JST
1830 :     Tag: blume-20000616-anchorenv
1831 :     Description:
1832 :    
1833 :     This patch implements the long anticipated (just kidding :) "anchor
1834 :     environment" mechanism. In the course of doing this, I also
1835 :     re-implemented CM's internal "SrcPath" module from scratch. The new
1836 :     one should be more robust in certain boundary cases. In any case, it
1837 :     is a lot cleaner than its predecessor (IMHO).
1838 :    
1839 :     This time, although there is yet another boot file format change, I
1840 :     kept the unpickler backward-compatible. As a result, no new bootfiles
1841 :     are necessary and bootstrapping is straightforward. (You cannot read
1842 :     new bootfiles into an old system, but the other way around is no
1843 :     problem.)
1844 :    
1845 :     Visible changes:
1846 :    
1847 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
1848 :     longer recognized at all. This means that such path names are not
1849 :     illegal either. For example, the name basis.cm simply refers to a
1850 :     local file called "basis.cm" (i.e, the name is an ordinary path
1851 :     relative to .cm-files directory). Or, to put it differently, only
1852 :     names that start with $ are anchored paths.
1853 : blume 666
1854 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1855 :     vanished.
1856 :    
1857 :     John (Reppy) had critizised this as soon as I originally proposed and
1858 : blume 666 implemented it, but at that time I did not really deeply believe
1859 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
1860 :     in another place where it cannot be seen as an abbreviation for
1861 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
1862 :     seems to mean (i.e., it "expands" into the corresponding anchor
1863 :     value).
1864 :    
1865 :     However, when paths are used as members in CM description files, it
1866 :     continues to be true that there must be at least another arc after the
1867 :     anchor. This is now enforced separately during semantic analysis
1868 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
1869 : blume 666
1870 :     ** 2. The "cm" class now accepts an option "bind". The option's value
1871 :     is a sub-option list of precisely two items -- one labeled "anchor"
1872 :     and the other one labeled "value". As you might expect, "anchor" is
1873 :     used to specify an anchor name to be bound, and "value" specifies what
1874 :     the anchor is being bound to.
1875 :    
1876 :     The value must be a directory name and can be given in either standard
1877 :     syntax (including the possibility that it is itself an anchored path)
1878 :     or native syntax.
1879 :    
1880 :     Examples:
1881 :    
1882 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
1883 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
1884 :    
1885 :     and so on.
1886 :    
1887 :     The meaning of this is that the .cm-file will be processed with an
1888 :     augmented anchor environment where the given anchor(s) is/are bound to
1889 :     the given values(s).
1890 :    
1891 :     The rationale for having this feature is this: Suppose you are trying
1892 :     to use two different (already stable) libraries a.cm and b.cm (that
1893 :     you perhaps didn't write yourself). Further, suppose each of these
1894 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
1895 :     Normally you would now have a problem because the anchor "lib" can not
1896 :     be bound to more than one value globally. Therefore, the project that
1897 :     uses both a.cm and b.cm must locally redirect the anchor to some other
1898 :     place:
1899 :    
1900 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1901 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1902 :    
1903 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1904 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1905 :    
1906 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1907 :     you when you do so at the time of CM.stabilize). Therefore, you can
1908 :     also use an anchored path as the value:
1909 :    
1910 :     a.cm (bind:(anchor:lib value:$a-lib))
1911 :     b.cm (bind:(anchor:lib value:$b-lib))
1912 :    
1913 :     Now you can globally configure (using the usual CM.Anchor.anchor or
1914 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
1915 :     itself is always locally bound, setting it globally is no longer
1916 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
1917 :     can still be used as a global anchor for separate purposes. As a
1918 :     matter of fact, one can locally define "lib" in terms of a global
1919 :     "lib":
1920 :    
1921 :     a.cm (bind:(anchor:lib value:$lib/a))
1922 :     b.cm (bind:(anchor:lib value:$lib/b))
1923 :    
1924 :     ** 3: The encoding of path names has changed. This affects the way
1925 :     path names are shown in CM's progress report and also the internal
1926 :     protocol encoding used for parallel make.
1927 :    
1928 :     The encoding now uses one or more ':'-separated segments. Each
1929 :     segments corresponds to a file that has been specified relative to the
1930 :     file given by its preceding segment. The first segment is either
1931 :     relative to the CWD, absolute, or anchored. Each segment itself is
1932 :     basically a Unix pathname; all segments but the first are relative.
1933 :    
1934 :     Example:
1935 :    
1936 :     $foo/bar/baz.cm:a/b/c.sml
1937 :    
1938 :     This path denotes the file bar/a/b/c.sml relative to the directory
1939 :     denoted by anchor "foo". Notice that the encoding also includes
1940 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
1941 :     paths are resolved relative to the .cm-files directory, so baz.cm must
1942 :     be ignored to get the "real" pathname.
1943 :    
1944 :     To make this fact more obvious, CM puts the names of such "virtual
1945 :     arcs" into parentheses when they appear in progress reports. (No
1946 :     parentheses will appear in the internal protocol encoding.) Thus,
1947 :     what you really see is:
1948 :    
1949 :     $foo/bar/(baz.cm):a/b/c.sml
1950 :    
1951 :     I find this notation to be much more informative than before.
1952 :    
1953 :     Another new feature of the encoding is that special characters
1954 :     including parentheses, colons, (back)slashes, and white space are
1955 :     written as \ddd (where ddd is the decimal encoding of the character).
1956 :    
1957 : blume 670 *** The CM manual still needs to be updated.
1958 :    
1959 : blume 666 ----------------------------------------------------------------------
1960 : leunga 660 Name: Allen Leung
1961 :     Date: 2000/06/15 00:38:00
1962 :     Tag: leunga-20000615-x86-peephole
1963 :    
1964 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
1965 :    
1966 :     ----------------------------------------------------------------------
1967 : blume 658 Name: Matthias Blume
1968 :     Date: 2000/06/12 11:40:00
1969 :     Tag: blume-20000612-parmakefix
1970 :     Description:
1971 :    
1972 :     More cleanup after changing the file naming scheme: This time I
1973 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
1974 :    
1975 :     ----------------------------------------------------------------------
1976 : leunga 657 Name: Allen Leung
1977 :     Date: 2000/06/09 01:25:00
1978 :     Tag: leunga-20000609-various
1979 :    
1980 :     None of these things should affect normal SML/NJ operations
1981 :    
1982 :     1. Peephole improvements provided by Fermin (c--)
1983 :     2. New annotation DEFUSE for adding extra dependence (moby)
1984 :     3. New X86 LOCK instructions (moby)
1985 :     4. New machine description language for reservation tables (scheduling)
1986 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
1987 :     trees etc.)
1988 :     6. I've changed the CM files so that they can work with versions
1989 :     110.0.6, 110.25 and 110.28
1990 :    
1991 :     ----------------------------------------------------------------------
1992 : blume 652 Name: Matthias Blume
1993 : blume 656 Date: 2000/06/09 12:40:00
1994 :     Tag: blume-20000609-log
1995 :     Description:
1996 :    
1997 :     - Removed all(?) remaining RCS Log entries from sources.
1998 :    
1999 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
2000 :     anchored paths).
2001 :    
2002 :     ----------------------------------------------------------------------
2003 :     Name: Matthias Blume
2004 : blume 653 Date: 2000/06/07 17:00:00 JST
2005 :     Tag: blume-20000607-no-implicit-anchors
2006 :     Description:
2007 :    
2008 :     1. This update changes the default setting for
2009 :     CM.Control.implicit_anchors from true to false. This means that
2010 :     implicit anchors are no longer permitted by default. I also tried to
2011 :     make sure that nothing else still relies on implicit anchors.
2012 :     (This is the next step on the schedule towards a CM that does not even
2013 :     have the notion of implicit anchors anymore.)
2014 :    
2015 :     2. More CM manual updates.
2016 :    
2017 :     3. I managed to track down and fix the pickling bug I mentioned last
2018 :     time. Because of the previously existing workaround, this entails no
2019 :     immediate practical changes.
2020 :    
2021 :     ----------------------------------------------------------------------
2022 :     Name: Matthias Blume
2023 : blume 652 Date: 2000/06/06 11:15:00 JST
2024 :     Tag: blume-20000606-lazierpickle
2025 :     Description:
2026 :    
2027 :     !!!! NEW BOOT FILES !!!!
2028 :    
2029 :     * The main purpose of this update is to make library pickles lazier in
2030 :     order to reduce the initial space penalty for autoloading a library.
2031 :     As a result, it is now possible to have $smlnj/compiler.cm
2032 :     pre-registered. This should take care of the many complaints or
2033 :     inquiries about missing structure Compiler. This required changes to
2034 :     CM's internal data structures and small tweaks to some algorithms.
2035 :    
2036 :     As a neat additional effect, it is no longer necessary (for the sake
2037 :     of lean heap image files) to distinguish between a "minimal" CM and a
2038 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
2039 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
2040 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
2041 :    
2042 :     To make the life of compiler-hackers easier, "makeml" now also
2043 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
2044 :     words, after you bootstrap a new sml for the first time, you will not
2045 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
2046 :     around you will still have to do it, though.)
2047 :    
2048 :     * A second change consists of major updates to the CM manual. There
2049 :     are now several appendices with summary information and also a full
2050 :     specification of the CM description file syntax.
2051 :    
2052 :     * In directory src/system I added the script "allcross". This script
2053 :     invokes sml and cross-compiles the compiler for all supported
2054 :     architectures. (Useful when providing a new set of boot files.)
2055 :    
2056 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
2057 :     added a small tweak to pickle-util.sml to work around this problem,
2058 :     but it is not a proper fix yet. I will investigate further. (The
2059 :     effect of the bug was an inflation of library pickle size.)
2060 :    
2061 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
2062 :    
2063 :     ----------------------------------------------------------------------
2064 : leunga 646 Name: Allen Leung
2065 : leunga 648 Date: 2000/05/25 17:28 EDT
2066 :     Tag: leunga-20000525-ra
2067 :     Description:
2068 :    
2069 :     Fixed a bug in freezing phase of the register allocator.
2070 :    
2071 :     ----------------------------------------------------------------------
2072 :     Name: Allen Leung
2073 : leunga 647 Date: 2000/05/15 22:53 EDT
2074 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
2075 :     Description:
2076 :    
2077 :     1. Alpha
2078 :    
2079 :     Slight cleanup. Removed the instruction SGNXL
2080 :    
2081 :     2. X86
2082 :    
2083 :     Added the following instructions to the instruction set:
2084 :    
2085 :     ROLx, RORx,
2086 :     BTx, BTSx, BTLx, BTRx,
2087 :     XCHGx, and variants with the LOCK prefix
2088 :    
2089 :     3. Register Allocation
2090 :    
2091 :     The module ra-rewrite-with-renaming has been improved.
2092 :    
2093 : leunga 647 These have no effect on SML/NJ.
2094 :    
2095 : leunga 646 ----------------------------------------------------------------------
2096 : blume 642 Name: Matthias Blume
2097 : blume 645 Date: 2000/05/15 16:20:00 JST
2098 :     Tag: blume-20000515-lightrebuild
2099 :     Description:
2100 :    
2101 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
2102 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
2103 :     defined. In effect, the command will not build any cross-compiler
2104 :     backends and therefore finish more quickly.
2105 :    
2106 :     The "fixpt" script also takes a "-light" switch to be able to use
2107 :     this new facility while compiling for a fixpoint.
2108 :    
2109 :     2. I replaced all mentions of anchored paths in group owner specifications
2110 :     with simple relative paths (usually starting with "..").
2111 :     The rationale is that a library's internal workings should not be
2112 :     compromised by the lack of some anchor. (An anchor is necessary
2113 :     for someone who wants to refer to the library by an anchored path,
2114 :     but it should not be necessary to build the same library in the first
2115 :     place.)
2116 :    
2117 :     3. I changed the way CM's tool mechanism determines the shell command
2118 :     string used for things like ml-yacc etc. so that it does not break
2119 :     when CM.Control.implicit_anchors is turned off.
2120 :    
2121 :     ----------------------------------------------------------------------
2122 :     Name: Matthias Blume
2123 : blume 644 Date: 2000/05/12 18:20:00 JST
2124 :     Tag: blume-20000512-ml-build
2125 :     Description:
2126 :    
2127 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
2128 :     getting installed properly (by config/install.sh).
2129 :    
2130 :     ----------------------------------------------------------------------
2131 :     Name: Matthias Blume
2132 : blume 643 Date: 2000/05/12 17:30:00 JST
2133 :     Tag: blume-20000512-anchors
2134 :     Description:
2135 :    
2136 :     !!! NEW BOOT FILES !!!
2137 :    
2138 :     This change is in preparation of fading out support for "implicitly
2139 :     anchored path names". I went through all sources and used the
2140 :     explicit (and relatively new) $-notation. See system/README and the
2141 :     CM manual for more info on this.
2142 :    
2143 :     I also modified the anchoring scheme for some things such as "smlnj",
2144 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
2145 :     anchors are more expressive: anchor name and first arc do not have to
2146 :     coincide. This entails the following user-visible change:
2147 :    
2148 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
2149 :     particular, when you fire up sml with a command-line argument, say,
2150 :     e.g.:
2151 :    
2152 :     sml '$smlnj/cmb.cm'
2153 :    
2154 :     At the ML toplevel prompt:
2155 :    
2156 :     CM.autoload "$smlnj/cmb.cm";
2157 :    
2158 :     There is also a new controller in CM.Control that can be used to turn
2159 :     off all remaining support for implicit anchors by saying:
2160 :    
2161 :     CM.autoload "$smlnj/
2162 :     #set CM.Control.implicit_anchors false;
2163 :    
2164 :     This causes CM to reject implicitly anchored paths. This is (for the
2165 :     time being) less permissive than the "final" version where there will
2166 :     be no more such implicit anchors and relative paths will be just that:
2167 :     relative.
2168 :    
2169 :     The next step (version after next version?) will be to make the
2170 :     default for CM.Control.implicit_anchors false. After the dust has
2171 :     settled, I can then produce the "final" version of this...
2172 :    
2173 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
2174 :    
2175 :     ----------------------------------------------------------------------
2176 :     Name: Matthias Blume
2177 : blume 642 Date: 2000/05/11 16:30:00 JST
2178 :     Tag: blume-20000511-sources
2179 :     Description:
2180 :    
2181 :     The main change is that I added function CM.sources as a generalized
2182 :     version of the earlier CM.makedepend. This entails the following
2183 :     additional changes:
2184 :    
2185 :     - CM.makedepend has been dropped.
2186 :    
2187 :     - CM manual has been updated.
2188 :    
2189 :     - TOOLS signature and API have been changed.
2190 :    
2191 :     ----------------------------------------------------------------------
2192 : leunga 640 Name: Allen Leung
2193 : leunga 641 Date: 2000/05/10 21:17 EDT
2194 :     Tag: leunga-20000510-moby-c--ssa
2195 :     Description:
2196 :    
2197 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2198 :     None of these affect SML/NJ.
2199 :    
2200 :     1. Register Allocation
2201 :    
2202 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2203 :     This module tries to remove local (i.e. basic block level) redundancies
2204 :     during spilling.
2205 :    
2206 :     b. A new framework for performing region based register allocation.
2207 :     Not yet entirely functional.
2208 :    
2209 :     2. X86
2210 :    
2211 :     a. DefUse for POP was missing the stack pointer [found by Lal]
2212 :     b. Reload for CALL was incorrect in X86Spill [found by John]
2213 :     c. Various fixes in X86Spill so that it can be used correctly for
2214 :     the new spilling module.
2215 :    
2216 :     3. SSA/IR
2217 :    
2218 :     a. New module ir/dj-dataflow.sml implements elimination based
2219 :     data flow analysis.
2220 :    
2221 :     4. MLRiscGen
2222 :    
2223 :     a. Fix for gc type annotation
2224 :    
2225 :     5. MDGen
2226 :    
2227 :     Various fixes for machine description -> ml code translation. For ssa
2228 :     only.
2229 :    
2230 :     ----------------------------------------------------------------------
2231 :     Name: Allen Leung
2232 : leunga 640 Date: 2000/05/08 22:17 EDT
2233 :     Tag: leunga-20000508-labexp
2234 :     Description:
2235 :    
2236 :     Fermin has found a few assembly problems with constant expressions
2237 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
2238 :     which choke on dumb assemblers. This is his fix.
2239 :    
2240 :     ----------------------------------------------------------------------
2241 : dbm 635 Name: Dave MacQueen
2242 :     Date: 2000/04/09 14:00 EDT
2243 :     Tag: dbm-20000502-Version_110_28
2244 :     Description:
2245 :    
2246 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2247 :    
2248 :     2. Updated config/version to 110.28
2249 :    
2250 :     3. Updated config/srcarchiveurl
2251 :    
2252 :     3. New boot files!
2253 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2254 :    
2255 :     ----------------------------------------------------------------------
2256 : blume 634 Name: Matthias Blume
2257 :     Date: 2000/05/01 19:05:00 JST
2258 :     Tag: blume-20000501-noweb
2259 :     Description:
2260 :    
2261 :     A new noweb tool has been added. The existing system is entirely
2262 :     unaffected by this, but some CM users have asked for renewed noweb
2263 :     support. Everything is documented in the CM manual.
2264 :    
2265 :     New (plugin) libraries:
2266 :    
2267 :     noweb-tool.cm
2268 :     nw-ext.cm
2269 :    
2270 :     ----------------------------------------------------------------------
2271 : dbm 633 Name: Dave MacQueen
2272 :     Date: 2000/04/30 12:40PM EDT
2273 :     Tag: dbm-20000430-bug_fixes
2274 :     Description:
2275 :    
2276 :     1. Fix for bug 1498
2277 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2278 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2279 :     added toRealArray function
2280 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
2281 :     added check for tag Obj.RealArray to array printing case in ppObj
2282 :    
2283 :     2. Fix for bug 1510
2284 :     smlnj/src/compiler/Semant/types/typesutil.sml
2285 :     fixed definition of dummyargs (used by equalTycon) so that
2286 :     dummy args are distinct types
2287 :    
2288 :     ----------------------------------------------------------------------
2289 : blume 629 Name: Matthias Blume
2290 : blume 632 Date: 2000/04/30 01:00:00 JST
2291 :     Tag: blume-20000430-versions
2292 :     Description:
2293 :    
2294 :     1. CM version numbering added. This is an implementation of Lal's
2295 :     proposal for adding version numbers and version checking to .cm
2296 :     files. Lal said that his proposal was just that -- a proposal.
2297 :     For the time being I went ahead and implemented it so that people
2298 :     can comment on it. Everything is completely backward-compatible
2299 :     (except for the stable library format, i.e., new bootfiles!).
2300 :    
2301 :     As usual, see the CM manual for details.
2302 :    
2303 :     2. An alternative syntax for anchored paths has been implemented.
2304 :     Dave has recently voiced the same concerns that I had when I did
2305 :     this, so there should be some support. My take is that eventually
2306 :     I will let support for the current syntax (where anchors are
2307 :     "implicit") fade out in favor of the new, explicit syntax.
2308 :     In order to be backward-compatible, both old and new syntax are
2309 :     currently supported.
2310 :    
2311 :     Again, see the CM manual for details.
2312 :    
2313 :     3. Parallel make is trying to be slightly smarter: When the master
2314 :     process finds a "bottleneck", i.e., when there is only one
2315 :     compilation unit that can be compiled and everybody else is
2316 :     waiting on it, then it will simply compile it directly instead
2317 :     of clumsily telling one of the slaves to do it.
2318 :    
2319 :     4. Support for "unsharing" added. This is necessary in order to be
2320 :     able to have two different versions of the same library running
2321 :     at the same time (e.g., for trying out a new MLRISC while still
2322 :     having the old MLRISC linked into the current compiler, etc.)
2323 :     See the CM manual.
2324 :    
2325 :     5. Simple "makedepend" functionality added for generating Makefile
2326 :     dependency information. (This is rather crude at the moment.
2327 :     Expect some changes here in the future.)
2328 :    
2329 :     6. ".fun" added as a recognized suffix for ML files. Also documented
2330 :     explicitly in the manual that the fallback behavior (unknown suffix
2331 :     -> ML file) is not an official feature!
2332 :    
2333 :     7. Small changes to the pickler for stable libraries.
2334 :    
2335 :     8. Several internal changes to CM (for cleanup/improvement).
2336 :    
2337 :    
2338 :     !!!! NEW BINFILES !!!!
2339 :    
2340 :     ----------------------------------------------------------------------
2341 :     Name: Matthias Blume
2342 : blume 631 Date: 2000/04/28 17:30:00 JST
2343 :     Tag: blume-20000428-pathconfig
2344 :     Description:
2345 :    
2346 :     1. I changed config/install.sh to remove duplicate entries from the
2347 :     lib/pathconfig file at the end. Moreover, the final version of
2348 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
2349 :     in src/system/installml.
2350 :    
2351 :     2. The config/install.sh script now consistently uses relative
2352 :     pathnames in lib/pathconfig whenever the anchor is in the lib
2353 :     directory. (So far this was true for the libraries that come
2354 :     pre-compiled and bundled as part of the bootfiles but not for
2355 :     libraries that are compiled by the script itself.)
2356 :    
2357 :     ----------------------------------------------------------------------
2358 :     Name: Matthias Blume
2359 : blume 629 Date: 2000/04/26 13:10:00 JST
2360 :     Tag: blume-20000426-fun_suffix
2361 :     Description:
2362 :    
2363 :     Added ".fun" as a recognized file name suffix (for ML code).
2364 :    
2365 :     ----------------------------------------------------------------------
2366 : leunga 628 Name: Allen Leung
2367 :     Date: 2000/04/25 17:00:00 EST
2368 :     Tag: leunga-20000425-alpha-ra
2369 :     Description:
2370 :    
2371 :     1. Alpha
2372 :    
2373 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
2374 :     in C--.
2375 :    
2376 :     2. RA
2377 :    
2378 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
2379 :     and the interference graph.
2380 :    
2381 :     ----------------------------------------------------------------------
2382 : dbm 627 Name: Dave MacQueen
2383 :     Date: 2000/04/25/
2384 :     Tag: dbm-20000425-mlyacc_doc_examples
2385 :     Description:
2386 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2387 :     Updated all three examples in src/ml-yacc/examples to run
2388 :     under 110.* using CM.make.
2389 :    
2390 :     ----------------------------------------------------------------------
2391 : leunga 624 Name: Allen Leung
2392 :     Date: 2000/04/20 23:04:00 EST
2393 :     Tag: leunga-20000420-ssa-c---stuff
2394 :     Description:
2395 :    
2396 :     This update synchronizes my repository with Yale's. Most of these
2397 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
2398 :    
2399 :     1. Register Allocator
2400 :    
2401 :     a. An improvement in the interference graph construction:
2402 :     Given a copy
2403 :    
2404 :     s <- t
2405 :    
2406 : leunga 628 no interference edge between s and t is added for this definition of s.
2407 : leunga 624
2408 :     b. I've added two new spill heuristic modules that Fermin and I developed
2409 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
2410 :     useful for others (Moby?)
2411 :    
2412 :     2. X86
2413 :    
2414 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
2415 :    
2416 :     3. Alpha
2417 :    
2418 :     a. Added the BSR instruction and code generation that goes with it [C--]
2419 :     b. Other fixes too numerous to recount provided by Fermin [C--]
2420 :    
2421 :     4. Regmaps
2422 :    
2423 :     a. The regmaps are not initialized with the identity physical bindings
2424 :     at creation time. This is unneeded.
2425 :    
2426 :     5. MLRISC Optimizations
2427 :    
2428 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
2429 :     intersects with liveness incrementally in linear time! Woohoo!
2430 :     This is now used in my new SSA construction algorithm.
2431 :    
2432 :     b. THe branch reorganization module is now smarter about linear chains of
2433 :     basic blocks.
2434 :    
2435 :    
2436 :     ----------------------------------------------------------------------
2437 : blume 615 Name: Matthias Blume
2438 :     Date: 2000/04/12 13:52:00 JST
2439 :     Tag: blume_main_v110p27_1
2440 :     Description:
2441 :    
2442 :     Changed install.sh script to handle archive files without version number
2443 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2444 :     name of the boot file archive.
2445 :    
2446 :     ----------------------------------------------------------------------
2447 : dbm 607 Name: Dave MacQueen
2448 :     Date: 2000/04/09 14:00 EDT
2449 :     Tag: dbm-20000410-Version_110_27
2450 :     Description:
2451 :    
2452 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2453 :    
2454 :     2. Updated src/config/version to 110.27
2455 :    
2456 :     3. New boot files!
2457 :    
2458 :     ----------------------------------------------------------------------
2459 : leunga 606 Name: Allen Leung
2460 :     Date: 2000/04/09 19:09:00 EST
2461 :     Tag: leunga-20000409-misc
2462 :     Description:
2463 :    
2464 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
2465 :    
2466 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
2467 :    
2468 :     ----------------------------------------------------------------------
2469 : monnier 605 Name: Stefan
2470 :     Date: 2000/04/07 10:00:00 EDT
2471 :     Tag: monnier-20000406-branch-handling
2472 :     Description:
2473 :    
2474 :     Improved handling of branches (mostly those generated from
2475 :     polymorphic equality), removed switchoff and changed the
2476 :     default optimization settings (more cpsopt and less flintopt).
2477 :    
2478 :     ----------------------------------------------------------------------
2479 : leunga 601 Name: Allen Leung
2480 : leunga 602 Date: 2000/04/06 01:30:00 EST
2481 :     Tag: leunga-20000406-peephole-x86-SSA-2
2482 :     Description:
2483 :    
2484 :     Forgot a few files.
2485 :    
2486 :     ----------------------------------------------------------------------
2487 :     Name: Allen Leung
2488 : leunga 601 Date: 2000/04/06 00:36:00 EST
2489 :     Tag: leunga-20000406-peephole-x86-SSA
2490 :     Description:
2491 :    
2492 :     1. New Peephole code
2493 :    
2494 :     2. Minor improvement to X86 instruction selection
2495 :    
2496 :     3. Various fixes to SSA and machine description -> code translator
2497 :    
2498 :     ----------------------------------------------------------------------
2499 : blume 595 Name: Matthias Blume
2500 :     Date: 2000/04/05 12:30:00 JST
2501 :     Tag: blume_main_v110p26p2_3
2502 :     Description:
2503 :    
2504 :     This update just merges three minor cosmetic updates to CM's sources
2505 :     to get ready for the 110.27 code freeze on Friday. No functionality
2506 :     has changed.
2507 :    
2508 :     ----------------------------------------------------------------------
2509 : leunga 593 Name: Allen Leung
2510 : leunga 594 Date: 2000/04/04 19:39:00 EST
2511 :     Tag: leunga-20000404-x86-asm
2512 :     Description:
2513 :    
2514 :     1. Fixed a problem in X86 assembly.
2515 :    
2516 :     Things like
2517 :    
2518 :     jmp %eax
2519 :     jmp (%eax)
2520 :    
2521 :     should be output as
2522 :    
2523 :     jmp *%eax
2524 :     jmp *(%eax)
2525 :    
2526 :     2. Assembly output
2527 :    
2528 :     Added a new flag
2529 :    
2530 :     "asm-indent-copies" (default to false)
2531 :    
2532 :     When this flag is on, parallel copies will be indented an extra level.
2533 :    
2534 :     ----------------------------------------------------------------------
2535 :     Name: Allen Leung
2536 : leunga 593 Date: 2000/04/04 03:18:00 EST
2537 :     Tag: leunga-20000404-C--Moby
2538 :     Description:
2539 :    
2540 :     All of these fixes are related to C--, Moby, and my own optimization
2541 :     stuff; so they shouldn't affect SML/NJ.
2542 :    
2543 :     1. X86
2544 :    
2545 :     Various fixes related floating point, and extensions.
2546 :    
2547 :     2. Alpha
2548 :    
2549 :     Some extra patterns related to loads with signed/zero extension
2550 :     provided by Fermin.
2551 :    
2552 :     3. Assembly
2553 :    
2554 : monnier 659 When generating assembly, resolve the value of client defined constants,
2555 : leunga 593 instead of generating symbolic values. This is controlled by the
2556 :     new flag "asm-resolve-constants", which is default to true.
2557 :    
2558 :     4. Machine Descriptions
2559 :    
2560 :     a. The precedence parser was slightly broken when parsing infixr symbols.
2561 :     b. The type generalizing code had the bound variables reversed, resulting
2562 :     in a problem during arity raising.
2563 :     c. Various fixes in machine descriptions.
2564 :    
2565 :     ----------------------------------------------------------------------
2566 : blume 592 Name: Matthias Blume
2567 :     Date: 2000/04/03 16:05:00 JST
2568 :     Tag: blume_main_v110p26p2_2
2569 :     Description:
2570 :    
2571 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
2572 :     now done via the ordinary static environment that is context to each
2573 :     compilation unit.
2574 :    
2575 :     To this end, I arranged that instead of "structure Core" as "structure
2576 :     _Core" is bound in the pervasive environment. Core access is done via
2577 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
2578 : blume 592 legal surface-syntax symbol).
2579 :    
2580 :     The current solution is much cleaner because the core environment is
2581 :     now simply part of the pervasive environment which is part of every
2582 :     compilation unit's context anyway. In particular, this eliminates all
2583 :     special-case handling that was necessary until now in order to deal
2584 :     with dynamic and symbolic parts of the core environment.
2585 :    
2586 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2587 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
2588 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
2589 :     more details.
2590 :    
2591 :     I also tried to track down all mentions of "Core" (as string argument
2592 :     to Symbol.strSymbol) in the compiler and replaced them with a
2593 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
2594 :     name appears in one place only.
2595 :    
2596 :     Binfile and bootfile format have not changed, but the switchover from
2597 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
2598 :     bootfiles anyway.
2599 :    
2600 :     ----------------------------------------------------------------------
2601 : leunga 591 Name: Allen Leung
2602 :     Date: 2000/04/02 21:17:00 EST
2603 :     Tag: leunga-20000402-mltree
2604 :     Description:
2605 : leunga 585
2606 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
2607 :     2. Added a bunch of files from my repository. These are currently
2608 :     used by other non-SMLNJ backends.
2609 :    
2610 : leunga 576 ----------------------------------------------------------------------
2611 : leunga 590 Name: Allen Leung
2612 :     Date: 2000/03/31 21:15:00 EST
2613 :     Tag: leunga-20000331-aliasing
2614 :     Description:
2615 :    
2616 :     This update contains a rewritten (and hopefully more correct) module
2617 :     for extracting aliasing information from CPS.
2618 :    
2619 :     To turn on this feature:
2620 :    
2621 :     Compiler.Control.CG.memDisambiguate := true
2622 :    
2623 :     To pretty print the region information with assembly
2624 :    
2625 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
2626 :    
2627 :     To control how many levels of aliasing information are printed, use:
2628 :    
2629 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
2630 :    
2631 :     The default of n is 3.
2632 :    
2633 :     ----------------------------------------------------------------------
2634 : dbm 589 Name: David MacQueen
2635 :     Date: 2000/03/31 11:15:00 EST
2636 :     Tag: dbm-20000331-runtime_fix
2637 :     Description:
2638 :    
2639 :     This update contains:
2640 :    
2641 :     1. runtime/c-lib/c-libraries.c
2642 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
2643 :    
2644 :     2. fix for bug 1556
2645 :     system/Basis/Implementation/NJ/internal-signals.sml
2646 :    
2647 :     ----------------------------------------------------------------------
2648 : blume 587 Name: Matthias Blume
2649 : blume 588 Date: 2000/03/31 18:00:00 JST
2650 :     Tag: blume_main_v110p26p2_1
2651 :     Description:
2652 :    
2653 :     This update contains:
2654 :    
2655 :     1. A small change to CM's handling of stable libraries:
2656 :     CM now maintains one "global" modmap that is used for all stable
2657 :     libraries. The use of such a global modmap maximizes sharing and
2658 :     minimizes the need for re-traversing parts of environments during
2659 :     modmap construction. (However, this has minor impact since modmap
2660 :     construction seems to account for just one percent or less of total
2661 :     compile time.)
2662 :    
2663 :     2. I added a "genmap" phase to the statistics. This is where I got the
2664 :     "one percent" number (see above).
2665 :    
2666 :     3. CM's new tool parameter mechanism just became _even_ better. :)
2667 :     - The parser understands named parameters and recursive options.
2668 :     - The "make" and "shell" tools use these new features.
2669 :     (This makes it a lot easier to cascade these tools.)
2670 :     - There is a small syntax change: named parameters use a
2671 :    
2672 :     <name> : ( <option> ... ) or
2673 :     <name> : <string>
2674 :    
2675 :     syntax. Previously, named parameters were implemented in an
2676 :     ad-hoc fashion by each tool individually (by parsing strings)
2677 :     and had the form
2678 :    
2679 :     <name>=<string>
2680 :    
2681 :     See the CM manual for a full description of these issues.
2682 :    
2683 :     ----------------------------------------------------------------------
2684 :     Name: Matthias Blume
2685 : blume 587 Date: 2000/03/30 18:00:00 JST
2686 :     Tag: blume_main_v110p26p2_0
2687 :     Description:
2688 :    
2689 :     !!!!! WARNING !!!!!!
2690 :     !! New binfiles !!
2691 :     !!!!!!!!!!!!!!!!!!!!
2692 :    
2693 :     This update contains:
2694 :    
2695 :     1. Moderate changes to CM:
2696 :    
2697 :     - Changes to CM's tools mechanism. In particular, it is now possible
2698 :     to have tools that accept additional "command line" parameters
2699 :     (specified in the .cm file at each instance where the tool's class is
2700 :     used).
2701 :    
2702 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
2703 :     facilitate fairly seamless hookup to portions of code managed using
2704 : blume 587 Makefiles or Shell scripts.
2705 :    
2706 :     There are no classes "shared" or "private" anymore. Instead, the
2707 :     sharing annotation is now a parameter to the "sml" class.
2708 :    
2709 :     There is a bit of generic machinery for implementing one's own
2710 :     tools that accept command-line parameters. However, I am not yet fully
2711 :     satisfied with that part, so expect changes here in the future.
2712 :    
2713 :     All existing tools are described in the CM manual.
2714 :    
2715 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
2716 : blume 587 error messages that tended to be more annoying than helpful.)
2717 :    
2718 :     2. Major changes to the compiler's static environment data structures.
2719 :    
2720 :     - no CMStaticEnv anymore.
2721 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
2722 :     but it is called Environment), no conversions between different
2723 :     kinds of static environments
2724 :    
2725 :     - There is still a notion of a "modmap", but such modmaps are generated
2726 :     on demand at the time when they are needed. This sounds slow, but I
2727 :     sped up the code that generates modmaps enough for this not to lead to
2728 :     a slowdown of the compiler (at least I didn't detect any).
2729 :    
2730 :     - To facilitate rapid modmap generation, static environments now
2731 :     contain an (optional) "modtree" structure. Modtree annotations are
2732 :     constructed by the unpickler during unpickling. (This means that
2733 :     the elaborator does not have to worry about modtrees at all.)
2734 :     Modtrees have the advantage that they are compositional in the same
2735 :     way as the environment data structure itself is compositional.
2736 :     As a result, modtrees never hang on to parts of an environment that
2737 :     has already been rendered "stale" by filtering or rebinding.
2738 :    
2739 :     - I went through many, many trials and errors before arriving at the
2740 :     current solution. (The initial idea of "linkpaths" did not work.)
2741 :     But the result of all this is that I have touched a lot of files that
2742 :     depend on the "modules" and "types" data structures (most of the
2743 :     elaborator). There were a lot of changes during my "linkpath" trials
2744 :     that could have been reverted to their original state but weren't.
2745 :     Please, don't be too harsh on me for messing with this code a bit more
2746 :     than what was strictly necessary... (I _did_ resist the tempation
2747 :     of doing any "global reformatting" to avoid an untimely death at
2748 :     Dave's hands. :)
2749 :    
2750 :     - One positive aspect of the previous point: At least I made sure that
2751 :     all files that I touched now compile without warnings (other than
2752 :     "polyEqual").
2753 :    
2754 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
2755 :     redundant modmaps)
2756 :    
2757 :     ----------------------------------------------------------------------
2758 : leunga 580 Name: Allen Leung
2759 : leunga 585 Date: 2000/03/29 18:00:00
2760 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
2761 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
2762 : leunga 585 Description:
2763 :    
2764 :     This update contains *MAJOR* changes to the way code is generated from CPS
2765 :     in the module mlriscGen, and in various backend modules.
2766 :    
2767 :     CHANGES
2768 :     =======
2769 :    
2770 :     1. MLRiscGen: forward propagation fix.
2771 :    
2772 :     There was a bug in forward propagation introduced at about the same time
2773 :     as the MLRISC x86 backend, which prohibits coalescing to be
2774 :     performed effectively in loops.
2775 :    
2776 :     Effect: speed up of loops in RISC architectures.
2777 :     By itself, this actually slowed down certain benchmarks on the x86.
2778 :    
2779 :     2. MLRiscGen: forward propagating addresses from consing.
2780 :    
2781 :     I've changed the way consing code is generated. Basically I separated
2782 :     out the initialization part:
2783 :    
2784 :     store tag, offset(allocptr)
2785 :     store elem1, offset+4(allocptr)
2786 :     store elem2, offset+8(allocptr)
2787 :     ...
2788 :     store elemn, offset+4n(allocptr)
2789 :    
2790 :     and the address computation part:
2791 :    
2792 :     celladdr <- offset+4+alloctpr
2793 :    
2794 :     and move the address computation part
2795 :    
2796 :     Effect: register pressure is generally lower as a result. This
2797 :     makes compilation of certain expressions much faster, such as
2798 :     long lists with non-trivial elements.
2799 :    
2800 :     [(0,0), (0,0), .... (0,0)]
2801 :    
2802 :     3. MLRiscGen: base pointer elimination.
2803 :    
2804 :     As part of the linkage mechanism, we generate the sequence:
2805 :    
2806 :     L: ... <- start of the code fragment
2807 :    
2808 :     L1:
2809 :     base pointer <- linkreg - L1 + L
2810 :    
2811 :     The base pointer was then used for computing relocatable addresses
2812 :     in the code fragment. Frequently (such as in lots of continuations)
2813 :     this is not needed. We now eliminate this sequence whenever possible.
2814 :    
2815 :     For compile time efficiency, I'm using a very stupid local heuristic.
2816 :     But in general, this should be done as a control flow analysis.
2817 :    
2818 :     Effect: Smaller code size. Speed up of most programs.
2819 :    
2820 :     4. Hppa back end
2821 :    
2822 :     Long jumps in span dependence resolution used to depend on the existence
2823 :     of the base pointer.
2824 :    
2825 :     A jump to a long label L was expanded into the following sequence:
2826 :    
2827 :     LDIL %hi(L-8192), %r29
2828 :     LDO %lo(L-8192)(%r29), %r29
2829 :     ADD %r29, baseptr, %r29
2830 :     BV,n %r0(%r29)
2831 :    
2832 :     In the presence of change (3) above, this will not work. I've changed
2833 :     it so that the following sequence of instructions are generated, which
2834 :     doesn't mention the base pointer at all:
2835 :    
2836 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
2837 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
2838 :     BV,n %r0(%r29) /* Jump */
2839 :    
2840 :     5. Alpha back end
2841 :    
2842 :     New alpha instructions LDB/LDW have been added, as per Fermin's
2843 :     suggestions. This is unrelated to all other changes.
2844 :    
2845 :     6. X86 back end
2846 :    
2847 :     I've changed andl to testl in the floating point test sequence
2848 :     whenever appropriate. The Intel optimization guide states that
2849 : monnier 659 testl is preferable to andl.
2850 : leunga 585
2851 :     7. RA (x86 only)
2852 :    
2853 :     I've improved the spill propagation algorithm, using an approximation
2854 :     of maximal weighted independent sets. This seems to be necessary to
2855 :     alleviate the negative effect in light of the slow down in (1).
2856 :    
2857 :     I'll write down the algorithm one of these days.
2858 :    
2859 :     8. MLRiscGen: frequencies
2860 :    
2861 :     I've added an annotation that states that all call gc blocks have zero
2862 :     execution frequencies. This improves register allocation on the x86.
2863 :    
2864 :     BENCHMARKS
2865 :     ==========
2866 :    
2867 :     I've only perform the comparison on 110.25.
2868 :    
2869 :     The platforms are:
2870 :    
2871 :     HPPA A four processor HP machine (E9000) with 5G of memory.
2872 :     X86 A 300Hhz Pentium II with 128M of memory, and
2873 :     SPARC An Ultra sparc 2 with 512M of memory.
2874 :    
2875 :     I used the following parameters for the SML benchmarks:
2876 :    
2877 :     @SMLalloc
2878 :     HPPA 256k
2879 :     SPARC 512k
2880 :     X86 256k
2881 :    
2882 :     COMPILATION TIME
2883 :     ----------------
2884 :     Here are the numbers comparing the compilation times of the compilers.
2885 :     I've only compared 110.25 compiling the new sources versus
2886 :     a fixpoint version of the new compiler compiling the same.
2887 :    
2888 :     110.25 New
2889 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
2890 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
2891 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
2892 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
2893 :    
2894 :     110.25 New
2895 :     Code Size Code Size
2896 :     HPPA 8596736 8561421
2897 :     SPARC 8974299 8785143
2898 :     X86 9029180 8716783
2899 :    
2900 :     So in summary, things are at least as good as before. Dramatic
2901 :     reduction in compilation is obtained on the Sparc; I can't explain it,
2902 :     but it is reproducible. Perhaps someone should try to reproduce this
2903 :     on their own machines.
2904 :    
2905 :     SML BENCHMARKS
2906 :     --------------
2907 :    
2908 :     On the average, all benchmarks perform at least as well as before.
2909 :    
2910 :     HPPA Compilation Time Spill+Reload Run Time
2911 :     110.25 New 110.25 New 110.25 New
2912 :    
2913 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
2914 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
2915 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
2916 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
2917 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
2918 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
2919 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
2920 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
2921 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
2922 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
2923 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
2924 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
2925 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
2926 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
2927 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
2928 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
2929 :     --------------------------------------------------------------------------
2930 :     Average 6.12% 4.09%
2931 :    
2932 :     SPARC Compilation Time Spill+Reload Run Time
2933 :     110.25 New 110.25 New 110.25 New
2934 :    
2935 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
2936 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
2937 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
2938 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
2939 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
2940 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
2941 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
2942 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
2943 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
2944 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
2945 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
2946 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
2947 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
2948 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
2949 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
2950 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
2951 :     ----------------------------------------------------------------------------
2952 :     Average 6.94% 2.64%
2953 :    
2954 :     X86 Compilation Time Spill+Reload Run Time
2955 :     110.25 New 110.25 New 110.25 New
2956 :    
2957 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
2958 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
2959 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
2960 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
2961 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
2962 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
2963 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
2964 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
2965 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
2966 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
2967 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
2968 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
2969 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
2970 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
2971 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
2972 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
2973 :     ----------------------------------------------------------------------------
2974 :     Average 1.22% 3.36%
2975 :    
2976 :     ----------------------------------------------------------------------
2977 :     Name: Allen Leung
2978 : leunga 583 Date: 2000/03/23 16:25:00
2979 :     Tag: leunga-20000323-fix_x86_alpha
2980 :     Description:
2981 :    
2982 :     1. X86 fixes/changes
2983 :    
2984 :     a. The old code generated for SETcc was completely wrong.
2985 :     The Intel optimization guide is VERY misleading.
2986 :    
2987 :     2. ALPHA fixes/changes
2988 :    
2989 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2990 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2991 :     c. Added reassociation code for address computation.
2992 :    
2993 :     ----------------------------------------------------------------------
2994 :     Name: Allen Leung
2995 : leunga 580 Date: 2000/03/22 01:23:00
2996 :     Tag: leunga-20000322-fix_x86_hppa_ra
2997 :     Description:
2998 :    
2999 :     1. X86 fixes/changes
3000 :    
3001 :     a. x86Rewrite bug with MUL3 (found by Lal)
3002 :     b. Added the instructions FSTS, FSTL
3003 :    
3004 :     2. PA-RISC fixes/changes
3005 :    
3006 :     a. B label should not be a delay slot candidate! Why did this work?
3007 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
3008 :     as it should be.
3009 :     c. The assembly syntax for fstds and fstdd was wrong.
3010 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
3011 :     operand variant of COMCLR/LDO.
3012 :    
3013 :     3. Generic MLRISC
3014 :    
3015 :     a. shuffle.sml rewritten to be slightly more efficient
3016 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
3017 :    
3018 :     4. Register Allocator
3019 :    
3020 :     a. I now release the interference graph earlier during spilling.
3021 :     May improve memory usage.
3022 :    
3023 :     ----------------------------------------------------------------------
3024 : blume 577 Name: Matthias Blume
3025 : blume 578 Date: 2000/03/14 14:15:32
3026 :     Tag: blume_main_v110p26p1_2
3027 :     Description:
3028 :    
3029 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
3030 :     additional argument called "template" which is an optional string that
3031 : monnier 659 specifies the layout of the tool command line. See the CM manual for
3032 : blume 578 explanation.
3033 :    
3034 : monnier 659 2. A special-purpose tool can be "registered" by simply dropping the
3035 : blume 578 corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
3036 :     directory where the .cm file lives that uses this tool. (The
3037 :     behavior/misfeature until now was to look for the tool description
3038 :     files in the current working directory.) As before, tool description
3039 :     files could also be anchored -- in which case they can live anywhere
3040 :     they like. Following the recent e-mail discussion, this change should
3041 :     make it easier to have special-purpose tools that are shipped together
3042 :     with the sources of the program that uses them.
3043 :    
3044 :     ----------------------------------------------------------------------
3045 :     Name: Matthias Blume
3046 : blume 577 Date: 2000/03/10 07:48:34
3047 :     Tag: blume_main_v110p26p1_1
3048 :     Description:
3049 :    
3050 :     I added a re-written version of Dave's fixpt script to src/system.
3051 :     Changes relative to the original version:
3052 :     - sh-ified (not everybody has ksh)
3053 :     - automatically figures out which architecture it runs on
3054 :     - uses ./makeml a bit more cleverly
3055 :     - never invokes ./installml (and, thus, does not clobber your
3056 :     good and working installation of sml in case something goes wrong)
3057 :     - accepts max iteration count using option "-iter <n>"
3058 :     - accepts a "base" name using option "-base <base>"
3059 :    
3060 :     It does not build any extraneous heap images but directly rebuilds
3061 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
3062 :     it can incorporate existing bin- and boot- hierarchies. For example,
3063 :     suppose the base is set to "sml" (which is the default). Then it
3064 :     successively builds
3065 :    
3066 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
3067 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
3068 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
3069 :     ...
3070 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
3071 :    
3072 :     and so on. If any of these already exist, it will just use what's
3073 :     there. In particular, many people will have the initial set of bin
3074 :     and boot files around, so this saves time for at least one full
3075 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
3076 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
3077 :     they might be the result of an earlier partial run of fixpt (which
3078 : monnier 659 perhaps got accidentally killed). In this case, fixpt will quickly
3079 : blume 577 move through what exists before continuing where it left off earlier,
3080 :     and, thus, saves a lot of time.
3081 :    
3082 :     ----------------------------------------------------------------------
3083 : leunga 576 Name: Allen Leung
3084 :     Date: 00/03/10 02:20:00
3085 :     Tag: leunga-20000310-fix_x86_asm_ra
3086 :     Description:
3087 : dbm 570
3088 : leunga 576 More assembly output problems involving the indexed addressing mode
3089 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
3090 :     fix.
3091 :    
3092 :     The interface and implementation of the register allocator have been changed
3093 :     slightly to accommodate the possibility to skip the register allocation
3094 :     phases completely and go directly to memory allocation. This is needed
3095 :     for C-- use.
3096 :    
3097 : dbm 570 ----------------------------------------------------------------------
3098 : blume 572 Name: Matthias Blume
3099 : blume 575 Date: 00/03/09 10:23:53
3100 :     Tag: blume_main_v110p26p1_0
3101 :     Description:
3102 :    
3103 :     * Complete re-organization of library names. Many libraries have been
3104 :     consolidated so that they share the same path anchor. For example,
3105 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
3106 :     are SML/NJ-specific are under "smlnj". Notice that names like
3107 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
3108 :     for a complete description of the new naming scheme. Quick reference:
3109 :    
3110 :     host-cmb.cm -> smlnj/cmb.cm
3111 :     host-compiler.cm -> smlnj/compiler.cm
3112 :     full-cm.cm -> smlnj/cm.cm
3113 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
3114 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
3115 :    
3116 :     * Bug fixes in CM.
3117 :     - exceptions in user code are being passed through (i.e., reach top level)
3118 :     - more bugs in paranoia mode fixed
3119 :     - bug related to checking group owners fixed
3120 :    
3121 :     * New install.sh script that automagically fetches archive files:
3122 :     The new file config/srcarchiveurl must contain the URL of the
3123 :     (remote) directory that contains bin files (or other source archives).
3124 :     If install.sh does not find the archive locally, it tries to get
3125 :     it from that remote directory.
3126 :     This should simplify installation further: For machines that have
3127 :     access to the internet, just fetch <version>-config.tgz, unpack it,
3128 : monnier 659 edit config/targets, and go (run config/install.sh). The script will
3129 : blume 575 fetch everything else that it might need all by itself.
3130 :    
3131 :     For CVS users, this mechanism is not relevant for source archives, but
3132 :     it is convenient for getting new sets of binfiles.
3133 :    
3134 :     Archives should be tar files compressed with either gzip, compress, or
3135 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
3136 :    
3137 :     ----------------------------------------------------------------------
3138 :     Name: Matthias Blume
3139 : blume 572 Date: 2000/03/07 04:01:04
3140 :     Tag: blume_main_v110_26_2
3141 : dbm 570 Description:
3142 : blume 572 - size info in BOOTLIST
3143 :     * no fixed upper limits for number of bootfiles or length of
3144 :     bootfile names in runtime
3145 :     * falling back to old behavior if no BOOTLIST size info found
3146 :     - allocation size heuristics in .run-sml
3147 :     * tries to read cache size from /proc/cpuinfo (this is important for
3148 :     small-cache Celeron systems!)
3149 :     - install.sh robustified
3150 :     - CM manual updates
3151 :     - paranoid mode
3152 :     * no more CMB.deliver() (i.e., all done by CMB.make())
3153 :     * can re-use existing sml.boot.* files
3154 :     * init.cmi now treated as library
3155 :     * library stamps for consistency checks
3156 :     - sml.boot.<arch>-<os>/PIDMAP file
3157 :     * This file is read by the CM startup code. This is used to minimize
3158 :     the amount of dynamic state that needs to be stowed away for the
3159 :     purpose of sharing between interactive system and user code.
3160 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
3161 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
3162 :     like most other CM state variables.
3163 :     - Compiler.CMSA eliminated
3164 :     * No longer supported by CM anyway.
3165 :     - fixed bugs in pickler that kept biting Stefan
3166 :     * past refs to past refs (was caused by the possibility that
3167 :     ad-hoc sharing is more discriminating than hash-cons sharing)
3168 :     * integer overflow on LargeInt.minInt
3169 :     - ml-{lex,yacc} build scripts now use new mechanism
3170 :     for building standalone programs
3171 :     - fixed several gcc -Wall warnings that were caused by missing header
3172 :     files, missing initializations, etc., in runtime (not all warnings
3173 :     eliminated, though)

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