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 1168 - (view) (download)

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 : george 1003 Date: yyyy/mm/dd
12 : dbm 570 Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 1142
15 : george 1136 ----------------------------------------------------------------------
16 : george 1168 Name: Lal George
17 :     Date: 2002/03/22 14:18:25 EST
18 :     Tag: blume-20020321-cps-branch-prob
19 :     Description:
20 :    
21 :     Implemented the Ball-Larus branch prediction-heuristic, and
22 :     incorporated graphical viewers for control flow graphs.
23 :    
24 :     Ball-Larus Heuristic:
25 :     ---------------------
26 :     See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
27 :    
28 :     By design it uses the Dempster-Shafer theory for combining
29 :     probabilities. For example, in the function:
30 :    
31 :     fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
32 :    
33 :     the ball-larus heuristics predicts that the n=0 is unlikely
34 :     (OH-heuristic), and the 'then' branch is unlikely because of the
35 :     RH-heuristic -- giving the 'then' branch an even lower combined
36 :     probability using the Dempster-Shater theory.
37 :    
38 :     Finally, John Reppy's loop analysis in MLRISC, further lowers the
39 :     probability of the 'then' branch because of the loop in the else
40 :     branch.
41 :    
42 :    
43 :     Graphical Viewing:
44 :     ------------------
45 :     I merely plugged in Allen's graphical viewers into the compiler. The
46 :     additional code is not much. At the top level, saying:
47 :    
48 :     Control.MLRISC.getFlag "cfg-graphical-view" := true;
49 :    
50 :     will display the graphical view of the control flow graph just before
51 :     back-patching. daVinci must be in your path for this to work. If
52 :     daVinci is not available, then the default viewer can be changed
53 :     using:
54 :    
55 :     Control.MLRISC.getString "viewer"
56 :    
57 :     which can be set to "dot" or "vcg" for the corresponding viewers. Of
58 :     course, these viewers must be in your path.
59 :    
60 :     The above will display the compilation unit at the level of clusters,
61 :     many of which are small, boring, and un-interesting. Also setting:
62 :    
63 :     Control.MLRISC.getInt "cfg-graphical-view_size"
64 :    
65 :     will display clusters that are larger than the value set by the above.
66 :    
67 :    
68 :     ----------------------------------------------------------------------
69 : blume 1159 Name: Matthias Blume
70 :     Date: 2002/03/21 22:20:00 EST
71 :     Tag: blume-20020321-kmp-bugfix
72 :     Description:
73 :    
74 :     Changed the interface to the KMP routine in PreString and fixed
75 :     a minor bug in one place where it was used.
76 :    
77 :     ----------------------------------------------------------------------
78 : leunga 1156 Name: Allen Leung
79 : leunga 1158 Date: 2002/03/21 20:30:00 EST
80 :     Tag: leunga-20020321-cfg
81 :     Description:
82 :    
83 :     Fixed a potential problem in cfg edge splitting.
84 :    
85 :     ----------------------------------------------------------------------
86 :     Name: Allen Leung
87 : leunga 1156 Date: 2002/03/21 17:15:00 EST
88 :     Tag: leunga-20020321-x86-fp-cfg
89 :     Description:
90 :    
91 :     1. Recoded the buggy parts of x86-fp.
92 :    
93 :     a. All the block reordering code has been removed.
94 :     We now depend on the block placement phases to do this work.
95 :    
96 :     b. Critical edge splitting code has been simplified and moved into the
97 :     CFG modules, as where they belong.
98 :    
99 :     Both of these were quite buggy and complex. The code is now much, much
100 :     simpler.
101 :    
102 :     2. X86 backend.
103 :    
104 :     a. Added instructions for 64-bit support. Instruction selection for
105 :     64-bit has not been committed, however, since that
106 :     requires changes to MLTREE which haven't been approved by
107 :     Lal and John.
108 :    
109 :     b. Added support for FUCOMI and FUCOMIP when generating code for
110 :     PentiumPro and above. We only generate these instructions in
111 :     the fast-fp mode.
112 :    
113 :     c. Added cases for JP and JNP in X86FreqProps.
114 :    
115 :     3. CFG
116 :    
117 :     CFG now has a bunch of methods for edge splitting and merging.
118 :    
119 :     4. Machine description.
120 :    
121 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
122 :     description things:
123 :    
124 :     rtl-build.{sig,sml} and hppa.mdl fixed.
125 :    
126 :     NOTE: the machine description stuff in the repository is still broken.
127 :     Again, I can't put my fixes in because that involves
128 :     changes to MLTREE.
129 :    
130 :     ----------------------------------------------------------------------
131 : blume 1145 Name: Matthias Blume
132 : blume 1155 Date: 2002/03/20 15:55:00 EST
133 :     Tag: blume-20020320-kmp
134 :     Description:
135 :    
136 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
137 :     it for String.isSubstring, Substring.isSubstring, and
138 :     Substring.position.
139 :    
140 :     (Might need some stress-testing. Simple examples worked fine.)
141 :    
142 :     ----------------------------------------------------------------------
143 :     Name: Matthias Blume
144 : blume 1152 Date: 2002/03/19 16:37:00 EST
145 :     Tag: blume-20020319-witnesses
146 :     Description:
147 :    
148 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
149 :    
150 :     This implements a generic mechanism for changing constness qualifiers
151 :     anywhere within big C types without resorting to outright "casts".
152 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
153 :     modify the constness at the outermost level.)
154 :     The implementation of "convert" is based on the idea of "witness"
155 :     values -- values that are not used by the operation but whose types
156 :     "testify" to their applicability. On the implementation side, "convert"
157 :     is simply a projection (returning its second curried argument). With
158 :     cross-module inlining, it should not result in any machine code being
159 :     generated.
160 :    
161 :     ----------------------------------------------------------------------
162 :     Name: Matthias Blume
163 : blume 1148 Date: 2002/03/15 16:40:00 EST
164 :     Tag: blume-20020315-basis
165 :     Description:
166 :    
167 :     Provided (preliminary?) implementations for
168 :    
169 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
170 :    
171 :     and
172 :    
173 :     Substring.full
174 :    
175 :     Those are in the Basis spec but they were missing in SML/NJ.
176 :    
177 :     ----------------------------------------------------------------------
178 :     Name: Matthias Blume
179 : blume 1145 Date: 2002/03/14 21:30:00 EST
180 :     Tag: blume-20020314-controls
181 :     Description:
182 :    
183 :     Controls:
184 :     ---------
185 :    
186 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
187 :     it into its own library $/controls-lib.cm. The source tree for
188 :     this is under src/smlnj-lib/Controls.
189 :    
190 :     2. Changed the names of types and functions in this interface, so they
191 :     make a bit more "sense":
192 :    
193 :     module -> registry
194 :     'a registry -> 'a group
195 :    
196 :     3. The interface now deals in ref cells only. The getter/setter interface
197 :     is (mostly) gone.
198 :    
199 :     4. Added a function that lets one register an already-existing ref cell.
200 :    
201 :     5. Made the corresponding modifications to the rest of the code so that
202 :     everything compiles again.
203 :    
204 :     6. Changed the implementation of Controls.MLRISC back to something closer
205 :     to the original. In particular, this module (and therefore MLRISC)
206 :     does not depend on Controls. There now is some link-time code in
207 :     int-sys.sml that registers the MLRISC controls with the Controls
208 :     module.
209 :    
210 :     CM:
211 :     ---
212 :    
213 :     * One can now specify the lambda-split aggressiveness in init.cmi.
214 :    
215 :     ----------------------------------------------------------------------
216 : leunga 1142 Name: Allen Leung
217 :     Date: 2002/03/13 17:30:00 EST
218 :     Tag: leunga-20020313-x86-fp-unary
219 :     Description:
220 :    
221 :     Bug fix for:
222 :    
223 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
224 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
225 :     > - fun f(x,(y,z)) = Real.~ y;
226 :     > [autoloading]
227 :     > [autoloading done]
228 :     > fchsl (%eax), 184(%esp)
229 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
230 :     >
231 :     > uncaught exception Error
232 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
233 :    
234 :     The problem was that the code generator did not generate any fp registers
235 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
236 :     translate the pseudo fp instruction. This only happened with unary fp
237 :     operators in certain situations.
238 :    
239 :     ----------------------------------------------------------------------
240 : blume 1137 Name: Matthias Blume
241 : blume 1140 Date: 2002/03/13 14:00:00 EST
242 :     Tag: blume-20020313-overload-etc
243 :     Description:
244 :    
245 :     1. Added _overload as a synonym for overload for backward compatibility.
246 :     (Control.overloadKW must be true for either version to be accepted.)
247 :    
248 :     2. Fixed bug in install script that caused more things to be installed
249 :     than what was requested in config/targets.
250 :    
251 :     3. Made CM aware of the (_)overload construct so that autoloading
252 :     works.
253 :    
254 :     ----------------------------------------------------------------------
255 :     Name: Matthias Blume
256 : blume 1138 Date: 2002/03/12 22:03:00 EST
257 :     Tag: blume-20020312-url
258 :     Description:
259 :    
260 :     Forgot to update BOOT and srcarchiveurl.
261 :    
262 :     ----------------------------------------------------------------------
263 :     Name: Matthias Blume
264 : blume 1137 Date: 2002/03/12 17:30:00 EST
265 :     Tag: blume-20020312-version110392
266 :     Description:
267 :    
268 :     Yet another version number bump (because of small changes to the
269 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
270 :    
271 :     Changes:
272 :    
273 :     The new pid generation scheme described a few weeks ago was overly
274 :     complicated. I implemented a new mechanism that is simpler and
275 :     provides a bit more "stability": Once CM has seen a compilation
276 :     unit, it keeps its identity constant (as long as you do not delete
277 :     those crucial CM/GUID/* files). This means that when you change
278 : blume 1145 an interface, compile, then go back to the old interface, and
279 : blume 1137 compile again, you arrive at the original pid.
280 :    
281 :     There now also is a mechanism that instructs CM to use the plain
282 :     environment hash as a module's pid (effectively making its GUID
283 :     the empty string). For this, "noguid" must be specified as an
284 :     option to the .sml file in question within its .cm file.
285 :     This is most useful for code that is being generated by tools such
286 :     as ml-nlffigen (because during development programmers tend to
287 :     erase the tool's entire output directory tree including CM's cached
288 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
289 :     revert to the old, broken behavior of SML/NJ, but in specific cases
290 :     where there is no danger of interface confusion, its use is ok
291 :     (I think).
292 :    
293 :     ml-nlffigen by default generates "noguid" annotations. They can be
294 :     turned off by specifying -guid in its command line.
295 :    
296 :     ----------------------------------------------------------------------
297 : george 1136 Name: Lal George
298 :     Date: 2002/03/12 12 14:42:36 EST
299 :     Tag: george-20020312-frequency-computation
300 :     Description:
301 :    
302 :     Integrated jump chaining and static block frequency into the
303 :     compiler. More details and numbers later.
304 : leunga 1025
305 : george 1003 ----------------------------------------------------------------------
306 : george 1133 Name: Lal George
307 :     Date: 2002/03/11 11 22:38:53 EST
308 :     Tag: george-20020311-jump-chain-elim
309 :     Description:
310 :    
311 :     Tested the jump chain elimination on all architectures (except the
312 :     hppa). This is on by default right now and is profitable for the
313 :     alpha and x86, however, it may not be profitable for the sparc and ppc
314 :     when compiling the compiler.
315 :    
316 :     The gc test will typically jump to a label at the end of the cluster,
317 :     where there is another jump to an external cluster containing the actual
318 :     code to invoke gc. This is to allow factoring of common gc invocation
319 :     sequences. That is to say, we generate:
320 :    
321 :     f:
322 :     testgc
323 :     ja L1 % jump if above to L1
324 :    
325 :     L1:
326 :     jmp L2
327 :    
328 :    
329 :     After jump chain elimination the 'ja L1' instructions is converted to
330 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
331 :     up being implemented in their long form (if L2 is far away) using:
332 :    
333 :     jbe L3 % jump if below or equal to L3
334 :     jmp L2
335 :     L3:
336 :     ...
337 :    
338 :    
339 :     For large compilation units L2 may be far away.
340 :    
341 :    
342 :     ----------------------------------------------------------------------
343 : blume 1132 Name: Matthias Blume
344 :     Date: 2002/03/11 13:30:00 EST
345 :     Tag: blume-20020311-mltreeeval
346 :     Description:
347 :    
348 :     A functor parameter was missing.
349 :    
350 :     ----------------------------------------------------------------------
351 : leunga 1130 Name: Allen Leung
352 : leunga 1131 Date: 2002/03/11 10:30:00 EST
353 : leunga 1142 Tag: leunga-20020311-runtime-string0
354 : leunga 1131 Description:
355 :    
356 :     The representation of the empty string now points to a
357 :     legal null terminated C string instead of unit. It is now possible
358 :     to convert an ML string into C string with InlineT.CharVector.getData.
359 :     This compiles into one single machine instruction.
360 :    
361 :     ----------------------------------------------------------------------
362 :     Name: Allen Leung
363 : leunga 1130 Date: 2002/03/10 23:55:00 EST
364 :     Tag: leunga-20020310-x86-call
365 :     Description:
366 :    
367 :     Added machine generation for CALL instruction (relative displacement mode)
368 :    
369 :     ----------------------------------------------------------------------
370 : blume 1128 Name: Matthias Blume
371 :     Date: 2002/03/08 16:05:00
372 :     Tag: blume-20020308-entrypoints
373 :     Description:
374 :    
375 :     Version number bumped to 110.39.1. NEW BOOTFILES!
376 :    
377 :     Entrypoints: non-zero offset into a code object where execution should begin.
378 :    
379 :     - Added the notion of an entrypoint to CodeObj.
380 :     - Added reading/writing of entrypoint info to Binfile.
381 :     - Made runtime system bootloader aware of entrypoints.
382 :     - Use the address of the label of the first function given to mlriscGen
383 :     as the entrypoint. This address is currently always 0, but it will
384 :     not be 0 once we turn on block placement.
385 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
386 :     with entry points) from mlriscGen.
387 :    
388 :     ----------------------------------------------------------------------
389 : leunga 1127 Name: Allen Leung
390 :     Date: 2002/03/07 20:45:00 EST
391 :     Tag: leunga-20020307-x86-cmov
392 :     Description:
393 :    
394 :     Bug fixes for CMOVcc on x86.
395 :    
396 :     1. Added machine code generation for CMOVcc
397 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
398 :     3. CMOVcc cannot have an immediate operand as argument.
399 :    
400 :     ----------------------------------------------------------------------
401 : blume 1126 Name: Matthias Blume
402 :     Date: 2002/03/07 16:15:00 EST
403 :     Tag: blume-20020307-controls
404 :     Description:
405 :    
406 :     This is a very large but mostly boring patch which makes (almost)
407 :     every tuneable compiler knob (i.e., pretty much everything under
408 :     Control.* plus a few other things) configurable via both the command
409 :     line and environment variables in the style CM did its configuration
410 :     until now.
411 :    
412 :     Try starting sml with '-h' (or, if you are brave, '-H')
413 :    
414 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
415 :     implements the underlying generic mechanism.
416 :    
417 :     The interface to some of the existing such facilities has changed somewhat.
418 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
419 :     (The getFoo interface is still there for backward-compatibility, but its
420 :     use is deprecated.)
421 :    
422 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
423 :     that one can now twiddle the compiler settings when using this "batch"
424 :     compiler.
425 :    
426 :     TODO items:
427 :    
428 :     We should go through and throw out all controls that are no longer
429 :     connected to anything. Moreover, we should go through and provide
430 :     meaningful (and correct!) documentation strings for those controls
431 :     that still are connected.
432 :    
433 :     Currently, multiple calls to Controls.new are accepted (only the first
434 :     has any effect). Eventually we should make sure that every control
435 :     is being made (via Controls.new) exactly once. Future access can then
436 :     be done using Controls.acc.
437 :    
438 :     Finally, it would probably be a good idea to use the getter-setter
439 :     interface to controls rather than ref cells. For the time being, both
440 :     styles are provided by the Controls module, but getter-setter pairs are
441 :     better if thread-safety is of any concern because they can be wrapped.
442 :    
443 :     *****************************************
444 :    
445 :     One bug fix: The function blockPlacement in three of the MLRISC
446 :     backpatch files used to be hard-wired to one of two possibilities at
447 :     link time (according to the value of the placementFlag). But (I
448 :     think) it should rather sense the flag every time.
449 :    
450 :     *****************************************
451 :    
452 :     Other assorted changes (by other people who did not supply a HISTORY entry):
453 :    
454 :     1. the cross-module inliner now works much better (Monnier)
455 :     2. representation of weights, frequencies, and probabilities in MLRISC
456 :     changed in preparation of using those for weighted block placement
457 :     (Reppy, George)
458 :    
459 :     ----------------------------------------------------------------------
460 : george 1116 Name: Lal George
461 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
462 :     Tag: george-20020307-weighted-block-placement
463 :    
464 :     Tested the weighted block placement optimization on all architectures
465 :     (except the hppa) using AMPL to generate the block and edge frequencies.
466 :     Changes were required in the machine properties to correctly
467 :     categorize trap instructions. There is an MLRISC flag
468 :     "weighted-block-placement" that can be used to enable weighted block
469 :     placement, but this will be ineffective without block/edge
470 :     frequencies (coming soon).
471 :    
472 :    
473 :     ----------------------------------------------------------------------
474 :     Name: Lal George
475 : george 1116 Date: 2002/03/05 17:24:48 EST
476 :     Tag: george-20020305-linkage-cluster
477 :    
478 : george 1124 In order to support the block placement optimization, a new cluster
479 :     is generated as the very first cluster (called the linkage cluster).
480 :     It contains a single jump to the 'real' entry point for the compilation
481 :     unit. Block placement has no effect on the linkage cluster itself, but
482 :     all the other clusters have full freedom in the manner in which they
483 :     reorder blocks or functions.
484 : george 1116
485 :     On the x86 the typical linkage code that is generated is:
486 :     ----------------------
487 :     .align 2
488 :     L0:
489 :     addl $L1-L0, 72(%esp)
490 : george 1124 jmp L1
491 : george 1116
492 :    
493 :     .align 2
494 :     L1:
495 :     ----------------------
496 :    
497 :     72(%esp) is the memory location for the stdlink register. This
498 :     must contain the address of the CPS function being called. In the
499 : george 1124 above example, it contains the address of L0; before
500 : george 1116 calling L1 (the real entry point for the compilation unit), it
501 :     must contain the address for L1, and hence
502 :    
503 :     addl $L1-L0, 72(%esp)
504 :    
505 : george 1124 I have tested this on all architectures except the hppa.The increase
506 :     in code size is of course negligible
507 : george 1116
508 :     ----------------------------------------------------------------------
509 : leunga 1115 Name: Allen Leung
510 :     Date: 2002/03/03 13:20:00 EST
511 :     Tag: leunga-20020303-mlrisc-tools
512 :    
513 :     Added #[ ... ] expressions to mlrisc tools
514 :    
515 :     ----------------------------------------------------------------------
516 : blume 1096 Name: Matthias Blume
517 : blume 1098 Date: 2002/02/27 12:29:00 EST
518 :     Tag: blume-20020227-cdebug
519 :     Description:
520 :    
521 :     - made types in structure C and C_Debug to be equal
522 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
523 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
524 :    
525 :     ----------------------------------------------------------------------
526 :     Name: Matthias Blume
527 : blume 1096 Date: 2002/02/26 12:00:00 EST
528 :     Tag: blume-20020226-ffi
529 :     Description:
530 :    
531 :     1. Fixed a minor bug in CM's "noweb" tool:
532 :     If numbering is turned off, then truly don't number (i.e., do not
533 :     supply the -L option to noweb). The previous behavior was to supply
534 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
535 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
536 :    
537 :     2. Once again, I reworked some aspects of the FFI:
538 :    
539 :     A. The incomplete/complete type business:
540 :    
541 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
542 :     gone!
543 :     - ML types representing an incomplete type are now *equal* to
544 :     ML types representing their corresponding complete types (just like
545 :     in C). This is still safe because ml-nlffigen will not generate
546 :     RTTI for incomplete types, nor will it generate functions that
547 :     require access to such RTTI. But when ML code generated from both
548 :     incomplete and complete versions of the C type meet, the ML types
549 :     are trivially interoperable.
550 :    
551 :     NOTE: These changes restore the full generality of the translation
552 :     (which was previously lost when I eliminated functorization)!
553 :    
554 :     B. Enum types:
555 :    
556 :     - Structure C now has a type constructor "enum" that is similar to
557 :     how the "su" constructor works. However, "enum" is not a phantom
558 :     type because each "T enum" has values (and is isomorphic to
559 :     MLRep.Signed.int).
560 :     - There are generic access operations for enum objects (using
561 :     MLRep.Signed.int).
562 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
563 :     * The structure contains the definition of type "mlrep" (the ML-side
564 :     representation type of the enum). Normally, mlrep is the same
565 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
566 :     then mlrep will be defined as a datatype -- thus facilitating
567 :     pattern matching on mlrep values.
568 :     ("-ec" will be suppressed if there are duplicate values in an
569 :     enumeration.)
570 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
571 :     will be generated for each C enum constant xxx.
572 :     * Conversion functions m2i and i2m convert between mlrep and
573 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
574 :     * Coversion functions c and ml convert between mlrep and "tag enum".
575 :     * Access functions (get/set) fetch and store mlrep values.
576 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
577 :     enumerations are merged into one single enumeration represented by
578 :     structure E_'.
579 :    
580 :     ----------------------------------------------------------------------
581 : leunga 1080 Name: Allen Leung
582 : leunga 1094 Date: 2002/02/25 04:45:00 EST
583 :     Tag: leunga-20020225-cps-spill
584 :    
585 :     This is a new implementation of the CPS spill phase.
586 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
587 :     In case of problems, replace it with the old file spill.sml
588 :    
589 :     The current compiler runs into some serious performance problems when
590 :     constructing a large record. This can happen when we try to compile a
591 :     structure with many items. Even a very simple structure like the following
592 :     makes the compiler slow down.
593 :    
594 :     structure Foo = struct
595 :     val x_1 = 0w1 : Word32.int
596 :     val x_2 = 0w2 : Word32.int
597 :     val x_3 = 0w3 : Word32.int
598 :     ...
599 :     val x_N = 0wN : Word32.int
600 :     end
601 :    
602 :     The following table shows the compile time, from N=1000 to N=4000,
603 :     with the old compiler:
604 :    
605 :     N
606 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
607 :     MLRISC ra 0.06u 0.00s 0.05g
608 :     (spills = 0 reloads = 0)
609 :     TOTAL 0.63u 0.07s 0.21g
610 :    
611 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
612 :     MLRISC ra 5.68u 0.59s 3.93g
613 :     (spills = 0 reloads = 0)
614 :     TOTAL 14.71u 0.99s 4.81g
615 :    
616 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
617 :     MLRISC ra 5.54u 0.65s 3.91g
618 :     (spills = 543 reloads = 1082)
619 :     TOTAL 65.40u 3.13s 6.00g
620 :    
621 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
622 :     MLRISC ra 0.80u 0.10s 0.55g
623 :     (spills = 42 reloads = 84)
624 :     TOTAL 129.42u 5.10s 4.13g
625 :    
626 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
627 :     MLRISC ra 2.69u 0.27s 1.38g
628 :     (spills = 62 reloads = 124)
629 :     TOTAL 682.48u 19.61s 13.99g
630 :    
631 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
632 :     MLRISC ra 4.96u 0.27s 2.72g
633 :     (spills = 85 reloads = 170)
634 :     TOTAL 2375.26u 57.21s 48.00g
635 :    
636 :     As you can see the old cps spill module suffers from some serious
637 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
638 :     instead of patching the problems up, I'm reimplementing it
639 : leunga 1094 with a different algorithm. The new code is more modular,
640 :     smaller when compiled, and substantially faster
641 :     (O(n log n) time and O(n) space). Timing of the new spill module:
642 :    
643 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
644 :     MLRISC ra 0.25u 0.02s 0.15g
645 :     (spills=1 reloads=3)
646 :     TOTAL 7.74u 0.34s 1.62g
647 :    
648 :     Implementation details:
649 :    
650 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
651 :     number of live variables at any program point (the bandwidth)
652 :     does not exceed a certain limit, which is determined by the
653 :     size of the spill area.
654 :    
655 :     When the bandwidth is too large, we decrease the register pressure by
656 :     packing live variables into spill records. How we achieve this is
657 :     completely different than what we did in the old code.
658 :    
659 :     First, there is something about the MLRiscGen code generator
660 :     that we should be aware of:
661 :    
662 :     o MLRiscGen performs code motion!
663 :    
664 :     In particular, it will move floating point computations and
665 :     address computations involving only the heap pointer to
666 :     their use sites (if there is only a single use).
667 :     What this means is that if we have a CPS record construction
668 :     statement
669 :    
670 :     RECORD(k,vl,w,e)
671 :    
672 :     we should never count the new record address w as live if w
673 :     has only one use (which is often the case).
674 :    
675 :     We should do something similar to floating point, but the transformation
676 :     there is much more complex, so I won't deal with that.
677 :    
678 :     Secondly, there are now two new cps primops at our disposal:
679 :    
680 :     1. rawrecord of record_kind option
681 :     This pure operator allocates some uninitialized storage from the heap.
682 :     There are two forms:
683 :    
684 :     rawrecord NONE [INT n] allocates a tagless record of length n
685 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
686 :     and initializes the tag.
687 :    
688 :     2. rawupdate of cty
689 :     rawupdate cty (v,i,x)
690 :     Assigns to x to the ith component of record v.
691 :     The storelist is not updated.
692 :    
693 :     We use these new primops for both spilling and increment record construction.
694 :    
695 :     1. Spilling.
696 :    
697 :     This is implemented with a linear scan algorithm (but generalized
698 :     to trees). The algorithm will create a single spill record at the
699 :     beginning of the cps function and use rawupdate to spill to it,
700 :     and SELECT or SELp to reload from it. So both spills and reloads
701 :     are fine-grain operations. In contrast, in the old algorithm
702 :     "spills" have to be bundled together in records.
703 :    
704 :     Ideally, we should sink the spill record construction to where
705 :     it is needed. We can even split the spill record into multiple ones
706 :     at the places where they are needed. But CPS is not a good
707 :     representation for global code motion, so I'll keep it simple and
708 :     am not attempting this.
709 :    
710 :     2. Incremental record construction (aka record splitting).
711 :    
712 :     Long records with many component values which are simulatenously live
713 :     (recall that single use record addresses are not considered to
714 :     be live) are constructed with rawrecord and rawupdate.
715 :     We allocate space on the heap with rawrecord first, then gradually
716 :     fill it in with rawupdate. This is the technique suggested to me
717 :     by Matthias.
718 :    
719 :     Some restrictions on when this is applicable:
720 :     1. It is not a VECTOR record. The code generator currently does not handle
721 :     this case. VECTOR record uses double indirection like arrays.
722 :     2. All the record component values are defined in the same "basic block"
723 :     as the record constructor. This is to prevent speculative
724 :     record construction.
725 :    
726 :     ----------------------------------------------------------------------
727 :     Name: Allen Leung
728 : leunga 1086 Date: 2002/02/22 01:02:00 EST
729 :     Tag: leunga-20020222-mlrisc-tools
730 :    
731 :     Minor bug fixes in the parser and rewriter
732 :    
733 :     ----------------------------------------------------------------------
734 :     Name: Allen Leung
735 : leunga 1085 Date: 2002/02/21 20:20:00 EST
736 :     Tag: leunga-20020221-peephole
737 :    
738 :     Regenerated the peephole files. Some contained typos in the specification
739 :     and some didn't compile because of pretty printing bugs in the old version
740 :     of 'nowhere'.
741 :    
742 :     ----------------------------------------------------------------------
743 :     Name: Allen Leung
744 : leunga 1079 Date: 2002/02/19 20:20:00 EST
745 :     Tag: leunga-20020219-mlrisc-tools
746 :     Description:
747 :    
748 :     Minor bug fixes to the mlrisc-tools library:
749 :    
750 :     1. Fixed up parsing colon suffixed keywords
751 :     2. Added the ability to shut the error messages up
752 :     3. Reimplemented the pretty printer and fixed up/improved
753 :     the pretty printing of handle and -> types.
754 :     4. Fixed up generation of literal symbols in the nowhere tool.
755 :     5. Added some SML keywords to to sml.sty
756 :    
757 :     ----------------------------------------------------------------------
758 :     Name: Matthias Blume
759 : blume 1078 Date: 2002/02/19 16:20:00 EST
760 :     Tag: blume-20020219-cmffi
761 :     Description:
762 :    
763 :     A wild mix of changes, some minor, some major:
764 :    
765 :     * All C FFI-related libraries are now anchored under $c:
766 :     $/c.cm --> $c/c.cm
767 :     $/c-int.cm --> $c/internals/c-int.cm
768 :     $/memory.cm --> $c/memory/memory.cm
769 :    
770 :     * "make" tool (in CM) now treats its argument pathname slightly
771 :     differently:
772 :     1. If the native expansion is an absolute name, then before invoking
773 :     the "make" command on it, CM will apply OS.Path.mkRelative
774 :     (with relativeTo = OS.FileSys.getDir()) to it.
775 :     2. The argument will be passed through to subsequent phases of CM
776 :     processing without "going native". In particular, if the argument
777 :     was an anchored path, then "make" will not lose track of that anchor.
778 :    
779 :     * Compiler backends now "know" their respective C calling conventions
780 :     instead of having to be told about it by ml-nlffigen. This relieves
781 :     ml-nlffigen from one of its burdens.
782 :    
783 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
784 :    
785 :     * Export C_DEBUG and C_Debug from $c/c.cm.
786 :    
787 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
788 :     subtyping relationship between incomplete pointers and their complete
789 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
790 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
791 :     the pointer target type is complete. In the incomplete case, 'o
792 :     will be instantiated with some "'c iobj" -- a type obtained by
793 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
794 :    
795 :     Operations that work on both incomplete and complete pointer types are
796 :     typed as taking an 'o ptr while operations that require the target to
797 :     be known are typed as taking some ('t, 'c) obj ptr.
798 :    
799 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
800 :     where void is an eqtype without any values. This makes it possible
801 :     to work on voidptr values using functions meant to operate on light
802 :     incomplete pointers.
803 :    
804 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
805 :     been vastly simplified.
806 :    
807 :     ----------------------------------------------------------------------
808 :     Name: Matthias Blume
809 : blume 1076 Date: 2002/02/19 10:48:00 EST
810 :     Tag: blume-20020219-pqfix
811 :     Description:
812 :    
813 :     Applied Chris Okasaki's bug fix for priority queues.
814 :    
815 :     ----------------------------------------------------------------------
816 :     Name: Matthias Blume
817 : blume 1073 Date: 2002/02/15 17:05:00
818 : blume 1069 Tag: Release_110_39
819 :     Description:
820 :    
821 : blume 1073 Last-minute retagging is becoming a tradition... :-(
822 :    
823 :     This is the working release 110.39.
824 :    
825 :     ----------------------------------------------------------------------
826 :     Name: Matthias Blume
827 :     Date: 2002/02/15 16:00:00 EST
828 :     Tag: Release_110_39-orig
829 :     Description:
830 :    
831 : blume 1069 Working release 110.39. New bootfiles.
832 :    
833 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
834 :     with all shells. So I retagged. -Matthias)
835 :    
836 : blume 1069 ----------------------------------------------------------------------
837 :     Name: Matthias Blume
838 : blume 1068 Date: 2002/02/15 14:17:00 EST
839 :     Tag: blume-20020215-showbindings
840 :     Description:
841 :    
842 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
843 :     the latter can be useful for exploring what bindings are available at
844 :     the interactive prompt. (The first function returns only the list
845 :     of symbols that are really bound, the second prints those but also the
846 :     ones that CM's autoloading mechanism knows about.)
847 :    
848 :     ----------------------------------------------------------------------
849 :     Name: Matthias Blume
850 : blume 1067 Date: 2002/02/15 12:08:00 EST
851 :     Tag: blume-20020215-iptrs
852 :     Description:
853 :    
854 :     Two improvements to ml-nlffigen:
855 :    
856 :     1. Write files only if they do not exist or if their current contents
857 :     do not coincide with what's being written. (That is, avoid messing
858 :     with the time stamps unless absolutely necessary.)
859 :    
860 :     2. Implement a "repository" mechanism for generated files related
861 :     to "incomplete pointer types". See the README file for details.
862 :    
863 :     ----------------------------------------------------------------------
864 :     Name: Matthias Blume
865 : blume 1066 Date: 2002/02/14 11:50:00 EST
866 :     Tag: blume-20020214-quote
867 :     Description:
868 :    
869 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
870 :     because of the new and improved tag generation scheme. (Thanks to Allen
871 :     Leung for pointing it out.)
872 :    
873 :     ----------------------------------------------------------------------
874 : george 1065 Name: Lal George
875 :     Date: 2002/02/14 09:55:27 EST 2002
876 :     Tag: george-20020214-isabelle-bug
877 :     Description:
878 :    
879 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
880 :     of Isabelle on the x86.
881 :    
882 :     From Allen:
883 :     -----------
884 :     I've found the problem:
885 :    
886 :     in ra-core.sml, I use the counter "blocked" to keep track of the
887 :     true number of elements in the freeze queue. When the counter goes
888 :     to zero, I skip examining the queue. But I've messed up the
889 :     bookkeeping in combine():
890 :    
891 :     else ();
892 :     case !ucol of
893 :     PSEUDO => (if !cntv > 0 then
894 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
895 :     ^^^^^^^^^^^^^^^^^^^^^^^
896 :     moveu := mergeMoveList(!movev, !moveu)
897 :     )
898 :     else ();
899 :    
900 :     combine() is called to coalesce two nodes u and v.
901 :     I think I was thinking that if the move counts of u and v are both
902 :     greater than zero then after they are coalesced then one node is
903 :     removed from the freeze queue. Apparently I was thinking that
904 :     both u and v are of low degree, but that's clearly not necessarily true.
905 :    
906 :    
907 :     02/12/2002:
908 :     Here's the patch. HOL now compiles.
909 :    
910 :     I don't know how this impact on performance (compile
911 :     time or runtime). This bug caused the RA (especially on the x86)
912 :     to go thru the potential spill phase when there are still nodes on the
913 :     freeze queue.
914 :    
915 :    
916 :    
917 :    
918 :     ----------------------------------------------------------------------
919 : blume 1058 Name: Matthias Blume
920 : blume 1064 Date: 2002/02/13 22:40:00 EST
921 :     Tag: blume-20020213-fptr-rtti
922 :     Description:
923 :    
924 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
925 :     updates.
926 :    
927 :     ----------------------------------------------------------------------
928 :     Name: Matthias Blume
929 : blume 1063 Date: 2002/02/13 16:41:00 EST
930 :     Tag: blume-20020213-cmlpq
931 :     Description:
932 :    
933 :     Added new priority queue export symbols (which have just been added to
934 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
935 :     not compile and the installer would choke.)
936 :    
937 :     ----------------------------------------------------------------------
938 :     Name: Matthias Blume
939 : blume 1062 Date: 2002/02/13 16:15:00 EST
940 :     Tag: blume-20020213-various
941 :     Description:
942 :    
943 :     1. More tweaks to ml-nlffigen:
944 :    
945 :     - better internal datastructures (resulting in slight speedup)
946 :     - "-match" option requires exact match
947 :     - "localized" gensym counters (untagged structs/unions nested within
948 :     other structs/unions or within typedefs get a fresh counter; their
949 :     tag will be prefixed by a concatenation of their parents' tags)
950 :     - bug fixes (related to calculation of transitive closure of types
951 :     to be included in the output)
952 :    
953 :     2. Minor Basis updates:
954 :    
955 :     - added implementations for List.collate and Option.app
956 :    
957 :     ----------------------------------------------------------------------
958 :     Name: Matthias Blume
959 : blume 1060 Date: 2002/02/11 15:55:00 EST
960 :     Tag: blume-20020211-gensym
961 :     Description:
962 :    
963 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
964 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
965 :     names (ML structure names that correspond to unnamed C structs, unions,
966 :     and enums), so that separate runs of ml-nlffigen do not clash.
967 :    
968 :     ----------------------------------------------------------------------
969 :     Name: Matthias Blume
970 : blume 1059 Date: 2002/02/11 12:05:00 EST
971 :     Tag: blume-20020211-gensml
972 :     Description:
973 :    
974 :     A quick fix for a problem with GenSML (in the pgraph-util library):
975 :     Make generation of toplevel "local" optional. (Strictly speaking,
976 :     signature definitions within "local" are not legal SML.)
977 :    
978 :     Other than that: updates to INSTALL and cm/TODO.
979 :    
980 :     ----------------------------------------------------------------------
981 :     Name: Matthias Blume
982 : blume 1058 Date: 2002/02/08 15:00:00 EST
983 :     Tag: blume-20020208-uniquepid
984 :     Description:
985 :    
986 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
987 :    
988 :     1. The installer (config/install.sh) has gotten smarter:
989 :    
990 :     - Configuration options are a bit easier to specify now
991 :     (in config/targets).
992 :     - Bug in recognizing .tar.bz2 files fixed.
993 :     - Installer automatically resolves dependencies between
994 :     configuration options (e.g., if you ask for eXene, you will
995 :     also get cml -- regardless whether you asked for it or not).
996 :     - Installer can run in "quieter mode" by setting the environment
997 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
998 :     "completely silent", though.
999 :     - Build HashCons library as part of smlnj-lib.
1000 :    
1001 :     2. A new scheme for assigning persistent identifiers to compilation
1002 :     units (and, by extension, to types etc.) has been put into place.
1003 :     This fixes a long-standing bug where types and even dynamic values
1004 :     can get internally confused, thereby compromising type safety
1005 :     (abstraction) and dynamic correctness. See
1006 :    
1007 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1008 :    
1009 :     for an example of how things could go wrong until now.
1010 :    
1011 :     The downside of the new scheme is that pids are not quite as
1012 :     persistent as they used to be: CM will generate a fresh pid
1013 :     for every compilation unit that it thinks it sees for the first
1014 :     time. That means that if you compile starting from a clean, fresh
1015 :     source tree at two different times, you end up with different
1016 :     binaries.
1017 :    
1018 :     Cutoff recompilation, however, has not been compromised because
1019 :     CM keeps pid information in special caches between runs.
1020 :    
1021 :     ----------------------------------------------------------------------
1022 : george 1053 Name: Lal George
1023 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
1024 :     Tag: <none>
1025 :     Description:
1026 :    
1027 :     Compilers that generate assembly code may produce global labels
1028 :     whose value is resolved at link time. The various peephole optimization
1029 :     modules did not take this in account.
1030 :    
1031 :     TODO. The Labels.addrOf function should really return an option
1032 :     type so that clients are forced to deal with this issue, rather
1033 :     than an exception being raised.
1034 :    
1035 :     ----------------------------------------------------------------------
1036 :     Name: Lal George
1037 : george 1053 Date: 2002/02/06 13:55:02 EST
1038 :     Tag: george-20020206-ra-breakup
1039 :     Description:
1040 :    
1041 :     1. A bug fix from Allen.
1042 :    
1043 :     A typo causes extra fstp %st(0)'s to be generated at compensation
1044 :     edges, which might cause stack underflow traps at runtime. This
1045 :     occurs in fft where there are extraneous fstps right before the 'into'
1046 :     trap instruction (in this case they are harmless since none of the
1047 :     integers overflow.)
1048 :    
1049 : george 1054 2. Pulled out various utility modules that were embedded in the modules
1050 : george 1053 of the register allocator. I need these modules for other purposes, but
1051 : george 1054 they are not complete enough to put into a library (just yet).
1052 : george 1053 ----------------------------------------------------------------------
1053 : blume 1049 Name: Matthias Blume
1054 :     Date: 2002/01/31 16:05:00 EST
1055 :     Tag: blume-20020131-sparc-ccalls
1056 :     Description:
1057 :    
1058 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1059 :     of extra stack space by mistake. Fixed.
1060 :    
1061 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1062 :    
1063 :     ----------------------------------------------------------------------
1064 : leunga 1048 Name: Allen Leung
1065 :     Date: 2002/01/30
1066 :     Tag: leunga-20020130-nowhere-bug-fix
1067 :     Description:
1068 :    
1069 :     MLRISC bug fixes:
1070 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1071 :     2. MachineInt.fromString was negating its value.
1072 :    
1073 :     ----------------------------------------------------------------------
1074 : blume 1039 Name: Matthias Blume
1075 : blume 1047 Date: 2002/01/29
1076 :     Tag: blume-20020129-INSTALL
1077 :     Description:
1078 :    
1079 :     - Added somewhat detailed installation instructions (file INSTALL).
1080 :     - Fixed curl-detection bug in config/install.sh.
1081 :     - It is now possible to select the URL getter using the URLGETTER
1082 :     environment variable:
1083 :    
1084 :     not set / "unknown" --> automatic detection (script tries wget,
1085 :     curl, and lynx)
1086 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1087 :     how to properly invoke them)
1088 :     other --> use $URLGETTER directly, it must take
1089 :     precisely two command-line arguments
1090 :     (source URL and destination file name)
1091 :    
1092 :     ----------------------------------------------------------------------
1093 :     Name: Matthias Blume
1094 : blume 1044 Date: 2002/01/28
1095 : blume 1045 Tag: blume-20020128-sparc-ccalls
1096 :     Description:
1097 :    
1098 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
1099 :     - Make use of the allocParam argument in sparc-c-calls.
1100 :    
1101 :     ----------------------------------------------------------------------
1102 :     Name: Matthias Blume
1103 :     Date: 2002/01/28
1104 : blume 1044 Tag: blume-20020128-allocParam
1105 :     Description:
1106 :    
1107 :     John Reppy: Changes c-calls API to accept client-callback for
1108 :     allocating extra stack space.
1109 :     me: Corresponding changes to mlriscGen (using a dummy argument that
1110 :     does not change the current behavior).
1111 :    
1112 :     ----------------------------------------------------------------------
1113 :     Name: Matthias Blume
1114 : blume 1040 Date: 2002/01/28 12:00:00
1115 :     Tag: Release_110_38
1116 :     Description:
1117 :    
1118 :     This time for real!!!
1119 :    
1120 :     ----------------------------------------------------------------------
1121 :     Name: Matthias Blume
1122 : blume 1039 Date: 2002/01/28 10:56:00 EST
1123 :     Tag: blume-20020128-retraction
1124 :     Description:
1125 :    
1126 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
1127 :     with blume-Release_110_38-retracted.)
1128 :    
1129 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
1130 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1131 :    
1132 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1133 :     and enums fixed. The naming of corresponding ML identifiers should
1134 :     now be consistent again.
1135 :    
1136 :     ----------------------------------------------------------------------
1137 : leunga 1038 Name: Allen Leung
1138 :     Date: 2002/01/27
1139 :     Tag: leunga-20020127-nowhere
1140 :     Description:
1141 :    
1142 :     Added a target called nowhere in the configuration scripts.
1143 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
1144 :     programs with where-clauses into legal SML code) during installation.
1145 :    
1146 :     ----------------------------------------------------------------------
1147 : blume 1036 Name: Matthias Blume
1148 : leunga 1038 Date: 2002/01/25 21:27:00 EST
1149 : blume 1039 Tag: blume-Release_110_38-retracted
1150 : blume 1037 Description:
1151 :    
1152 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
1153 :    
1154 :     README will be added later.
1155 :    
1156 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
1157 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
1158 :    
1159 : blume 1037 ----------------------------------------------------------------------
1160 :     Name: Matthias Blume
1161 : blume 1036 Date: 2002/01/25
1162 :     Tag: blume-20020125-ffi
1163 :     Description:
1164 :    
1165 :     A large number of tweaks and improvements to ml-nlffi-lib and
1166 :     ml-nlffigen:
1167 :    
1168 :     - ML represenation types have been streamlined
1169 :     - getter and setter functions work with concrete values, not abstract
1170 :     ones where possible
1171 :     - ml-nlffigen command line more flexible (see README file there)
1172 :     - some bugs have been fixed (hopefully)
1173 :    
1174 :     ----------------------------------------------------------------------
1175 : george 1034 Name: Lal George
1176 :     Date: 2002/01/24
1177 :     Tag: george-20020124-risc-ra-interface
1178 :     Description:
1179 :    
1180 :     There is a dramatic simplification in the interface to the
1181 :     register allocator for RISC architectures as a result of making
1182 :     parallel copy instructions explicit.
1183 :    
1184 :     ----------------------------------------------------------------------
1185 : blume 1028 Name: Matthias Blume
1186 : blume 1032 Date: 2002/01/22
1187 :     Tag: blume-20020122-x86-ccalls
1188 :     Description:
1189 :    
1190 :     Bug fix for c-calls on x86 (having to do with how char- and
1191 :     short-arguments are being handled).
1192 :    
1193 :     ----------------------------------------------------------------------
1194 :     Name: Matthias Blume
1195 : blume 1031 Date: 2002/01/21
1196 :     Tag: blume-20020121-ff
1197 :     Description:
1198 :    
1199 :     Another day of fiddling with the FFI...
1200 :    
1201 :     1. Bug fix/workaround: CKIT does not complain about negative array
1202 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1203 :     (Otherwise a negative dimension would send it into an infinite loop.)
1204 :    
1205 :     2. Some of the abstract types (light objects, light pointers, most "base"
1206 :     types) in structure C are now eqtypes.
1207 :    
1208 :     3. Added constructors and test functions for NULL function pointers.
1209 :    
1210 :     ----------------------------------------------------------------------
1211 :     Name: Matthias Blume
1212 : blume 1028 Date: 2002/01/18
1213 : blume 1030 Tag: blume-20020118-ready-for-new-release
1214 :     Description:
1215 :    
1216 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1217 :     files shortly.)
1218 :    
1219 :     Maybe we christen this to be 110.38?
1220 :    
1221 :     ----------------------------------------------------------------------
1222 :     Name: Matthias Blume
1223 :     Date: 2002/01/18
1224 : blume 1029 Tag: blume-20020118-more-ffifiddle
1225 :     Description:
1226 :    
1227 :     Today's FFI fiddling:
1228 :    
1229 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1230 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1231 :     of abstract ones.
1232 :    
1233 :     - Provided word-style bit operations etc. for "int" representation
1234 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1235 :     and Long.
1236 :    
1237 :     ----------------------------------------------------------------------
1238 :     Name: Matthias Blume
1239 :     Date: 2002/01/18
1240 : blume 1028 Tag: blume-20020118-use-x86-fp
1241 :     Description:
1242 :    
1243 :     Now that x86-fast-fp seems to be working, I turned it back on again
1244 :     by default. (Seems to work fine now, even with the FFI.)
1245 :    
1246 :     Other than that, I added some documentation about the FFI to
1247 :     src/ml-nlffigen/README and updated the FFI test examples in
1248 :     src/ml-nlffi-lib/Tests/*.
1249 :    
1250 :     ----------------------------------------------------------------------
1251 : leunga 1025 Name: Allen Leung
1252 :     Date: 2002/01/17
1253 :     Tag: leunga-20020117-x86-fast-fp-call
1254 :     Description:
1255 :    
1256 :     1. Fixed a problem with handling return fp values when x86's fast fp
1257 :     mode is turned on.
1258 :    
1259 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1260 :     of %f32.
1261 :    
1262 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1263 :    
1264 :     ----------------------------------------------------------------------
1265 : blume 1021 Name: Matthias Blume
1266 :     Date: 2002/01/16
1267 :     Tag: blume-20020116-ffifiddle
1268 :     Description:
1269 :    
1270 :     More fiddling with the FFI interface:
1271 :    
1272 :     - Make constness 'c instead of rw wherever possible. This eliminates
1273 :     the need for certain explicit coercions. (However, due to ML's
1274 :     value polymorphism, there will still be many cases where explicit
1275 :     coercions are necessary. Phantom types are not the whole answer
1276 :     to modeling a subtyping relationship in ML.)
1277 :    
1278 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1279 :    
1280 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1281 :     clumsy workaround for varargs problem.)
1282 :    
1283 :     ----------------------------------------------------------------------
1284 : george 1016 Name: Lal George
1285 :     Date: 2002/01/15
1286 :     Tag: <none>
1287 :     Description:
1288 :    
1289 :     1. Since COPY instructions are no longer native to the architecture,
1290 :     a generic functor can be used to implement the expandCopies function.
1291 :    
1292 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1293 :     TEXT segment.
1294 :    
1295 :     ----------------------------------------------------------------------
1296 : blume 1011 Name: Matthias Blume
1297 : blume 1015 Date: 2002/01/15
1298 :     Tag: blume-20020115-ffiupdates
1299 :     Description:
1300 :    
1301 :     1. Fix for bug resulting in single-precision float values being returned
1302 :     incorrectly from FFI calls.
1303 :    
1304 :     2. Small modifications to C FFI API:
1305 :    
1306 :     - memory-allocation routines return straight objects (no options)
1307 :     and raise an exception in out-of-memory situations
1308 :     - unsafe extensions to cast between function pointers and pointers
1309 :     from/to ints
1310 :     - added structure C_Debug as an alternative to structure C where
1311 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1312 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1313 :     but also takes a (possibly empty) list of existing library handles
1314 :     that the current library depends on
1315 :    
1316 :     ----------------------------------------------------------------------
1317 :     Name: Matthias Blume
1318 : blume 1011 Date: 2002/01/10
1319 :     Tag: blume-20020110-newffigen
1320 :     Description:
1321 :    
1322 :     1. Updates to portable graph code.
1323 :    
1324 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1325 :     more scalable now so that even huge interfaces such as the one
1326 :     for GTK compile in finite time and space. :-)
1327 :     See src/ml-nlffigen/README for details on what's new.
1328 :    
1329 :     ----------------------------------------------------------------------
1330 : george 1003 Name: Lal George
1331 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1332 :     Tag: george-20011206-rm-native-copy
1333 :     Description:
1334 :    
1335 :     Removed the native COPY and FCOPY instructions
1336 :     from all the architectures and replaced it with the
1337 :     explicit COPY instruction from the previous commit.
1338 :    
1339 :     It is now possible to simplify many of the optimizations
1340 :     modules that manipulate copies. This has not been
1341 :     done in this change.
1342 :    
1343 :     ----------------------------------------------------------------------
1344 :     Name: Lal George
1345 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1346 :     Tag: george-20011206-mlrisc-instruction
1347 :     Description:
1348 : george 824
1349 : george 1003 Changed the representation of instructions from being fully abstract
1350 :     to being partially concrete. That is to say:
1351 :    
1352 :     from
1353 :     type instruction
1354 :    
1355 :     to
1356 :     type instr (* machine instruction *)
1357 :    
1358 :     datatype instruction =
1359 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1360 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1361 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1362 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1363 :     | INSTR of instr
1364 :    
1365 :     This makes the handling of certain special instructions that appear on
1366 :     all architectures easier and uniform.
1367 :    
1368 :     LIVE and KILL say that a list of registers are live or killed at the
1369 :     program point where they appear. No spill code is generated when an
1370 :     element of the 'regs' field is spilled, but the register is moved to
1371 :     the 'spilled' (which is present, more for debugging than anything else).
1372 :    
1373 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1374 :     We used to generate:
1375 :    
1376 :     DEFFREG f1
1377 :     f1 := f2 + f3
1378 :     trapb
1379 :    
1380 :     but now generate:
1381 :    
1382 :     f1 := f2 + f3
1383 :     trapb
1384 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1385 :    
1386 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1387 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1388 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1389 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1390 :     this architecture).
1391 :    
1392 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1393 :     all the architectures. This will result in further simplification of the
1394 :     register allocator that must be aware of them for coalescing purposes, and
1395 :     will also simplify certain aspects of the machine description that provides
1396 :     callbacks related to parallel copies.
1397 :    
1398 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1399 :     machine instruction set!
1400 :    
1401 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1402 :     functions for making porting easier -- essentially converting upper case
1403 :     to lower case. All machine instructions (of type instr) are in upper case,
1404 :     and the lower case form generates an MLRISC instruction. For example on
1405 :     the alpha we have:
1406 :    
1407 :     datatype instr =
1408 :     LDA of {r:cell, b:cell, d:operand}
1409 :     | ...
1410 :    
1411 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1412 :     ...
1413 :    
1414 :     where lda is just (INSTR o LDA), etc.
1415 :    
1416 : leunga 815 ----------------------------------------------------------------------
1417 : blume 986 Name: Matthias Blume
1418 : blume 993 Date: 2001/11/22 21:40:00 EST
1419 :     Tag: Release_110_37
1420 :     Description:
1421 :    
1422 :     Release 110.37. This time for real.
1423 :    
1424 :     ----------------------------------------------------------------------
1425 :     Name: Matthias Blume
1426 : blume 988 Date: 2001/11/21 16:35:00 EST
1427 :     Tag: blume-20011121-foot-in-mouth
1428 :     Description:
1429 :    
1430 :     Removed the "Release_110_37" tag because of a serious bug.
1431 :     This will be re-tagged once the bug is fixed.
1432 :    
1433 :     ----------------------------------------------------------------------
1434 :     Name: Matthias Blume
1435 : blume 987 Date: 2001/11/21 16:14:00 EST
1436 :     Tag: blume-20011121-forgottenfile
1437 :     Description:
1438 :    
1439 :     Forgot to add a file. (Just a .tex-file -- part of
1440 :     the CM manual source.)
1441 :    
1442 :     ----------------------------------------------------------------------
1443 :     Name: Matthias Blume
1444 : blume 986 Date: 2001/11/21 16:10:00 EST
1445 : blume 988 Tag: blume-20011121-invalid_110_37
1446 : blume 986 Description:
1447 :    
1448 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
1449 :     because we found a serious bug in all non-x86 backends.
1450 :     - Matthias
1451 :    
1452 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
1453 :     so that code object name strings are directly inserted into code
1454 :     objects at code generation time. The only business the runtime system
1455 :     has with this is now to read the name strings on occasions.
1456 :     (The encoding of the name string has also changed somewhat.)
1457 :    
1458 :     2. CM now implements a simple "set calculus" for specifying export lists.
1459 :     In particular, it is now possible to refer to the export lists of
1460 :     other libraries/groups/sources and form unions as well as differences.
1461 :     See the latest CM manual for details.
1462 :    
1463 :     3. An separate notion of "proxy" libraries has again be eliminated from
1464 :     CM's model. (Proxy libraries are now simply a special case of using
1465 :     the export list calculus.)
1466 :    
1467 :     4. Some of the existing libraries now take advantage of the new set
1468 :     calculus.
1469 :     (Notice that not all libraries have been converted because some
1470 :     of the existing .cm-files are supposed to be backward compatible
1471 :     with 110.0.x.)
1472 :    
1473 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
1474 :     "General.exnMessage"! The former relies on a certain hook to be
1475 :     initialized, and that often does not happen in the stand-alone case.)
1476 :    
1477 :     ----------------------------------------------------------------------
1478 : george 984 Name: Lal George
1479 :     Date: 2001/11/21 13:56:18 EST
1480 :     Tag: george-2001121-pseudo-ops
1481 :     Description:
1482 :    
1483 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
1484 :     ought to never be any question of incompatabilities with
1485 :     pseudo-op syntax expected by host assemblers.
1486 :    
1487 :     For now, only modules supporting GAS syntax are implemented
1488 :     but more should follow, such as MASM, and vendor assembler
1489 :     syntax, e.g. IBM as, Sun as, etc.
1490 :    
1491 :     ----------------------------------------------------------------------
1492 : blume 977 Name: Matthias Blume
1493 :     Date: 2001/11/14 11:52:00 EST
1494 :     Tag: blume-20011114-srcname
1495 :     Description:
1496 :    
1497 :     1. Routed the name of the current source file to mlriscgen where it
1498 :     should be directly emitted into the code object. (This last part
1499 :     is yet to be done.)
1500 :    
1501 :     2. Some cleanup of the pgraph code to make it match the proposal that
1502 :     I put out the other day. (The proposal notwithstanding, things are
1503 :     still in flux here.)
1504 :    
1505 :     ----------------------------------------------------------------------
1506 : george 976 Name: Lal George
1507 :     Date: 2001/11/14 09:44:04 EST
1508 :     Tag:
1509 :     Description:
1510 :    
1511 :     Fix for a backpatching bug reported by Allen.
1512 :    
1513 :     Because the boundary between short and long span-dependent
1514 :     instructions is +/- 128, there are an astounding number of
1515 :     span-dependent instructions whose size is over estimated.
1516 :    
1517 :     Allen came up with the idea of letting the size of span
1518 :     dependent instructions be non-monotonic, for a maxIter
1519 :     number of times, after which the size must be monotonically
1520 :     increasing.
1521 :    
1522 :     This table shows the number of span-dependent instructions
1523 :     whose size was over-estimated as a function of maxIter, for the
1524 :     file Parse/parse/ml.grm.sml:
1525 :    
1526 :     maxIter # of instructions:
1527 :     10 687
1528 :     20 438
1529 :     30 198
1530 :     40 0
1531 :    
1532 :     In compiling the compiler, there is no significant difference in
1533 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1534 :     my measurements showed that maxIter=40 was a tad faster than
1535 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1536 :     point within 13 iterations, so fixing maxIter at 40, while high,
1537 :     is okay.
1538 :    
1539 :     ----------------------------------------------------------------------
1540 : blume 975 Name: Matthias Blume
1541 :     Date: 2001/10/31 15:25:00 EST
1542 :     Tag: blume-20011031-pgraph
1543 :     Description:
1544 :    
1545 :     CKIT:
1546 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1547 :     argument identifiers.
1548 :     * Changed the return type of TypeUtil.getFunction accordingly.
1549 :     * Type equality ignores the argument names.
1550 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1551 :     if there is a mismatch.
1552 :    
1553 :     installation script:
1554 :     * attempts to use "curl" if available (unless "wget" is available as well)
1555 :    
1556 :     CM:
1557 :     * has an experimental implementation of "portable graphs" which I will
1558 :     soon propose as an implementation-independent library format
1559 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1560 :    
1561 :     NLFFI-LIB:
1562 :     * some cleanup (all cosmetic)
1563 :    
1564 :     NLFFIGEN:
1565 :     * temporarily disabled the mechanism that suppresses ML output for
1566 :     C definitions whose identifiers start with an underscore character
1567 :     * generate val bindings for enum constants
1568 :     * user can request that only one style (light or heavy) is being used;
1569 :     default is to use both (command-line arguments: -heavy and -light)
1570 :     * fixed bug in handling of function types involving incomplete pointers
1571 :     * generate ML entry points that take record arguments (i.e., using
1572 :     named arguments) for C functions that have a prototype with named
1573 :     arguments
1574 :     (see changes to CKIT)
1575 :    
1576 :     ----------------------------------------------------------------------
1577 : leunga 972 Name: Allen Leung
1578 :     Date: 2001/10/27 20:34:00 EDT
1579 :     Tag: leunga-20011027-x86-fast-fp-call
1580 :     Description:
1581 :    
1582 :     Fixed the bug described in blume-20010920-slowfp.
1583 :    
1584 :     The fix involves
1585 :     1. generating FCOPYs in FSTP in ia32-svid
1586 :     2. marking a CALL with the appropriate annotation
1587 :    
1588 :     ----------------------------------------------------------------------
1589 : blume 961 Name: Matthias Blume
1590 :     Date: 2001/10/16 11:32:00 EDT
1591 :     Tag: blume-20011016-netbsd
1592 :     Description:
1593 :    
1594 :     Underscore patch from Chris Richards (fixing problem with compiling
1595 :     runtime system under recent NetBSD).
1596 :    
1597 :     ----------------------------------------------------------------------
1598 : george 959 Name: Allen Leung
1599 :     Date: 2001/10/12 17:18:32 EDT 2001
1600 :     Tag: leung-20011012-x86-printflowgraph
1601 :     Description:
1602 :    
1603 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1604 :    
1605 :     ----------------------------------------------------------------------
1606 : george 952 Name: Lal George
1607 : george 958 Date: 2001/10/11 23:51:34 EDT
1608 :     Tag: george-20011011-too-many-instrs
1609 :     Description:
1610 :    
1611 :     The representation of a program point never expected to see more
1612 :     than 65536 instructions in a basic block!
1613 :    
1614 :     ----------------------------------------------------------------------
1615 :     Name: Lal George
1616 : george 952 Date: 2001/10/09 09:41:37 EDT
1617 :     Tag: george-20011008-mlrisc-labels
1618 :     Description:
1619 :    
1620 :     Changed the machine description files to support printing of
1621 :     local and global labels in assembly code, based on host assembler
1622 :     conventions.
1623 :    
1624 :     ----------------------------------------------------------------------
1625 : blume 916 Name: Matthias Blume
1626 : blume 939 Date: 2001/09/25 15:25:00 EDT
1627 :     Tag: blume-20010925-exninfo
1628 :     Description:
1629 :    
1630 :     I provided a non-hook implementation of exnName (at the toplevel) and
1631 :     made the "dummy" implementation of exnMessage (at the toplevel) more
1632 :     useful: if nothing gets "hooked in", then at least you are going to
1633 :     see the exception name and a message indicating why you don't see more.
1634 :    
1635 :     [For the time being, programs that need exnMessage and want to use
1636 :     ml-build should either use General.exnMessage (strongly recommended) or
1637 :     refer to structure General at some other point so that CM sees a
1638 :     static dependency.]
1639 :    
1640 :     [Similar remarks go for "print" and "use": If you want to use their
1641 :     functionality in stand-alone programs generated by ml-build, then use
1642 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1643 :    
1644 :     ----------------------------------------------------------------------
1645 :     Name: Matthias Blume
1646 : blume 938 Date: 2001/09/20 17:28:00 EDT
1647 :     Tag: blume-20010920-slowfp
1648 :     Description:
1649 :    
1650 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1651 :    
1652 :     ----------------------------------------------------------------------
1653 :     Name: Matthias Blume
1654 : blume 937 Date: 2001/09/20 17:20:00 EDT
1655 :     Tag: blume-20010920-canonicalpaths
1656 :     Description:
1657 :    
1658 :     0. Updated the BOOT file (something that I forgot to do earlier).
1659 :    
1660 :     1. Small internal change to CM so that it avoids "/../" in filenames
1661 :     as much as possible (but only where it is safe).
1662 :    
1663 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1664 :     that arguments that contain delimiters are passed through correctly.
1665 :     This change also means that all "special" arguments of the form
1666 :     @SMLxxx... must come first.
1667 :    
1668 :     3. Changed install script to put relative anchor names for tool commands
1669 :     into pathconfig.
1670 :    
1671 :     ----------------------------------------------------------------------
1672 :     Name: Matthias Blume
1673 : blume 932 Date: 2001/09/18 15:35:00 EDT
1674 :     Tag: blume-20010918-readme11036
1675 :     Description:
1676 :    
1677 :     Added README files.
1678 :    
1679 :     ----------------------------------------------------------------------
1680 :     Name: Matthias Blume
1681 : blume 929 Date: 2001/09/18 11:45:00 EDT
1682 :     Tag: Release_110_36 (retag)
1683 :     Description:
1684 :    
1685 :     Fixed mistake in config/preloads. Retagged as 110.36.
1686 :    
1687 :     ----------------------------------------------------------------------
1688 :     Name: Matthias Blume
1689 : blume 928 Date: 2001/09/18 09:40:00 EDT
1690 : blume 929 Tag: Release_110_36_orig (tag changed)
1691 : blume 928 Description:
1692 :    
1693 :     New version (110.36). New bootfiles.
1694 :    
1695 :     ----------------------------------------------------------------------
1696 :     Name: Matthias Blume
1697 : blume 927 Date: 2001/09/14 16:15:00 EDT
1698 :     Tag: blume-20010914-x86fastfp
1699 :     Description:
1700 :    
1701 :     John committed some changes that Allen made, in particular a (hopefully)
1702 :     correctly working version of the x86-fp module.
1703 :    
1704 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1705 :     flag to "true". Everything seems to compile to a fixpoint ok, and
1706 :     "mandelbrot" speeds up by about 15%.
1707 :    
1708 :     ----------------------------------------------------------------------
1709 :     Name: Matthias Blume
1710 : blume 923 Date: 2001/09/13 11:20:00 EDT
1711 :     Tag: blume-20010913-minimal
1712 :     Description:
1713 :    
1714 :     1. Stefan Monnier's patch to fix a miscompilation problem that
1715 :     was brought to light by John Reppy's work on Moby.
1716 :    
1717 :     2. Implemented a minimal "structure Compiler" that contains just
1718 :     "version" and "architecture". The minimal version will be
1719 :     available when the full version is not. This is for backward-
1720 :     compatibility with code that wants to test Compiler.version.
1721 :    
1722 :     ----------------------------------------------------------------------
1723 :     Name: Matthias Blume
1724 : blume 920 Date: 2001/08/28 14:03:00 EDT
1725 :     Tag: blume-20010828-ml-lex
1726 :     Description:
1727 :    
1728 :     Fix for bug 1581, received from Neophytos Michael.
1729 :    
1730 :     ----------------------------------------------------------------------
1731 :     Name: Matthias Blume
1732 : blume 918 Date: 2001/08/27 11:20:00 EDT
1733 :     Tag: blume-20010827-readme11035
1734 :     Description:
1735 :    
1736 :     Fleshed out the README file for 110.35.
1737 :    
1738 :     ----------------------------------------------------------------------
1739 :     Name: Matthias Blume
1740 : blume 916 Date: 2001/08/24 17:10:00 EDT
1741 :     Tag: Release_110_35
1742 :     Description:
1743 :    
1744 :     New version number (110.35). New bootfiles.
1745 :    
1746 :     ----------------------------------------------------------------------
1747 : george 915 Name: Lal George
1748 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1749 :     Tag: george-20010824-MLRISC-graphs
1750 :     Description:
1751 :    
1752 :     removed clusters from MLRISC completely and replaced with graphs.
1753 :    
1754 :     ----------------------------------------------------------------------
1755 :     Name: Matthias Blume
1756 : blume 905 Date: 2001/08/23 17:50:00 EDT
1757 :     Tag: blume-20010823-toplevel
1758 :     Description:
1759 :    
1760 :     - some reorganization of the code that implements various kinds of
1761 :     environments in the compiler (static, dynamic, symbolic, combined)
1762 :     - re-implemented the EnvRef module so that evalStream works properly
1763 :     (if the stream contains references to "use", "CM.make", etc.)
1764 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1765 :    
1766 :     ----------------------------------------------------------------------
1767 :     Name: Matthias Blume
1768 : blume 904 Date: 2001/08/20 15:50 EDT
1769 :     Tag: blume20010820-slipup
1770 :     Description:
1771 :    
1772 :     I forgot to commit a few files. Here they are...
1773 :    
1774 :     ----------------------------------------------------------------------
1775 :     Name: Matthias Blume
1776 : blume 903 Date: 2001/08/20 15:35:00 EDT
1777 :     Tag: blume-20010820-debugprof
1778 :     Description:
1779 :    
1780 :     !!!! NEW BOOTFILES !!!!
1781 :    
1782 :     This is another round of reorganizing the compiler sources. This
1783 :     time the main goal was to factor out all the "instrumentation"
1784 :     passes (for profiling and backtracing) into their own library.
1785 :     The difficulty was to do it in such a way that it does not depend
1786 :     on elaborate.cm but only on elabdata.cm.
1787 :    
1788 :     Therefore there have been further changes to both elaborate.cm and
1789 :     elabdata.cm -- more "generic" things have been moved from the former
1790 :     to the latter. As a result, I was forced to split the assignment
1791 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1792 :     SML/NJ-specific. Since it would have been awkward to maintain,
1793 :     I bit the bullet and actually _changed_ the mapping between these
1794 :     numbers and primtycs. The bottom line of this is that you need
1795 :     a new set of bin- and bootfiles.
1796 :    
1797 :     I have built new bootfiles for all architectures, so doing a fresh
1798 :     checkout and config/install.sh should be all you need.
1799 :    
1800 :     The newly created library's name is
1801 :    
1802 :     $smlnj/viscomp/debugprof.cm
1803 :    
1804 :     and its sources live under
1805 :    
1806 :     src/compiler/DebugProf
1807 :    
1808 :     ----------------------------------------------------------------------
1809 :     Name: Matthias Blume
1810 : blume 902 Date: 2001/08/15 17:15:00 EDT
1811 :     Tag: blume-20010815-compreorg
1812 :     Description:
1813 :    
1814 :     This is a first cut at reorganizing the CM libraries that make up the
1815 :     core of the compiler. The idea is to separate out pieces that could
1816 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1817 :    
1818 :     The current status is a step in this direction, but it is not quite
1819 :     satisfactory yet. Expect more changes in the future.
1820 :    
1821 :     Here is the current (new) organization...
1822 :    
1823 :     What used to be $smlnj/viscomp/core.cm is now divided into
1824 :     six CM libraries:
1825 :    
1826 :     $smlnj/viscomp/basics.cm
1827 :     /parser.cm
1828 :     /elabdata.cm
1829 :     /elaborate.cm
1830 :     /execute.cm
1831 :     /core.cm
1832 :    
1833 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1834 :     All these libraries are proxy libraries that contain precisely
1835 :     one CM library component. Here are the locations of the components
1836 :     (all within the src/compiler tree):
1837 :    
1838 :     Basics/basics.cm
1839 :     Parse/parser.cm
1840 :     ElabData/elabdata.cm
1841 :     Elaborator/elaborate.cm
1842 :     Execution/execute.cm
1843 :     core.cm
1844 :    
1845 :     [This organization is the same that has been used already
1846 :     for a while for the architecture-specific parts of the visible
1847 :     compiler and for the old version of core.cm.]
1848 :    
1849 :     As you will notice, many source files have been moved from their
1850 :     respective original locations to a new home in one of the above
1851 :     subtrees.
1852 :    
1853 :     The division of labor between the new libraries is the following:
1854 :    
1855 :     basics.cm:
1856 :     - Simple, basic definitions that pertain to many (or all) of
1857 :     the other libraries.
1858 :     parser.cm:
1859 :     - The SML parser, producing output of type Ast.dec.
1860 :     - The type family for Ast is also defined and exported here.
1861 :     elabdata.cm:
1862 :     - The datatypes that describe input and output of the elaborator.
1863 :     This includes types, absyn, and static environments.
1864 :     elaborator.cm:
1865 :     - The SML/NJ type checker and elaborator.
1866 :     This maps an Ast.dec (with a given static environment) to
1867 :     an Absyn.dec (with a new static environment).
1868 :     - This libraries implements certain modules that used to be
1869 :     structures as functors (to remove dependencies on FLINT).
1870 :     execute.cm:
1871 :     - Everything having to do with executing binary code objects.
1872 :     - Dynamic environments.
1873 :     core.cm:
1874 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1875 :     - Top-level modules.
1876 :     - FLINT (this should eventually become its own library)
1877 :    
1878 :     Notes:
1879 :    
1880 :     I am not 100% happy with the way I separated the elaborator (and its
1881 :     data structures) from FLINT. Two instances of the same problem:
1882 :    
1883 :     1. Data structures contain certain fields that carry FLINT-specific
1884 :     information. I hacked around this using exn and the property list
1885 :     module from smlnj-lib. But the fact that there are middle-end
1886 :     specific fields around at all is a bit annoying.
1887 :    
1888 :     2. The elaborator calculates certain FLINT-related information. I tried
1889 :     to make this as abstract as I could using functorization, but, again,
1890 :     the fact that the elaborator has to perform calculations on behalf
1891 :     of the middle-end at all is not nice.
1892 :    
1893 :     3. Having to used exn and property lists is unfortunate because it
1894 :     weakens type checking. The other alternative (parameterizing
1895 :     nearly *everything*) is not appealing, though.
1896 :    
1897 :     I removed the "rebinding =" warning hack because due to the new organization
1898 :     it was awkward to maintain it. As a result, the compiler now issues some of
1899 :     these warnings when compiling init.cmi during bootstrap compilation. On
1900 :     the plus side, you also get a warning when you do, for example:
1901 :     val op = = Int32.+
1902 :     which was not the case up to now.
1903 :    
1904 :     I placed "assign" and "deref" into the _Core structure so that the
1905 :     code that deals with the "lazy" keyword can find them there. This
1906 :     removes the need for having access to the primitive environment
1907 :     during elaboration.
1908 :    
1909 :     ----------------------------------------------------------------------
1910 :     Name: Matthias Blume
1911 : blume 898 Date: 2001/08/13
1912 :     Tag: blume-20010813-closures
1913 :     Description:
1914 :    
1915 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1916 :     performance of certain loops by avoiding needless closure allocation.
1917 :    
1918 :     ----------------------------------------------------------------------
1919 : george 890 Name: Lal George
1920 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1921 :     Tag: george-20010731-x86-fmalloc
1922 :     Description: Fixed bug in x86 calls
1923 :    
1924 :     There was a bug where call instructions would mysteriously
1925 :     vanish. The call instruction had to be one that returned
1926 :     a floating point value.
1927 :    
1928 :     ----------------------------------------------------------------------
1929 :     Name: Lal George
1930 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1931 :     Tag: george-20010719-simple-cells
1932 :     Description:
1933 :    
1934 :     I have dramatically simplified the interface for CELLS in MLRISC.
1935 :    
1936 :     In summary, the cells interface is broken up into three parts:
1937 :    
1938 :     1. CellsBasis : CELLS_BASIS
1939 :    
1940 :     CellsBasis is a top level structure and common for all
1941 :     architectures. it contains the definitions of basic datatypes
1942 :     and utility functions over these types.
1943 :    
1944 :     2. functor Cells() : CELLS
1945 :    
1946 :     Cells generates an interface for CELLS that incorporates the
1947 :     specific resources on the target architecture, such as the
1948 :     presence of special register classes, their number and size,
1949 :     and various useful substructures.
1950 :    
1951 :     3. <ARCH>CELLS
1952 :    
1953 :     e.g. SparcCells: SPARCCELLS
1954 :    
1955 :     <ARCH>CELLS usually contains additional bindings for special
1956 :     registers on the architecture, such as:
1957 :    
1958 :     val r0 : cell (* register zero *)
1959 :     val y : cell (* Y register *)
1960 :     val psr : cell (* processor status register *)
1961 :     ...
1962 :    
1963 :     The structure returned by applying the Cells functor is opened
1964 :     in this interface.
1965 :    
1966 :     The main implication of all this is that the datatypes for cells is
1967 :     split between CellsBasis and CELLS -- a fairly simple change for user
1968 :     code.
1969 :    
1970 :     In the old scheme the CELLS interface had a definitional binding of
1971 :     the form:
1972 :    
1973 :     signature CELLS = sig
1974 :    
1975 :     structure CellsBasis = CellsBasis
1976 :    
1977 :     ...
1978 :    
1979 :     end
1980 :    
1981 :     With all the sharing constraints that goes on in MLRISC, this old
1982 :     design quickly leads to errors such as:
1983 :    
1984 :     "structure definition spec inside of sharing ... "
1985 :    
1986 :    
1987 :     and appears to require an unacceptable amount of sharing and where
1988 :     constraint hackery.
1989 :    
1990 :     I think this error message (the interaction of definitional specs and
1991 :     sharing) requires more explanation on our web page.
1992 :    
1993 :     ----------------------------------------------------------------------
1994 : blume 826 Name: Matthias Blume
1995 : blume 879 Date: 2001/07/19 15:00:00 EDT
1996 :     Tag: blume-20010719-libreorg
1997 :     Description:
1998 :    
1999 :     This update puts together a fairly extensive but straightforward change
2000 :     to the way the libraries that implement the interactive system are
2001 :     organized:
2002 :    
2003 :     The biggest change is the elimination of structure Compiler. As a
2004 :     replacement for this structure, there is now a CM library
2005 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2006 :     that exports all the substructures of the original structure Compiler
2007 :     directly. So instead of saying Compiler.Foo.bar one now simply
2008 :     says Foo.bar. (The CM libraries actually export a collection of
2009 :     structures that is richer than the collection of substructures of
2010 :     structure Compiler.)
2011 :    
2012 :     To make the transition smooth, there is a separate library called
2013 :     $smlnj/compiler/compiler.cm which puts together and exports the
2014 :     original structure Compiler (or at least something very close to it).
2015 :    
2016 :     There are five members of the original structure Compiler
2017 :     that are not exported directly but which instead became members
2018 :     of a new structure Backend (described by signature BACKEND). These are:
2019 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2020 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
2021 :     architecture (: string).
2022 :    
2023 :     Structure Compiler.Version has become structure CompilerVersion.
2024 :    
2025 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2026 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2027 :     or x86, respectively.
2028 :     Each of these exports the same frontend structures that
2029 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
2030 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2031 :     Hppa, PPC, Sparc, or X86, respectively.
2032 :    
2033 :     Library $smlnj/compiler/all.cm exports the union of the exports of
2034 :     $smlnj/compiler/<arch>.cm
2035 :    
2036 :     There are no structures <Arch>Compiler anymore, use
2037 :     $smlnj/compiler/<arch>.cm instead.
2038 :    
2039 :     Library host-compiler-0.cm is gone. Instead, the internal library
2040 :     that instantiates CM is now called cm0.cm. Selection of the host
2041 :     compiler (backend) is no longer done here but. (Responsibility for it
2042 :     now lies with $smlnj/compiler/current.cm. This seems to be more
2043 :     logical.)
2044 :    
2045 :     Many individual files have been moved or renamed. Some files have
2046 :     been split into multiple files, and some "dead" files have been deleted.
2047 :    
2048 :     Aside from these changes to library organization, there are also changes
2049 :     to the way the code itself is organized:
2050 :    
2051 :     Structure Binfile has been re-implemented in such a way that it no
2052 :     longer needs any knowledge of the compiler. It exclusively deals
2053 :     with the details of binfile layout. It no longer invokes the
2054 :     compiler (for the purpose of creating new prospective binfile
2055 :     content), and it no longer has any knowledge of how to interpret
2056 :     pickles.
2057 :    
2058 :     Structure Compile (: COMPILE) has been stripped down to the bare
2059 :     essentials of compilation. It no longer deals with linking/execution.
2060 :     The interface has been cleaned up considerably.
2061 :    
2062 :     Utility routines for dealing with linking and execution have been
2063 :     moved into their own substructures.
2064 :    
2065 :     (The ultimate goal of these changes is to provide a light-weight
2066 :     binfile loader/linker (at least for, e.g., stable libraries) that
2067 :     does not require CM or the compiler to be present.)
2068 :    
2069 :     CM documentation has been updated to reflect the changes to library
2070 :     organization.
2071 :    
2072 :     ----------------------------------------------------------------------
2073 :     Name: Matthias Blume
2074 : blume 875 Date: 2001/07/10 17:30:00 EDT
2075 :     Tag: Release_110_34
2076 :     Description:
2077 :    
2078 : blume 876 Minor tweak to 110.34 (re-tagged):
2079 :    
2080 :     - README.html file added to CVS repository
2081 :     - runtime compiles properly under FreeBSD 3.X and 4.X
2082 :    
2083 :     ----------------------------------------------------------------------
2084 :     Name: Matthias Blume
2085 :     Date: 2001/07/10 17:30:00 EDT
2086 :     Tag: Release_110_34
2087 :     Description:
2088 :    
2089 : blume 875 New version number (110.34). New bootfiles.
2090 :    
2091 :     ----------------------------------------------------------------------
2092 :     Name: Matthias Blume
2093 : blume 874 Date: 2001/07/09 16:00:00 EDT
2094 :     Tag: blume-20010709-more-varargs
2095 :     Description:
2096 :    
2097 :     I changed the handling of varargs in ml-nlffigen again:
2098 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
2099 :    
2100 :     The immediate effect is that you can actually call a varargs function
2101 :     from ML -- but you can't actually supply any arguments beyond the ones
2102 :     specified explicitly. (For example, you can call printf with its format
2103 :     string, but you cannot pass additional arguments.)
2104 :    
2105 :     This behavior is only marginally more useful than the one before, but
2106 :     it has the advantage that a function or, more importantly, a function
2107 :     type never gets dropped on the floor, thus avoiding follow-up problems with
2108 :     other types that refer to the offending one.
2109 :    
2110 :     ----------------------------------------------------------------------
2111 :     Name: Matthias Blume
2112 : blume 873 Date: 2001/07/09 11:25:00 EDT
2113 :     Tag: blume-20010709-varargs
2114 :     Description:
2115 :    
2116 :     1. ckit-lib.cm now exports structure Error
2117 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2118 :     with a warning accompanied by a source location. Moreover, it
2119 :     merely skips the offending function or type and proceeds with the
2120 :     rest of its work.u As a result, one can safely feed C code containing
2121 :     "..." to ml-nlffigen.
2122 :     3. There are some internal improvements to CM, providing slightly
2123 :     more general string substitutions in the tools subsystem.
2124 :    
2125 :     ----------------------------------------------------------------------
2126 :     Name: Matthias Blume
2127 : blume 854 Date: 2001/06/27 15:10:00 EDT
2128 :     Tag: blume-20010627-concur
2129 :     Description:
2130 :    
2131 :     Fixed a small bug in CM's handling of parallel compilation.
2132 :     (You could observe the bug by Control-C-interrupting an ordinary
2133 :     CMB.make or CM.stabilize and then attaching some compile servers.
2134 :     The result was that all of a sudden the previously interrupted
2135 :     compilation would continue on its own. This was because of
2136 :     an over-optimization: CM did not bother to clean out certain queues
2137 :     when no servers were attached "anyway", resulting in the contents
2138 :     of these queues to grab control when new servers did get attached.)
2139 :    
2140 :     There is also another minor update to the CM manual.
2141 :    
2142 :     ----------------------------------------------------------------------
2143 :     Name: Matthias Blume
2144 : blume 853 Date: 2001/06/26 16:15:00 EDT
2145 :     Tag: blume-20010626-cmdoc
2146 :     Description:
2147 :    
2148 :     Minor typo fixed in CM manual (syntax diagram for libraries).
2149 :    
2150 :     ----------------------------------------------------------------------
2151 :     Name: Matthias Blume
2152 : blume 852 Date: 2001/06/25 22:55:00 EDT
2153 :     Tag: blume-20010625-x86pc
2154 :     Description:
2155 :    
2156 :     Fixed a nasty bug in the X86 assembly code that caused signal
2157 :     handlers to fail (crash) randomly.
2158 :    
2159 :     ----------------------------------------------------------------------
2160 :     Name: Matthias Blume
2161 : blume 846 Date: 2001/06/25 12:05:00 EDT
2162 :     Tag: blume-20010625-nlffigen
2163 :     Description:
2164 :    
2165 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
2166 :     Nick Carter <nbc@andrew.cmu.edu>.
2167 :    
2168 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2169 :     2. Default names for generated files are now derived from the name of
2170 :     the C file *without its directory*. In particular, this causes generated
2171 :     files to be placed locally even if the C file is in some system directory.
2172 :     3. Default names for generated signatures and structures are also derived
2173 :     from the C file name without its directory. This avoids silly things
2174 :     like "structure GL/GL".
2175 :     (Other silly names are still possible because ml-nlffigen does not do
2176 : blume 847 a thorough check of whether generated names are legal ML identifiers.
2177 :     When in doubt, use command line arguments to force particular names.)
2178 : blume 846
2179 :     ----------------------------------------------------------------------
2180 :     Name: Matthias Blume
2181 : blume 845 Date: 2001/06/21 12:25:00 EDT
2182 :     Tag: blume-20010621-eXene
2183 :     Description:
2184 :    
2185 :     eXene now compiles and (sort of) works again.
2186 :    
2187 :     The library name (for version > 110.33) is $/eXene.cm.
2188 :    
2189 :     I also added an new example in src/eXene/examples/nbody. See the
2190 :     README file there for details.
2191 :    
2192 :     ----------------------------------------------------------------------
2193 :     Name: Matthias Blume
2194 : blume 844 Date: 2001/06/20 16:40:00 EDT
2195 :     Tag: blume-20010620-cml
2196 :     Description:
2197 :    
2198 :     CML now compiles and works again.
2199 :    
2200 :     Libraries (for version > 110.33):
2201 :    
2202 :     $cml/cml.cm Main CML library.
2203 :     $cml/basis.cm CML's version of $/basis.cm.
2204 :     $cml/cml-internal.cm Internal helper library.
2205 :     $cml/core-cml.cm Internal helper library.
2206 :     $cml-lib/trace-cml.cm Tracing facility.
2207 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2208 :    
2209 :     The installer (config/install.sh) has been taught how to properly
2210 :     install this stuff.
2211 :    
2212 :     ----------------------------------------------------------------------
2213 :     Name: Matthias Blume
2214 : blume 843 Date: 2001/06/19 17:55:00 EDT
2215 :     Tag: blume-20010619-instantiate
2216 :     Description:
2217 :    
2218 :     This un-breaks the fix for bug 1432.
2219 :     (The bug was originally fixed in 110.9 but I broke it again some
2220 :     time after that.)
2221 :    
2222 :     ----------------------------------------------------------------------
2223 :     Name: Matthias Blume
2224 : blume 842 Date: 2001/06/19 17:25:00 EDT
2225 :     Tag: blume-20010619-signals
2226 :     Description:
2227 :    
2228 :     This should (hopefully) fix the long-standing signal handling bug.
2229 :     (The runtime system was constructing a continuation record with an
2230 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2231 :    
2232 :     ----------------------------------------------------------------------
2233 :     Name: Matthias Blume
2234 : blume 840 Date: 2001/06/15 15:05:00 EDT
2235 : blume 841 Tag: blume-20010615-moresparc
2236 :     Description:
2237 :    
2238 :     Here is a short late-hour update related to Sparc c-calls:
2239 :    
2240 :     -- made handling of double-word arguments a bit smarter
2241 :    
2242 :     -- instruction selection phase tries to collapse certain clumsily
2243 :     constructed ML-Trees; typical example:
2244 :    
2245 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2246 :    
2247 :     This currently has no further impact on SML/NJ since mlriscGen does
2248 :     not seem to generate such patterns in the first place, and c-calls
2249 :     (which did generate them in the beginning) has meanwhile been fixed
2250 :     so as to avoid them as well.
2251 :    
2252 :     ----------------------------------------------------------------------
2253 :     Name: Matthias Blume
2254 :     Date: 2001/06/15 15:05:00 EDT
2255 : blume 840 Tag: blume-20010615-sparc
2256 :     Description:
2257 :    
2258 :     The purpose of this update is to provide an implementation of NLFFI
2259 :     on Sparc machines.
2260 :    
2261 :     Here are the changes in detail:
2262 :    
2263 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2264 :     the Sparc implementation of the c-calls API.
2265 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2266 :     for accessing the ML frame. Thus, we have a real frame pointer and
2267 :     can freely modify %sp without need for an omit-frame-ptr phase.
2268 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2269 :     this case.
2270 :     * ml-nlffigen has been taught to produce code for different architectures
2271 :     and calling conventions.
2272 :     * In a way similar to what was done in the x86 case, the Sparc
2273 :     backend uses its own specific extension to mltree. (For example,
2274 :     it needs to be able to generate UNIMP instructions which are part
2275 :     of the calling convention.)
2276 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2277 :     to make it easier to plug in new machine- and os-dependent parts).
2278 :    
2279 :     There are some other fairly unrelated bug fixes and cleanups as well:
2280 :    
2281 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2282 :     that they properly share their libraries with existing SML/NJ libraries.
2283 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2284 :     follow-up error messages.
2285 :     * Updates to CM/CMB documentation.
2286 :    
2287 :     TODO items:
2288 :    
2289 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2290 :     (The current %o2 is a really bad choice because it is part of the
2291 :     calling conventions, so things might interfere in unexpected ways.)
2292 :    
2293 :     ----------------------------------------------------------------------
2294 :     Name: Matthias Blume
2295 : blume 839 Date: 2001/06/07
2296 :     Tag: blume-20010607-calls
2297 :     Description:
2298 :    
2299 :     A number of internal changes related to C calls and calling conventions:
2300 :    
2301 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2302 :     number of bytes popped implicitly (by the callee). In most cases
2303 :     this field is 0 but on x86/win32 it is some non-zero value. This
2304 :     is information provided for the benefit of the "omit-frameptr" pass.
2305 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2306 :     The instruction selection phase copies its value from the ML-Tree
2307 :     CALL statement.
2308 :     3. On all other architectures, the instruction selection phase checks
2309 :     whether "pops=0" and complains if not.
2310 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2311 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2312 :     up after the call and pops is set to 0. For "stdcall", the caller
2313 :     does nothing, leaving the cleanup to the callee; pops is set to
2314 :     the number of bytes that were pushed onto the stack.
2315 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2316 :     distinguish between "ccall" and "stdcall".
2317 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2318 :     set. (This is needed for implementing the official C calling convention
2319 :     on this architecture.)
2320 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2321 :     work with the latest CM.
2322 :    
2323 :     ----------------------------------------------------------------------
2324 :     Name: Matthias Blume
2325 : blume 838 Date: 2001/06/05 15:10:00 EDT
2326 :     Tag: blume-20010605-cm-index
2327 :     Description:
2328 :    
2329 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2330 :    
2331 :     1. CM can now generate "index files". These are human-readable files
2332 :     that list on a per-.cm-file basis each toplevel symbol defined or
2333 :     imported. The location of the index file for
2334 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2335 :     To enable index-file generation, set CM.Control.generate_index to true
2336 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2337 :    
2338 :     The CM manual has been updated accordingly.
2339 :    
2340 :     2. I made some slight modifications to the c-calls API in MLRISC.
2341 :    
2342 :     a) There is now a callback to support saving/restoring of
2343 :     dedicated but caller-save registers around the actual call
2344 :     instruction.
2345 :     b) One can optionally specify a comment-annotation for the
2346 :     call instruction.
2347 :    
2348 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2349 :     (For example, the comment annotation shows the C prototype of
2350 :     the function being called.)
2351 :    
2352 :     ----------------------------------------------------------------------
2353 :     Name: Matthias Blume
2354 : blume 837 Date: 2001/06/01 13:30:00 EDT
2355 :     Tag: blume-20010601-nlffi-cleanup
2356 :     Description:
2357 :    
2358 :     This is mostly a cleanup of MLFFI stuff:
2359 :    
2360 :     - some signature files have been put into a more exposed place
2361 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2362 :     - ml-nlffigen changed accordingly
2363 :     - tutorial updated
2364 :    
2365 :     Other changes:
2366 :    
2367 :     - author's affiliation in CM manual(s) updated
2368 :     - some more recognized keywords added to Allen's sml.sty
2369 :    
2370 :     ----------------------------------------------------------------------
2371 :     Name: Matthias Blume
2372 : blume 836 Date: 2001/05/25 15:30:00 EDT
2373 :     Tag: blume-20010525-iptr
2374 :     Description:
2375 :    
2376 :     - put the official 110.33-README (as it appears on the ftp server) under
2377 :     CVS
2378 :     - fixed a small bug related to incomplete pointer types in
2379 :     ml-nlffigen
2380 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2381 :     (it does not need the 'f type parameter)
2382 :    
2383 :     ----------------------------------------------------------------------
2384 :     Name: Matthias Blume
2385 : blume 834 Date: 2001/05/23 14:30:00 EDT
2386 :     Tag: Release_110_33
2387 :     Description:
2388 :    
2389 :     New version number (110.33). New bootfiles.
2390 :    
2391 :     ----------------------------------------------------------------------
2392 :     Name: Matthias Blume
2393 : blume 833 Date: 2001/05/22 18:06:00 EDT
2394 :     Tag: blume-20010522-targets
2395 :     Description:
2396 :    
2397 :     Made install.sh use file config/targets.customized if it exists, falling
2398 :     back to config/targets if it doesn't. This way one can have a customized
2399 :     version of the targets file without touching the "real thing", thus
2400 :     eliminating the constant fear of accidentally checking something bogus
2401 :     back into the CVS repository... (File config/targets.customized must
2402 :     not be added to the repository!)
2403 :    
2404 :     ----------------------------------------------------------------------
2405 :     Name: Matthias Blume
2406 : blume 832 Date: 2001/05/22 16:30:00 EDT
2407 :     Tag: blume-20010522-minitut
2408 :     Description:
2409 :    
2410 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2411 :     struct returns.
2412 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2413 :     incomplete, preliminary documentation for NLFFI.
2414 :    
2415 :     ----------------------------------------------------------------------
2416 :     Name: Matthias Blume
2417 : blume 829 Date: 2001/05/14 11:30:00 EDT
2418 :     Tag: blume-20010514-script
2419 :     Description:
2420 :    
2421 :     Some bugs in install script fixed.
2422 :    
2423 :     In addition to that I also made a slight change to the NLFFI API:
2424 :     Functors generated by ml-nlffigen now take the dynamic library as a
2425 :     straight functor argument, not as a suspended one. (The original
2426 :     functor code used to force the suspension right away anyway, so there
2427 :     was nothing gained by this complication of the interface.)
2428 :    
2429 :     ----------------------------------------------------------------------
2430 :     Name: Matthias Blume
2431 : blume 828 Date: 2001/05/11 14:35:00 EDT
2432 :     Tag: blume-20010511-ml-nlffi
2433 :     Description:
2434 :    
2435 :     I finally took the plunge and added my new FFI code to the main
2436 :     repository. For x86-linux it is now ready for prime-time.
2437 :    
2438 :     There are two new subdirectories of "src":
2439 :    
2440 :     - ml-nlffi-lib:
2441 :     The utility library for programs using the FFI interface.
2442 :     Here is the implementation of $/c.cm and its associated low-level
2443 :     partners $/c-int.cm and $/memory.cm.
2444 :     - ml-nlffigen:
2445 :     A stand-alone program for generating ML glue code from C source
2446 :     code.
2447 :    
2448 :     Building ml-nlffigen requires $/ckit-lib.cm.
2449 :    
2450 :     The config/install.sh script has been updates to do the Right Thing
2451 :     (hopefully).
2452 :    
2453 :     Notice that the source tree for the C-Kit will not be put under "src"
2454 :     but directly under the installation root directory. (This is the
2455 :     structure that currently exists on the CVS server when you check out
2456 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
2457 :    
2458 :     Bugs: No documentation yet.
2459 :    
2460 :     ----------------------------------------------------------------------
2461 :     Name: Matthias Blume
2462 : blume 826 Date: 2001/05/09 16:35:00 EDT
2463 :     Tag: blume-20010509-cpscontract
2464 :     Description:
2465 :    
2466 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
2467 :     wrapper/unwrapper elimination did not decrement usage counts and some
2468 :     dead variables got overlooked by the dead-up logic.)
2469 :    
2470 :     ----------------------------------------------------------------------
2471 : george 824 Name: Lal George
2472 :     Date: 2001/05/08 17:26:09 EDT
2473 :     Tag: george-20010508-omit-frameptr
2474 :     Description:
2475 :    
2476 :     Changes to implement the omit-frame-pointer optimization to support
2477 :     raw C calls. For now, there is only support on the Intel x86, but
2478 :     other architectures will follow as more experience is gained with this.
2479 :    
2480 :    
2481 :     ----------------------------------------------------------------------
2482 : blume 818 Name: Matthias Blume
2483 : blume 819 Date: 2001/05/07 14:40:00 EDT
2484 :     Tag: blume-20010507-proxies
2485 :     Description:
2486 :    
2487 :     I made into "proxy libraries" all libraries that qualify for such a
2488 :     change. (A qualifying library is a library that has another library or
2489 :     groups as its sole member and repeats that member's export list
2490 :     verbatim. A proxy library avoids this repetition by omitting its export
2491 :     list, effectively inheriting the list that its (only) member exports.
2492 :     See the CM manual for more explanation.)
2493 :     The main effect is that explicit export lists for these libraries
2494 :     do not have to be kepts in sync, making maintenance a bit easier.
2495 :    
2496 :     I also added copyright notices to many .cm-files.
2497 :    
2498 :     Last but not least, I made a new set of bootfiles.
2499 :    
2500 :     ----------------------------------------------------------------------
2501 :     Name: Matthias Blume
2502 : blume 818 Date: 2001/05/04 17:00:00 EDT
2503 :     Tag: blume-20010504-cm-lsplit
2504 :     Description:
2505 :    
2506 :     0. John merged pending changes to $/smlnj-lib.cm
2507 :    
2508 :     1. Allen's previous change accidentally backed out of one of Lal's
2509 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2510 :    
2511 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2512 :     within the compiler where applicable. There is some code simplification
2513 :     because of that.
2514 :    
2515 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2516 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2517 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2518 :     it can now also be controlled on a per-source basis: CM has been taught
2519 :     a new tool parameter applicable to ML source files.
2520 :    
2521 :     - To turn lambda-splitting off completely:
2522 :     local open Compiler.Control.LambdaSplitting in
2523 :     val _ = set Off
2524 :     end
2525 :     - To make "no lambda-splitting" the global default (but allow per-source
2526 :     overriding); this is the initial setting:
2527 :     local open Compiler.Control.LambdaSplitting in
2528 :     val _ = set (Default NONE)
2529 :     end
2530 :     - To make "lambda-splitting with aggressiveness a" the global default
2531 :     (and allow per-source overriding):
2532 :     local open Compiler.Control.LambdaSplitting in
2533 :     val _ = set (Default (SOME a))
2534 :     end
2535 :    
2536 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2537 :     write (in the respective .cm-file):
2538 :     a.sml (lambdasplitting:off)
2539 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2540 :     a.sml (lambdasplitting:on)
2541 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2542 :     <a> is a decimal non-negative integer):
2543 :     a.sml (lambdasplitting:<a>)
2544 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2545 :     a.sml (lambdasplitting:infinity)
2546 :     - To use the global default for a.sml:
2547 :     a.sml (lambdasplitting:default)
2548 :     or simply
2549 :     a.sml
2550 :    
2551 :     ----------------------------------------------------------------------
2552 : leunga 815 Name: Allen Leung
2553 :     Date: 2001/05/04 01:57:00 EDT
2554 :     Tag: leunga-20010504-sync
2555 :     Description:
2556 : leunga 788
2557 : leunga 815 MLRISC features.
2558 :    
2559 :     1. Fix to CMPXCHG instructions.
2560 :     2. Changed RA interface to allow annotations in callbacks.
2561 :     3. Added a new method to the stream interface to allow annotations updates.
2562 :    
2563 : george 749 ----------------------------------------------------------------------
2564 : blume 812 Name: Matthias Blume
2565 : blume 814 Date: 2001/05/01 11:45:00 EDT
2566 :     Tag: blume-20010501-pcedittmp
2567 :     Description:
2568 :    
2569 :     Changed install.sh to use the current working directory instead of
2570 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2571 :     of /usr/tmp caused trouble with MacOS X because of file premission
2572 :     problems.
2573 :    
2574 :     ----------------------------------------------------------------------
2575 :     Name: Matthias Blume
2576 : blume 812 Date: 2001/04/20 11:10:00 EDT
2577 :     Tag: blume-20010420-inMLflag
2578 :     Description:
2579 :    
2580 :     - added vp_limitPtrMask to vproc-state.h
2581 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2582 :     handling)
2583 :     - made the ML compiler aware of various data-structure offsets so it
2584 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2585 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2586 :     raw C-calls
2587 :    
2588 :     ----------------------------------------------------------------------
2589 : george 810 Name: Lal George
2590 :     Date: 2001/04/20 09:15:28 EDT
2591 :     Tag: george-20010420-macosX
2592 :     Description:
2593 :    
2594 :     - Changes to port to Mac OS X; Darwin.
2595 :    
2596 :     - In the process I found that sqrt was broken on the PPC, because the
2597 :     fsqrt instruction is not implemented.
2598 :    
2599 :     ----------------------------------------------------------------------
2600 : blume 799 Name: Matthias Blume
2601 : blume 808 Date: 2001/04/18 12:45:00 EDT
2602 :     Tag: blume-20010418-ccalls
2603 :     Description:
2604 :    
2605 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2606 :     (this bug prevented structure arguments containing pointers from being
2607 :     passed correctly)
2608 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
2609 :     structure arguments are represented as a pointer to the beginning
2610 :     of the structure (instead of having a series of synthesized arguments,
2611 :     one for each structure member)
2612 :    
2613 :     - made makeml script's verbosity level configurable via environment
2614 :     variable (MAKEML_VERBOSITY)
2615 :    
2616 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2617 : blume 808 in rawmem-x86.sml; we are now using the real thing
2618 :    
2619 :     ----------------------------------------------------------------------
2620 :     Name: Matthias Blume
2621 : blume 806 Date: 2001/03/22 16:25:00 EST
2622 :     Tag: blume-20010322-bootfiles
2623 :     Description:
2624 :    
2625 :     Created a new set of bootfiles (for your automatic installation convenience).
2626 :    
2627 :     ----------------------------------------------------------------------
2628 :     Name: Matthias Blume
2629 : blume 805 Date: 2001/03/22 15:10:00 EST
2630 :     Tag: blume-20010322-rawmem-parcm
2631 :     Description:
2632 :    
2633 :     1. All "raw memory access" primitives for the new FFI are implemented now
2634 :     (at least on the x86).
2635 :     2. Some further cleanup of CM's parallel make mechanism.
2636 :    
2637 :     ----------------------------------------------------------------------
2638 :     Name: Matthias Blume
2639 : blume 801 Date: 2001/03/19 17:53:00 EST
2640 :     Tag: blume-20010319-parallel
2641 :     Description:
2642 :    
2643 :     Parallel make (using compile servers) now works again.
2644 :    
2645 :     To this end, CM.stabilize and CMB.make have been modified to work in
2646 :     two passes when compile servers are attached:
2647 :     1. Compile everything, do not perform stabilization; this pass
2648 :     uses compile servers
2649 :     2. Stabilize everything; this pass does not use compile servers
2650 :     If there are no compile servers, the two passes are combined into one
2651 :     (as before). Splitting the passes increases the inherent parallelism
2652 :     in the dependency graph because the entire graph including all
2653 :     libraries is available at the same time. This, in turn, improves
2654 :     server utilization. The downside is that the master process will
2655 :     have to do some extra work after compilation is done (because for
2656 :     technical reasons it must re-read all the binfiles during stabilization).
2657 :    
2658 :     ----------------------------------------------------------------------
2659 :     Name: Matthias Blume
2660 : blume 800 Date: 2001/03/16 12:22:00 EST
2661 :     Tag: blume-20010316-bootfiles
2662 :     Description:
2663 :    
2664 :     Created a new set of bootfiles (for your automatic installation convenience).
2665 :    
2666 :     ----------------------------------------------------------------------
2667 :     Name: Matthias Blume
2668 : blume 799 Date: 2001/03/16 11:00:00 EST
2669 :     Tag: blume-20010316-MLTREE-fixup
2670 :     Description:
2671 :    
2672 :     This is a minor fixup for an (untagged) earlier commit by Allen.
2673 :     (A file was missing).
2674 :    
2675 :     ----------------------------------------------------------------------
2676 : leunga 788 Name: Allen Leung
2677 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
2678 :     Tag: leunga-20010305-cut-support
2679 :    
2680 :     1. New support for alternative control-flow in MLTREE.
2681 :     Currently we support
2682 :    
2683 :     FLOW_TO(CALL ...., [k1,...,kn])
2684 :    
2685 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
2686 :     in Moby
2687 :    
2688 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2689 :    
2690 :     2. Register Allocator
2691 :    
2692 :     Changes in interface [from Fermin, John]
2693 :    
2694 :     3. Alpha 8-bit SLL support [Fermin]
2695 :    
2696 :     4. All architectures
2697 :    
2698 :     A new module (ClusterExpandCopies) for expanding parallel copies.
2699 :    
2700 :     ----------------------------------------------------------------------
2701 :     Name: Allen Leung
2702 : leunga 788 Date: 2001/02/27 23:07:00 EST
2703 :     Tag: leunga-20010227-minor-stuff
2704 :    
2705 :     1. Alpha bug fix for CMOVNE
2706 :     2. Handle mltree COND(..,FCMP ...,...)
2707 :     3. Bug fix in simplifier
2708 :    
2709 :     ----------------------------------------------------------------------
2710 : blume 777 Name: Matthias Blume
2711 : blume 781 Date: 2001/01/30 17:50:00 EST
2712 :     Tag: blume-20010130-sync
2713 :     Description:
2714 :    
2715 :     This is just a minor update to sync my devel branch with the main brach.
2716 :     The only visible change is the addition of some README files.
2717 :    
2718 :     ----------------------------------------------------------------------
2719 :     Name: Matthias Blume
2720 : blume 778 Date: 2001/01/12 23:30:00 JST
2721 :     Tag: blume-20010112-bootfiles
2722 :     Description:
2723 :    
2724 :     Made a new set of bootfiles that goes with the current state of the
2725 :     repository.
2726 :    
2727 :     ----------------------------------------------------------------------
2728 :     Name: Matthias Blume
2729 : blume 777 Date: 2001/01/12 21:20:00 JST
2730 :     Tag: blume-20010112-sync
2731 :     Description:
2732 :    
2733 :     I am just flushing out some minor changes that had accumulated in
2734 :     my private branch in order to sync with the main tree. (This is
2735 :     mainly because I had CVS trouble when trying to merge _into_ my
2736 :     private branch.)
2737 :    
2738 :     Most people should be completely unaffected by this.
2739 :    
2740 :     ----------------------------------------------------------------------
2741 : leunga 775 Name: Allen Leung
2742 :     Date: Thu Jan 11 21:03:00 EST 2001
2743 :     Tag: leunga-20010111-labexp=mltree
2744 :     Description:
2745 :    
2746 :     1. Removed the type LabelExp and replace it by MLTree.
2747 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2748 :     3. There were some bugs in alpha code generator which would break
2749 :     64-bit code generation.
2750 :     4. Redo the tools to generate code with the
2751 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2752 :     are now generated by perl scripts.
2753 :    
2754 :     ----------------------------------------------------------------------
2755 : blume 771 Name: Matthias Blume
2756 : blume 774 Date: 2001/01/10 21:55:00 JST
2757 :     Tag: blume-20010110-rcc
2758 :     Description:
2759 :    
2760 :     The RCC stuff now seems to work (but only on the x86).
2761 :     This required hacking of the c-calls interface (and -implementation) in
2762 :     MLRISC.
2763 :    
2764 :     Normal compiler users should be unaffected.
2765 :    
2766 :     ----------------------------------------------------------------------
2767 :     Name: Matthias Blume
2768 : blume 773 Date: 2001/01/09 01:20:00 JST
2769 :     Tag: blume-20010109-rcc
2770 :     Description:
2771 :    
2772 :     This is a fairly big patch, flushing out a large number of pending
2773 :     changes that I made to my development copy over the last couple of days.
2774 :    
2775 :     Of practical relevance at this moment is a workaround for a pickling
2776 :     bug that Allen ran into the other day. The cause of the bug itself is
2777 :     still unknown and it might be hard to fix it properly, but the
2778 :     workaround has some merits of its own (namely somewhat reducing pickling
2779 :     overhead for certain libraries). Therefore, I think this solution should
2780 :     be satisfactory at this time.
2781 :    
2782 :     The rest of the changes (i.e., the vast majority) has to do with my
2783 :     ongoing efforts of providing direct support for C function calls from
2784 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2785 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2786 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2787 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2788 :     handle RCC.
2789 :    
2790 :     The last part (the changes to mlriscGen.sml) are still known to be
2791 :     wrong on the x86 and not implemented on all other architectures. But
2792 :     the infrastructure is in place. I had to change a few functor
2793 :     signatures in the backend to be able to route the CCalls interface
2794 :     from MLRISC there, and I had to specialize the mltree type (on the
2795 :     x86) to include the necessary extensions. (The extensions themselves
2796 :     were already there and redy to go in MLRISC/x86).
2797 :    
2798 :     Everything should be very happy as soon as someone helps me with
2799 :     mlriscGen.sml...
2800 :    
2801 :     In any case, nothing of this should matter to anyone as long as the
2802 :     new primop is not being used (which is going to be the case unless you
2803 :     find it where I hid it :). The rest of the compiler is completely
2804 :     unaffected.
2805 :    
2806 :     ----------------------------------------------------------------------
2807 :     Name: Matthias Blume
2808 : blume 772 Date: 2001/01/05 00:30:00 JST
2809 :     Tag: blume-20010105-primops
2810 :     Description:
2811 :    
2812 :     Added some experimental support for work that I am doing right now.
2813 :     These changes mostly concern added primops, but there is also a new
2814 :     experimental C library in the runtime system (but currently not enabled
2815 :     anywhere except on Linux/X86).
2816 :    
2817 :     In the course of adding primops (and playing with them), I discovered that
2818 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2819 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2820 :     To recover, I made type information in INL_PRIM mandatory and changed
2821 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2822 :     has complete, correct type information (i.e., no bottom types).
2823 :    
2824 :     Since all these changes mean that we need new binfiles, I also bumped the
2825 :     version number to 110.32.1.
2826 :    
2827 :     ----------------------------------------------------------------------
2828 :     Name: Matthias Blume
2829 : blume 771 Date: 2000/12/30 22:10:00 JST
2830 :     Tag: blume-20001230-various
2831 :     Description:
2832 :    
2833 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2834 :     to each other using path anchors. (See CM manual for explanation.)
2835 :    
2836 :     Updated CM documentation.
2837 :    
2838 :     Fixed some bugs in CM.
2839 :    
2840 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2841 :    
2842 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2843 :    
2844 :     Added cleanup handler for exportML to reset timers and compiler stats.
2845 :    
2846 :     ----------------------------------------------------------------------
2847 : george 761 Name: Lal George
2848 :     Date: 2000/12/22 22:22:58 EST 2000
2849 :     Tag: Release_110_32
2850 :     Description:
2851 :    
2852 :     Infinite precision used throughout MLRISC.
2853 :     see MLRISC/mltree/machine-int.sig
2854 :    
2855 :     ----------------------------------------------------------------------
2856 : blume 756 Name: Matthias Blume
2857 : blume 760 Date: 2000/12/22 23:16:00 JST
2858 :     Tag: blume-20001222-warn
2859 :     Description:
2860 : blume 759
2861 : blume 760 Corrected wording and formatting of some CM warning message which I
2862 :     broke in my previous patch.
2863 : blume 759
2864 :     ----------------------------------------------------------------------
2865 :     Name: Matthias Blume
2866 : blume 763 Date: 2000/12/22 21:20:00 JST
2867 :     Tag: blume-20001222-anchorenv
2868 :     Description:
2869 :    
2870 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2871 :    
2872 :     ----------------------------------------------------------------------
2873 :     Name: Matthias Blume
2874 : blume 758 Date: 2000/12/22 13:15:00 JST
2875 :     Tag: blume-20001222-cleanup
2876 :     Description:
2877 :    
2878 :     Removed src/cm/ffi which does not (and did not) belong here.
2879 :    
2880 :     ----------------------------------------------------------------------
2881 :     Name: Matthias Blume
2882 : blume 757 Date: 2000/12/21 23:55:00 JST
2883 :     Tag: blume-20001221-exn
2884 :     Description:
2885 :    
2886 :     Probably most important: CM no longer silently swallows all exceptions
2887 :     in the compiler.
2888 :     Plus: some other minor CM changes. For example, CM now reports some
2889 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2890 :    
2891 :     ----------------------------------------------------------------------
2892 :     Name: Matthias Blume
2893 : blume 756 Date: 2000/12/15 00:01:05 JST
2894 :     Tag: blume-20001215-dirtool
2895 :     Description:
2896 :    
2897 :     - "dir" tool added.
2898 :     - improvements and cleanup to Tools structure
2899 :     - documentation updates
2900 :    
2901 :     ----------------------------------------------------------------------
2902 : leunga 752 Name: Allen Leung
2903 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2904 :     Description:
2905 :     Tag: leunga-20001214-int-inf
2906 :     Description:
2907 :    
2908 :     In IntInf, added these standard functions, which are missing from our
2909 :     implementation:
2910 :    
2911 :     andb : int * int -> int
2912 :     xorb : int * int -> int
2913 :     orb : int * int -> int
2914 :     notb : int -> int
2915 :     << : int * word -> int
2916 :     ~>> : int * word -> int
2917 :    
2918 :     Not tested, I hope they are correct.
2919 :    
2920 :     ----------------------------------------------------------------------
2921 :     Name: Allen Leung
2922 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2923 :     Description:
2924 :     Tag: leunga-20001208-nowhere
2925 :     Description:
2926 :    
2927 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2928 :     to generate better error messages etc. Plus a brief manual.
2929 :    
2930 :     ----------------------------------------------------------------------
2931 : george 749 Name: Lal George
2932 :     Date: 2000/12/08 09:54:02 EST 2000
2933 :     Tag: Release_110_31
2934 :     Description:
2935 : george 705
2936 : george 749 - Version 110.31
2937 : leunga 591 ----------------------------------------------------------------------
2938 : leunga 744 Name: Allen Leung
2939 :     Date: Thu Dec 7 22:01:04 EST 2000
2940 :     Tag: leunga-20001207-cell-monster-hack
2941 :     Description:
2942 :    
2943 :     Major MLRISC internal changes. Affect all clients.
2944 :     Summary:
2945 :    
2946 :     1. Type CELLS.cell = int is now replaced by a datatype.
2947 :     As a result, the old regmap is now gone. Almost all interfaces
2948 :     in MLRISC change as a consequence.
2949 :    
2950 :     2. A new brand version of machine description tool (v3.0) that generates
2951 :     modules expecting the new interface. The old version is removed.
2952 :    
2953 :     3. The RA interface has been further abstracted into two new functors.
2954 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2955 :     [See also directory MLRISC/demo.]
2956 :    
2957 :     4. Some other new source->source code generation tools are available:
2958 :    
2959 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2960 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2961 :     I use this tool to generate the peephole optimizers---with the new
2962 :     cell type changes, peephole rules are becoming difficult to write
2963 :     without conditional pattern matching.
2964 :    
2965 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2966 :     cover the entire MLRISC source tree so many things broke.
2967 :    
2968 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2969 :     They are moved because there are a lot of them and they clutter up the
2970 :     root dir.
2971 :    
2972 :     7. More detailed documentation to come...
2973 :    
2974 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2975 :     a makeml -rebuild first. This is because of other other
2976 :     changes that Matthias has made (see below).
2977 :    
2978 :    
2979 :     ----------------------------------------------------------------------
2980 : blume 733 Name: Matthias Blume
2981 : blume 742 Date: 2000/11/30 23:12:00 JST
2982 :     Tag: blume-20001130-filereorg
2983 :     Description:
2984 :    
2985 :     Some manual updates and some file reorganizations in CM.
2986 :    
2987 :     ----------------------------------------------------------------------
2988 :     Name: Matthias Blume
2989 : blume 737 Date: 2000/11/24 17:45:00 JST
2990 :     Tag: blume-20001124-link
2991 :     Description:
2992 :    
2993 :     Drastically improved link traversal code for the case that the dynamic
2994 :     value was already loaded at bootstrap time. As a result, CM and CMB
2995 :     now both load blazingly fast -- even on a very slow machine. Also,
2996 :     memory consumption has been further reduced by this.
2997 :    
2998 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2999 :     bootstrap you have to do this:
3000 :    
3001 :     1. Run CMB.make
3002 :     2. Make a symbolic link for the boot directory:
3003 :     ln -s sml.boot.ARCH-OS xxx
3004 :     3. "Rebuild" the boot directory:
3005 :     ./makeml -boot xxx -rebuild sml ; rm xxx
3006 :     4. Boot normally:
3007 :     ./makeml
3008 :    
3009 :     ----------------------------------------------------------------------
3010 :     Name: Matthias Blume
3011 : blume 735 Date: 2000/11/21 21:20:00 JST
3012 :     Tag: blume-20001121-tools
3013 :     Description:
3014 :    
3015 :     Continued hacking on autoloading problem -- with success this time.
3016 :     Also changed tool-plugin mechanism. See new CM manual.
3017 :    
3018 :     ----------------------------------------------------------------------
3019 :     Name: Matthias Blume
3020 : blume 734 Date: 2000/11/19 14:30:00 JST
3021 :     Tag: blume-20001119-autoload
3022 :     Description:
3023 :    
3024 :     Some hacking to make autoloading faster. Success for CMB, no success
3025 :     so far for CM. There is a reduced structure CM' that autoloads faster.
3026 :     (This is a temporary, non-documented hack to be eliminated again when
3027 :     the general problem is solved.)
3028 :    
3029 :     ----------------------------------------------------------------------
3030 :     Name: Matthias Blume
3031 : blume 733 Date: 2000/11/17 14:10:00 JST
3032 :     Tag: blume-20001117-pickle-lib
3033 :     Description:
3034 :    
3035 :     1. Eliminated comp-lib.cm
3036 :     2. Made pickle-lib.cm
3037 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
3038 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
3039 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
3040 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
3041 :     Duplicated intmap modules thrown out.
3042 :     6. Hunted down all duplicated SCC code and replaced it with
3043 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
3044 :     7. Rewrote Feedback module.
3045 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
3046 :     should be thrown out and equivalent modules from smlnj-lib.cm
3047 :     should be used (IntRedBlackSet, IntListSet, ...).
3048 :    
3049 :     Confirmed that compiler compiles to fixpoint.
3050 :    
3051 :     ----------------------------------------------------------------------
3052 : leunga 731 Name: Allen Leung
3053 :     Date: 2000/11/10 18:00:00
3054 :     Tag: leunga-20001110-new-x86-fp
3055 :    
3056 :     A new x86 floating point code generator has been added.
3057 :     By default this is turned off. To turn this on, do:
3058 :    
3059 :     CM.autoload "$smlnj/compiler.cm";
3060 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3061 :    
3062 :     Changes:
3063 :    
3064 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
3065 :     2. Changed the extension callback for FTANGENT to generate:
3066 :    
3067 :     fptan
3068 :     fstp %st(0)
3069 :     instead of
3070 :     fptan
3071 :     fstpl ftempmem
3072 :    
3073 :     3. Numerous assembly fixes for x86.
3074 :    
3075 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
3076 :     support for a whole bunch of instructions and addressing modes:
3077 :    
3078 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
3079 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
3080 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
3081 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3082 :     fxch %st(n)
3083 :     fld %st(n)
3084 :     fst %st(n)
3085 :     fst mem
3086 :     fstp %st(n)
3087 :     fucom %st(n)
3088 :     fucomp %st(n)
3089 :    
3090 :     All these are now generated when the fast fp mode is turned on.
3091 :    
3092 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3093 :    
3094 :     ----------------------------------------------------------------------
3095 : blume 729 Name: Matthias Blume
3096 :     Date: 2000/11/09 11:20:00 JST
3097 :     Tag: blume-20001109-scc
3098 :     Description:
3099 :    
3100 :     Eliminated some code duplication:
3101 :    
3102 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
3103 :     (Otherwise the functor is useless.)
3104 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
3105 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
3106 :    
3107 :     ----------------------------------------------------------------------
3108 : george 721 Name: Lal George
3109 :     Date: 2000/11/06 09:02:21 EST 2000
3110 :     Tag: Release_110_30
3111 :     Description:
3112 :    
3113 :     - Version 110.30
3114 :     ----------------------------------------------------------------------
3115 : blume 715 Name: Matthias Blume
3116 : blume 716 Date: 2000/11/04 14:45:00
3117 :     Tag: blume-20001104-mlbuild
3118 :     Description:
3119 :    
3120 :     - Made ml-build faster on startup.
3121 :     - Documentation fixes.
3122 :    
3123 :     ----------------------------------------------------------------------
3124 :     Name: Matthias Blume
3125 : blume 715 Date: 2000/11/02 17:00:00 JST
3126 :     Tag: blume-20001102-condcomp
3127 :     Description:
3128 :    
3129 :     - Small tweaks to pickler -- new BOOTFILES!
3130 :     - Version bumped to 110.29.2.
3131 :     - Added conditional compilation facility to init.cmi (see comment there).
3132 :     ----------------------------------------------------------------------
3133 : leunga 713 Name: Allen Leung
3134 :     Date: 2000/10/23 19:31:00
3135 :     Tag: leunga-20001023-demo-ra
3136 :    
3137 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
3138 :     2. Test programs for the graph library updated
3139 :     3. Some new MLRISC demo programs added
3140 :    
3141 :     ----------------------------------------------------------------------
3142 : blume 710 Name: Matthias Blume
3143 : blume 711 Date: 2000/08/31 22:15:00 JST
3144 :     Tag: blume-20001017-errmsg
3145 :     Description:
3146 :    
3147 :     More error message grief: Where there used to be no messages, there
3148 :     now were some that had bogus error regions. Fixed.
3149 :    
3150 :     ----------------------------------------------------------------------
3151 :     Name: Matthias Blume
3152 : blume 710 Date: 2000/08/31 17:30:00 JST
3153 :     Tag: blume-20001017-v110p29p1
3154 :     Description:
3155 :    
3156 :     I made a version 110.29.1 with new bootfiles.
3157 :    
3158 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
3159 :     CM documentation changes and a small bugfix in CM's error reporting.
3160 :    
3161 :     ----------------------------------------------------------------------
3162 : george 705 Name: Lal George
3163 :     Date: 2000/09/27 14:42:35 EDT
3164 :     Tag: george-20000927-nodestatus
3165 :     Description:
3166 :    
3167 :     Changed the type of the nodestatus, so that:
3168 :    
3169 :     SPILLED(~1) is now SPILLED
3170 :     SPILLED(m) where m>=0 is now MEMREG(m)
3171 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
3172 :    
3173 :     ----------------------------------------------------------------------
3174 : blume 697 Name: Matthias Blume
3175 : blume 703 Date: 2000/09/07 14:45:00 JST
3176 :     Tag: blume-20000907-cmerrmsg
3177 :     Description:
3178 :    
3179 :     Small tweak to CM to avoid getting ML syntax error messages twice.
3180 :    
3181 :     ----------------------------------------------------------------------
3182 :     Name: Matthias Blume
3183 : blume 701 Date: 2000/08/31 18:00:00 JST
3184 :     Tag: blume-20000831-cvsbootfiles
3185 :     Description:
3186 :    
3187 :     New URL for boot files (because the 110.29 files on the BL server do
3188 :     now work correctly with my updated install scripts for yacc and lex).
3189 :    
3190 :     ----------------------------------------------------------------------
3191 :     Name: Matthias Blume
3192 : blume 697 Date: 2000/08/08 12:33:00 JST
3193 :     Tag: blume-20000808-manual
3194 :     Description:
3195 :    
3196 :     Tiny update to CM manual.
3197 :    
3198 :     ----------------------------------------------------------------------
3199 : leunga 695 Name: Allen Leung
3200 :     Date: 2000/08/7 19:31:00
3201 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3202 :    
3203 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3204 :     and Mobdy.
3205 :    
3206 :     1. x86
3207 :    
3208 :     a. Fixes to peephole module by John and Dan.
3209 :     b. Assembly fix to SETcc by Allen.
3210 :     c. Fix to c-call by John.
3211 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3212 :     e. Instruction selection optimization to SETcc as suggested by John.
3213 :    
3214 :     For example,
3215 :    
3216 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3217 :    
3218 :     should generate:
3219 :    
3220 :     MOVL a, x
3221 :     SUBL b, x
3222 :     SHRL 31, x
3223 :    
3224 :     2. IR stuff
3225 :    
3226 :     A bunch of new DJ-graph related algorithms added. These
3227 :     speed up SSA construction.
3228 :    
3229 :     3. SSA + Scheduling
3230 :    
3231 :     Added code for SSA and scheduling to the repository
3232 :    
3233 :     ----------------------------------------------------------------------
3234 : george 694 Name: Lal George
3235 :     Date: 2000/07/27 11:53:14 EDT
3236 :    
3237 :     Tag: lal-20000727-linux-ppc
3238 :     Description:
3239 :    
3240 :     Made changes to support Linux PPC.
3241 :     p.s. I have confirmation that the 110.29 boot files work fine.
3242 :    
3243 :     ----------------------------------------------------------------------
3244 : blume 684 Name: Matthias Blume
3245 : blume 692 Date: 2000/07/27 17:40:00 JST
3246 :     Tag: blume-20000727-scripts
3247 :     Description:
3248 :    
3249 :     !!!! WARNING !!!!
3250 :     You must recompile the runtime system!
3251 :     !!!! WARNING !!!!
3252 :    
3253 :     This is basically another round of script-enhancements:
3254 :    
3255 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3256 :     and undefine CM preprocessor symbols.
3257 :    
3258 :     2. ml-build avoids generating a new heap image if it finds that the
3259 :     existing one is still ok. (The condition is that no ML file had to
3260 :     be recompiled and all ML files are found to be older that the heap
3261 :     file.)
3262 :    
3263 :     To make this work smoothly, I also hacked the runtime system as
3264 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3265 :     (.sparc-solaris, ...) that is currently being used.
3266 :    
3267 :     Moreover, the signature of CM.mk_standalone has changed. See the
3268 :     CM manual.
3269 :    
3270 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3271 :     CM manual for details.)
3272 :    
3273 :     4. More CM manual updates:
3274 :     - all of the above has been documented.
3275 :     - there is now a section describing the (CM-related) command line
3276 :     arguments that are accepted by the "sml" command
3277 :    
3278 :     ----------------------------------------------------------------------
3279 :     Name: Matthias Blume
3280 : blume 691 Date: 2000/07/25 16:20:00 JST
3281 :     Tag: blume-20000725-makedepend
3282 :     Description:
3283 :    
3284 :     Added a script called ml-makedepend. This can be used in makefiles
3285 :     for Unix' make in a way very similar to the "makedepend" command for
3286 :     C.
3287 :    
3288 :     The script internally uses function CM.sources.
3289 :    
3290 :     Synopsis:
3291 :    
3292 :     ml-makedepend [-f makefile] cmfile targetname
3293 :    
3294 :     The default for the makefile is "makefile" (or "Makefile" should
3295 :     "makefile" not exist).
3296 :    
3297 :     ml-makedepend adds a cmfile/targetname-specific section to this
3298 :     makefile (after removing the previous version of this section). The
3299 :     section contains a single dependency specification with targetname on
3300 :     the LHS (targetname is an arbitrary name), and a list of files derived
3301 :     from the cmfile on the RHS. Some of the files on the RHS are
3302 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3303 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3304 :     path names. The makefile writer is responsible for making sure that
3305 :     these variables have correct at the time "make" is invoked.
3306 :    
3307 :     ----------------------------------------------------------------------
3308 :     Name: Matthias Blume
3309 : blume 690 Date: 2000/07/22 23:30:00 JST
3310 :     Tag: blume-20000722-urlupdate
3311 :     Description:
3312 :    
3313 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3314 :    
3315 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3316 :    
3317 :     ----------------------------------------------------------------------
3318 :     Name: Matthias Blume
3319 : blume 686 Date: 2000/07/18 18:00:00 JST
3320 :     Tag: blume-20000718-Version_110_29
3321 :     Description:
3322 :    
3323 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3324 :    
3325 :     2. Updated config/version to 110.29
3326 :    
3327 :     3. Updated config/srcarchiveurl
3328 :    
3329 :     3. New boot files!
3330 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3331 :    
3332 :     ----------------------------------------------------------------------
3333 :     Name: Matthias Blume
3334 : blume 684 Date: 2000/07/11 13:58:00 JST
3335 :     Tag: blume-20000711-doctypo
3336 :     Description:
3337 :    
3338 :     Fixed a few typos in CM manual.
3339 :    
3340 :     ----------------------------------------------------------------------
3341 : leunga 683 Name: Allen Leung
3342 :     Date: 2000/06/15 00:38:00
3343 :     Tag: leunga-20000704-sparc-x86
3344 :    
3345 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3346 :     2. fix to x86 RET bug [found by Dan Grossman]
3347 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3348 :    
3349 :     Affects c-- and moby only
3350 :    
3351 :     ----------------------------------------------------------------------
3352 : blume 680 Name: Matthias Blume
3353 : blume 682 Date: 2000/07/04 15:26:00
3354 :     Tag: blume-20000704-trigger
3355 :     Description:
3356 :    
3357 :     1. Improvements to CM manual.
3358 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3359 :     of getting a back-trace. The function, when called, raises an
3360 :     internal exception which explicitly carries the full back-trace history,
3361 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3362 :     or not). The interactive loop will print that history once it arrives
3363 :     at top level.
3364 :     Short of having all exceptions implicitly carry the full history, the
3365 :     recommended way of using this facility is:
3366 :     - compile your program with instrumentation "on"
3367 :     - run it, when it raises an exception, look at the history
3368 :     - if the history is "cut off" because of some handler, go and modify
3369 :     your program so that it explicitly calls BTrace.trigger
3370 :     - recompile (still instrumented), and rerun; look at the full history
3371 :    
3372 :     ----------------------------------------------------------------------
3373 :     Name: Matthias Blume
3374 :     Date: 2000/07/03 15:36:00 JST
3375 : blume 680 Tag: blume-20000702-manual
3376 :     Description:
3377 :    
3378 :     Small corrections and updates to CM manual.
3379 :    
3380 :     ----------------------------------------------------------------------
3381 : blume 682 Name: Matthias Blume
3382 : blume 679 Date: 2000/06/29 16:04:00 JST
3383 :     Tag: blume-20000629-yacctool
3384 :     Description:
3385 :    
3386 :     Changes:
3387 :    
3388 :     1. Class "mlyacc" now takes separate arguments to pass options to
3389 :     generated .sml- and .sig-files independently.
3390 :     2. Corresponding CM manual updates.
3391 :     3. BTrace module now also reports call sites. (However, for loop clusters
3392 :     it only shows from where the cluster was entered.) There are associated
3393 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3394 :    
3395 :     ----------------------------------------------------------------------
3396 : blume 675 Name: Matthias Blume
3397 : blume 678 Date: 2000/06/27 16:51:00 JST
3398 :     Tag: blume-20000627-noweb
3399 :     Description:
3400 :    
3401 :     Changes:
3402 :    
3403 :     1. Implemented "subdir" and "witness" options for noweb tool.
3404 :     This caused some slight internal changes in CM's tool implementation.
3405 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3406 :     some remaining issues from earlier path anchor changes.
3407 :     3. Updated CM manual accordingly.
3408 :    
3409 :     4. Changed implementation of back-tracing so that I now consider it
3410 :     ready for prime-time.
3411 :    
3412 :     In particular, you don't have to explicitly trigger the back-trace
3413 :     anymore. Instead, if you are running BTrace-instrumented code and
3414 :     there is an uncaught exception (regardless of whether or not it was
3415 :     raised in instrumented code), the top-level evalloop will print
3416 :     the back-trace.
3417 :    
3418 :     Features:
3419 :    
3420 :     - Instrumented and uninstrumented code work together seemlessly.
3421 :     (Of course, uninstrumented code is never mentioned in actual
3422 :     back-traces.)
3423 :    
3424 :     - Asymptotic time- and space-complexity of instrumented code is
3425 :     equal to that of uninstrumented code. (This means that
3426 :     tail-recursion is preserved by the instrumentation phase.)
3427 :    
3428 :     - Modules whose code has been instrumented in different sessions
3429 :     work together without problem.
3430 :    
3431 :     - There is no penalty whatsoever on uninstrumented code.
3432 :    
3433 :     - There is no penalty on "raise" expressions, even in
3434 :     instrumented code.
3435 :    
3436 :     A potential bug (or perhaps it is a feature, too):
3437 :    
3438 :     A back-trace reaches no further than the outermost instrumented
3439 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3440 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3441 :    
3442 :     <exp> handle e => raise e
3443 :    
3444 :     and even
3445 :    
3446 :     <exp> handle Foo => raise Bar
3447 :    
3448 :     and, of course, through
3449 :    
3450 :     <exp> handle Foo => ...
3451 :    
3452 :     if the exception was not Foo.
3453 :    
3454 :     Back-traces always reach right through any un-instrumented code
3455 :     including any of its "handle" expressions, trivial or not.
3456 :    
3457 :     To try this out, do the following:
3458 :    
3459 :     - Erase all existing binfiles for your program.
3460 :     (You may keep binfiles for those modules where you think you
3461 :     definitely don't need back-tracing.)
3462 :     - Turn on back-trace instrumentation:
3463 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3464 :     - Recompile your program. (I.e., run "CM.make" or "use".)
3465 :     - You may now turn instrumentation off again (if you want):
3466 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3467 :     - Run your program as usual. If it raises an exception that
3468 :     reaches the interactive toplevel, then a back-trace will
3469 :     automatically be printed. After that, the toplevel loop
3470 :     will print the exception history as usual.
3471 :    
3472 :     ----------------------------------------------------------------------
3473 :     Name: Matthias Blume
3474 : blume 677 Date: 2000/06/26 09:56:46 JST
3475 :     Tag: blume-20000626-setup
3476 :     Description:
3477 :    
3478 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
3479 :     ML code before and after compiling a file (e.g., to set compiler
3480 :     flags)
3481 :    
3482 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
3483 :     - associated changes to btrace.sml (BTrace instrumentation pass)
3484 :     - cleaner implementation of btimp.sml (BTrace tracing and report
3485 :     module)
3486 :    
3487 :     CM manual: * new path encoding documented
3488 :     * description of setup-parameter to "sml" added
3489 :    
3490 :     The biggest user-visible change to back-tracing is that it is no
3491 :     longer necessary to compile all traced modules within the same
3492 :     session. (This was a real limitation.)
3493 :    
3494 :     ----------------------------------------------------------------------
3495 :     Name: Matthias Blume
3496 : blume 676 Date: 2000/06/24 12:40:00 JST
3497 :     Tag: blume-20000624-startup
3498 :     Description:
3499 :    
3500 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
3501 :     bit too often -- to put it mildly. :)
3502 :    
3503 :     ----------------------------------------------------------------------
3504 :     Name: Matthias Blume
3505 : blume 675 Date: 2000/06/23 18:20:00 JST
3506 :     Tag: blume-20000623-btrace
3507 :     Description:
3508 :    
3509 :     This updates adds a backtrace facility to aid programmers in debugging
3510 :     their programs. This involves the following changes:
3511 :    
3512 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3513 :     keeping track of the current call stack. When programs are compiled
3514 :     in a special mode, the compiler will insert calls to these hooks
3515 :     into the user program.
3516 :     "Hook" means that it is possible for different implementations of
3517 :     back-tracing to register themselves (at different times).
3518 :    
3519 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3520 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3521 :     It can be turned on using this call:
3522 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3523 :     Turning it off again:
3524 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3525 :     Querying the current status:
3526 :     SMLofNJ.Internals.BTrace.mode NONE;
3527 :     Annotated programs are about twice as big as normal ones, and they
3528 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3529 :     where all hooks do nothing). The slowdown with a plugin that is
3530 :     actually useful (such as the one supplied by default) is even greater,
3531 :     but in the case of the default plugin it is still only an constant
3532 :     factor (amortized).
3533 :    
3534 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3535 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3536 :     the above-mentioned function "mode" controls whether the annotation
3537 :     phase is invoked by the compiler. Another important function is
3538 :     "trigger": when called it aborts the current execution and causes
3539 :     the top-level loop to print a full back-trace.
3540 :    
3541 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3542 :     for back-tracing. It keeps track of the dynamic call stack and in
3543 :     addition to that it keeps a partial history at each "level" of that
3544 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3545 :     d and b (at separate times, dynamically), then the report will show:
3546 :    
3547 :     GOTO d
3548 :     /c
3549 :     GOTO \b
3550 :     CALL a
3551 :    
3552 :     This shows that there was an initial non-tail call of a, then a
3553 :     tail-call to b or c, looping behavior in a cluster of functions that
3554 :     consist of b and c, and then a goto from that cluster (i.e., either from
3555 :     b or from c) to d.
3556 :    
3557 :     Note that (depending on the user program) the amount of information
3558 :     that the back-trace module has to keep track of at each level is bounded
3559 :     by a constant. Thus, the whole implementation has the same asymptotical
3560 :     complexity as the original program (both in space and in time).
3561 :    
3562 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3563 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3564 :     which is raised by the "trigger" function mentioned above.
3565 :    
3566 :     Notes on usage:
3567 :    
3568 :     - Annotated code works well together with unannotated code:
3569 :     Unannotated calls simply do not show up at all in the backtrace.
3570 :    
3571 :     - It is not a good idea to let modules that were annotated during
3572 :     different sessions run at the same time. This is because the compiler
3573 :     chooses small integers to identify individual functions, and there
3574 :     will be clashes if different modules were compiled in separate sessions.
3575 :     (Nothing will crash, and you will even be told about the clashes, but
3576 :     back-trace information will in general not be useful.)
3577 :    
3578 :     - Back-tracing can be confused by callcc and capture.
3579 :    
3580 :     - The only way of getting a back-trace right now is to explicitly
3581 :     invoke the "trigger" function from your user program. Eventually, we
3582 :     should make every exception carry back-trace information (if
3583 :     available). But since this creates more overhead at "raise"-time
3584 :     (similar to the current exnHistory overhead), I have not yet
3585 :     implemented this. (The implementation will be rather easy.) With
3586 :     exceptions carrying back-trace information, this facility will be even
3587 :     more useful because users don't need to modify their programs...
3588 :    
3589 :     - While it is possible to compile the compiler with back-trace
3590 :     annotations turned on (I did it to get some confidence in
3591 :     correctness), you must make absolutely sure that core.sml and
3592 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3593 :     be compiled with annotation because there is no core access yet, but
3594 :     if you compile btimp.sml with annotation, then the system will go into
3595 :     an infinite recursion and crash.)
3596 :     Since CM currently does not know about BTrace, the only way to turn
3597 :     annotations on and off for different modules of the compiler is to
3598 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3599 :     this is awkward and clumsy.
3600 :    
3601 :     Sample sessions:
3602 :    
3603 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3604 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3605 :     [autoloading]
3606 :     [autoloading done]
3607 :     val it = false : bool
3608 :     - structure X = struct
3609 :     - fun main n = let
3610 :     - fun a (x, 0) = d x
3611 :     - | a (x, n) = b (x, n - 1)
3612 :     - and b (x, n) = c (x, n)
3613 :     - and c (x, n) = a (x, n)
3614 :     - and d x = e (x, 3)
3615 :     - and e (x, 0) = f x
3616 :     - | e (x, n) = e (x, n - 1)
3617 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
3618 :     - | f n = n * g (n - 1)
3619 :     - and g n = a (n, 3)
3620 :     - in
3621 :     - f n
3622 :     - end
3623 :     - end;
3624 :     structure X : sig val main : int -> int end
3625 :     - X.main 3;
3626 :     *** BACK-TRACE ***
3627 :     GOTO stdIn:4.2-13.20: X.main[2].f
3628 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3629 :     GOTO stdIn:4.2-13.20: X.main[2].d
3630 :     / stdIn:4.2-13.20: X.main[2].a
3631 :     | stdIn:4.2-13.20: X.main[2].b
3632 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3633 :     CALL stdIn:4.2-13.20: X.main[2].g
3634 :     GOTO stdIn:4.2-13.20: X.main[2].f
3635 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3636 :     GOTO stdIn:4.2-13.20: X.main[2].d
3637 :     / stdIn:4.2-13.20: X.main[2].a
3638 :     | stdIn:4.2-13.20: X.main[2].b
3639 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3640 :     CALL stdIn:4.2-13.20: X.main[2].g
3641 :     GOTO stdIn:4.2-13.20: X.main[2].f
3642 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3643 :     GOTO stdIn:4.2-13.20: X.main[2].d
3644 :     / stdIn:4.2-13.20: X.main[2].a
3645 :     | stdIn:4.2-13.20: X.main[2].b
3646 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3647 :     CALL stdIn:4.2-13.20: X.main[2].g
3648 :     GOTO stdIn:4.2-13.20: X.main[2].f
3649 :     CALL stdIn:2.15-17.4: X.main[2]
3650 :     -
3651 :    
3652 :     (Note that because of a FLINt bug the above code currently does not
3653 :     compile without BTrace turned on.)
3654 :    
3655 :     Here is another example, using my modified Tiger compiler:
3656 :    
3657 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3658 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3659 :     [autoloading]
3660 :     [autoloading done]
3661 :     val it = false : bool
3662 :     - CM.make "sources.cm";
3663 :     [autoloading]
3664 :     ...
3665 :     [autoloading done]
3666 :     [scanning sources.cm]
3667 :     [parsing (sources.cm):parse.sml]
3668 :     [creating directory CM/SKEL ...]
3669 :     [parsing (sources.cm):tiger.lex.sml]
3670 :     ...
3671 :     [wrote CM/sparc-unix/semant.sml]
3672 :     [compiling (sources.cm):main.sml]
3673 :     [wrote CM/sparc-unix/main.sml]
3674 :     [New bindings added.]
3675 :     val it = true : bool
3676 :     - Main.compile ("../testcases/merge.tig", "foo.out");
3677 :     *** BACK-TRACE ***
3678 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
3679 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3680 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3681 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3682 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3683 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3684 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
3685 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
3686 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
3687 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3688 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
3689 :     CALL main.sml:1.18-118.4: Main.compile[2]
3690 :     -
3691 :    
3692 :     ----------------------------------------------------------------------
3693 : blume 673 Name: Matthias Blumen
3694 :     Date: 2000/06/21 18:00:00 JST
3695 :     Tag: blume-20000621-manual
3696 :     Description:
3697 :    
3698 :     CM manual update: Path environments documented.
3699 :    
3700 :     ----------------------------------------------------------------------
3701 : blume 666 Name: Matthias Blume
3702 : blume 672 Date: 2000/06/19 13:40:00
3703 :     Tag: blume-20000619-manual
3704 :     Description:
3705 :    
3706 :     CM manual and system/README update. This only covers the fact that
3707 :     there are no more implicit anchors. (Path environments and the "bind"
3708 :     option to "cm" have yet to be documented.)
3709 :    
3710 :     ----------------------------------------------------------------------
3711 :     Name: Matthias Blume
3712 : blume 671 Date: 2000/06/19 11:05:00 JST
3713 :     Tag: blume-20000619-chdir-bugfix
3714 :     Description:
3715 :    
3716 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
3717 :    
3718 :     ----------------------------------------------------------------------
3719 :     Name: Matthias Blume
3720 : blume 670 Date: 2000/06/18 22:00:10 JST
3721 :     Tag: blume-20000618-implicit-anchors-really-gone
3722 :     Description:
3723 :    
3724 :     I updates the previous HISTORY entry where I forgot to mention that
3725 :     implicit anchors are no longer with us.
3726 :    
3727 :     The current update also gets rid of the (now useless) controller
3728 :     CM.Control.implicit_anchors.
3729 :    
3730 :     ----------------------------------------------------------------------
3731 :     Name: Matthias Blume
3732 : blume 666 Date: 2000/06/16 17:30:00 JST
3733 :     Tag: blume-20000616-anchorenv
3734 :     Description:
3735 :    
3736 :     This patch implements the long anticipated (just kidding :) "anchor
3737 :     environment" mechanism. In the course of doing this, I also
3738 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3739 :     one should be more robust in certain boundary cases. In any case, it
3740 :     is a lot cleaner than its predecessor (IMHO).
3741 :    
3742 :     This time, although there is yet another boot file format change, I
3743 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3744 :     are necessary and bootstrapping is straightforward. (You cannot read
3745 :     new bootfiles into an old system, but the other way around is no
3746 :     problem.)
3747 :    
3748 :     Visible changes:
3749 :    
3750 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3751 :     longer recognized at all. This means that such path names are not
3752 :     illegal either. For example, the name basis.cm simply refers to a
3753 :     local file called "basis.cm" (i.e, the name is an ordinary path
3754 :     relative to .cm-files directory). Or, to put it differently, only
3755 :     names that start with $ are anchored paths.
3756 : blume 666
3757 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3758 :     vanished.
3759 :    
3760 :     John (Reppy) had critizised this as soon as I originally proposed and
3761 : blume 666 implemented it, but at that time I did not really deeply believe
3762 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3763 :     in another place where it cannot be seen as an abbreviation for
3764 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3765 :     seems to mean (i.e., it "expands" into the corresponding anchor
3766 :     value).
3767 :    
3768 :     However, when paths are used as members in CM description files, it
3769 :     continues to be true that there must be at least another arc after the
3770 :     anchor. This is now enforced separately during semantic analysis
3771 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3772 : blume 666
3773 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3774 :     is a sub-option list of precisely two items -- one labeled "anchor"
3775 :     and the other one labeled "value". As you might expect, "anchor" is
3776 :     used to specify an anchor name to be bound, and "value" specifies what
3777 :     the anchor is being bound to.
3778 :    
3779 :     The value must be a directory name and can be given in either standard
3780 :     syntax (including the possibility that it is itself an anchored path)
3781 :     or native syntax.
3782 :    
3783 :     Examples:
3784 :    
3785 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3786 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3787 :    
3788 :     and so on.
3789 :    
3790 :     The meaning of this is that the .cm-file will be processed with an
3791 :     augmented anchor environment where the given anchor(s) is/are bound to
3792 :     the given values(s).
3793 :    
3794 :     The rationale for having this feature is this: Suppose you are trying
3795 :     to use two different (already stable) libraries a.cm and b.cm (that
3796 :     you perhaps didn't write yourself). Further, suppose each of these
3797 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3798 :     Normally you would now have a problem because the anchor "lib" can not
3799 :     be bound to more than one value globally. Therefore, the project that
3800 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3801 :     place:
3802 :    
3803 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3804 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3805 :    
3806 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3807 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3808 :    
3809 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3810 :     you when you do so at the time of CM.stabilize). Therefore, you can
3811 :     also use an anchored path as the value:
3812 :    
3813 :     a.cm (bind:(anchor:lib value:$a-lib))
3814 :     b.cm (bind:(anchor:lib value:$b-lib))
3815 :    
3816 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3817 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3818 :     itself is always locally bound, setting it globally is no longer
3819 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3820 :     can still be used as a global anchor for separate purposes. As a
3821 :     matter of fact, one can locally define "lib" in terms of a global
3822 :     "lib":
3823 :    
3824 :     a.cm (bind:(anchor:lib value:$lib/a))
3825 :     b.cm (bind:(anchor:lib value:$lib/b))
3826 :    
3827 :     ** 3: The encoding of path names has changed. This affects the way
3828 :     path names are shown in CM's progress report and also the internal
3829 :     protocol encoding used for parallel make.
3830 :    
3831 :     The encoding now uses one or more ':'-separated segments. Each
3832 :     segments corresponds to a file that has been specified relative to the
3833 :     file given by its preceding segment. The first segment is either
3834 :     relative to the CWD, absolute, or anchored. Each segment itself is
3835 :     basically a Unix pathname; all segments but the first are relative.
3836 :    
3837 :     Example:
3838 :    
3839 :     $foo/bar/baz.cm:a/b/c.sml
3840 :    
3841 :     This path denotes the file bar/a/b/c.sml relative to the directory
3842 :     denoted by anchor "foo". Notice that the encoding also includes
3843 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3844 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3845 :     be ignored to get the "real" pathname.
3846 :    
3847 :     To make this fact more obvious, CM puts the names of such "virtual
3848 :     arcs" into parentheses when they appear in progress reports. (No
3849 :     parentheses will appear in the internal protocol encoding.) Thus,
3850 :     what you really see is:
3851 :    
3852 :     $foo/bar/(baz.cm):a/b/c.sml
3853 :    
3854 :     I find this notation to be much more informative than before.
3855 :    
3856 :     Another new feature of the encoding is that special characters
3857 :     including parentheses, colons, (back)slashes, and white space are
3858 :     written as \ddd (where ddd is the decimal encoding of the character).
3859 :    
3860 : blume 670 *** The CM manual still needs to be updated.
3861 :    
3862 : blume 666 ----------------------------------------------------------------------
3863 : leunga 660 Name: Allen Leung
3864 :     Date: 2000/06/15 00:38:00
3865 :     Tag: leunga-20000615-x86-peephole
3866 :    
3867 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3868 :    
3869 :     ----------------------------------------------------------------------
3870 : blume 658 Name: Matthias Blume
3871 :     Date: 2000/06/12 11:40:00
3872 :     Tag: blume-20000612-parmakefix
3873 :     Description:
3874 :    
3875 :     More cleanup after changing the file naming scheme: This time I
3876 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3877 :    
3878 :     ----------------------------------------------------------------------
3879 : leunga 657 Name: Allen Leung
3880 :     Date: 2000/06/09 01:25:00
3881 :     Tag: leunga-20000609-various
3882 :    
3883 :     None of these things should affect normal SML/NJ operations
3884 :    
3885 :     1. Peephole improvements provided by Fermin (c--)
3886 :     2. New annotation DEFUSE for adding extra dependence (moby)
3887 :     3. New X86 LOCK instructions (moby)
3888 :     4. New machine description language for reservation tables (scheduling)
3889 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3890 :     trees etc.)
3891 :     6. I've changed the CM files so that they can work with versions
3892 :     110.0.6, 110.25 and 110.28
3893 :    
3894 :     ----------------------------------------------------------------------
3895 : blume 652 Name: Matthias Blume
3896 : blume 656 Date: 2000/06/09 12:40:00
3897 :     Tag: blume-20000609-log
3898 :     Description:
3899 :    
3900 :     - Removed all(?) remaining RCS Log entries from sources.
3901 :    
3902 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3903 :     anchored paths).
3904 :    
3905 :     ----------------------------------------------------------------------
3906 :     Name: Matthias Blume
3907 : blume 653 Date: 2000/06/07 17:00:00 JST
3908 :     Tag: blume-20000607-no-implicit-anchors
3909 :     Description:
3910 :    
3911 :     1. This update changes the default setting for
3912 :     CM.Control.implicit_anchors from true to false. This means that
3913 :     implicit anchors are no longer permitted by default. I also tried to
3914 :     make sure that nothing else still relies on implicit anchors.
3915 :     (This is the next step on the schedule towards a CM that does not even
3916 :     have the notion of implicit anchors anymore.)
3917 :    
3918 :     2. More CM manual updates.
3919 :    
3920 :     3. I managed to track down and fix the pickling bug I mentioned last
3921 :     time. Because of the previously existing workaround, this entails no
3922 :     immediate practical changes.
3923 :    
3924 :     ----------------------------------------------------------------------
3925 :     Name: Matthias Blume
3926 : blume 652 Date: 2000/06/06 11:15:00 JST
3927 :     Tag: blume-20000606-lazierpickle
3928 :     Description:
3929 :    
3930 :     !!!! NEW BOOT FILES !!!!
3931 :    
3932 :     * The main purpose of this update is to make library pickles lazier in
3933 :     order to reduce the initial space penalty for autoloading a library.
3934 :     As a result, it is now possible to have $smlnj/compiler.cm
3935 :     pre-registered. This should take care of the many complaints or
3936 :     inquiries about missing structure Compiler. This required changes to
3937 :     CM's internal data structures and small tweaks to some algorithms.
3938 :    
3939 :     As a neat additional effect, it is no longer necessary (for the sake
3940 :     of lean heap image files) to distinguish between a "minimal" CM and a
3941 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3942 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3943 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3944 :    
3945 :     To make the life of compiler-hackers easier, "makeml" now also
3946 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3947 :     words, after you bootstrap a new sml for the first time, you will not
3948 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3949 :     around you will still have to do it, though.)
3950 :    
3951 :     * A second change consists of major updates to the CM manual. There
3952 :     are now several appendices with summary information and also a full
3953 :     specification of the CM description file syntax.
3954 :    
3955 :     * In directory src/system I added the script "allcross". This script
3956 :     invokes sml and cross-compiles the compiler for all supported
3957 :     architectures. (Useful when providing a new set of boot files.)
3958 :    
3959 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3960 :     added a small tweak to pickle-util.sml to work around this problem,
3961 :     but it is not a proper fix yet. I will investigate further. (The
3962 :     effect of the bug was an inflation of library pickle size.)
3963 :    
3964 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3965 :    
3966 :     ----------------------------------------------------------------------
3967 : leunga 646 Name: Allen Leung
3968 : leunga 648 Date: 2000/05/25 17:28 EDT
3969 :     Tag: leunga-20000525-ra
3970 :     Description:
3971 :    
3972 :     Fixed a bug in freezing phase of the register allocator.
3973 :    
3974 :     ----------------------------------------------------------------------
3975 :     Name: Allen Leung
3976 : leunga 647 Date: 2000/05/15 22:53 EDT
3977 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3978 :     Description:
3979 :    
3980 :     1. Alpha
3981 :    
3982 :     Slight cleanup. Removed the instruction SGNXL
3983 :    
3984 :     2. X86
3985 :    
3986 :     Added the following instructions to the instruction set:
3987 :    
3988 :     ROLx, RORx,
3989 :     BTx, BTSx, BTLx, BTRx,
3990 :     XCHGx, and variants with the LOCK prefix
3991 :    
3992 :     3. Register Allocation
3993 :    
3994 :     The module ra-rewrite-with-renaming has been improved.
3995 :    
3996 : leunga 647 These have no effect on SML/NJ.
3997 :    
3998 : leunga 646 ----------------------------------------------------------------------
3999 : blume 642 Name: Matthias Blume
4000 : blume 645 Date: 2000/05/15 16:20:00 JST
4001 :     Tag: blume-20000515-lightrebuild
4002 :     Description:
4003 :    
4004 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
4005 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
4006 :     defined. In effect, the command will not build any cross-compiler
4007 :     backends and therefore finish more quickly.
4008 :    
4009 :     The "fixpt" script also takes a "-light" switch to be able to use
4010 :     this new facility while compiling for a fixpoint.
4011 :    
4012 :     2. I replaced all mentions of anchored paths in group owner specifications
4013 :     with simple relative paths (usually starting with "..").
4014 :     The rationale is that a library's internal workings should not be
4015 :     compromised by the lack of some anchor. (An anchor is necessary
4016 :     for someone who wants to refer to the library by an anchored path,
4017 :     but it should not be necessary to build the same library in the first
4018 :     place.)
4019 :    
4020 :     3. I changed the way CM's tool mechanism determines the shell command
4021 :     string used for things like ml-yacc etc. so that it does not break
4022 :     when CM.Control.implicit_anchors is turned off.
4023 :    
4024 :     ----------------------------------------------------------------------
4025 :     Name: Matthias Blume
4026 : blume 644 Date: 2000/05/12 18:20:00 JST
4027 :     Tag: blume-20000512-ml-build
4028 :     Description:
4029 :    
4030 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
4031 :     getting installed properly (by config/install.sh).
4032 :    
4033 :     ----------------------------------------------------------------------
4034 :     Name: Matthias Blume
4035 : blume 643 Date: 2000/05/12 17:30:00 JST
4036 :     Tag: blume-20000512-anchors
4037 :     Description:
4038 :    
4039 :     !!! NEW BOOT FILES !!!
4040 :    
4041 :     This change is in preparation of fading out support for "implicitly
4042 :     anchored path names". I went through all sources and used the
4043 :     explicit (and relatively new) $-notation. See system/README and the
4044 :     CM manual for more info on this.
4045 :    
4046 :     I also modified the anchoring scheme for some things such as "smlnj",
4047 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
4048 :     anchors are more expressive: anchor name and first arc do not have to
4049 :     coincide. This entails the following user-visible change:
4050 :    
4051 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
4052 :     particular, when you fire up sml with a command-line argument, say,
4053 :     e.g.:
4054 :    
4055 :     sml '$smlnj/cmb.cm'
4056 :    
4057 :     At the ML toplevel prompt:
4058 :    
4059 :     CM.autoload "$smlnj/cmb.cm";
4060 :    
4061 :     There is also a new controller in CM.Control that can be used to turn
4062 :     off all remaining support for implicit anchors by saying:
4063 :    
4064 :     CM.autoload "$smlnj/
4065 :     #set CM.Control.implicit_anchors false;
4066 :    
4067 :     This causes CM to reject implicitly anchored paths. This is (for the
4068 :     time being) less permissive than the "final" version where there will
4069 :     be no more such implicit anchors and relative paths will be just that:
4070 :     relative.
4071 :    
4072 :     The next step (version after next version?) will be to make the
4073 :     default for CM.Control.implicit_anchors false. After the dust has
4074 :     settled, I can then produce the "final" version of this...
4075 :    
4076 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
4077 :    
4078 :     ----------------------------------------------------------------------
4079 :     Name: Matthias Blume
4080 : blume 642 Date: 2000/05/11 16:30:00 JST
4081 :     Tag: blume-20000511-sources
4082 :     Description:
4083 :    
4084 :     The main change is that I added function CM.sources as a generalized
4085 :     version of the earlier CM.makedepend. This entails the following
4086 :     additional changes:
4087 :    
4088 :     - CM.makedepend has been dropped.
4089 :    
4090 :     - CM manual has been updated.
4091 :    
4092 :     - TOOLS signature and API have been changed.
4093 :    
4094 :     ----------------------------------------------------------------------
4095 : leunga 640 Name: Allen Leung
4096 : leunga 641 Date: 2000/05/10 21:17 EDT
4097 :     Tag: leunga-20000510-moby-c--ssa
4098 :     Description:
4099 :    
4100 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
4101 :     None of these affect SML/NJ.
4102 :    
4103 :     1. Register Allocation
4104 :    
4105 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
4106 :     This module tries to remove local (i.e. basic block level) redundancies
4107 :     during spilling.
4108 :    
4109 :     b. A new framework for performing region based register allocation.
4110 :     Not yet entirely functional.
4111 :    
4112 :     2. X86
4113 :    
4114 :     a. DefUse for POP was missing the stack pointer [found by Lal]
4115 :     b. Reload for CALL was incorrect in X86Spill [found by John]
4116 :     c. Various fixes in X86Spill so that it can be used correctly for
4117 :     the new spilling module.
4118 :    
4119 :     3. SSA/IR
4120 :    
4121 :     a. New module ir/dj-dataflow.sml implements elimination based
4122 :     data flow analysis.
4123 :    
4124 :     4. MLRiscGen
4125 :    
4126 :     a. Fix for gc type annotation
4127 :    
4128 :     5. MDGen
4129 :    
4130 :     Various fixes for machine description -> ml code translation. For ssa
4131 :     only.
4132 :    
4133 :     ----------------------------------------------------------------------
4134 :     Name: Allen Leung
4135 : leunga 640 Date: 2000/05/08 22:17 EDT
4136 :     Tag: leunga-20000508-labexp
4137 :     Description:
4138 :    
4139 :     Fermin has found a few assembly problems with constant expressions
4140 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
4141 :     which choke on dumb assemblers. This is his fix.
4142 :    
4143 :     ----------------------------------------------------------------------
4144 : dbm 635 Name: Dave MacQueen
4145 :     Date: 2000/04/09 14:00 EDT
4146 :     Tag: dbm-20000502-Version_110_28
4147 :     Description:
4148 :    
4149 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
4150 :    
4151 :     2. Updated config/version to 110.28
4152 :    
4153 :     3. Updated config/srcarchiveurl
4154 :    
4155 :     3. New boot files!
4156 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
4157 :    
4158 :     ----------------------------------------------------------------------
4159 : blume 634 Name: Matthias Blume
4160 :     Date: 2000/05/01 19:05:00 JST
4161 :     Tag: blume-20000501-noweb
4162 :     Description:
4163 :    
4164 :     A new noweb tool has been added. The existing system is entirely
4165 :     unaffected by this, but some CM users have asked for renewed noweb
4166 :     support. Everything is documented in the CM manual.
4167 :    
4168 :     New (plugin) libraries:
4169 :    
4170 :     noweb-tool.cm
4171 :     nw-ext.cm
4172 :    
4173 :     ----------------------------------------------------------------------
4174 : dbm 633 Name: Dave MacQueen
4175 :     Date: 2000/04/30 12:40PM EDT
4176 :     Tag: dbm-20000430-bug_fixes
4177 :     Description:
4178 :    
4179 :     1. Fix for bug 1498
4180 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
4181 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
4182 :     added toRealArray function
4183 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
4184 :     added check for tag Obj.RealArray to array printing case in ppObj
4185 :    
4186 :     2. Fix for bug 1510
4187 :     smlnj/src/compiler/Semant/types/typesutil.sml
4188 :     fixed definition of dummyargs (used by equalTycon) so that
4189 :     dummy args are distinct types
4190 :    
4191 :     ----------------------------------------------------------------------
4192 : blume 629 Name: Matthias Blume
4193 : blume 632 Date: 2000/04/30 01:00:00 JST
4194 :     Tag: blume-20000430-versions
4195 :     Description:
4196 :    
4197 :     1. CM version numbering added. This is an implementation of Lal's
4198 :     proposal for adding version numbers and version checking to .cm
4199 :     files. Lal said that his proposal was just that -- a proposal.
4200 :     For the time being I went ahead and implemented it so that people
4201 :     can comment on it. Everything is completely backward-compatible
4202 :     (except for the stable library format, i.e., new bootfiles!).
4203 :    
4204 :     As usual, see the CM manual for details.
4205 :    
4206 :     2. An alternative syntax for anchored paths has been implemented.
4207 :     Dave has recently voiced the same concerns that I had when I did
4208 :     this, so there should be some support. My take is that eventually
4209 :     I will let support for the current syntax (where anchors are
4210 :     "implicit") fade out in favor of the new, explicit syntax.
4211 :     In order to be backward-compatible, both old and new syntax are
4212 :     currently supported.
4213 :    
4214 :     Again, see the CM manual for details.
4215 :    
4216 :     3. Parallel make is trying to be slightly smarter: When the master
4217 :     process finds a "bottleneck", i.e., when there is only one
4218 :     compilation unit that can be compiled and everybody else is
4219 :     waiting on it, then it will simply compile it directly instead
4220 :     of clumsily telling one of the slaves to do it.
4221 :    
4222 :     4. Support for "unsharing" added. This is necessary in order to be
4223 :     able to have two different versions of the same library running
4224 :     at the same time (e.g., for trying out a new MLRISC while still
4225 :     having the old MLRISC linked into the current compiler, etc.)
4226 :     See the CM manual.
4227 :    
4228 :     5. Simple "makedepend" functionality added for generating Makefile
4229 :     dependency information. (This is rather crude at the moment.
4230 :     Expect some changes here in the future.)
4231 :    
4232 :     6. ".fun" added as a recognized suffix for ML files. Also documented
4233 :     explicitly in the manual that the fallback behavior (unknown suffix
4234 :     -> ML file) is not an official feature!
4235 :    
4236 :     7. Small changes to the pickler for stable libraries.
4237 :    
4238 :     8. Several internal changes to CM (for cleanup/improvement).
4239 :    
4240 :    
4241 :     !!!! NEW BINFILES !!!!
4242 :    
4243 :     ----------------------------------------------------------------------
4244 :     Name: Matthias Blume
4245 : blume 631 Date: 2000/04/28 17:30:00 JST
4246 :     Tag: blume-20000428-pathconfig
4247 :     Description:
4248 :    
4249 :     1. I changed config/install.sh to remove duplicate entries from the
4250 :     lib/pathconfig file at the end. Moreover, the final version of
4251 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
4252 :     in src/system/installml.
4253 :    
4254 :     2. The config/install.sh script now consistently uses relative
4255 :     pathnames in lib/pathconfig whenever the anchor is in the lib
4256 :     directory. (So far this was true for the libraries that come
4257 :     pre-compiled and bundled as part of the bootfiles but not for
4258 :     libraries that are compiled by the script itself.)
4259 :    
4260 :     ----------------------------------------------------------------------
4261 :     Name: Matthias Blume
4262 : blume 629 Date: 2000/04/26 13:10:00 JST
4263 :     Tag: blume-20000426-fun_suffix
4264 :     Description:
4265 :    
4266 :     Added ".fun" as a recognized file name suffix (for ML code).
4267 :    
4268 :     ----------------------------------------------------------------------
4269 : leunga 628 Name: Allen Leung
4270 :     Date: 2000/04/25 17:00:00 EST
4271 :     Tag: leunga-20000425-alpha-ra
4272 :     Description:
4273 :    
4274 :     1. Alpha
4275 :    
4276 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
4277 :     in C--.
4278 :    
4279 :     2. RA
4280 :    
4281 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
4282 :     and the interference graph.
4283 :    
4284 :     ----------------------------------------------------------------------
4285 : dbm 627 Name: Dave MacQueen
4286 :     Date: 2000/04/25/
4287 :     Tag: dbm-20000425-mlyacc_doc_examples
4288 :     Description:
4289 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
4290 :     Updated all three examples in src/ml-yacc/examples to run
4291 :     under 110.* using CM.make.
4292 :    
4293 :     ----------------------------------------------------------------------
4294 : leunga 624 Name: Allen Leung
4295 :     Date: 2000/04/20 23:04:00 EST
4296 :     Tag: leunga-20000420-ssa-c---stuff
4297 :     Description:
4298 :    
4299 :     This update synchronizes my repository with Yale's. Most of these
4300 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
4301 :    
4302 :     1. Register Allocator
4303 :    
4304 :     a. An improvement in the interference graph construction:
4305 :     Given a copy
4306 :    
4307 :     s <- t
4308 :    
4309 : leunga 628 no interference edge between s and t is added for this definition of s.
4310 : leunga 624
4311 :     b. I've added two new spill heuristic modules that Fermin and I developed
4312 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
4313 :     useful for others (Moby?)
4314 :    
4315 :     2. X86
4316 :    
4317 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
4318 :    
4319 :     3. Alpha
4320 :    
4321 :     a. Added the BSR instruction and code generation that goes with it [C--]
4322 :     b. Other fixes too numerous to recount provided by Fermin [C--]
4323 :    
4324 :     4. Regmaps
4325 :    
4326 :     a. The regmaps are not initialized with the identity physical bindings
4327 :     at creation time. This is unneeded.
4328 :    
4329 :     5. MLRISC Optimizations
4330 :    
4331 :