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 1159 - (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 : blume 1159 Name: Matthias Blume
17 :     Date: 2002/03/21 22:20:00 EST
18 :     Tag: blume-20020321-kmp-bugfix
19 :     Description:
20 :    
21 :     Changed the interface to the KMP routine in PreString and fixed
22 :     a minor bug in one place where it was used.
23 :    
24 :     ----------------------------------------------------------------------
25 : leunga 1156 Name: Allen Leung
26 : leunga 1158 Date: 2002/03/21 20:30:00 EST
27 :     Tag: leunga-20020321-cfg
28 :     Description:
29 :    
30 :     Fixed a potential problem in cfg edge splitting.
31 :    
32 :     ----------------------------------------------------------------------
33 :     Name: Allen Leung
34 : leunga 1156 Date: 2002/03/21 17:15:00 EST
35 :     Tag: leunga-20020321-x86-fp-cfg
36 :     Description:
37 :    
38 :     1. Recoded the buggy parts of x86-fp.
39 :    
40 :     a. All the block reordering code has been removed.
41 :     We now depend on the block placement phases to do this work.
42 :    
43 :     b. Critical edge splitting code has been simplified and moved into the
44 :     CFG modules, as where they belong.
45 :    
46 :     Both of these were quite buggy and complex. The code is now much, much
47 :     simpler.
48 :    
49 :     2. X86 backend.
50 :    
51 :     a. Added instructions for 64-bit support. Instruction selection for
52 :     64-bit has not been committed, however, since that
53 :     requires changes to MLTREE which haven't been approved by
54 :     Lal and John.
55 :    
56 :     b. Added support for FUCOMI and FUCOMIP when generating code for
57 :     PentiumPro and above. We only generate these instructions in
58 :     the fast-fp mode.
59 :    
60 :     c. Added cases for JP and JNP in X86FreqProps.
61 :    
62 :     3. CFG
63 :    
64 :     CFG now has a bunch of methods for edge splitting and merging.
65 :    
66 :     4. Machine description.
67 :    
68 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
69 :     description things:
70 :    
71 :     rtl-build.{sig,sml} and hppa.mdl fixed.
72 :    
73 :     NOTE: the machine description stuff in the repository is still broken.
74 :     Again, I can't put my fixes in because that involves
75 :     changes to MLTREE.
76 :    
77 :     ----------------------------------------------------------------------
78 : blume 1145 Name: Matthias Blume
79 : blume 1155 Date: 2002/03/20 15:55:00 EST
80 :     Tag: blume-20020320-kmp
81 :     Description:
82 :    
83 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
84 :     it for String.isSubstring, Substring.isSubstring, and
85 :     Substring.position.
86 :    
87 :     (Might need some stress-testing. Simple examples worked fine.)
88 :    
89 :     ----------------------------------------------------------------------
90 :     Name: Matthias Blume
91 : blume 1152 Date: 2002/03/19 16:37:00 EST
92 :     Tag: blume-20020319-witnesses
93 :     Description:
94 :    
95 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
96 :    
97 :     This implements a generic mechanism for changing constness qualifiers
98 :     anywhere within big C types without resorting to outright "casts".
99 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
100 :     modify the constness at the outermost level.)
101 :     The implementation of "convert" is based on the idea of "witness"
102 :     values -- values that are not used by the operation but whose types
103 :     "testify" to their applicability. On the implementation side, "convert"
104 :     is simply a projection (returning its second curried argument). With
105 :     cross-module inlining, it should not result in any machine code being
106 :     generated.
107 :    
108 :     ----------------------------------------------------------------------
109 :     Name: Matthias Blume
110 : blume 1148 Date: 2002/03/15 16:40:00 EST
111 :     Tag: blume-20020315-basis
112 :     Description:
113 :    
114 :     Provided (preliminary?) implementations for
115 :    
116 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
117 :    
118 :     and
119 :    
120 :     Substring.full
121 :    
122 :     Those are in the Basis spec but they were missing in SML/NJ.
123 :    
124 :     ----------------------------------------------------------------------
125 :     Name: Matthias Blume
126 : blume 1145 Date: 2002/03/14 21:30:00 EST
127 :     Tag: blume-20020314-controls
128 :     Description:
129 :    
130 :     Controls:
131 :     ---------
132 :    
133 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
134 :     it into its own library $/controls-lib.cm. The source tree for
135 :     this is under src/smlnj-lib/Controls.
136 :    
137 :     2. Changed the names of types and functions in this interface, so they
138 :     make a bit more "sense":
139 :    
140 :     module -> registry
141 :     'a registry -> 'a group
142 :    
143 :     3. The interface now deals in ref cells only. The getter/setter interface
144 :     is (mostly) gone.
145 :    
146 :     4. Added a function that lets one register an already-existing ref cell.
147 :    
148 :     5. Made the corresponding modifications to the rest of the code so that
149 :     everything compiles again.
150 :    
151 :     6. Changed the implementation of Controls.MLRISC back to something closer
152 :     to the original. In particular, this module (and therefore MLRISC)
153 :     does not depend on Controls. There now is some link-time code in
154 :     int-sys.sml that registers the MLRISC controls with the Controls
155 :     module.
156 :    
157 :     CM:
158 :     ---
159 :    
160 :     * One can now specify the lambda-split aggressiveness in init.cmi.
161 :    
162 :     ----------------------------------------------------------------------
163 : leunga 1142 Name: Allen Leung
164 :     Date: 2002/03/13 17:30:00 EST
165 :     Tag: leunga-20020313-x86-fp-unary
166 :     Description:
167 :    
168 :     Bug fix for:
169 :    
170 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
171 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
172 :     > - fun f(x,(y,z)) = Real.~ y;
173 :     > [autoloading]
174 :     > [autoloading done]
175 :     > fchsl (%eax), 184(%esp)
176 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
177 :     >
178 :     > uncaught exception Error
179 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
180 :    
181 :     The problem was that the code generator did not generate any fp registers
182 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
183 :     translate the pseudo fp instruction. This only happened with unary fp
184 :     operators in certain situations.
185 :    
186 :     ----------------------------------------------------------------------
187 : blume 1137 Name: Matthias Blume
188 : blume 1140 Date: 2002/03/13 14:00:00 EST
189 :     Tag: blume-20020313-overload-etc
190 :     Description:
191 :    
192 :     1. Added _overload as a synonym for overload for backward compatibility.
193 :     (Control.overloadKW must be true for either version to be accepted.)
194 :    
195 :     2. Fixed bug in install script that caused more things to be installed
196 :     than what was requested in config/targets.
197 :    
198 :     3. Made CM aware of the (_)overload construct so that autoloading
199 :     works.
200 :    
201 :     ----------------------------------------------------------------------
202 :     Name: Matthias Blume
203 : blume 1138 Date: 2002/03/12 22:03:00 EST
204 :     Tag: blume-20020312-url
205 :     Description:
206 :    
207 :     Forgot to update BOOT and srcarchiveurl.
208 :    
209 :     ----------------------------------------------------------------------
210 :     Name: Matthias Blume
211 : blume 1137 Date: 2002/03/12 17:30:00 EST
212 :     Tag: blume-20020312-version110392
213 :     Description:
214 :    
215 :     Yet another version number bump (because of small changes to the
216 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
217 :    
218 :     Changes:
219 :    
220 :     The new pid generation scheme described a few weeks ago was overly
221 :     complicated. I implemented a new mechanism that is simpler and
222 :     provides a bit more "stability": Once CM has seen a compilation
223 :     unit, it keeps its identity constant (as long as you do not delete
224 :     those crucial CM/GUID/* files). This means that when you change
225 : blume 1145 an interface, compile, then go back to the old interface, and
226 : blume 1137 compile again, you arrive at the original pid.
227 :    
228 :     There now also is a mechanism that instructs CM to use the plain
229 :     environment hash as a module's pid (effectively making its GUID
230 :     the empty string). For this, "noguid" must be specified as an
231 :     option to the .sml file in question within its .cm file.
232 :     This is most useful for code that is being generated by tools such
233 :     as ml-nlffigen (because during development programmers tend to
234 :     erase the tool's entire output directory tree including CM's cached
235 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
236 :     revert to the old, broken behavior of SML/NJ, but in specific cases
237 :     where there is no danger of interface confusion, its use is ok
238 :     (I think).
239 :    
240 :     ml-nlffigen by default generates "noguid" annotations. They can be
241 :     turned off by specifying -guid in its command line.
242 :    
243 :     ----------------------------------------------------------------------
244 : george 1136 Name: Lal George
245 :     Date: 2002/03/12 12 14:42:36 EST
246 :     Tag: george-20020312-frequency-computation
247 :     Description:
248 :    
249 :     Integrated jump chaining and static block frequency into the
250 :     compiler. More details and numbers later.
251 : leunga 1025
252 : george 1003 ----------------------------------------------------------------------
253 : george 1133 Name: Lal George
254 :     Date: 2002/03/11 11 22:38:53 EST
255 :     Tag: george-20020311-jump-chain-elim
256 :     Description:
257 :    
258 :     Tested the jump chain elimination on all architectures (except the
259 :     hppa). This is on by default right now and is profitable for the
260 :     alpha and x86, however, it may not be profitable for the sparc and ppc
261 :     when compiling the compiler.
262 :    
263 :     The gc test will typically jump to a label at the end of the cluster,
264 :     where there is another jump to an external cluster containing the actual
265 :     code to invoke gc. This is to allow factoring of common gc invocation
266 :     sequences. That is to say, we generate:
267 :    
268 :     f:
269 :     testgc
270 :     ja L1 % jump if above to L1
271 :    
272 :     L1:
273 :     jmp L2
274 :    
275 :    
276 :     After jump chain elimination the 'ja L1' instructions is converted to
277 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
278 :     up being implemented in their long form (if L2 is far away) using:
279 :    
280 :     jbe L3 % jump if below or equal to L3
281 :     jmp L2
282 :     L3:
283 :     ...
284 :    
285 :    
286 :     For large compilation units L2 may be far away.
287 :    
288 :    
289 :     ----------------------------------------------------------------------
290 : blume 1132 Name: Matthias Blume
291 :     Date: 2002/03/11 13:30:00 EST
292 :     Tag: blume-20020311-mltreeeval
293 :     Description:
294 :    
295 :     A functor parameter was missing.
296 :    
297 :     ----------------------------------------------------------------------
298 : leunga 1130 Name: Allen Leung
299 : leunga 1131 Date: 2002/03/11 10:30:00 EST
300 : leunga 1142 Tag: leunga-20020311-runtime-string0
301 : leunga 1131 Description:
302 :    
303 :     The representation of the empty string now points to a
304 :     legal null terminated C string instead of unit. It is now possible
305 :     to convert an ML string into C string with InlineT.CharVector.getData.
306 :     This compiles into one single machine instruction.
307 :    
308 :     ----------------------------------------------------------------------
309 :     Name: Allen Leung
310 : leunga 1130 Date: 2002/03/10 23:55:00 EST
311 :     Tag: leunga-20020310-x86-call
312 :     Description:
313 :    
314 :     Added machine generation for CALL instruction (relative displacement mode)
315 :    
316 :     ----------------------------------------------------------------------
317 : blume 1128 Name: Matthias Blume
318 :     Date: 2002/03/08 16:05:00
319 :     Tag: blume-20020308-entrypoints
320 :     Description:
321 :    
322 :     Version number bumped to 110.39.1. NEW BOOTFILES!
323 :    
324 :     Entrypoints: non-zero offset into a code object where execution should begin.
325 :    
326 :     - Added the notion of an entrypoint to CodeObj.
327 :     - Added reading/writing of entrypoint info to Binfile.
328 :     - Made runtime system bootloader aware of entrypoints.
329 :     - Use the address of the label of the first function given to mlriscGen
330 :     as the entrypoint. This address is currently always 0, but it will
331 :     not be 0 once we turn on block placement.
332 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
333 :     with entry points) from mlriscGen.
334 :    
335 :     ----------------------------------------------------------------------
336 : leunga 1127 Name: Allen Leung
337 :     Date: 2002/03/07 20:45:00 EST
338 :     Tag: leunga-20020307-x86-cmov
339 :     Description:
340 :    
341 :     Bug fixes for CMOVcc on x86.
342 :    
343 :     1. Added machine code generation for CMOVcc
344 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
345 :     3. CMOVcc cannot have an immediate operand as argument.
346 :    
347 :     ----------------------------------------------------------------------
348 : blume 1126 Name: Matthias Blume
349 :     Date: 2002/03/07 16:15:00 EST
350 :     Tag: blume-20020307-controls
351 :     Description:
352 :    
353 :     This is a very large but mostly boring patch which makes (almost)
354 :     every tuneable compiler knob (i.e., pretty much everything under
355 :     Control.* plus a few other things) configurable via both the command
356 :     line and environment variables in the style CM did its configuration
357 :     until now.
358 :    
359 :     Try starting sml with '-h' (or, if you are brave, '-H')
360 :    
361 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
362 :     implements the underlying generic mechanism.
363 :    
364 :     The interface to some of the existing such facilities has changed somewhat.
365 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
366 :     (The getFoo interface is still there for backward-compatibility, but its
367 :     use is deprecated.)
368 :    
369 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
370 :     that one can now twiddle the compiler settings when using this "batch"
371 :     compiler.
372 :    
373 :     TODO items:
374 :    
375 :     We should go through and throw out all controls that are no longer
376 :     connected to anything. Moreover, we should go through and provide
377 :     meaningful (and correct!) documentation strings for those controls
378 :     that still are connected.
379 :    
380 :     Currently, multiple calls to Controls.new are accepted (only the first
381 :     has any effect). Eventually we should make sure that every control
382 :     is being made (via Controls.new) exactly once. Future access can then
383 :     be done using Controls.acc.
384 :    
385 :     Finally, it would probably be a good idea to use the getter-setter
386 :     interface to controls rather than ref cells. For the time being, both
387 :     styles are provided by the Controls module, but getter-setter pairs are
388 :     better if thread-safety is of any concern because they can be wrapped.
389 :    
390 :     *****************************************
391 :    
392 :     One bug fix: The function blockPlacement in three of the MLRISC
393 :     backpatch files used to be hard-wired to one of two possibilities at
394 :     link time (according to the value of the placementFlag). But (I
395 :     think) it should rather sense the flag every time.
396 :    
397 :     *****************************************
398 :    
399 :     Other assorted changes (by other people who did not supply a HISTORY entry):
400 :    
401 :     1. the cross-module inliner now works much better (Monnier)
402 :     2. representation of weights, frequencies, and probabilities in MLRISC
403 :     changed in preparation of using those for weighted block placement
404 :     (Reppy, George)
405 :    
406 :     ----------------------------------------------------------------------
407 : george 1116 Name: Lal George
408 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
409 :     Tag: george-20020307-weighted-block-placement
410 :    
411 :     Tested the weighted block placement optimization on all architectures
412 :     (except the hppa) using AMPL to generate the block and edge frequencies.
413 :     Changes were required in the machine properties to correctly
414 :     categorize trap instructions. There is an MLRISC flag
415 :     "weighted-block-placement" that can be used to enable weighted block
416 :     placement, but this will be ineffective without block/edge
417 :     frequencies (coming soon).
418 :    
419 :    
420 :     ----------------------------------------------------------------------
421 :     Name: Lal George
422 : george 1116 Date: 2002/03/05 17:24:48 EST
423 :     Tag: george-20020305-linkage-cluster
424 :    
425 : george 1124 In order to support the block placement optimization, a new cluster
426 :     is generated as the very first cluster (called the linkage cluster).
427 :     It contains a single jump to the 'real' entry point for the compilation
428 :     unit. Block placement has no effect on the linkage cluster itself, but
429 :     all the other clusters have full freedom in the manner in which they
430 :     reorder blocks or functions.
431 : george 1116
432 :     On the x86 the typical linkage code that is generated is:
433 :     ----------------------
434 :     .align 2
435 :     L0:
436 :     addl $L1-L0, 72(%esp)
437 : george 1124 jmp L1
438 : george 1116
439 :    
440 :     .align 2
441 :     L1:
442 :     ----------------------
443 :    
444 :     72(%esp) is the memory location for the stdlink register. This
445 :     must contain the address of the CPS function being called. In the
446 : george 1124 above example, it contains the address of L0; before
447 : george 1116 calling L1 (the real entry point for the compilation unit), it
448 :     must contain the address for L1, and hence
449 :    
450 :     addl $L1-L0, 72(%esp)
451 :    
452 : george 1124 I have tested this on all architectures except the hppa.The increase
453 :     in code size is of course negligible
454 : george 1116
455 :     ----------------------------------------------------------------------
456 : leunga 1115 Name: Allen Leung
457 :     Date: 2002/03/03 13:20:00 EST
458 :     Tag: leunga-20020303-mlrisc-tools
459 :    
460 :     Added #[ ... ] expressions to mlrisc tools
461 :    
462 :     ----------------------------------------------------------------------
463 : blume 1096 Name: Matthias Blume
464 : blume 1098 Date: 2002/02/27 12:29:00 EST
465 :     Tag: blume-20020227-cdebug
466 :     Description:
467 :    
468 :     - made types in structure C and C_Debug to be equal
469 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
470 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
471 :    
472 :     ----------------------------------------------------------------------
473 :     Name: Matthias Blume
474 : blume 1096 Date: 2002/02/26 12:00:00 EST
475 :     Tag: blume-20020226-ffi
476 :     Description:
477 :    
478 :     1. Fixed a minor bug in CM's "noweb" tool:
479 :     If numbering is turned off, then truly don't number (i.e., do not
480 :     supply the -L option to noweb). The previous behavior was to supply
481 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
482 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
483 :    
484 :     2. Once again, I reworked some aspects of the FFI:
485 :    
486 :     A. The incomplete/complete type business:
487 :    
488 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
489 :     gone!
490 :     - ML types representing an incomplete type are now *equal* to
491 :     ML types representing their corresponding complete types (just like
492 :     in C). This is still safe because ml-nlffigen will not generate
493 :     RTTI for incomplete types, nor will it generate functions that
494 :     require access to such RTTI. But when ML code generated from both
495 :     incomplete and complete versions of the C type meet, the ML types
496 :     are trivially interoperable.
497 :    
498 :     NOTE: These changes restore the full generality of the translation
499 :     (which was previously lost when I eliminated functorization)!
500 :    
501 :     B. Enum types:
502 :    
503 :     - Structure C now has a type constructor "enum" that is similar to
504 :     how the "su" constructor works. However, "enum" is not a phantom
505 :     type because each "T enum" has values (and is isomorphic to
506 :     MLRep.Signed.int).
507 :     - There are generic access operations for enum objects (using
508 :     MLRep.Signed.int).
509 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
510 :     * The structure contains the definition of type "mlrep" (the ML-side
511 :     representation type of the enum). Normally, mlrep is the same
512 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
513 :     then mlrep will be defined as a datatype -- thus facilitating
514 :     pattern matching on mlrep values.
515 :     ("-ec" will be suppressed if there are duplicate values in an
516 :     enumeration.)
517 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
518 :     will be generated for each C enum constant xxx.
519 :     * Conversion functions m2i and i2m convert between mlrep and
520 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
521 :     * Coversion functions c and ml convert between mlrep and "tag enum".
522 :     * Access functions (get/set) fetch and store mlrep values.
523 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
524 :     enumerations are merged into one single enumeration represented by
525 :     structure E_'.
526 :    
527 :     ----------------------------------------------------------------------
528 : leunga 1080 Name: Allen Leung
529 : leunga 1094 Date: 2002/02/25 04:45:00 EST
530 :     Tag: leunga-20020225-cps-spill
531 :    
532 :     This is a new implementation of the CPS spill phase.
533 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
534 :     In case of problems, replace it with the old file spill.sml
535 :    
536 :     The current compiler runs into some serious performance problems when
537 :     constructing a large record. This can happen when we try to compile a
538 :     structure with many items. Even a very simple structure like the following
539 :     makes the compiler slow down.
540 :    
541 :     structure Foo = struct
542 :     val x_1 = 0w1 : Word32.int
543 :     val x_2 = 0w2 : Word32.int
544 :     val x_3 = 0w3 : Word32.int
545 :     ...
546 :     val x_N = 0wN : Word32.int
547 :     end
548 :    
549 :     The following table shows the compile time, from N=1000 to N=4000,
550 :     with the old compiler:
551 :    
552 :     N
553 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
554 :     MLRISC ra 0.06u 0.00s 0.05g
555 :     (spills = 0 reloads = 0)
556 :     TOTAL 0.63u 0.07s 0.21g
557 :    
558 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
559 :     MLRISC ra 5.68u 0.59s 3.93g
560 :     (spills = 0 reloads = 0)
561 :     TOTAL 14.71u 0.99s 4.81g
562 :    
563 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
564 :     MLRISC ra 5.54u 0.65s 3.91g
565 :     (spills = 543 reloads = 1082)
566 :     TOTAL 65.40u 3.13s 6.00g
567 :    
568 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
569 :     MLRISC ra 0.80u 0.10s 0.55g
570 :     (spills = 42 reloads = 84)
571 :     TOTAL 129.42u 5.10s 4.13g
572 :    
573 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
574 :     MLRISC ra 2.69u 0.27s 1.38g
575 :     (spills = 62 reloads = 124)
576 :     TOTAL 682.48u 19.61s 13.99g
577 :    
578 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
579 :     MLRISC ra 4.96u 0.27s 2.72g
580 :     (spills = 85 reloads = 170)
581 :     TOTAL 2375.26u 57.21s 48.00g
582 :    
583 :     As you can see the old cps spill module suffers from some serious
584 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
585 :     instead of patching the problems up, I'm reimplementing it
586 : leunga 1094 with a different algorithm. The new code is more modular,
587 :     smaller when compiled, and substantially faster
588 :     (O(n log n) time and O(n) space). Timing of the new spill module:
589 :    
590 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
591 :     MLRISC ra 0.25u 0.02s 0.15g
592 :     (spills=1 reloads=3)
593 :     TOTAL 7.74u 0.34s 1.62g
594 :    
595 :     Implementation details:
596 :    
597 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
598 :     number of live variables at any program point (the bandwidth)
599 :     does not exceed a certain limit, which is determined by the
600 :     size of the spill area.
601 :    
602 :     When the bandwidth is too large, we decrease the register pressure by
603 :     packing live variables into spill records. How we achieve this is
604 :     completely different than what we did in the old code.
605 :    
606 :     First, there is something about the MLRiscGen code generator
607 :     that we should be aware of:
608 :    
609 :     o MLRiscGen performs code motion!
610 :    
611 :     In particular, it will move floating point computations and
612 :     address computations involving only the heap pointer to
613 :     their use sites (if there is only a single use).
614 :     What this means is that if we have a CPS record construction
615 :     statement
616 :    
617 :     RECORD(k,vl,w,e)
618 :    
619 :     we should never count the new record address w as live if w
620 :     has only one use (which is often the case).
621 :    
622 :     We should do something similar to floating point, but the transformation
623 :     there is much more complex, so I won't deal with that.
624 :    
625 :     Secondly, there are now two new cps primops at our disposal:
626 :    
627 :     1. rawrecord of record_kind option
628 :     This pure operator allocates some uninitialized storage from the heap.
629 :     There are two forms:
630 :    
631 :     rawrecord NONE [INT n] allocates a tagless record of length n
632 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
633 :     and initializes the tag.
634 :    
635 :     2. rawupdate of cty
636 :     rawupdate cty (v,i,x)
637 :     Assigns to x to the ith component of record v.
638 :     The storelist is not updated.
639 :    
640 :     We use these new primops for both spilling and increment record construction.
641 :    
642 :     1. Spilling.
643 :    
644 :     This is implemented with a linear scan algorithm (but generalized
645 :     to trees). The algorithm will create a single spill record at the
646 :     beginning of the cps function and use rawupdate to spill to it,
647 :     and SELECT or SELp to reload from it. So both spills and reloads
648 :     are fine-grain operations. In contrast, in the old algorithm
649 :     "spills" have to be bundled together in records.
650 :    
651 :     Ideally, we should sink the spill record construction to where
652 :     it is needed. We can even split the spill record into multiple ones
653 :     at the places where they are needed. But CPS is not a good
654 :     representation for global code motion, so I'll keep it simple and
655 :     am not attempting this.
656 :    
657 :     2. Incremental record construction (aka record splitting).
658 :    
659 :     Long records with many component values which are simulatenously live
660 :     (recall that single use record addresses are not considered to
661 :     be live) are constructed with rawrecord and rawupdate.
662 :     We allocate space on the heap with rawrecord first, then gradually
663 :     fill it in with rawupdate. This is the technique suggested to me
664 :     by Matthias.
665 :    
666 :     Some restrictions on when this is applicable:
667 :     1. It is not a VECTOR record. The code generator currently does not handle
668 :     this case. VECTOR record uses double indirection like arrays.
669 :     2. All the record component values are defined in the same "basic block"
670 :     as the record constructor. This is to prevent speculative
671 :     record construction.
672 :    
673 :     ----------------------------------------------------------------------
674 :     Name: Allen Leung
675 : leunga 1086 Date: 2002/02/22 01:02:00 EST
676 :     Tag: leunga-20020222-mlrisc-tools
677 :    
678 :     Minor bug fixes in the parser and rewriter
679 :    
680 :     ----------------------------------------------------------------------
681 :     Name: Allen Leung
682 : leunga 1085 Date: 2002/02/21 20:20:00 EST
683 :     Tag: leunga-20020221-peephole
684 :    
685 :     Regenerated the peephole files. Some contained typos in the specification
686 :     and some didn't compile because of pretty printing bugs in the old version
687 :     of 'nowhere'.
688 :    
689 :     ----------------------------------------------------------------------
690 :     Name: Allen Leung
691 : leunga 1079 Date: 2002/02/19 20:20:00 EST
692 :     Tag: leunga-20020219-mlrisc-tools
693 :     Description:
694 :    
695 :     Minor bug fixes to the mlrisc-tools library:
696 :    
697 :     1. Fixed up parsing colon suffixed keywords
698 :     2. Added the ability to shut the error messages up
699 :     3. Reimplemented the pretty printer and fixed up/improved
700 :     the pretty printing of handle and -> types.
701 :     4. Fixed up generation of literal symbols in the nowhere tool.
702 :     5. Added some SML keywords to to sml.sty
703 :    
704 :     ----------------------------------------------------------------------
705 :     Name: Matthias Blume
706 : blume 1078 Date: 2002/02/19 16:20:00 EST
707 :     Tag: blume-20020219-cmffi
708 :     Description:
709 :    
710 :     A wild mix of changes, some minor, some major:
711 :    
712 :     * All C FFI-related libraries are now anchored under $c:
713 :     $/c.cm --> $c/c.cm
714 :     $/c-int.cm --> $c/internals/c-int.cm
715 :     $/memory.cm --> $c/memory/memory.cm
716 :    
717 :     * "make" tool (in CM) now treats its argument pathname slightly
718 :     differently:
719 :     1. If the native expansion is an absolute name, then before invoking
720 :     the "make" command on it, CM will apply OS.Path.mkRelative
721 :     (with relativeTo = OS.FileSys.getDir()) to it.
722 :     2. The argument will be passed through to subsequent phases of CM
723 :     processing without "going native". In particular, if the argument
724 :     was an anchored path, then "make" will not lose track of that anchor.
725 :    
726 :     * Compiler backends now "know" their respective C calling conventions
727 :     instead of having to be told about it by ml-nlffigen. This relieves
728 :     ml-nlffigen from one of its burdens.
729 :    
730 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
731 :    
732 :     * Export C_DEBUG and C_Debug from $c/c.cm.
733 :    
734 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
735 :     subtyping relationship between incomplete pointers and their complete
736 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
737 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
738 :     the pointer target type is complete. In the incomplete case, 'o
739 :     will be instantiated with some "'c iobj" -- a type obtained by
740 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
741 :    
742 :     Operations that work on both incomplete and complete pointer types are
743 :     typed as taking an 'o ptr while operations that require the target to
744 :     be known are typed as taking some ('t, 'c) obj ptr.
745 :    
746 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
747 :     where void is an eqtype without any values. This makes it possible
748 :     to work on voidptr values using functions meant to operate on light
749 :     incomplete pointers.
750 :    
751 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
752 :     been vastly simplified.
753 :    
754 :     ----------------------------------------------------------------------
755 :     Name: Matthias Blume
756 : blume 1076 Date: 2002/02/19 10:48:00 EST
757 :     Tag: blume-20020219-pqfix
758 :     Description:
759 :    
760 :     Applied Chris Okasaki's bug fix for priority queues.
761 :    
762 :     ----------------------------------------------------------------------
763 :     Name: Matthias Blume
764 : blume 1073 Date: 2002/02/15 17:05:00
765 : blume 1069 Tag: Release_110_39
766 :     Description:
767 :    
768 : blume 1073 Last-minute retagging is becoming a tradition... :-(
769 :    
770 :     This is the working release 110.39.
771 :    
772 :     ----------------------------------------------------------------------
773 :     Name: Matthias Blume
774 :     Date: 2002/02/15 16:00:00 EST
775 :     Tag: Release_110_39-orig
776 :     Description:
777 :    
778 : blume 1069 Working release 110.39. New bootfiles.
779 :    
780 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
781 :     with all shells. So I retagged. -Matthias)
782 :    
783 : blume 1069 ----------------------------------------------------------------------
784 :     Name: Matthias Blume
785 : blume 1068 Date: 2002/02/15 14:17:00 EST
786 :     Tag: blume-20020215-showbindings
787 :     Description:
788 :    
789 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
790 :     the latter can be useful for exploring what bindings are available at
791 :     the interactive prompt. (The first function returns only the list
792 :     of symbols that are really bound, the second prints those but also the
793 :     ones that CM's autoloading mechanism knows about.)
794 :    
795 :     ----------------------------------------------------------------------
796 :     Name: Matthias Blume
797 : blume 1067 Date: 2002/02/15 12:08:00 EST
798 :     Tag: blume-20020215-iptrs
799 :     Description:
800 :    
801 :     Two improvements to ml-nlffigen:
802 :    
803 :     1. Write files only if they do not exist or if their current contents
804 :     do not coincide with what's being written. (That is, avoid messing
805 :     with the time stamps unless absolutely necessary.)
806 :    
807 :     2. Implement a "repository" mechanism for generated files related
808 :     to "incomplete pointer types". See the README file for details.
809 :    
810 :     ----------------------------------------------------------------------
811 :     Name: Matthias Blume
812 : blume 1066 Date: 2002/02/14 11:50:00 EST
813 :     Tag: blume-20020214-quote
814 :     Description:
815 :    
816 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
817 :     because of the new and improved tag generation scheme. (Thanks to Allen
818 :     Leung for pointing it out.)
819 :    
820 :     ----------------------------------------------------------------------
821 : george 1065 Name: Lal George
822 :     Date: 2002/02/14 09:55:27 EST 2002
823 :     Tag: george-20020214-isabelle-bug
824 :     Description:
825 :    
826 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
827 :     of Isabelle on the x86.
828 :    
829 :     From Allen:
830 :     -----------
831 :     I've found the problem:
832 :    
833 :     in ra-core.sml, I use the counter "blocked" to keep track of the
834 :     true number of elements in the freeze queue. When the counter goes
835 :     to zero, I skip examining the queue. But I've messed up the
836 :     bookkeeping in combine():
837 :    
838 :     else ();
839 :     case !ucol of
840 :     PSEUDO => (if !cntv > 0 then
841 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
842 :     ^^^^^^^^^^^^^^^^^^^^^^^
843 :     moveu := mergeMoveList(!movev, !moveu)
844 :     )
845 :     else ();
846 :    
847 :     combine() is called to coalesce two nodes u and v.
848 :     I think I was thinking that if the move counts of u and v are both
849 :     greater than zero then after they are coalesced then one node is
850 :     removed from the freeze queue. Apparently I was thinking that
851 :     both u and v are of low degree, but that's clearly not necessarily true.
852 :    
853 :    
854 :     02/12/2002:
855 :     Here's the patch. HOL now compiles.
856 :    
857 :     I don't know how this impact on performance (compile
858 :     time or runtime). This bug caused the RA (especially on the x86)
859 :     to go thru the potential spill phase when there are still nodes on the
860 :     freeze queue.
861 :    
862 :    
863 :    
864 :    
865 :     ----------------------------------------------------------------------
866 : blume 1058 Name: Matthias Blume
867 : blume 1064 Date: 2002/02/13 22:40:00 EST
868 :     Tag: blume-20020213-fptr-rtti
869 :     Description:
870 :    
871 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
872 :     updates.
873 :    
874 :     ----------------------------------------------------------------------
875 :     Name: Matthias Blume
876 : blume 1063 Date: 2002/02/13 16:41:00 EST
877 :     Tag: blume-20020213-cmlpq
878 :     Description:
879 :    
880 :     Added new priority queue export symbols (which have just been added to
881 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
882 :     not compile and the installer would choke.)
883 :    
884 :     ----------------------------------------------------------------------
885 :     Name: Matthias Blume
886 : blume 1062 Date: 2002/02/13 16:15:00 EST
887 :     Tag: blume-20020213-various
888 :     Description:
889 :    
890 :     1. More tweaks to ml-nlffigen:
891 :    
892 :     - better internal datastructures (resulting in slight speedup)
893 :     - "-match" option requires exact match
894 :     - "localized" gensym counters (untagged structs/unions nested within
895 :     other structs/unions or within typedefs get a fresh counter; their
896 :     tag will be prefixed by a concatenation of their parents' tags)
897 :     - bug fixes (related to calculation of transitive closure of types
898 :     to be included in the output)
899 :    
900 :     2. Minor Basis updates:
901 :    
902 :     - added implementations for List.collate and Option.app
903 :    
904 :     ----------------------------------------------------------------------
905 :     Name: Matthias Blume
906 : blume 1060 Date: 2002/02/11 15:55:00 EST
907 :     Tag: blume-20020211-gensym
908 :     Description:
909 :    
910 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
911 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
912 :     names (ML structure names that correspond to unnamed C structs, unions,
913 :     and enums), so that separate runs of ml-nlffigen do not clash.
914 :    
915 :     ----------------------------------------------------------------------
916 :     Name: Matthias Blume
917 : blume 1059 Date: 2002/02/11 12:05:00 EST
918 :     Tag: blume-20020211-gensml
919 :     Description:
920 :    
921 :     A quick fix for a problem with GenSML (in the pgraph-util library):
922 :     Make generation of toplevel "local" optional. (Strictly speaking,
923 :     signature definitions within "local" are not legal SML.)
924 :    
925 :     Other than that: updates to INSTALL and cm/TODO.
926 :    
927 :     ----------------------------------------------------------------------
928 :     Name: Matthias Blume
929 : blume 1058 Date: 2002/02/08 15:00:00 EST
930 :     Tag: blume-20020208-uniquepid
931 :     Description:
932 :    
933 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
934 :    
935 :     1. The installer (config/install.sh) has gotten smarter:
936 :    
937 :     - Configuration options are a bit easier to specify now
938 :     (in config/targets).
939 :     - Bug in recognizing .tar.bz2 files fixed.
940 :     - Installer automatically resolves dependencies between
941 :     configuration options (e.g., if you ask for eXene, you will
942 :     also get cml -- regardless whether you asked for it or not).
943 :     - Installer can run in "quieter mode" by setting the environment
944 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
945 :     "completely silent", though.
946 :     - Build HashCons library as part of smlnj-lib.
947 :    
948 :     2. A new scheme for assigning persistent identifiers to compilation
949 :     units (and, by extension, to types etc.) has been put into place.
950 :     This fixes a long-standing bug where types and even dynamic values
951 :     can get internally confused, thereby compromising type safety
952 :     (abstraction) and dynamic correctness. See
953 :    
954 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
955 :    
956 :     for an example of how things could go wrong until now.
957 :    
958 :     The downside of the new scheme is that pids are not quite as
959 :     persistent as they used to be: CM will generate a fresh pid
960 :     for every compilation unit that it thinks it sees for the first
961 :     time. That means that if you compile starting from a clean, fresh
962 :     source tree at two different times, you end up with different
963 :     binaries.
964 :    
965 :     Cutoff recompilation, however, has not been compromised because
966 :     CM keeps pid information in special caches between runs.
967 :    
968 :     ----------------------------------------------------------------------
969 : george 1053 Name: Lal George
970 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
971 :     Tag: <none>
972 :     Description:
973 :    
974 :     Compilers that generate assembly code may produce global labels
975 :     whose value is resolved at link time. The various peephole optimization
976 :     modules did not take this in account.
977 :    
978 :     TODO. The Labels.addrOf function should really return an option
979 :     type so that clients are forced to deal with this issue, rather
980 :     than an exception being raised.
981 :    
982 :     ----------------------------------------------------------------------
983 :     Name: Lal George
984 : george 1053 Date: 2002/02/06 13:55:02 EST
985 :     Tag: george-20020206-ra-breakup
986 :     Description:
987 :    
988 :     1. A bug fix from Allen.
989 :    
990 :     A typo causes extra fstp %st(0)'s to be generated at compensation
991 :     edges, which might cause stack underflow traps at runtime. This
992 :     occurs in fft where there are extraneous fstps right before the 'into'
993 :     trap instruction (in this case they are harmless since none of the
994 :     integers overflow.)
995 :    
996 : george 1054 2. Pulled out various utility modules that were embedded in the modules
997 : george 1053 of the register allocator. I need these modules for other purposes, but
998 : george 1054 they are not complete enough to put into a library (just yet).
999 : george 1053 ----------------------------------------------------------------------
1000 : blume 1049 Name: Matthias Blume
1001 :     Date: 2002/01/31 16:05:00 EST
1002 :     Tag: blume-20020131-sparc-ccalls
1003 :     Description:
1004 :    
1005 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1006 :     of extra stack space by mistake. Fixed.
1007 :    
1008 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1009 :    
1010 :     ----------------------------------------------------------------------
1011 : leunga 1048 Name: Allen Leung
1012 :     Date: 2002/01/30
1013 :     Tag: leunga-20020130-nowhere-bug-fix
1014 :     Description:
1015 :    
1016 :     MLRISC bug fixes:
1017 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1018 :     2. MachineInt.fromString was negating its value.
1019 :    
1020 :     ----------------------------------------------------------------------
1021 : blume 1039 Name: Matthias Blume
1022 : blume 1047 Date: 2002/01/29
1023 :     Tag: blume-20020129-INSTALL
1024 :     Description:
1025 :    
1026 :     - Added somewhat detailed installation instructions (file INSTALL).
1027 :     - Fixed curl-detection bug in config/install.sh.
1028 :     - It is now possible to select the URL getter using the URLGETTER
1029 :     environment variable:
1030 :    
1031 :     not set / "unknown" --> automatic detection (script tries wget,
1032 :     curl, and lynx)
1033 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1034 :     how to properly invoke them)
1035 :     other --> use $URLGETTER directly, it must take
1036 :     precisely two command-line arguments
1037 :     (source URL and destination file name)
1038 :    
1039 :     ----------------------------------------------------------------------
1040 :     Name: Matthias Blume
1041 : blume 1044 Date: 2002/01/28
1042 : blume 1045 Tag: blume-20020128-sparc-ccalls
1043 :     Description:
1044 :    
1045 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
1046 :     - Make use of the allocParam argument in sparc-c-calls.
1047 :    
1048 :     ----------------------------------------------------------------------
1049 :     Name: Matthias Blume
1050 :     Date: 2002/01/28
1051 : blume 1044 Tag: blume-20020128-allocParam
1052 :     Description:
1053 :    
1054 :     John Reppy: Changes c-calls API to accept client-callback for
1055 :     allocating extra stack space.
1056 :     me: Corresponding changes to mlriscGen (using a dummy argument that
1057 :     does not change the current behavior).
1058 :    
1059 :     ----------------------------------------------------------------------
1060 :     Name: Matthias Blume
1061 : blume 1040 Date: 2002/01/28 12:00:00
1062 :     Tag: Release_110_38
1063 :     Description:
1064 :    
1065 :     This time for real!!!
1066 :    
1067 :     ----------------------------------------------------------------------
1068 :     Name: Matthias Blume
1069 : blume 1039 Date: 2002/01/28 10:56:00 EST
1070 :     Tag: blume-20020128-retraction
1071 :     Description:
1072 :    
1073 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
1074 :     with blume-Release_110_38-retracted.)
1075 :    
1076 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
1077 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1078 :    
1079 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1080 :     and enums fixed. The naming of corresponding ML identifiers should
1081 :     now be consistent again.
1082 :    
1083 :     ----------------------------------------------------------------------
1084 : leunga 1038 Name: Allen Leung
1085 :     Date: 2002/01/27
1086 :     Tag: leunga-20020127-nowhere
1087 :     Description:
1088 :    
1089 :     Added a target called nowhere in the configuration scripts.
1090 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
1091 :     programs with where-clauses into legal SML code) during installation.
1092 :    
1093 :     ----------------------------------------------------------------------
1094 : blume 1036 Name: Matthias Blume
1095 : leunga 1038 Date: 2002/01/25 21:27:00 EST
1096 : blume 1039 Tag: blume-Release_110_38-retracted
1097 : blume 1037 Description:
1098 :    
1099 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
1100 :    
1101 :     README will be added later.
1102 :    
1103 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
1104 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
1105 :    
1106 : blume 1037 ----------------------------------------------------------------------
1107 :     Name: Matthias Blume
1108 : blume 1036 Date: 2002/01/25
1109 :     Tag: blume-20020125-ffi
1110 :     Description:
1111 :    
1112 :     A large number of tweaks and improvements to ml-nlffi-lib and
1113 :     ml-nlffigen:
1114 :    
1115 :     - ML represenation types have been streamlined
1116 :     - getter and setter functions work with concrete values, not abstract
1117 :     ones where possible
1118 :     - ml-nlffigen command line more flexible (see README file there)
1119 :     - some bugs have been fixed (hopefully)
1120 :    
1121 :     ----------------------------------------------------------------------
1122 : george 1034 Name: Lal George
1123 :     Date: 2002/01/24
1124 :     Tag: george-20020124-risc-ra-interface
1125 :     Description:
1126 :    
1127 :     There is a dramatic simplification in the interface to the
1128 :     register allocator for RISC architectures as a result of making
1129 :     parallel copy instructions explicit.
1130 :    
1131 :     ----------------------------------------------------------------------
1132 : blume 1028 Name: Matthias Blume
1133 : blume 1032 Date: 2002/01/22
1134 :     Tag: blume-20020122-x86-ccalls
1135 :     Description:
1136 :    
1137 :     Bug fix for c-calls on x86 (having to do with how char- and
1138 :     short-arguments are being handled).
1139 :    
1140 :     ----------------------------------------------------------------------
1141 :     Name: Matthias Blume
1142 : blume 1031 Date: 2002/01/21
1143 :     Tag: blume-20020121-ff
1144 :     Description:
1145 :    
1146 :     Another day of fiddling with the FFI...
1147 :    
1148 :     1. Bug fix/workaround: CKIT does not complain about negative array
1149 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1150 :     (Otherwise a negative dimension would send it into an infinite loop.)
1151 :    
1152 :     2. Some of the abstract types (light objects, light pointers, most "base"
1153 :     types) in structure C are now eqtypes.
1154 :    
1155 :     3. Added constructors and test functions for NULL function pointers.
1156 :    
1157 :     ----------------------------------------------------------------------
1158 :     Name: Matthias Blume
1159 : blume 1028 Date: 2002/01/18
1160 : blume 1030 Tag: blume-20020118-ready-for-new-release
1161 :     Description:
1162 :    
1163 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1164 :     files shortly.)
1165 :    
1166 :     Maybe we christen this to be 110.38?
1167 :    
1168 :     ----------------------------------------------------------------------
1169 :     Name: Matthias Blume
1170 :     Date: 2002/01/18
1171 : blume 1029 Tag: blume-20020118-more-ffifiddle
1172 :     Description:
1173 :    
1174 :     Today's FFI fiddling:
1175 :    
1176 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1177 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1178 :     of abstract ones.
1179 :    
1180 :     - Provided word-style bit operations etc. for "int" representation
1181 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1182 :     and Long.
1183 :    
1184 :     ----------------------------------------------------------------------
1185 :     Name: Matthias Blume
1186 :     Date: 2002/01/18
1187 : blume 1028 Tag: blume-20020118-use-x86-fp
1188 :     Description:
1189 :    
1190 :     Now that x86-fast-fp seems to be working, I turned it back on again
1191 :     by default. (Seems to work fine now, even with the FFI.)
1192 :    
1193 :     Other than that, I added some documentation about the FFI to
1194 :     src/ml-nlffigen/README and updated the FFI test examples in
1195 :     src/ml-nlffi-lib/Tests/*.
1196 :    
1197 :     ----------------------------------------------------------------------
1198 : leunga 1025 Name: Allen Leung
1199 :     Date: 2002/01/17
1200 :     Tag: leunga-20020117-x86-fast-fp-call
1201 :     Description:
1202 :    
1203 :     1. Fixed a problem with handling return fp values when x86's fast fp
1204 :     mode is turned on.
1205 :    
1206 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1207 :     of %f32.
1208 :    
1209 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1210 :    
1211 :     ----------------------------------------------------------------------
1212 : blume 1021 Name: Matthias Blume
1213 :     Date: 2002/01/16
1214 :     Tag: blume-20020116-ffifiddle
1215 :     Description:
1216 :    
1217 :     More fiddling with the FFI interface:
1218 :    
1219 :     - Make constness 'c instead of rw wherever possible. This eliminates
1220 :     the need for certain explicit coercions. (However, due to ML's
1221 :     value polymorphism, there will still be many cases where explicit
1222 :     coercions are necessary. Phantom types are not the whole answer
1223 :     to modeling a subtyping relationship in ML.)
1224 :    
1225 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1226 :    
1227 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1228 :     clumsy workaround for varargs problem.)
1229 :    
1230 :     ----------------------------------------------------------------------
1231 : george 1016 Name: Lal George
1232 :     Date: 2002/01/15
1233 :     Tag: <none>
1234 :     Description:
1235 :    
1236 :     1. Since COPY instructions are no longer native to the architecture,
1237 :     a generic functor can be used to implement the expandCopies function.
1238 :    
1239 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1240 :     TEXT segment.
1241 :    
1242 :     ----------------------------------------------------------------------
1243 : blume 1011 Name: Matthias Blume
1244 : blume 1015 Date: 2002/01/15
1245 :     Tag: blume-20020115-ffiupdates
1246 :     Description:
1247 :    
1248 :     1. Fix for bug resulting in single-precision float values being returned
1249 :     incorrectly from FFI calls.
1250 :    
1251 :     2. Small modifications to C FFI API:
1252 :    
1253 :     - memory-allocation routines return straight objects (no options)
1254 :     and raise an exception in out-of-memory situations
1255 :     - unsafe extensions to cast between function pointers and pointers
1256 :     from/to ints
1257 :     - added structure C_Debug as an alternative to structure C where
1258 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1259 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1260 :     but also takes a (possibly empty) list of existing library handles
1261 :     that the current library depends on
1262 :    
1263 :     ----------------------------------------------------------------------
1264 :     Name: Matthias Blume
1265 : blume 1011 Date: 2002/01/10
1266 :     Tag: blume-20020110-newffigen
1267 :     Description:
1268 :    
1269 :     1. Updates to portable graph code.
1270 :    
1271 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1272 :     more scalable now so that even huge interfaces such as the one
1273 :     for GTK compile in finite time and space. :-)
1274 :     See src/ml-nlffigen/README for details on what's new.
1275 :    
1276 :     ----------------------------------------------------------------------
1277 : george 1003 Name: Lal George
1278 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1279 :     Tag: george-20011206-rm-native-copy
1280 :     Description:
1281 :    
1282 :     Removed the native COPY and FCOPY instructions
1283 :     from all the architectures and replaced it with the
1284 :     explicit COPY instruction from the previous commit.
1285 :    
1286 :     It is now possible to simplify many of the optimizations
1287 :     modules that manipulate copies. This has not been
1288 :     done in this change.
1289 :    
1290 :     ----------------------------------------------------------------------
1291 :     Name: Lal George
1292 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1293 :     Tag: george-20011206-mlrisc-instruction
1294 :     Description:
1295 : george 824
1296 : george 1003 Changed the representation of instructions from being fully abstract
1297 :     to being partially concrete. That is to say:
1298 :    
1299 :     from
1300 :     type instruction
1301 :    
1302 :     to
1303 :     type instr (* machine instruction *)
1304 :    
1305 :     datatype instruction =
1306 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1307 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1308 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1309 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1310 :     | INSTR of instr
1311 :    
1312 :     This makes the handling of certain special instructions that appear on
1313 :     all architectures easier and uniform.
1314 :    
1315 :     LIVE and KILL say that a list of registers are live or killed at the
1316 :     program point where they appear. No spill code is generated when an
1317 :     element of the 'regs' field is spilled, but the register is moved to
1318 :     the 'spilled' (which is present, more for debugging than anything else).
1319 :    
1320 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1321 :     We used to generate:
1322 :    
1323 :     DEFFREG f1
1324 :     f1 := f2 + f3
1325 :     trapb
1326 :    
1327 :     but now generate:
1328 :    
1329 :     f1 := f2 + f3
1330 :     trapb
1331 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1332 :    
1333 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1334 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1335 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1336 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1337 :     this architecture).
1338 :    
1339 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1340 :     all the architectures. This will result in further simplification of the
1341 :     register allocator that must be aware of them for coalescing purposes, and
1342 :     will also simplify certain aspects of the machine description that provides
1343 :     callbacks related to parallel copies.
1344 :    
1345 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1346 :     machine instruction set!
1347 :    
1348 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1349 :     functions for making porting easier -- essentially converting upper case
1350 :     to lower case. All machine instructions (of type instr) are in upper case,
1351 :     and the lower case form generates an MLRISC instruction. For example on
1352 :     the alpha we have:
1353 :    
1354 :     datatype instr =
1355 :     LDA of {r:cell, b:cell, d:operand}
1356 :     | ...
1357 :    
1358 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1359 :     ...
1360 :    
1361 :     where lda is just (INSTR o LDA), etc.
1362 :    
1363 : leunga 815 ----------------------------------------------------------------------
1364 : blume 986 Name: Matthias Blume
1365 : blume 993 Date: 2001/11/22 21:40:00 EST
1366 :     Tag: Release_110_37
1367 :     Description:
1368 :    
1369 :     Release 110.37. This time for real.
1370 :    
1371 :     ----------------------------------------------------------------------
1372 :     Name: Matthias Blume
1373 : blume 988 Date: 2001/11/21 16:35:00 EST
1374 :     Tag: blume-20011121-foot-in-mouth
1375 :     Description:
1376 :    
1377 :     Removed the "Release_110_37" tag because of a serious bug.
1378 :     This will be re-tagged once the bug is fixed.
1379 :    
1380 :     ----------------------------------------------------------------------
1381 :     Name: Matthias Blume
1382 : blume 987 Date: 2001/11/21 16:14:00 EST
1383 :     Tag: blume-20011121-forgottenfile
1384 :     Description:
1385 :    
1386 :     Forgot to add a file. (Just a .tex-file -- part of
1387 :     the CM manual source.)
1388 :    
1389 :     ----------------------------------------------------------------------
1390 :     Name: Matthias Blume
1391 : blume 986 Date: 2001/11/21 16:10:00 EST
1392 : blume 988 Tag: blume-20011121-invalid_110_37
1393 : blume 986 Description:
1394 :    
1395 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
1396 :     because we found a serious bug in all non-x86 backends.
1397 :     - Matthias
1398 :    
1399 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
1400 :     so that code object name strings are directly inserted into code
1401 :     objects at code generation time. The only business the runtime system
1402 :     has with this is now to read the name strings on occasions.
1403 :     (The encoding of the name string has also changed somewhat.)
1404 :    
1405 :     2. CM now implements a simple "set calculus" for specifying export lists.
1406 :     In particular, it is now possible to refer to the export lists of
1407 :     other libraries/groups/sources and form unions as well as differences.
1408 :     See the latest CM manual for details.
1409 :    
1410 :     3. An separate notion of "proxy" libraries has again be eliminated from
1411 :     CM's model. (Proxy libraries are now simply a special case of using
1412 :     the export list calculus.)
1413 :    
1414 :     4. Some of the existing libraries now take advantage of the new set
1415 :     calculus.
1416 :     (Notice that not all libraries have been converted because some
1417 :     of the existing .cm-files are supposed to be backward compatible
1418 :     with 110.0.x.)
1419 :    
1420 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
1421 :     "General.exnMessage"! The former relies on a certain hook to be
1422 :     initialized, and that often does not happen in the stand-alone case.)
1423 :    
1424 :     ----------------------------------------------------------------------
1425 : george 984 Name: Lal George
1426 :     Date: 2001/11/21 13:56:18 EST
1427 :     Tag: george-2001121-pseudo-ops
1428 :     Description:
1429 :    
1430 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
1431 :     ought to never be any question of incompatabilities with
1432 :     pseudo-op syntax expected by host assemblers.
1433 :    
1434 :     For now, only modules supporting GAS syntax are implemented
1435 :     but more should follow, such as MASM, and vendor assembler
1436 :     syntax, e.g. IBM as, Sun as, etc.
1437 :    
1438 :     ----------------------------------------------------------------------
1439 : blume 977 Name: Matthias Blume
1440 :     Date: 2001/11/14 11:52:00 EST
1441 :     Tag: blume-20011114-srcname
1442 :     Description:
1443 :    
1444 :     1. Routed the name of the current source file to mlriscgen where it
1445 :     should be directly emitted into the code object. (This last part
1446 :     is yet to be done.)
1447 :    
1448 :     2. Some cleanup of the pgraph code to make it match the proposal that
1449 :     I put out the other day. (The proposal notwithstanding, things are
1450 :     still in flux here.)
1451 :    
1452 :     ----------------------------------------------------------------------
1453 : george 976 Name: Lal George
1454 :     Date: 2001/11/14 09:44:04 EST
1455 :     Tag:
1456 :     Description:
1457 :    
1458 :     Fix for a backpatching bug reported by Allen.
1459 :    
1460 :     Because the boundary between short and long span-dependent
1461 :     instructions is +/- 128, there are an astounding number of
1462 :     span-dependent instructions whose size is over estimated.
1463 :    
1464 :     Allen came up with the idea of letting the size of span
1465 :     dependent instructions be non-monotonic, for a maxIter
1466 :     number of times, after which the size must be monotonically
1467 :     increasing.
1468 :    
1469 :     This table shows the number of span-dependent instructions
1470 :     whose size was over-estimated as a function of maxIter, for the
1471 :     file Parse/parse/ml.grm.sml:
1472 :    
1473 :     maxIter # of instructions:
1474 :     10 687
1475 :     20 438
1476 :     30 198
1477 :     40 0
1478 :    
1479 :     In compiling the compiler, there is no significant difference in
1480 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1481 :     my measurements showed that maxIter=40 was a tad faster than
1482 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1483 :     point within 13 iterations, so fixing maxIter at 40, while high,
1484 :     is okay.
1485 :    
1486 :     ----------------------------------------------------------------------
1487 : blume 975 Name: Matthias Blume
1488 :     Date: 2001/10/31 15:25:00 EST
1489 :     Tag: blume-20011031-pgraph
1490 :     Description:
1491 :    
1492 :     CKIT:
1493 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1494 :     argument identifiers.
1495 :     * Changed the return type of TypeUtil.getFunction accordingly.
1496 :     * Type equality ignores the argument names.
1497 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1498 :     if there is a mismatch.
1499 :    
1500 :     installation script:
1501 :     * attempts to use "curl" if available (unless "wget" is available as well)
1502 :    
1503 :     CM:
1504 :     * has an experimental implementation of "portable graphs" which I will
1505 :     soon propose as an implementation-independent library format
1506 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1507 :    
1508 :     NLFFI-LIB:
1509 :     * some cleanup (all cosmetic)
1510 :    
1511 :     NLFFIGEN:
1512 :     * temporarily disabled the mechanism that suppresses ML output for
1513 :     C definitions whose identifiers start with an underscore character
1514 :     * generate val bindings for enum constants
1515 :     * user can request that only one style (light or heavy) is being used;
1516 :     default is to use both (command-line arguments: -heavy and -light)
1517 :     * fixed bug in handling of function types involving incomplete pointers
1518 :     * generate ML entry points that take record arguments (i.e., using
1519 :     named arguments) for C functions that have a prototype with named
1520 :     arguments
1521 :     (see changes to CKIT)
1522 :    
1523 :     ----------------------------------------------------------------------
1524 : leunga 972 Name: Allen Leung
1525 :     Date: 2001/10/27 20:34:00 EDT
1526 :     Tag: leunga-20011027-x86-fast-fp-call
1527 :     Description:
1528 :    
1529 :     Fixed the bug described in blume-20010920-slowfp.
1530 :    
1531 :     The fix involves
1532 :     1. generating FCOPYs in FSTP in ia32-svid
1533 :     2. marking a CALL with the appropriate annotation
1534 :    
1535 :     ----------------------------------------------------------------------
1536 : blume 961 Name: Matthias Blume
1537 :     Date: 2001/10/16 11:32:00 EDT
1538 :     Tag: blume-20011016-netbsd
1539 :     Description:
1540 :    
1541 :     Underscore patch from Chris Richards (fixing problem with compiling
1542 :     runtime system under recent NetBSD).
1543 :    
1544 :     ----------------------------------------------------------------------
1545 : george 959 Name: Allen Leung
1546 :     Date: 2001/10/12 17:18:32 EDT 2001
1547 :     Tag: leung-20011012-x86-printflowgraph
1548 :     Description:
1549 :    
1550 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1551 :    
1552 :     ----------------------------------------------------------------------
1553 : george 952 Name: Lal George
1554 : george 958 Date: 2001/10/11 23:51:34 EDT
1555 :     Tag: george-20011011-too-many-instrs
1556 :     Description:
1557 :    
1558 :     The representation of a program point never expected to see more
1559 :     than 65536 instructions in a basic block!
1560 :    
1561 :     ----------------------------------------------------------------------
1562 :     Name: Lal George
1563 : george 952 Date: 2001/10/09 09:41:37 EDT
1564 :     Tag: george-20011008-mlrisc-labels
1565 :     Description:
1566 :    
1567 :     Changed the machine description files to support printing of
1568 :     local and global labels in assembly code, based on host assembler
1569 :     conventions.
1570 :    
1571 :     ----------------------------------------------------------------------
1572 : blume 916 Name: Matthias Blume
1573 : blume 939 Date: 2001/09/25 15:25:00 EDT
1574 :     Tag: blume-20010925-exninfo
1575 :     Description:
1576 :    
1577 :     I provided a non-hook implementation of exnName (at the toplevel) and
1578 :     made the "dummy" implementation of exnMessage (at the toplevel) more
1579 :     useful: if nothing gets "hooked in", then at least you are going to
1580 :     see the exception name and a message indicating why you don't see more.
1581 :    
1582 :     [For the time being, programs that need exnMessage and want to use
1583 :     ml-build should either use General.exnMessage (strongly recommended) or
1584 :     refer to structure General at some other point so that CM sees a
1585 :     static dependency.]
1586 :    
1587 :     [Similar remarks go for "print" and "use": If you want to use their
1588 :     functionality in stand-alone programs generated by ml-build, then use
1589 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1590 :    
1591 :     ----------------------------------------------------------------------
1592 :     Name: Matthias Blume
1593 : blume 938 Date: 2001/09/20 17:28:00 EDT
1594 :     Tag: blume-20010920-slowfp
1595 :     Description:
1596 :    
1597 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1598 :    
1599 :     ----------------------------------------------------------------------
1600 :     Name: Matthias Blume
1601 : blume 937 Date: 2001/09/20 17:20:00 EDT
1602 :     Tag: blume-20010920-canonicalpaths
1603 :     Description:
1604 :    
1605 :     0. Updated the BOOT file (something that I forgot to do earlier).
1606 :    
1607 :     1. Small internal change to CM so that it avoids "/../" in filenames
1608 :     as much as possible (but only where it is safe).
1609 :    
1610 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1611 :     that arguments that contain delimiters are passed through correctly.
1612 :     This change also means that all "special" arguments of the form
1613 :     @SMLxxx... must come first.
1614 :    
1615 :     3. Changed install script to put relative anchor names for tool commands
1616 :     into pathconfig.
1617 :    
1618 :     ----------------------------------------------------------------------
1619 :     Name: Matthias Blume
1620 : blume 932 Date: 2001/09/18 15:35:00 EDT
1621 :     Tag: blume-20010918-readme11036
1622 :     Description:
1623 :    
1624 :     Added README files.
1625 :    
1626 :     ----------------------------------------------------------------------
1627 :     Name: Matthias Blume
1628 : blume 929 Date: 2001/09/18 11:45:00 EDT
1629 :     Tag: Release_110_36 (retag)
1630 :     Description:
1631 :    
1632 :     Fixed mistake in config/preloads. Retagged as 110.36.
1633 :    
1634 :     ----------------------------------------------------------------------
1635 :     Name: Matthias Blume
1636 : blume 928 Date: 2001/09/18 09:40:00 EDT
1637 : blume 929 Tag: Release_110_36_orig (tag changed)
1638 : blume 928 Description:
1639 :    
1640 :     New version (110.36). New bootfiles.
1641 :    
1642 :     ----------------------------------------------------------------------
1643 :     Name: Matthias Blume
1644 : blume 927 Date: 2001/09/14 16:15:00 EDT
1645 :     Tag: blume-20010914-x86fastfp
1646 :     Description:
1647 :    
1648 :     John committed some changes that Allen made, in particular a (hopefully)
1649 :     correctly working version of the x86-fp module.
1650 :    
1651 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1652 :     flag to "true". Everything seems to compile to a fixpoint ok, and
1653 :     "mandelbrot" speeds up by about 15%.
1654 :    
1655 :     ----------------------------------------------------------------------
1656 :     Name: Matthias Blume
1657 : blume 923 Date: 2001/09/13 11:20:00 EDT
1658 :     Tag: blume-20010913-minimal
1659 :     Description:
1660 :    
1661 :     1. Stefan Monnier's patch to fix a miscompilation problem that
1662 :     was brought to light by John Reppy's work on Moby.
1663 :    
1664 :     2. Implemented a minimal "structure Compiler" that contains just
1665 :     "version" and "architecture". The minimal version will be
1666 :     available when the full version is not. This is for backward-
1667 :     compatibility with code that wants to test Compiler.version.
1668 :    
1669 :     ----------------------------------------------------------------------
1670 :     Name: Matthias Blume
1671 : blume 920 Date: 2001/08/28 14:03:00 EDT
1672 :     Tag: blume-20010828-ml-lex
1673 :     Description:
1674 :    
1675 :     Fix for bug 1581, received from Neophytos Michael.
1676 :    
1677 :     ----------------------------------------------------------------------
1678 :     Name: Matthias Blume
1679 : blume 918 Date: 2001/08/27 11:20:00 EDT
1680 :     Tag: blume-20010827-readme11035
1681 :     Description:
1682 :    
1683 :     Fleshed out the README file for 110.35.
1684 :    
1685 :     ----------------------------------------------------------------------
1686 :     Name: Matthias Blume
1687 : blume 916 Date: 2001/08/24 17:10:00 EDT
1688 :     Tag: Release_110_35
1689 :     Description:
1690 :    
1691 :     New version number (110.35). New bootfiles.
1692 :    
1693 :     ----------------------------------------------------------------------
1694 : george 915 Name: Lal George
1695 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1696 :     Tag: george-20010824-MLRISC-graphs
1697 :     Description:
1698 :    
1699 :     removed clusters from MLRISC completely and replaced with graphs.
1700 :    
1701 :     ----------------------------------------------------------------------
1702 :     Name: Matthias Blume
1703 : blume 905 Date: 2001/08/23 17:50:00 EDT
1704 :     Tag: blume-20010823-toplevel
1705 :     Description:
1706 :    
1707 :     - some reorganization of the code that implements various kinds of
1708 :     environments in the compiler (static, dynamic, symbolic, combined)
1709 :     - re-implemented the EnvRef module so that evalStream works properly
1710 :     (if the stream contains references to "use", "CM.make", etc.)
1711 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1712 :    
1713 :     ----------------------------------------------------------------------
1714 :     Name: Matthias Blume
1715 : blume 904 Date: 2001/08/20 15:50 EDT
1716 :     Tag: blume20010820-slipup
1717 :     Description:
1718 :    
1719 :     I forgot to commit a few files. Here they are...
1720 :    
1721 :     ----------------------------------------------------------------------
1722 :     Name: Matthias Blume
1723 : blume 903 Date: 2001/08/20 15:35:00 EDT
1724 :     Tag: blume-20010820-debugprof
1725 :     Description:
1726 :    
1727 :     !!!! NEW BOOTFILES !!!!
1728 :    
1729 :     This is another round of reorganizing the compiler sources. This
1730 :     time the main goal was to factor out all the "instrumentation"
1731 :     passes (for profiling and backtracing) into their own library.
1732 :     The difficulty was to do it in such a way that it does not depend
1733 :     on elaborate.cm but only on elabdata.cm.
1734 :    
1735 :     Therefore there have been further changes to both elaborate.cm and
1736 :     elabdata.cm -- more "generic" things have been moved from the former
1737 :     to the latter. As a result, I was forced to split the assignment
1738 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1739 :     SML/NJ-specific. Since it would have been awkward to maintain,
1740 :     I bit the bullet and actually _changed_ the mapping between these
1741 :     numbers and primtycs. The bottom line of this is that you need
1742 :     a new set of bin- and bootfiles.
1743 :    
1744 :     I have built new bootfiles for all architectures, so doing a fresh
1745 :     checkout and config/install.sh should be all you need.
1746 :    
1747 :     The newly created library's name is
1748 :    
1749 :     $smlnj/viscomp/debugprof.cm
1750 :    
1751 :     and its sources live under
1752 :    
1753 :     src/compiler/DebugProf
1754 :    
1755 :     ----------------------------------------------------------------------
1756 :     Name: Matthias Blume
1757 : blume 902 Date: 2001/08/15 17:15:00 EDT
1758 :     Tag: blume-20010815-compreorg
1759 :     Description:
1760 :    
1761 :     This is a first cut at reorganizing the CM libraries that make up the
1762 :     core of the compiler. The idea is to separate out pieces that could
1763 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1764 :    
1765 :     The current status is a step in this direction, but it is not quite
1766 :     satisfactory yet. Expect more changes in the future.
1767 :    
1768 :     Here is the current (new) organization...
1769 :    
1770 :     What used to be $smlnj/viscomp/core.cm is now divided into
1771 :     six CM libraries:
1772 :    
1773 :     $smlnj/viscomp/basics.cm
1774 :     /parser.cm
1775 :     /elabdata.cm
1776 :     /elaborate.cm
1777 :     /execute.cm
1778 :     /core.cm
1779 :    
1780 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1781 :     All these libraries are proxy libraries that contain precisely
1782 :     one CM library component. Here are the locations of the components
1783 :     (all within the src/compiler tree):
1784 :    
1785 :     Basics/basics.cm
1786 :     Parse/parser.cm
1787 :     ElabData/elabdata.cm
1788 :     Elaborator/elaborate.cm
1789 :     Execution/execute.cm
1790 :     core.cm
1791 :    
1792 :     [This organization is the same that has been used already
1793 :     for a while for the architecture-specific parts of the visible
1794 :     compiler and for the old version of core.cm.]
1795 :    
1796 :     As you will notice, many source files have been moved from their
1797 :     respective original locations to a new home in one of the above
1798 :     subtrees.
1799 :    
1800 :     The division of labor between the new libraries is the following:
1801 :    
1802 :     basics.cm:
1803 :     - Simple, basic definitions that pertain to many (or all) of
1804 :     the other libraries.
1805 :     parser.cm:
1806 :     - The SML parser, producing output of type Ast.dec.
1807 :     - The type family for Ast is also defined and exported here.
1808 :     elabdata.cm:
1809 :     - The datatypes that describe input and output of the elaborator.
1810 :     This includes types, absyn, and static environments.
1811 :     elaborator.cm:
1812 :     - The SML/NJ type checker and elaborator.
1813 :     This maps an Ast.dec (with a given static environment) to
1814 :     an Absyn.dec (with a new static environment).
1815 :     - This libraries implements certain modules that used to be
1816 :     structures as functors (to remove dependencies on FLINT).
1817 :     execute.cm:
1818 :     - Everything having to do with executing binary code objects.
1819 :     - Dynamic environments.
1820 :     core.cm:
1821 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1822 :     - Top-level modules.
1823 :     - FLINT (this should eventually become its own library)
1824 :    
1825 :     Notes:
1826 :    
1827 :     I am not 100% happy with the way I separated the elaborator (and its
1828 :     data structures) from FLINT. Two instances of the same problem:
1829 :    
1830 :     1. Data structures contain certain fields that carry FLINT-specific
1831 :     information. I hacked around this using exn and the property list
1832 :     module from smlnj-lib. But the fact that there are middle-end
1833 :     specific fields around at all is a bit annoying.
1834 :    
1835 :     2. The elaborator calculates certain FLINT-related information. I tried
1836 :     to make this as abstract as I could using functorization, but, again,
1837 :     the fact that the elaborator has to perform calculations on behalf
1838 :     of the middle-end at all is not nice.
1839 :    
1840 :     3. Having to used exn and property lists is unfortunate because it
1841 :     weakens type checking. The other alternative (parameterizing
1842 :     nearly *everything*) is not appealing, though.
1843 :    
1844 :     I removed the "rebinding =" warning hack because due to the new organization
1845 :     it was awkward to maintain it. As a result, the compiler now issues some of
1846 :     these warnings when compiling init.cmi during bootstrap compilation. On
1847 :     the plus side, you also get a warning when you do, for example:
1848 :     val op = = Int32.+
1849 :     which was not the case up to now.
1850 :    
1851 :     I placed "assign" and "deref" into the _Core structure so that the
1852 :     code that deals with the "lazy" keyword can find them there. This
1853 :     removes the need for having access to the primitive environment
1854 :     during elaboration.
1855 :    
1856 :     ----------------------------------------------------------------------
1857 :     Name: Matthias Blume
1858 : blume 898 Date: 2001/08/13
1859 :     Tag: blume-20010813-closures
1860 :     Description:
1861 :    
1862 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1863 :     performance of certain loops by avoiding needless closure allocation.
1864 :    
1865 :     ----------------------------------------------------------------------
1866 : george 890 Name: Lal George
1867 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1868 :     Tag: george-20010731-x86-fmalloc
1869 :     Description: Fixed bug in x86 calls
1870 :    
1871 :     There was a bug where call instructions would mysteriously
1872 :     vanish. The call instruction had to be one that returned
1873 :     a floating point value.
1874 :    
1875 :     ----------------------------------------------------------------------
1876 :     Name: Lal George
1877 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1878 :     Tag: george-20010719-simple-cells
1879 :     Description:
1880 :    
1881 :     I have dramatically simplified the interface for CELLS in MLRISC.
1882 :    
1883 :     In summary, the cells interface is broken up into three parts:
1884 :    
1885 :     1. CellsBasis : CELLS_BASIS
1886 :    
1887 :     CellsBasis is a top level structure and common for all
1888 :     architectures. it contains the definitions of basic datatypes
1889 :     and utility functions over these types.
1890 :    
1891 :     2. functor Cells() : CELLS
1892 :    
1893 :     Cells generates an interface for CELLS that incorporates the
1894 :     specific resources on the target architecture, such as the
1895 :     presence of special register classes, their number and size,
1896 :     and various useful substructures.
1897 :    
1898 :     3. <ARCH>CELLS
1899 :    
1900 :     e.g. SparcCells: SPARCCELLS
1901 :    
1902 :     <ARCH>CELLS usually contains additional bindings for special
1903 :     registers on the architecture, such as:
1904 :    
1905 :     val r0 : cell (* register zero *)
1906 :     val y : cell (* Y register *)
1907 :     val psr : cell (* processor status register *)
1908 :     ...
1909 :    
1910 :     The structure returned by applying the Cells functor is opened
1911 :     in this interface.
1912 :    
1913 :     The main implication of all this is that the datatypes for cells is
1914 :     split between CellsBasis and CELLS -- a fairly simple change for user
1915 :     code.
1916 :    
1917 :     In the old scheme the CELLS interface had a definitional binding of
1918 :     the form:
1919 :    
1920 :     signature CELLS = sig
1921 :    
1922 :     structure CellsBasis = CellsBasis
1923 :    
1924 :     ...
1925 :    
1926 :     end
1927 :    
1928 :     With all the sharing constraints that goes on in MLRISC, this old
1929 :     design quickly leads to errors such as:
1930 :    
1931 :     "structure definition spec inside of sharing ... "
1932 :    
1933 :    
1934 :     and appears to require an unacceptable amount of sharing and where
1935 :     constraint hackery.
1936 :    
1937 :     I think this error message (the interaction of definitional specs and
1938 :     sharing) requires more explanation on our web page.
1939 :    
1940 :     ----------------------------------------------------------------------
1941 : blume 826 Name: Matthias Blume
1942 : blume 879 Date: 2001/07/19 15:00:00 EDT
1943 :     Tag: blume-20010719-libreorg
1944 :     Description:
1945 :    
1946 :     This update puts together a fairly extensive but straightforward change
1947 :     to the way the libraries that implement the interactive system are
1948 :     organized:
1949 :    
1950 :     The biggest change is the elimination of structure Compiler. As a
1951 :     replacement for this structure, there is now a CM library
1952 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1953 :     that exports all the substructures of the original structure Compiler
1954 :     directly. So instead of saying Compiler.Foo.bar one now simply
1955 :     says Foo.bar. (The CM libraries actually export a collection of
1956 :     structures that is richer than the collection of substructures of
1957 :     structure Compiler.)
1958 :    
1959 :     To make the transition smooth, there is a separate library called
1960 :     $smlnj/compiler/compiler.cm which puts together and exports the
1961 :     original structure Compiler (or at least something very close to it).
1962 :    
1963 :     There are five members of the original structure Compiler
1964 :     that are not exported directly but which instead became members
1965 :     of a new structure Backend (described by signature BACKEND). These are:
1966 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1967 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
1968 :     architecture (: string).
1969 :    
1970 :     Structure Compiler.Version has become structure CompilerVersion.
1971 :    
1972 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1973 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1974 :     or x86, respectively.
1975 :     Each of these exports the same frontend structures that
1976 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
1977 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1978 :     Hppa, PPC, Sparc, or X86, respectively.
1979 :    
1980 :     Library $smlnj/compiler/all.cm exports the union of the exports of
1981 :     $smlnj/compiler/<arch>.cm
1982 :    
1983 :     There are no structures <Arch>Compiler anymore, use
1984 :     $smlnj/compiler/<arch>.cm instead.
1985 :    
1986 :     Library host-compiler-0.cm is gone. Instead, the internal library
1987 :     that instantiates CM is now called cm0.cm. Selection of the host
1988 :     compiler (backend) is no longer done here but. (Responsibility for it
1989 :     now lies with $smlnj/compiler/current.cm. This seems to be more
1990 :     logical.)
1991 :    
1992 :     Many individual files have been moved or renamed. Some files have
1993 :     been split into multiple files, and some "dead" files have been deleted.
1994 :    
1995 :     Aside from these changes to library organization, there are also changes
1996 :     to the way the code itself is organized:
1997 :    
1998 :     Structure Binfile has been re-implemented in such a way that it no
1999 :     longer needs any knowledge of the compiler. It exclusively deals
2000 :     with the details of binfile layout. It no longer invokes the
2001 :     compiler (for the purpose of creating new prospective binfile
2002 :     content), and it no longer has any knowledge of how to interpret
2003 :     pickles.
2004 :    
2005 :     Structure Compile (: COMPILE) has been stripped down to the bare
2006 :     essentials of compilation. It no longer deals with linking/execution.
2007 :     The interface has been cleaned up considerably.
2008 :    
2009 :     Utility routines for dealing with linking and execution have been
2010 :     moved into their own substructures.
2011 :    
2012 :     (The ultimate goal of these changes is to provide a light-weight
2013 :     binfile loader/linker (at least for, e.g., stable libraries) that
2014 :     does not require CM or the compiler to be present.)
2015 :    
2016 :     CM documentation has been updated to reflect the changes to library
2017 :     organization.
2018 :    
2019 :     ----------------------------------------------------------------------
2020 :     Name: Matthias Blume
2021 : blume 875 Date: 2001/07/10 17:30:00 EDT
2022 :     Tag: Release_110_34
2023 :     Description:
2024 :    
2025 : blume 876 Minor tweak to 110.34 (re-tagged):
2026 :    
2027 :     - README.html file added to CVS repository
2028 :     - runtime compiles properly under FreeBSD 3.X and 4.X
2029 :    
2030 :     ----------------------------------------------------------------------
2031 :     Name: Matthias Blume
2032 :     Date: 2001/07/10 17:30:00 EDT
2033 :     Tag: Release_110_34
2034 :     Description:
2035 :    
2036 : blume 875 New version number (110.34). New bootfiles.
2037 :    
2038 :     ----------------------------------------------------------------------
2039 :     Name: Matthias Blume
2040 : blume 874 Date: 2001/07/09 16:00:00 EDT
2041 :     Tag: blume-20010709-more-varargs
2042 :     Description:
2043 :    
2044 :     I changed the handling of varargs in ml-nlffigen again:
2045 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
2046 :    
2047 :     The immediate effect is that you can actually call a varargs function
2048 :     from ML -- but you can't actually supply any arguments beyond the ones
2049 :     specified explicitly. (For example, you can call printf with its format
2050 :     string, but you cannot pass additional arguments.)
2051 :    
2052 :     This behavior is only marginally more useful than the one before, but
2053 :     it has the advantage that a function or, more importantly, a function
2054 :     type never gets dropped on the floor, thus avoiding follow-up problems with
2055 :     other types that refer to the offending one.
2056 :    
2057 :     ----------------------------------------------------------------------
2058 :     Name: Matthias Blume
2059 : blume 873 Date: 2001/07/09 11:25:00 EDT
2060 :     Tag: blume-20010709-varargs
2061 :     Description:
2062 :    
2063 :     1. ckit-lib.cm now exports structure Error
2064 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2065 :     with a warning accompanied by a source location. Moreover, it
2066 :     merely skips the offending function or type and proceeds with the
2067 :     rest of its work.u As a result, one can safely feed C code containing
2068 :     "..." to ml-nlffigen.
2069 :     3. There are some internal improvements to CM, providing slightly
2070 :     more general string substitutions in the tools subsystem.
2071 :    
2072 :     ----------------------------------------------------------------------
2073 :     Name: Matthias Blume
2074 : blume 854 Date: 2001/06/27 15:10:00 EDT
2075 :     Tag: blume-20010627-concur
2076 :     Description:
2077 :    
2078 :     Fixed a small bug in CM's handling of parallel compilation.
2079 :     (You could observe the bug by Control-C-interrupting an ordinary
2080 :     CMB.make or CM.stabilize and then attaching some compile servers.
2081 :     The result was that all of a sudden the previously interrupted
2082 :     compilation would continue on its own. This was because of
2083 :     an over-optimization: CM did not bother to clean out certain queues
2084 :     when no servers were attached "anyway", resulting in the contents
2085 :     of these queues to grab control when new servers did get attached.)
2086 :    
2087 :     There is also another minor update to the CM manual.
2088 :    
2089 :     ----------------------------------------------------------------------
2090 :     Name: Matthias Blume
2091 : blume 853 Date: 2001/06/26 16:15:00 EDT
2092 :     Tag: blume-20010626-cmdoc
2093 :     Description:
2094 :    
2095 :     Minor typo fixed in CM manual (syntax diagram for libraries).
2096 :    
2097 :     ----------------------------------------------------------------------
2098 :     Name: Matthias Blume
2099 : blume 852 Date: 2001/06/25 22:55:00 EDT
2100 :     Tag: blume-20010625-x86pc
2101 :     Description:
2102 :    
2103 :     Fixed a nasty bug in the X86 assembly code that caused signal
2104 :     handlers to fail (crash) randomly.
2105 :    
2106 :     ----------------------------------------------------------------------
2107 :     Name: Matthias Blume
2108 : blume 846 Date: 2001/06/25 12:05:00 EDT
2109 :     Tag: blume-20010625-nlffigen
2110 :     Description:
2111 :    
2112 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
2113 :     Nick Carter <nbc@andrew.cmu.edu>.
2114 :    
2115 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2116 :     2. Default names for generated files are now derived from the name of
2117 :     the C file *without its directory*. In particular, this causes generated
2118 :     files to be placed locally even if the C file is in some system directory.
2119 :     3. Default names for generated signatures and structures are also derived
2120 :     from the C file name without its directory. This avoids silly things
2121 :     like "structure GL/GL".
2122 :     (Other silly names are still possible because ml-nlffigen does not do
2123 : blume 847 a thorough check of whether generated names are legal ML identifiers.
2124 :     When in doubt, use command line arguments to force particular names.)
2125 : blume 846
2126 :     ----------------------------------------------------------------------
2127 :     Name: Matthias Blume
2128 : blume 845 Date: 2001/06/21 12:25:00 EDT
2129 :     Tag: blume-20010621-eXene
2130 :     Description:
2131 :    
2132 :     eXene now compiles and (sort of) works again.
2133 :    
2134 :     The library name (for version > 110.33) is $/eXene.cm.
2135 :    
2136 :     I also added an new example in src/eXene/examples/nbody. See the
2137 :     README file there for details.
2138 :    
2139 :     ----------------------------------------------------------------------
2140 :     Name: Matthias Blume
2141 : blume 844 Date: 2001/06/20 16:40:00 EDT
2142 :     Tag: blume-20010620-cml
2143 :     Description:
2144 :    
2145 :     CML now compiles and works again.
2146 :    
2147 :     Libraries (for version > 110.33):
2148 :    
2149 :     $cml/cml.cm Main CML library.
2150 :     $cml/basis.cm CML's version of $/basis.cm.
2151 :     $cml/cml-internal.cm Internal helper library.
2152 :     $cml/core-cml.cm Internal helper library.
2153 :     $cml-lib/trace-cml.cm Tracing facility.
2154 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2155 :    
2156 :     The installer (config/install.sh) has been taught how to properly
2157 :     install this stuff.
2158 :    
2159 :     ----------------------------------------------------------------------
2160 :     Name: Matthias Blume
2161 : blume 843 Date: 2001/06/19 17:55:00 EDT
2162 :     Tag: blume-20010619-instantiate
2163 :     Description:
2164 :    
2165 :     This un-breaks the fix for bug 1432.
2166 :     (The bug was originally fixed in 110.9 but I broke it again some
2167 :     time after that.)
2168 :    
2169 :     ----------------------------------------------------------------------
2170 :     Name: Matthias Blume
2171 : blume 842 Date: 2001/06/19 17:25:00 EDT
2172 :     Tag: blume-20010619-signals
2173 :     Description:
2174 :    
2175 :     This should (hopefully) fix the long-standing signal handling bug.
2176 :     (The runtime system was constructing a continuation record with an
2177 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2178 :    
2179 :     ----------------------------------------------------------------------
2180 :     Name: Matthias Blume
2181 : blume 840 Date: 2001/06/15 15:05:00 EDT
2182 : blume 841 Tag: blume-20010615-moresparc
2183 :     Description:
2184 :    
2185 :     Here is a short late-hour update related to Sparc c-calls:
2186 :    
2187 :     -- made handling of double-word arguments a bit smarter
2188 :    
2189 :     -- instruction selection phase tries to collapse certain clumsily
2190 :     constructed ML-Trees; typical example:
2191 :    
2192 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2193 :    
2194 :     This currently has no further impact on SML/NJ since mlriscGen does
2195 :     not seem to generate such patterns in the first place, and c-calls
2196 :     (which did generate them in the beginning) has meanwhile been fixed
2197 :     so as to avoid them as well.
2198 :    
2199 :     ----------------------------------------------------------------------
2200 :     Name: Matthias Blume
2201 :     Date: 2001/06/15 15:05:00 EDT
2202 : blume 840 Tag: blume-20010615-sparc
2203 :     Description:
2204 :    
2205 :     The purpose of this update is to provide an implementation of NLFFI
2206 :     on Sparc machines.
2207 :    
2208 :     Here are the changes in detail:
2209 :    
2210 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2211 :     the Sparc implementation of the c-calls API.
2212 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2213 :     for accessing the ML frame. Thus, we have a real frame pointer and
2214 :     can freely modify %sp without need for an omit-frame-ptr phase.
2215 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2216 :     this case.
2217 :     * ml-nlffigen has been taught to produce code for different architectures
2218 :     and calling conventions.
2219 :     * In a way similar to what was done in the x86 case, the Sparc
2220 :     backend uses its own specific extension to mltree. (For example,
2221 :     it needs to be able to generate UNIMP instructions which are part
2222 :     of the calling convention.)
2223 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2224 :     to make it easier to plug in new machine- and os-dependent parts).
2225 :    
2226 :     There are some other fairly unrelated bug fixes and cleanups as well:
2227 :    
2228 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2229 :     that they properly share their libraries with existing SML/NJ libraries.
2230 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2231 :     follow-up error messages.
2232 :     * Updates to CM/CMB documentation.
2233 :    
2234 :     TODO items:
2235 :    
2236 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2237 :     (The current %o2 is a really bad choice because it is part of the
2238 :     calling conventions, so things might interfere in unexpected ways.)
2239 :    
2240 :     ----------------------------------------------------------------------
2241 :     Name: Matthias Blume
2242 : blume 839 Date: 2001/06/07
2243 :     Tag: blume-20010607-calls
2244 :     Description:
2245 :    
2246 :     A number of internal changes related to C calls and calling conventions:
2247 :    
2248 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2249 :     number of bytes popped implicitly (by the callee). In most cases
2250 :     this field is 0 but on x86/win32 it is some non-zero value. This
2251 :     is information provided for the benefit of the "omit-frameptr" pass.
2252 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2253 :     The instruction selection phase copies its value from the ML-Tree
2254 :     CALL statement.
2255 :     3. On all other architectures, the instruction selection phase checks
2256 :     whether "pops=0" and complains if not.
2257 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2258 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2259 :     up after the call and pops is set to 0. For "stdcall", the caller
2260 :     does nothing, leaving the cleanup to the callee; pops is set to
2261 :     the number of bytes that were pushed onto the stack.
2262 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2263 :     distinguish between "ccall" and "stdcall".
2264 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2265 :     set. (This is needed for implementing the official C calling convention
2266 :     on this architecture.)
2267 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2268 :     work with the latest CM.
2269 :    
2270 :     ----------------------------------------------------------------------
2271 :     Name: Matthias Blume
2272 : blume 838 Date: 2001/06/05 15:10:00 EDT
2273 :     Tag: blume-20010605-cm-index
2274 :     Description:
2275 :    
2276 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2277 :    
2278 :     1. CM can now generate "index files". These are human-readable files
2279 :     that list on a per-.cm-file basis each toplevel symbol defined or
2280 :     imported. The location of the index file for
2281 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2282 :     To enable index-file generation, set CM.Control.generate_index to true
2283 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2284 :    
2285 :     The CM manual has been updated accordingly.
2286 :    
2287 :     2. I made some slight modifications to the c-calls API in MLRISC.
2288 :    
2289 :     a) There is now a callback to support saving/restoring of
2290 :     dedicated but caller-save registers around the actual call
2291 :     instruction.
2292 :     b) One can optionally specify a comment-annotation for the
2293 :     call instruction.
2294 :    
2295 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2296 :     (For example, the comment annotation shows the C prototype of
2297 :     the function being called.)
2298 :    
2299 :     ----------------------------------------------------------------------
2300 :     Name: Matthias Blume
2301 : blume 837 Date: 2001/06/01 13:30:00 EDT
2302 :     Tag: blume-20010601-nlffi-cleanup
2303 :     Description:
2304 :    
2305 :     This is mostly a cleanup of MLFFI stuff:
2306 :    
2307 :     - some signature files have been put into a more exposed place
2308 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2309 :     - ml-nlffigen changed accordingly
2310 :     - tutorial updated
2311 :    
2312 :     Other changes:
2313 :    
2314 :     - author's affiliation in CM manual(s) updated
2315 :     - some more recognized keywords added to Allen's sml.sty
2316 :    
2317 :     ----------------------------------------------------------------------
2318 :     Name: Matthias Blume
2319 : blume 836 Date: 2001/05/25 15:30:00 EDT
2320 :     Tag: blume-20010525-iptr
2321 :     Description:
2322 :    
2323 :     - put the official 110.33-README (as it appears on the ftp server) under
2324 :     CVS
2325 :     - fixed a small bug related to incomplete pointer types in
2326 :     ml-nlffigen
2327 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2328 :     (it does not need the 'f type parameter)
2329 :    
2330 :     ----------------------------------------------------------------------
2331 :     Name: Matthias Blume
2332 : blume 834 Date: 2001/05/23 14:30:00 EDT
2333 :     Tag: Release_110_33
2334 :     Description:
2335 :    
2336 :     New version number (110.33). New bootfiles.
2337 :    
2338 :     ----------------------------------------------------------------------
2339 :     Name: Matthias Blume
2340 : blume 833 Date: 2001/05/22 18:06:00 EDT
2341 :     Tag: blume-20010522-targets
2342 :     Description:
2343 :    
2344 :     Made install.sh use file config/targets.customized if it exists, falling
2345 :     back to config/targets if it doesn't. This way one can have a customized
2346 :     version of the targets file without touching the "real thing", thus
2347 :     eliminating the constant fear of accidentally checking something bogus
2348 :     back into the CVS repository... (File config/targets.customized must
2349 :     not be added to the repository!)
2350 :    
2351 :     ----------------------------------------------------------------------
2352 :     Name: Matthias Blume
2353 : blume 832 Date: 2001/05/22 16:30:00 EDT
2354 :     Tag: blume-20010522-minitut
2355 :     Description:
2356 :    
2357 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2358 :     struct returns.
2359 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2360 :     incomplete, preliminary documentation for NLFFI.
2361 :    
2362 :     ----------------------------------------------------------------------
2363 :     Name: Matthias Blume
2364 : blume 829 Date: 2001/05/14 11:30:00 EDT
2365 :     Tag: blume-20010514-script
2366 :     Description:
2367 :    
2368 :     Some bugs in install script fixed.
2369 :    
2370 :     In addition to that I also made a slight change to the NLFFI API:
2371 :     Functors generated by ml-nlffigen now take the dynamic library as a
2372 :     straight functor argument, not as a suspended one. (The original
2373 :     functor code used to force the suspension right away anyway, so there
2374 :     was nothing gained by this complication of the interface.)
2375 :    
2376 :     ----------------------------------------------------------------------
2377 :     Name: Matthias Blume
2378 : blume 828 Date: 2001/05/11 14:35:00 EDT
2379 :     Tag: blume-20010511-ml-nlffi
2380 :     Description:
2381 :    
2382 :     I finally took the plunge and added my new FFI code to the main
2383 :     repository. For x86-linux it is now ready for prime-time.
2384 :    
2385 :     There are two new subdirectories of "src":
2386 :    
2387 :     - ml-nlffi-lib:
2388 :     The utility library for programs using the FFI interface.
2389 :     Here is the implementation of $/c.cm and its associated low-level
2390 :     partners $/c-int.cm and $/memory.cm.
2391 :     - ml-nlffigen:
2392 :     A stand-alone program for generating ML glue code from C source
2393 :     code.
2394 :    
2395 :     Building ml-nlffigen requires $/ckit-lib.cm.
2396 :    
2397 :     The config/install.sh script has been updates to do the Right Thing
2398 :     (hopefully).
2399 :    
2400 :     Notice that the source tree for the C-Kit will not be put under "src"
2401 :     but directly under the installation root directory. (This is the
2402 :     structure that currently exists on the CVS server when you check out
2403 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
2404 :    
2405 :     Bugs: No documentation yet.
2406 :    
2407 :     ----------------------------------------------------------------------
2408 :     Name: Matthias Blume
2409 : blume 826 Date: 2001/05/09 16:35:00 EDT
2410 :     Tag: blume-20010509-cpscontract
2411 :     Description:
2412 :    
2413 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
2414 :     wrapper/unwrapper elimination did not decrement usage counts and some
2415 :     dead variables got overlooked by the dead-up logic.)
2416 :    
2417 :     ----------------------------------------------------------------------
2418 : george 824 Name: Lal George
2419 :     Date: 2001/05/08 17:26:09 EDT
2420 :     Tag: george-20010508-omit-frameptr
2421 :     Description:
2422 :    
2423 :     Changes to implement the omit-frame-pointer optimization to support
2424 :     raw C calls. For now, there is only support on the Intel x86, but
2425 :     other architectures will follow as more experience is gained with this.
2426 :    
2427 :    
2428 :     ----------------------------------------------------------------------
2429 : blume 818 Name: Matthias Blume
2430 : blume 819 Date: 2001/05/07 14:40:00 EDT
2431 :     Tag: blume-20010507-proxies
2432 :     Description:
2433 :    
2434 :     I made into "proxy libraries" all libraries that qualify for such a
2435 :     change. (A qualifying library is a library that has another library or
2436 :     groups as its sole member and repeats that member's export list
2437 :     verbatim. A proxy library avoids this repetition by omitting its export
2438 :     list, effectively inheriting the list that its (only) member exports.
2439 :     See the CM manual for more explanation.)
2440 :     The main effect is that explicit export lists for these libraries
2441 :     do not have to be kepts in sync, making maintenance a bit easier.
2442 :    
2443 :     I also added copyright notices to many .cm-files.
2444 :    
2445 :     Last but not least, I made a new set of bootfiles.
2446 :    
2447 :     ----------------------------------------------------------------------
2448 :     Name: Matthias Blume
2449 : blume 818 Date: 2001/05/04 17:00:00 EDT
2450 :     Tag: blume-20010504-cm-lsplit
2451 :     Description:
2452 :    
2453 :     0. John merged pending changes to $/smlnj-lib.cm
2454 :    
2455 :     1. Allen's previous change accidentally backed out of one of Lal's
2456 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2457 :    
2458 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2459 :     within the compiler where applicable. There is some code simplification
2460 :     because of that.
2461 :    
2462 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2463 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2464 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2465 :     it can now also be controlled on a per-source basis: CM has been taught
2466 :     a new tool parameter applicable to ML source files.
2467 :    
2468 :     - To turn lambda-splitting off completely:
2469 :     local open Compiler.Control.LambdaSplitting in
2470 :     val _ = set Off
2471 :     end
2472 :     - To make "no lambda-splitting" the global default (but allow per-source
2473 :     overriding); this is the initial setting:
2474 :     local open Compiler.Control.LambdaSplitting in
2475 :     val _ = set (Default NONE)
2476 :     end
2477 :     - To make "lambda-splitting with aggressiveness a" the global default
2478 :     (and allow per-source overriding):
2479 :     local open Compiler.Control.LambdaSplitting in
2480 :     val _ = set (Default (SOME a))
2481 :     end
2482 :    
2483 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2484 :     write (in the respective .cm-file):
2485 :     a.sml (lambdasplitting:off)
2486 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2487 :     a.sml (lambdasplitting:on)
2488 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2489 :     <a> is a decimal non-negative integer):
2490 :     a.sml (lambdasplitting:<a>)
2491 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2492 :     a.sml (lambdasplitting:infinity)
2493 :     - To use the global default for a.sml:
2494 :     a.sml (lambdasplitting:default)
2495 :     or simply
2496 :     a.sml
2497 :    
2498 :     ----------------------------------------------------------------------
2499 : leunga 815 Name: Allen Leung
2500 :     Date: 2001/05/04 01:57:00 EDT
2501 :     Tag: leunga-20010504-sync
2502 :     Description:
2503 : leunga 788
2504 : leunga 815 MLRISC features.
2505 :    
2506 :     1. Fix to CMPXCHG instructions.
2507 :     2. Changed RA interface to allow annotations in callbacks.
2508 :     3. Added a new method to the stream interface to allow annotations updates.
2509 :    
2510 : george 749 ----------------------------------------------------------------------
2511 : blume 812 Name: Matthias Blume
2512 : blume 814 Date: 2001/05/01 11:45:00 EDT
2513 :     Tag: blume-20010501-pcedittmp
2514 :     Description:
2515 :    
2516 :     Changed install.sh to use the current working directory instead of
2517 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2518 :     of /usr/tmp caused trouble with MacOS X because of file premission
2519 :     problems.
2520 :    
2521 :     ----------------------------------------------------------------------
2522 :     Name: Matthias Blume
2523 : blume 812 Date: 2001/04/20 11:10:00 EDT
2524 :     Tag: blume-20010420-inMLflag
2525 :     Description:
2526 :    
2527 :     - added vp_limitPtrMask to vproc-state.h
2528 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2529 :     handling)
2530 :     - made the ML compiler aware of various data-structure offsets so it
2531 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2532 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2533 :     raw C-calls
2534 :    
2535 :     ----------------------------------------------------------------------
2536 : george 810 Name: Lal George
2537 :     Date: 2001/04/20 09:15:28 EDT
2538 :     Tag: george-20010420-macosX
2539 :     Description:
2540 :    
2541 :     - Changes to port to Mac OS X; Darwin.
2542 :    
2543 :     - In the process I found that sqrt was broken on the PPC, because the
2544 :     fsqrt instruction is not implemented.
2545 :    
2546 :     ----------------------------------------------------------------------
2547 : blume 799 Name: Matthias Blume
2548 : blume 808 Date: 2001/04/18 12:45:00 EDT
2549 :     Tag: blume-20010418-ccalls
2550 :     Description:
2551 :    
2552 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2553 :     (this bug prevented structure arguments containing pointers from being
2554 :     passed correctly)
2555 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
2556 :     structure arguments are represented as a pointer to the beginning
2557 :     of the structure (instead of having a series of synthesized arguments,
2558 :     one for each structure member)
2559 :    
2560 :     - made makeml script's verbosity level configurable via environment
2561 :     variable (MAKEML_VERBOSITY)
2562 :    
2563 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2564 : blume 808 in rawmem-x86.sml; we are now using the real thing
2565 :    
2566 :     ----------------------------------------------------------------------
2567 :     Name: Matthias Blume
2568 : blume 806 Date: 2001/03/22 16:25:00 EST
2569 :     Tag: blume-20010322-bootfiles
2570 :     Description:
2571 :    
2572 :     Created a new set of bootfiles (for your automatic installation convenience).
2573 :    
2574 :     ----------------------------------------------------------------------
2575 :     Name: Matthias Blume
2576 : blume 805 Date: 2001/03/22 15:10:00 EST
2577 :     Tag: blume-20010322-rawmem-parcm
2578 :     Description:
2579 :    
2580 :     1. All "raw memory access" primitives for the new FFI are implemented now
2581 :     (at least on the x86).
2582 :     2. Some further cleanup of CM's parallel make mechanism.
2583 :    
2584 :     ----------------------------------------------------------------------
2585 :     Name: Matthias Blume
2586 : blume 801 Date: 2001/03/19 17:53:00 EST
2587 :     Tag: blume-20010319-parallel
2588 :     Description:
2589 :    
2590 :     Parallel make (using compile servers) now works again.
2591 :    
2592 :     To this end, CM.stabilize and CMB.make have been modified to work in
2593 :     two passes when compile servers are attached:
2594 :     1. Compile everything, do not perform stabilization; this pass
2595 :     uses compile servers
2596 :     2. Stabilize everything; this pass does not use compile servers
2597 :     If there are no compile servers, the two passes are combined into one
2598 :     (as before). Splitting the passes increases the inherent parallelism
2599 :     in the dependency graph because the entire graph including all
2600 :     libraries is available at the same time. This, in turn, improves
2601 :     server utilization. The downside is that the master process will
2602 :     have to do some extra work after compilation is done (because for
2603 :     technical reasons it must re-read all the binfiles during stabilization).
2604 :    
2605 :     ----------------------------------------------------------------------
2606 :     Name: Matthias Blume
2607 : blume 800 Date: 2001/03/16 12:22:00 EST
2608 :     Tag: blume-20010316-bootfiles
2609 :     Description:
2610 :    
2611 :     Created a new set of bootfiles (for your automatic installation convenience).
2612 :    
2613 :     ----------------------------------------------------------------------
2614 :     Name: Matthias Blume
2615 : blume 799 Date: 2001/03/16 11:00:00 EST
2616 :     Tag: blume-20010316-MLTREE-fixup
2617 :     Description:
2618 :    
2619 :     This is a minor fixup for an (untagged) earlier commit by Allen.
2620 :     (A file was missing).
2621 :    
2622 :     ----------------------------------------------------------------------
2623 : leunga 788 Name: Allen Leung
2624 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
2625 :     Tag: leunga-20010305-cut-support
2626 :    
2627 :     1. New support for alternative control-flow in MLTREE.
2628 :     Currently we support
2629 :    
2630 :     FLOW_TO(CALL ...., [k1,...,kn])
2631 :    
2632 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
2633 :     in Moby
2634 :    
2635 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2636 :    
2637 :     2. Register Allocator
2638 :    
2639 :     Changes in interface [from Fermin, John]
2640 :    
2641 :     3. Alpha 8-bit SLL support [Fermin]
2642 :    
2643 :     4. All architectures
2644 :    
2645 :     A new module (ClusterExpandCopies) for expanding parallel copies.
2646 :    
2647 :     ----------------------------------------------------------------------
2648 :     Name: Allen Leung
2649 : leunga 788 Date: 2001/02/27 23:07:00 EST
2650 :     Tag: leunga-20010227-minor-stuff
2651 :    
2652 :     1. Alpha bug fix for CMOVNE
2653 :     2. Handle mltree COND(..,FCMP ...,...)
2654 :     3. Bug fix in simplifier
2655 :    
2656 :     ----------------------------------------------------------------------
2657 : blume 777 Name: Matthias Blume
2658 : blume 781 Date: 2001/01/30 17:50:00 EST
2659 :     Tag: blume-20010130-sync
2660 :     Description:
2661 :    
2662 :     This is just a minor update to sync my devel branch with the main brach.
2663 :     The only visible change is the addition of some README files.
2664 :    
2665 :     ----------------------------------------------------------------------
2666 :     Name: Matthias Blume
2667 : blume 778 Date: 2001/01/12 23:30:00 JST
2668 :     Tag: blume-20010112-bootfiles
2669 :     Description:
2670 :    
2671 :     Made a new set of bootfiles that goes with the current state of the
2672 :     repository.
2673 :    
2674 :     ----------------------------------------------------------------------
2675 :     Name: Matthias Blume
2676 : blume 777 Date: 2001/01/12 21:20:00 JST
2677 :     Tag: blume-20010112-sync
2678 :     Description:
2679 :    
2680 :     I am just flushing out some minor changes that had accumulated in
2681 :     my private branch in order to sync with the main tree. (This is
2682 :     mainly because I had CVS trouble when trying to merge _into_ my
2683 :     private branch.)
2684 :    
2685 :     Most people should be completely unaffected by this.
2686 :    
2687 :     ----------------------------------------------------------------------
2688 : leunga 775 Name: Allen Leung
2689 :     Date: Thu Jan 11 21:03:00 EST 2001
2690 :     Tag: leunga-20010111-labexp=mltree
2691 :     Description:
2692 :    
2693 :     1. Removed the type LabelExp and replace it by MLTree.
2694 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2695 :     3. There were some bugs in alpha code generator which would break
2696 :     64-bit code generation.
2697 :     4. Redo the tools to generate code with the
2698 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2699 :     are now generated by perl scripts.
2700 :    
2701 :     ----------------------------------------------------------------------
2702 : blume 771 Name: Matthias Blume
2703 : blume 774 Date: 2001/01/10 21:55:00 JST
2704 :     Tag: blume-20010110-rcc
2705 :     Description:
2706 :    
2707 :     The RCC stuff now seems to work (but only on the x86).
2708 :     This required hacking of the c-calls interface (and -implementation) in
2709 :     MLRISC.
2710 :    
2711 :     Normal compiler users should be unaffected.
2712 :    
2713 :     ----------------------------------------------------------------------
2714 :     Name: Matthias Blume
2715 : blume 773 Date: 2001/01/09 01:20:00 JST
2716 :     Tag: blume-20010109-rcc
2717 :     Description:
2718 :    
2719 :     This is a fairly big patch, flushing out a large number of pending
2720 :     changes that I made to my development copy over the last couple of days.
2721 :    
2722 :     Of practical relevance at this moment is a workaround for a pickling
2723 :     bug that Allen ran into the other day. The cause of the bug itself is
2724 :     still unknown and it might be hard to fix it properly, but the
2725 :     workaround has some merits of its own (namely somewhat reducing pickling
2726 :     overhead for certain libraries). Therefore, I think this solution should
2727 :     be satisfactory at this time.
2728 :    
2729 :     The rest of the changes (i.e., the vast majority) has to do with my
2730 :     ongoing efforts of providing direct support for C function calls from
2731 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2732 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2733 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2734 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2735 :     handle RCC.
2736 :    
2737 :     The last part (the changes to mlriscGen.sml) are still known to be
2738 :     wrong on the x86 and not implemented on all other architectures. But
2739 :     the infrastructure is in place. I had to change a few functor
2740 :     signatures in the backend to be able to route the CCalls interface
2741 :     from MLRISC there, and I had to specialize the mltree type (on the
2742 :     x86) to include the necessary extensions. (The extensions themselves
2743 :     were already there and redy to go in MLRISC/x86).
2744 :    
2745 :     Everything should be very happy as soon as someone helps me with
2746 :     mlriscGen.sml...
2747 :    
2748 :     In any case, nothing of this should matter to anyone as long as the
2749 :     new primop is not being used (which is going to be the case unless you
2750 :     find it where I hid it :). The rest of the compiler is completely
2751 :     unaffected.
2752 :    
2753 :     ----------------------------------------------------------------------
2754 :     Name: Matthias Blume
2755 : blume 772 Date: 2001/01/05 00:30:00 JST
2756 :     Tag: blume-20010105-primops
2757 :     Description:
2758 :    
2759 :     Added some experimental support for work that I am doing right now.
2760 :     These changes mostly concern added primops, but there is also a new
2761 :     experimental C library in the runtime system (but currently not enabled
2762 :     anywhere except on Linux/X86).
2763 :    
2764 :     In the course of adding primops (and playing with them), I discovered that
2765 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2766 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2767 :     To recover, I made type information in INL_PRIM mandatory and changed
2768 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2769 :     has complete, correct type information (i.e., no bottom types).
2770 :    
2771 :     Since all these changes mean that we need new binfiles, I also bumped the
2772 :     version number to 110.32.1.
2773 :    
2774 :     ----------------------------------------------------------------------
2775 :     Name: Matthias Blume
2776 : blume 771 Date: 2000/12/30 22:10:00 JST
2777 :     Tag: blume-20001230-various
2778 :     Description:
2779 :    
2780 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2781 :     to each other using path anchors. (See CM manual for explanation.)
2782 :    
2783 :     Updated CM documentation.
2784 :    
2785 :     Fixed some bugs in CM.
2786 :    
2787 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2788 :    
2789 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2790 :    
2791 :     Added cleanup handler for exportML to reset timers and compiler stats.
2792 :    
2793 :     ----------------------------------------------------------------------
2794 : george 761 Name: Lal George
2795 :     Date: 2000/12/22 22:22:58 EST 2000
2796 :     Tag: Release_110_32
2797 :     Description:
2798 :    
2799 :     Infinite precision used throughout MLRISC.
2800 :     see MLRISC/mltree/machine-int.sig
2801 :    
2802 :     ----------------------------------------------------------------------
2803 : blume 756 Name: Matthias Blume
2804 : blume 760 Date: 2000/12/22 23:16:00 JST
2805 :     Tag: blume-20001222-warn
2806 :     Description:
2807 : blume 759
2808 : blume 760 Corrected wording and formatting of some CM warning message which I
2809 :     broke in my previous patch.
2810 : blume 759
2811 :     ----------------------------------------------------------------------
2812 :     Name: Matthias Blume
2813 : blume 763 Date: 2000/12/22 21:20:00 JST
2814 :     Tag: blume-20001222-anchorenv
2815 :     Description:
2816 :    
2817 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2818 :    
2819 :     ----------------------------------------------------------------------
2820 :     Name: Matthias Blume
2821 : blume 758 Date: 2000/12/22 13:15:00 JST
2822 :     Tag: blume-20001222-cleanup
2823 :     Description:
2824 :    
2825 :     Removed src/cm/ffi which does not (and did not) belong here.
2826 :    
2827 :     ----------------------------------------------------------------------
2828 :     Name: Matthias Blume
2829 : blume 757 Date: 2000/12/21 23:55:00 JST
2830 :     Tag: blume-20001221-exn
2831 :     Description:
2832 :    
2833 :     Probably most important: CM no longer silently swallows all exceptions
2834 :     in the compiler.
2835 :     Plus: some other minor CM changes. For example, CM now reports some
2836 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2837 :    
2838 :     ----------------------------------------------------------------------
2839 :     Name: Matthias Blume
2840 : blume 756 Date: 2000/12/15 00:01:05 JST
2841 :     Tag: blume-20001215-dirtool
2842 :     Description:
2843 :    
2844 :     - "dir" tool added.
2845 :     - improvements and cleanup to Tools structure
2846 :     - documentation updates
2847 :    
2848 :     ----------------------------------------------------------------------
2849 : leunga 752 Name: Allen Leung
2850 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2851 :     Description:
2852 :     Tag: leunga-20001214-int-inf
2853 :     Description:
2854 :    
2855 :     In IntInf, added these standard functions, which are missing from our
2856 :     implementation:
2857 :    
2858 :     andb : int * int -> int
2859 :     xorb : int * int -> int
2860 :     orb : int * int -> int
2861 :     notb : int -> int
2862 :     << : int * word -> int
2863 :     ~>> : int * word -> int
2864 :    
2865 :     Not tested, I hope they are correct.
2866 :    
2867 :     ----------------------------------------------------------------------
2868 :     Name: Allen Leung
2869 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2870 :     Description:
2871 :     Tag: leunga-20001208-nowhere
2872 :     Description:
2873 :    
2874 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2875 :     to generate better error messages etc. Plus a brief manual.
2876 :    
2877 :     ----------------------------------------------------------------------
2878 : george 749 Name: Lal George
2879 :     Date: 2000/12/08 09:54:02 EST 2000
2880 :     Tag: Release_110_31
2881 :     Description:
2882 : george 705
2883 : george 749 - Version 110.31
2884 : leunga 591 ----------------------------------------------------------------------
2885 : leunga 744 Name: Allen Leung
2886 :     Date: Thu Dec 7 22:01:04 EST 2000
2887 :     Tag: leunga-20001207-cell-monster-hack
2888 :     Description:
2889 :    
2890 :     Major MLRISC internal changes. Affect all clients.
2891 :     Summary:
2892 :    
2893 :     1. Type CELLS.cell = int is now replaced by a datatype.
2894 :     As a result, the old regmap is now gone. Almost all interfaces
2895 :     in MLRISC change as a consequence.
2896 :    
2897 :     2. A new brand version of machine description tool (v3.0) that generates
2898 :     modules expecting the new interface. The old version is removed.
2899 :    
2900 :     3. The RA interface has been further abstracted into two new functors.
2901 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2902 :     [See also directory MLRISC/demo.]
2903 :    
2904 :     4. Some other new source->source code generation tools are available:
2905 :    
2906 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2907 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2908 :     I use this tool to generate the peephole optimizers---with the new
2909 :     cell type changes, peephole rules are becoming difficult to write
2910 :     without conditional pattern matching.
2911 :    
2912 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2913 :     cover the entire MLRISC source tree so many things broke.
2914 :    
2915 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2916 :     They are moved because there are a lot of them and they clutter up the
2917 :     root dir.
2918 :    
2919 :     7. More detailed documentation to come...
2920 :    
2921 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2922 :     a makeml -rebuild first. This is because of other other
2923 :     changes that Matthias has made (see below).
2924 :    
2925 :    
2926 :     ----------------------------------------------------------------------
2927 : blume 733 Name: Matthias Blume
2928 : blume 742 Date: 2000/11/30 23:12:00 JST
2929 :     Tag: blume-20001130-filereorg
2930 :     Description:
2931 :    
2932 :     Some manual updates and some file reorganizations in CM.
2933 :    
2934 :     ----------------------------------------------------------------------
2935 :     Name: Matthias Blume
2936 : blume 737 Date: 2000/11/24 17:45:00 JST
2937 :     Tag: blume-20001124-link
2938 :     Description:
2939 :    
2940 :     Drastically improved link traversal code for the case that the dynamic
2941 :     value was already loaded at bootstrap time. As a result, CM and CMB
2942 :     now both load blazingly fast -- even on a very slow machine. Also,
2943 :     memory consumption has been further reduced by this.
2944 :    
2945 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2946 :     bootstrap you have to do this:
2947 :    
2948 :     1. Run CMB.make
2949 :     2. Make a symbolic link for the boot directory:
2950 :     ln -s sml.boot.ARCH-OS xxx
2951 :     3. "Rebuild" the boot directory:
2952 :     ./makeml -boot xxx -rebuild sml ; rm xxx
2953 :     4. Boot normally:
2954 :     ./makeml
2955 :    
2956 :     ----------------------------------------------------------------------
2957 :     Name: Matthias Blume
2958 : blume 735 Date: 2000/11/21 21:20:00 JST
2959 :     Tag: blume-20001121-tools
2960 :     Description:
2961 :    
2962 :     Continued hacking on autoloading problem -- with success this time.
2963 :     Also changed tool-plugin mechanism. See new CM manual.
2964 :    
2965 :     ----------------------------------------------------------------------
2966 :     Name: Matthias Blume
2967 : blume 734 Date: 2000/11/19 14:30:00 JST
2968 :     Tag: blume-20001119-autoload
2969 :     Description:
2970 :    
2971 :     Some hacking to make autoloading faster. Success for CMB, no success
2972 :     so far for CM. There is a reduced structure CM' that autoloads faster.
2973 :     (This is a temporary, non-documented hack to be eliminated again when
2974 :     the general problem is solved.)
2975 :    
2976 :     ----------------------------------------------------------------------
2977 :     Name: Matthias Blume
2978 : blume 733 Date: 2000/11/17 14:10:00 JST
2979 :     Tag: blume-20001117-pickle-lib
2980 :     Description:
2981 :    
2982 :     1. Eliminated comp-lib.cm
2983 :     2. Made pickle-lib.cm
2984 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
2985 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
2986 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
2987 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
2988 :     Duplicated intmap modules thrown out.
2989 :     6. Hunted down all duplicated SCC code and replaced it with
2990 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
2991 :     7. Rewrote Feedback module.
2992 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
2993 :     should be thrown out and equivalent modules from smlnj-lib.cm
2994 :     should be used (IntRedBlackSet, IntListSet, ...).
2995 :    
2996 :     Confirmed that compiler compiles to fixpoint.
2997 :    
2998 :     ----------------------------------------------------------------------
2999 : leunga 731 Name: Allen Leung
3000 :     Date: 2000/11/10 18:00:00
3001 :     Tag: leunga-20001110-new-x86-fp
3002 :    
3003 :     A new x86 floating point code generator has been added.
3004 :     By default this is turned off. To turn this on, do:
3005 :    
3006 :     CM.autoload "$smlnj/compiler.cm";
3007 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3008 :    
3009 :     Changes:
3010 :    
3011 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
3012 :     2. Changed the extension callback for FTANGENT to generate:
3013 :    
3014 :     fptan
3015 :     fstp %st(0)
3016 :     instead of
3017 :     fptan
3018 :     fstpl ftempmem
3019 :    
3020 :     3. Numerous assembly fixes for x86.
3021 :    
3022 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
3023 :     support for a whole bunch of instructions and addressing modes:
3024 :    
3025 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
3026 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
3027 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
3028 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3029 :     fxch %st(n)
3030 :     fld %st(n)
3031 :     fst %st(n)
3032 :     fst mem
3033 :     fstp %st(n)
3034 :     fucom %st(n)
3035 :     fucomp %st(n)
3036 :    
3037 :     All these are now generated when the fast fp mode is turned on.
3038 :    
3039 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3040 :    
3041 :     ----------------------------------------------------------------------
3042 : blume 729 Name: Matthias Blume
3043 :     Date: 2000/11/09 11:20:00 JST
3044 :     Tag: blume-20001109-scc
3045 :     Description:
3046 :    
3047 :     Eliminated some code duplication:
3048 :    
3049 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
3050 :     (Otherwise the functor is useless.)
3051 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
3052 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
3053 :    
3054 :     ----------------------------------------------------------------------
3055 : george 721 Name: Lal George
3056 :     Date: 2000/11/06 09:02:21 EST 2000
3057 :     Tag: Release_110_30
3058 :     Description:
3059 :    
3060 :     - Version 110.30
3061 :     ----------------------------------------------------------------------
3062 : blume 715 Name: Matthias Blume
3063 : blume 716 Date: 2000/11/04 14:45:00
3064 :     Tag: blume-20001104-mlbuild
3065 :     Description:
3066 :    
3067 :     - Made ml-build faster on startup.
3068 :     - Documentation fixes.
3069 :    
3070 :     ----------------------------------------------------------------------
3071 :     Name: Matthias Blume
3072 : blume 715 Date: 2000/11/02 17:00:00 JST
3073 :     Tag: blume-20001102-condcomp
3074 :     Description:
3075 :    
3076 :     - Small tweaks to pickler -- new BOOTFILES!
3077 :     - Version bumped to 110.29.2.
3078 :     - Added conditional compilation facility to init.cmi (see comment there).
3079 :     ----------------------------------------------------------------------
3080 : leunga 713 Name: Allen Leung
3081 :     Date: 2000/10/23 19:31:00
3082 :     Tag: leunga-20001023-demo-ra
3083 :    
3084 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
3085 :     2. Test programs for the graph library updated
3086 :     3. Some new MLRISC demo programs added
3087 :    
3088 :     ----------------------------------------------------------------------
3089 : blume 710 Name: Matthias Blume
3090 : blume 711 Date: 2000/08/31 22:15:00 JST
3091 :     Tag: blume-20001017-errmsg
3092 :     Description:
3093 :    
3094 :     More error message grief: Where there used to be no messages, there
3095 :     now were some that had bogus error regions. Fixed.
3096 :    
3097 :     ----------------------------------------------------------------------
3098 :     Name: Matthias Blume
3099 : blume 710 Date: 2000/08/31 17:30:00 JST
3100 :     Tag: blume-20001017-v110p29p1
3101 :     Description:
3102 :    
3103 :     I made a version 110.29.1 with new bootfiles.
3104 :    
3105 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
3106 :     CM documentation changes and a small bugfix in CM's error reporting.
3107 :    
3108 :     ----------------------------------------------------------------------
3109 : george 705 Name: Lal George
3110 :     Date: 2000/09/27 14:42:35 EDT
3111 :     Tag: george-20000927-nodestatus
3112 :     Description:
3113 :    
3114 :     Changed the type of the nodestatus, so that:
3115 :    
3116 :     SPILLED(~1) is now SPILLED
3117 :     SPILLED(m) where m>=0 is now MEMREG(m)
3118 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
3119 :    
3120 :     ----------------------------------------------------------------------
3121 : blume 697 Name: Matthias Blume
3122 : blume 703 Date: 2000/09/07 14:45:00 JST
3123 :     Tag: blume-20000907-cmerrmsg
3124 :     Description:
3125 :    
3126 :     Small tweak to CM to avoid getting ML syntax error messages twice.
3127 :    
3128 :     ----------------------------------------------------------------------
3129 :     Name: Matthias Blume
3130 : blume 701 Date: 2000/08/31 18:00:00 JST
3131 :     Tag: blume-20000831-cvsbootfiles
3132 :     Description:
3133 :    
3134 :     New URL for boot files (because the 110.29 files on the BL server do
3135 :     now work correctly with my updated install scripts for yacc and lex).
3136 :    
3137 :     ----------------------------------------------------------------------
3138 :     Name: Matthias Blume
3139 : blume 697 Date: 2000/08/08 12:33:00 JST
3140 :     Tag: blume-20000808-manual
3141 :     Description:
3142 :    
3143 :     Tiny update to CM manual.
3144 :    
3145 :     ----------------------------------------------------------------------
3146 : leunga 695 Name: Allen Leung
3147 :     Date: 2000/08/7 19:31:00
3148 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3149 :    
3150 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3151 :     and Mobdy.
3152 :    
3153 :     1. x86
3154 :    
3155 :     a. Fixes to peephole module by John and Dan.
3156 :     b. Assembly fix to SETcc by Allen.
3157 :     c. Fix to c-call by John.
3158 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3159 :     e. Instruction selection optimization to SETcc as suggested by John.
3160 :    
3161 :     For example,
3162 :    
3163 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3164 :    
3165 :     should generate:
3166 :    
3167 :     MOVL a, x
3168 :     SUBL b, x
3169 :     SHRL 31, x
3170 :    
3171 :     2. IR stuff
3172 :    
3173 :     A bunch of new DJ-graph related algorithms added. These
3174 :     speed up SSA construction.
3175 :    
3176 :     3. SSA + Scheduling
3177 :    
3178 :     Added code for SSA and scheduling to the repository
3179 :    
3180 :     ----------------------------------------------------------------------
3181 : george 694 Name: Lal George
3182 :     Date: 2000/07/27 11:53:14 EDT
3183 :    
3184 :     Tag: lal-20000727-linux-ppc
3185 :     Description:
3186 :    
3187 :     Made changes to support Linux PPC.
3188 :     p.s. I have confirmation that the 110.29 boot files work fine.
3189 :    
3190 :     ----------------------------------------------------------------------
3191 : blume 684 Name: Matthias Blume
3192 : blume 692 Date: 2000/07/27 17:40:00 JST
3193 :     Tag: blume-20000727-scripts
3194 :     Description:
3195 :    
3196 :     !!!! WARNING !!!!
3197 :     You must recompile the runtime system!
3198 :     !!!! WARNING !!!!
3199 :    
3200 :     This is basically another round of script-enhancements:
3201 :    
3202 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3203 :     and undefine CM preprocessor symbols.
3204 :    
3205 :     2. ml-build avoids generating a new heap image if it finds that the
3206 :     existing one is still ok. (The condition is that no ML file had to
3207 :     be recompiled and all ML files are found to be older that the heap
3208 :     file.)
3209 :    
3210 :     To make this work smoothly, I also hacked the runtime system as
3211 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3212 :     (.sparc-solaris, ...) that is currently being used.
3213 :    
3214 :     Moreover, the signature of CM.mk_standalone has changed. See the
3215 :     CM manual.
3216 :    
3217 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3218 :     CM manual for details.)
3219 :    
3220 :     4. More CM manual updates:
3221 :     - all of the above has been documented.
3222 :     - there is now a section describing the (CM-related) command line
3223 :     arguments that are accepted by the "sml" command
3224 :    
3225 :     ----------------------------------------------------------------------
3226 :     Name: Matthias Blume
3227 : blume 691 Date: 2000/07/25 16:20:00 JST
3228 :     Tag: blume-20000725-makedepend
3229 :     Description:
3230 :    
3231 :     Added a script called ml-makedepend. This can be used in makefiles
3232 :     for Unix' make in a way very similar to the "makedepend" command for
3233 :     C.
3234 :    
3235 :     The script internally uses function CM.sources.
3236 :    
3237 :     Synopsis:
3238 :    
3239 :     ml-makedepend [-f makefile] cmfile targetname
3240 :    
3241 :     The default for the makefile is "makefile" (or "Makefile" should
3242 :     "makefile" not exist).
3243 :    
3244 :     ml-makedepend adds a cmfile/targetname-specific section to this
3245 :     makefile (after removing the previous version of this section). The
3246 :     section contains a single dependency specification with targetname on
3247 :     the LHS (targetname is an arbitrary name), and a list of files derived
3248 :     from the cmfile on the RHS. Some of the files on the RHS are
3249 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3250 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3251 :     path names. The makefile writer is responsible for making sure that
3252 :     these variables have correct at the time "make" is invoked.
3253 :    
3254 :     ----------------------------------------------------------------------
3255 :     Name: Matthias Blume
3256 : blume 690 Date: 2000/07/22 23:30:00 JST
3257 :     Tag: blume-20000722-urlupdate
3258 :     Description:
3259 :    
3260 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3261 :    
3262 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3263 :    
3264 :     ----------------------------------------------------------------------
3265 :     Name: Matthias Blume
3266 : blume 686 Date: 2000/07/18 18:00:00 JST
3267 :     Tag: blume-20000718-Version_110_29
3268 :     Description:
3269 :    
3270 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3271 :    
3272 :     2. Updated config/version to 110.29
3273 :    
3274 :     3. Updated config/srcarchiveurl
3275 :    
3276 :     3. New boot files!
3277 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3278 :    
3279 :     ----------------------------------------------------------------------
3280 :     Name: Matthias Blume
3281 : blume 684 Date: 2000/07/11 13:58:00 JST
3282 :     Tag: blume-20000711-doctypo
3283 :     Description:
3284 :    
3285 :     Fixed a few typos in CM manual.
3286 :    
3287 :     ----------------------------------------------------------------------
3288 : leunga 683 Name: Allen Leung
3289 :     Date: 2000/06/15 00:38:00
3290 :     Tag: leunga-20000704-sparc-x86
3291 :    
3292 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3293 :     2. fix to x86 RET bug [found by Dan Grossman]
3294 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3295 :    
3296 :     Affects c-- and moby only
3297 :    
3298 :     ----------------------------------------------------------------------
3299 : blume 680 Name: Matthias Blume
3300 : blume 682 Date: 2000/07/04 15:26:00
3301 :     Tag: blume-20000704-trigger
3302 :     Description:
3303 :    
3304 :     1. Improvements to CM manual.
3305 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3306 :     of getting a back-trace. The function, when called, raises an
3307 :     internal exception which explicitly carries the full back-trace history,
3308 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3309 :     or not). The interactive loop will print that history once it arrives
3310 :     at top level.
3311 :     Short of having all exceptions implicitly carry the full history, the
3312 :     recommended way of using this facility is:
3313 :     - compile your program with instrumentation "on"
3314 :     - run it, when it raises an exception, look at the history
3315 :     - if the history is "cut off" because of some handler, go and modify
3316 :     your program so that it explicitly calls BTrace.trigger
3317 :     - recompile (still instrumented), and rerun; look at the full history
3318 :    
3319 :     ----------------------------------------------------------------------
3320 :     Name: Matthias Blume
3321 :     Date: 2000/07/03 15:36:00 JST
3322 : blume 680 Tag: blume-20000702-manual
3323 :     Description:
3324 :    
3325 :     Small corrections and updates to CM manual.
3326 :    
3327 :     ----------------------------------------------------------------------
3328 : blume 682 Name: Matthias Blume
3329 : blume 679 Date: 2000/06/29 16:04:00 JST
3330 :     Tag: blume-20000629-yacctool
3331 :     Description:
3332 :    
3333 :     Changes:
3334 :    
3335 :     1. Class "mlyacc" now takes separate arguments to pass options to
3336 :     generated .sml- and .sig-files independently.
3337 :     2. Corresponding CM manual updates.
3338 :     3. BTrace module now also reports call sites. (However, for loop clusters
3339 :     it only shows from where the cluster was entered.) There are associated
3340 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3341 :    
3342 :     ----------------------------------------------------------------------
3343 : blume 675 Name: Matthias Blume
3344 : blume 678 Date: 2000/06/27 16:51:00 JST
3345 :     Tag: blume-20000627-noweb
3346 :     Description:
3347 :    
3348 :     Changes:
3349 :    
3350 :     1. Implemented "subdir" and "witness" options for noweb tool.
3351 :     This caused some slight internal changes in CM's tool implementation.
3352 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3353 :     some remaining issues from earlier path anchor changes.
3354 :     3. Updated CM manual accordingly.
3355 :    
3356 :     4. Changed implementation of back-tracing so that I now consider it
3357 :     ready for prime-time.
3358 :    
3359 :     In particular, you don't have to explicitly trigger the back-trace
3360 :     anymore. Instead, if you are running BTrace-instrumented code and
3361 :     there is an uncaught exception (regardless of whether or not it was
3362 :     raised in instrumented code), the top-level evalloop will print
3363 :     the back-trace.
3364 :    
3365 :     Features:
3366 :    
3367 :     - Instrumented and uninstrumented code work together seemlessly.
3368 :     (Of course, uninstrumented code is never mentioned in actual
3369 :     back-traces.)
3370 :    
3371 :     - Asymptotic time- and space-complexity of instrumented code is
3372 :     equal to that of uninstrumented code. (This means that
3373 :     tail-recursion is preserved by the instrumentation phase.)
3374 :    
3375 :     - Modules whose code has been instrumented in different sessions
3376 :     work together without problem.
3377 :    
3378 :     - There is no penalty whatsoever on uninstrumented code.
3379 :    
3380 :     - There is no penalty on "raise" expressions, even in
3381 :     instrumented code.
3382 :    
3383 :     A potential bug (or perhaps it is a feature, too):
3384 :    
3385 :     A back-trace reaches no further than the outermost instrumented
3386 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3387 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3388 :    
3389 :     <exp> handle e => raise e
3390 :    
3391 :     and even
3392 :    
3393 :     <exp> handle Foo => raise Bar
3394 :    
3395 :     and, of course, through
3396 :    
3397 :     <exp> handle Foo => ...
3398 :    
3399 :     if the exception was not Foo.
3400 :    
3401 :     Back-traces always reach right through any un-instrumented code
3402 :     including any of its "handle" expressions, trivial or not.
3403 :    
3404 :     To try this out, do the following:
3405 :    
3406 :     - Erase all existing binfiles for your program.
3407 :     (You may keep binfiles for those modules where you think you
3408 :     definitely don't need back-tracing.)
3409 :     - Turn on back-trace instrumentation:
3410 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3411 :     - Recompile your program. (I.e., run "CM.make" or "use".)
3412 :     - You may now turn instrumentation off again (if you want):
3413 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3414 :     - Run your program as usual. If it raises an exception that
3415 :     reaches the interactive toplevel, then a back-trace will
3416 :     automatically be printed. After that, the toplevel loop
3417 :     will print the exception history as usual.
3418 :    
3419 :     ----------------------------------------------------------------------
3420 :     Name: Matthias Blume
3421 : blume 677 Date: 2000/06/26 09:56:46 JST
3422 :     Tag: blume-20000626-setup
3423 :     Description:
3424 :    
3425 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
3426 :     ML code before and after compiling a file (e.g., to set compiler
3427 :     flags)
3428 :    
3429 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
3430 :     - associated changes to btrace.sml (BTrace instrumentation pass)
3431 :     - cleaner implementation of btimp.sml (BTrace tracing and report
3432 :     module)
3433 :    
3434 :     CM manual: * new path encoding documented
3435 :     * description of setup-parameter to "sml" added
3436 :    
3437 :     The biggest user-visible change to back-tracing is that it is no
3438 :     longer necessary to compile all traced modules within the same
3439 :     session. (This was a real limitation.)
3440 :    
3441 :     ----------------------------------------------------------------------
3442 :     Name: Matthias Blume
3443 : blume 676 Date: 2000/06/24 12:40:00 JST
3444 :     Tag: blume-20000624-startup
3445 :     Description:
3446 :    
3447 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
3448 :     bit too often -- to put it mildly. :)
3449 :    
3450 :     ----------------------------------------------------------------------
3451 :     Name: Matthias Blume
3452 : blume 675 Date: 2000/06/23 18:20:00 JST
3453 :     Tag: blume-20000623-btrace
3454 :     Description:
3455 :    
3456 :     This updates adds a backtrace facility to aid programmers in debugging
3457 :     their programs. This involves the following changes:
3458 :    
3459 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3460 :     keeping track of the current call stack. When programs are compiled
3461 :     in a special mode, the compiler will insert calls to these hooks
3462 :     into the user program.
3463 :     "Hook" means that it is possible for different implementations of
3464 :     back-tracing to register themselves (at different times).
3465 :    
3466 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3467 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3468 :     It can be turned on using this call:
3469 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3470 :     Turning it off again:
3471 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3472 :     Querying the current status:
3473 :     SMLofNJ.Internals.BTrace.mode NONE;
3474 :     Annotated programs are about twice as big as normal ones, and they
3475 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3476 :     where all hooks do nothing). The slowdown with a plugin that is
3477 :     actually useful (such as the one supplied by default) is even greater,
3478 :     but in the case of the default plugin it is still only an constant
3479 :     factor (amortized).
3480 :    
3481 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3482 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3483 :     the above-mentioned function "mode" controls whether the annotation
3484 :     phase is invoked by the compiler. Another important function is
3485 :     "trigger": when called it aborts the current execution and causes
3486 :     the top-level loop to print a full back-trace.
3487 :    
3488 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3489 :     for back-tracing. It keeps track of the dynamic call stack and in
3490 :     addition to that it keeps a partial history at each "level" of that
3491 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3492 :     d and b (at separate times, dynamically), then the report will show:
3493 :    
3494 :     GOTO d
3495 :     /c
3496 :     GOTO \b
3497 :     CALL a
3498 :    
3499 :     This shows that there was an initial non-tail call of a, then a
3500 :     tail-call to b or c, looping behavior in a cluster of functions that
3501 :     consist of b and c, and then a goto from that cluster (i.e., either from
3502 :     b or from c) to d.
3503 :    
3504 :     Note that (depending on the user program) the amount of information
3505 :     that the back-trace module has to keep track of at each level is bounded
3506 :     by a constant. Thus, the whole implementation has the same asymptotical
3507 :     complexity as the original program (both in space and in time).
3508 :    
3509 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3510 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3511 :     which is raised by the "trigger" function mentioned above.
3512 :    
3513 :     Notes on usage:
3514 :    
3515 :     - Annotated code works well together with unannotated code:
3516 :     Unannotated calls simply do not show up at all in the backtrace.
3517 :    
3518 :     - It is not a good idea to let modules that were annotated during
3519 :     different sessions run at the same time. This is because the compiler
3520 :     chooses small integers to identify individual functions, and there
3521 :     will be clashes if different modules were compiled in separate sessions.
3522 :     (Nothing will crash, and you will even be told about the clashes, but
3523 :     back-trace information will in general not be useful.)
3524 :    
3525 :     - Back-tracing can be confused by callcc and capture.
3526 :    
3527 :     - The only way of getting a back-trace right now is to explicitly
3528 :     invoke the "trigger" function from your user program. Eventually, we
3529 :     should make every exception carry back-trace information (if
3530 :     available). But since this creates more overhead at "raise"-time
3531 :     (similar to the current exnHistory overhead), I have not yet
3532 :     implemented this. (The implementation will be rather easy.) With
3533 :     exceptions carrying back-trace information, this facility will be even
3534 :     more useful because users don't need to modify their programs...
3535 :    
3536 :     - While it is possible to compile the compiler with back-trace
3537 :     annotations turned on (I did it to get some confidence in
3538 :     correctness), you must make absolutely sure that core.sml and
3539 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3540 :     be compiled with annotation because there is no core access yet, but
3541 :     if you compile btimp.sml with annotation, then the system will go into
3542 :     an infinite recursion and crash.)
3543 :     Since CM currently does not know about BTrace, the only way to turn
3544 :     annotations on and off for different modules of the compiler is to
3545 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3546 :     this is awkward and clumsy.
3547 :    
3548 :     Sample sessions:
3549 :    
3550 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3551 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3552 :     [autoloading]
3553 :     [autoloading done]
3554 :     val it = false : bool
3555 :     - structure X = struct
3556 :     - fun main n = let
3557 :     - fun a (x, 0) = d x
3558 :     - | a (x, n) = b (x, n - 1)
3559 :     - and b (x, n) = c (x, n)
3560 :     - and c (x, n) = a (x, n)
3561 :     - and d x = e (x, 3)
3562 :     - and e (x, 0) = f x
3563 :     - | e (x, n) = e (x, n - 1)
3564 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
3565 :     - | f n = n * g (n - 1)
3566 :     - and g n = a (n, 3)
3567 :     - in
3568 :     - f n
3569 :     - end
3570 :     - end;
3571 :     structure X : sig val main : int -> int end
3572 :     - X.main 3;
3573 :     *** BACK-TRACE ***
3574 :     GOTO stdIn:4.2-13.20: X.main[2].f
3575 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3576 :     GOTO stdIn:4.2-13.20: X.main[2].d
3577 :     / stdIn:4.2-13.20: X.main[2].a
3578 :     | stdIn:4.2-13.20: X.main[2].b
3579 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3580 :     CALL stdIn:4.2-13.20: X.main[2].g
3581 :     GOTO stdIn:4.2-13.20: X.main[2].f
3582 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3583 :     GOTO stdIn:4.2-13.20: X.main[2].d
3584 :     / stdIn:4.2-13.20: X.main[2].a
3585 :     | stdIn:4.2-13.20: X.main[2].b
3586 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3587 :     CALL stdIn:4.2-13.20: X.main[2].g
3588 :     GOTO stdIn:4.2-13.20: X.main[2].f
3589 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3590 :     GOTO stdIn:4.2-13.20: X.main[2].d
3591 :     / stdIn:4.2-13.20: X.main[2].a
3592 :     | stdIn:4.2-13.20: X.main[2].b
3593 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3594 :     CALL stdIn:4.2-13.20: X.main[2].g
3595 :     GOTO stdIn:4.2-13.20: X.main[2].f
3596 :     CALL stdIn:2.15-17.4: X.main[2]
3597 :     -
3598 :    
3599 :     (Note that because of a FLINt bug the above code currently does not
3600 :     compile without BTrace turned on.)
3601 :    
3602 :     Here is another example, using my modified Tiger compiler:
3603 :    
3604 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3605 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3606 :     [autoloading]
3607 :     [autoloading done]
3608 :     val it = false : bool
3609 :     - CM.make "sources.cm";
3610 :     [autoloading]
3611 :     ...
3612 :     [autoloading done]
3613 :     [scanning sources.cm]
3614 :     [parsing (sources.cm):parse.sml]
3615 :     [creating directory CM/SKEL ...]
3616 :     [parsing (sources.cm):tiger.lex.sml]
3617 :     ...
3618 :     [wrote CM/sparc-unix/semant.sml]
3619 :     [compiling (sources.cm):main.sml]
3620 :     [wrote CM/sparc-unix/main.sml]
3621 :     [New bindings added.]
3622 :     val it = true : bool
3623 :     - Main.compile ("../testcases/merge.tig", "foo.out");
3624 :     *** BACK-TRACE ***
3625 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
3626 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3627 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3628 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3629 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3630 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3631 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
3632 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
3633 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
3634 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3635 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
3636 :     CALL main.sml:1.18-118.4: Main.compile[2]
3637 :     -
3638 :    
3639 :     ----------------------------------------------------------------------
3640 : blume 673 Name: Matthias Blumen
3641 :     Date: 2000/06/21 18:00:00 JST
3642 :     Tag: blume-20000621-manual
3643 :     Description:
3644 :    
3645 :     CM manual update: Path environments documented.
3646 :    
3647 :     ----------------------------------------------------------------------
3648 : blume 666 Name: Matthias Blume
3649 : blume 672 Date: 2000/06/19 13:40:00
3650 :     Tag: blume-20000619-manual
3651 :     Description:
3652 :    
3653 :     CM manual and system/README update. This only covers the fact that
3654 :     there are no more implicit anchors. (Path environments and the "bind"
3655 :     option to "cm" have yet to be documented.)
3656 :    
3657 :     ----------------------------------------------------------------------
3658 :     Name: Matthias Blume
3659 : blume 671 Date: 2000/06/19 11:05:00 JST
3660 :     Tag: blume-20000619-chdir-bugfix
3661 :     Description:
3662 :    
3663 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
3664 :    
3665 :     ----------------------------------------------------------------------
3666 :     Name: Matthias Blume
3667 : blume 670 Date: 2000/06/18 22:00:10 JST
3668 :     Tag: blume-20000618-implicit-anchors-really-gone
3669 :     Description:
3670 :    
3671 :     I updates the previous HISTORY entry where I forgot to mention that
3672 :     implicit anchors are no longer with us.
3673 :    
3674 :     The current update also gets rid of the (now useless) controller
3675 :     CM.Control.implicit_anchors.
3676 :    
3677 :     ----------------------------------------------------------------------
3678 :     Name: Matthias Blume
3679 : blume 666 Date: 2000/06/16 17:30:00 JST
3680 :     Tag: blume-20000616-anchorenv
3681 :     Description:
3682 :    
3683 :     This patch implements the long anticipated (just kidding :) "anchor
3684 :     environment" mechanism. In the course of doing this, I also
3685 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3686 :     one should be more robust in certain boundary cases. In any case, it
3687 :     is a lot cleaner than its predecessor (IMHO).
3688 :    
3689 :     This time, although there is yet another boot file format change, I
3690 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3691 :     are necessary and bootstrapping is straightforward. (You cannot read
3692 :     new bootfiles into an old system, but the other way around is no
3693 :     problem.)
3694 :    
3695 :     Visible changes:
3696 :    
3697 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3698 :     longer recognized at all. This means that such path names are not
3699 :     illegal either. For example, the name basis.cm simply refers to a
3700 :     local file called "basis.cm" (i.e, the name is an ordinary path
3701 :     relative to .cm-files directory). Or, to put it differently, only
3702 :     names that start with $ are anchored paths.
3703 : blume 666
3704 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3705 :     vanished.
3706 :    
3707 :     John (Reppy) had critizised this as soon as I originally proposed and
3708 : blume 666 implemented it, but at that time I did not really deeply believe
3709 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3710 :     in another place where it cannot be seen as an abbreviation for
3711 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3712 :     seems to mean (i.e., it "expands" into the corresponding anchor
3713 :     value).
3714 :    
3715 :     However, when paths are used as members in CM description files, it
3716 :     continues to be true that there must be at least another arc after the
3717 :     anchor. This is now enforced separately during semantic analysis
3718 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3719 : blume 666
3720 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3721 :     is a sub-option list of precisely two items -- one labeled "anchor"
3722 :     and the other one labeled "value". As you might expect, "anchor" is
3723 :     used to specify an anchor name to be bound, and "value" specifies what
3724 :     the anchor is being bound to.
3725 :    
3726 :     The value must be a directory name and can be given in either standard
3727 :     syntax (including the possibility that it is itself an anchored path)
3728 :     or native syntax.
3729 :    
3730 :     Examples:
3731 :    
3732 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3733 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3734 :    
3735 :     and so on.
3736 :    
3737 :     The meaning of this is that the .cm-file will be processed with an
3738 :     augmented anchor environment where the given anchor(s) is/are bound to
3739 :     the given values(s).
3740 :    
3741 :     The rationale for having this feature is this: Suppose you are trying
3742 :     to use two different (already stable) libraries a.cm and b.cm (that
3743 :     you perhaps didn't write yourself). Further, suppose each of these
3744 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3745 :     Normally you would now have a problem because the anchor "lib" can not
3746 :     be bound to more than one value globally. Therefore, the project that
3747 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3748 :     place:
3749 :    
3750 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3751 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3752 :    
3753 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3754 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3755 :    
3756 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3757 :     you when you do so at the time of CM.stabilize). Therefore, you can
3758 :     also use an anchored path as the value:
3759 :    
3760 :     a.cm (bind:(anchor:lib value:$a-lib))
3761 :     b.cm (bind:(anchor:lib value:$b-lib))
3762 :    
3763 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3764 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3765 :     itself is always locally bound, setting it globally is no longer
3766 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3767 :     can still be used as a global anchor for separate purposes. As a
3768 :     matter of fact, one can locally define "lib" in terms of a global
3769 :     "lib":
3770 :    
3771 :     a.cm (bind:(anchor:lib value:$lib/a))
3772 :     b.cm (bind:(anchor:lib value:$lib/b))
3773 :    
3774 :     ** 3: The encoding of path names has changed. This affects the way
3775 :     path names are shown in CM's progress report and also the internal
3776 :     protocol encoding used for parallel make.
3777 :    
3778 :     The encoding now uses one or more ':'-separated segments. Each
3779 :     segments corresponds to a file that has been specified relative to the
3780 :     file given by its preceding segment. The first segment is either
3781 :     relative to the CWD, absolute, or anchored. Each segment itself is
3782 :     basically a Unix pathname; all segments but the first are relative.
3783 :    
3784 :     Example:
3785 :    
3786 :     $foo/bar/baz.cm:a/b/c.sml
3787 :    
3788 :     This path denotes the file bar/a/b/c.sml relative to the directory
3789 :     denoted by anchor "foo". Notice that the encoding also includes
3790 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3791 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3792 :     be ignored to get the "real" pathname.
3793 :    
3794 :     To make this fact more obvious, CM puts the names of such "virtual
3795 :     arcs" into parentheses when they appear in progress reports. (No
3796 :     parentheses will appear in the internal protocol encoding.) Thus,
3797 :     what you really see is:
3798 :    
3799 :     $foo/bar/(baz.cm):a/b/c.sml
3800 :    
3801 :     I find this notation to be much more informative than before.
3802 :    
3803 :     Another new feature of the encoding is that special characters
3804 :     including parentheses, colons, (back)slashes, and white space are
3805 :     written as \ddd (where ddd is the decimal encoding of the character).
3806 :    
3807 : blume 670 *** The CM manual still needs to be updated.
3808 :    
3809 : blume 666 ----------------------------------------------------------------------
3810 : leunga 660 Name: Allen Leung
3811 :     Date: 2000/06/15 00:38:00
3812 :     Tag: leunga-20000615-x86-peephole
3813 :    
3814 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3815 :    
3816 :     ----------------------------------------------------------------------
3817 : blume 658 Name: Matthias Blume
3818 :     Date: 2000/06/12 11:40:00
3819 :     Tag: blume-20000612-parmakefix
3820 :     Description:
3821 :    
3822 :     More cleanup after changing the file naming scheme: This time I
3823 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3824 :    
3825 :     ----------------------------------------------------------------------
3826 : leunga 657 Name: Allen Leung
3827 :     Date: 2000/06/09 01:25:00
3828 :     Tag: leunga-20000609-various
3829 :    
3830 :     None of these things should affect normal SML/NJ operations
3831 :    
3832 :     1. Peephole improvements provided by Fermin (c--)
3833 :     2. New annotation DEFUSE for adding extra dependence (moby)
3834 :     3. New X86 LOCK instructions (moby)
3835 :     4. New machine description language for reservation tables (scheduling)
3836 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3837 :     trees etc.)
3838 :     6. I've changed the CM files so that they can work with versions
3839 :     110.0.6, 110.25 and 110.28
3840 :    
3841 :     ----------------------------------------------------------------------
3842 : blume 652 Name: Matthias Blume
3843 : blume 656 Date: 2000/06/09 12:40:00
3844 :     Tag: blume-20000609-log
3845 :     Description:
3846 :    
3847 :     - Removed all(?) remaining RCS Log entries from sources.
3848 :    
3849 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3850 :     anchored paths).
3851 :    
3852 :     ----------------------------------------------------------------------
3853 :     Name: Matthias Blume
3854 : blume 653 Date: 2000/06/07 17:00:00 JST
3855 :     Tag: blume-20000607-no-implicit-anchors
3856 :     Description:
3857 :    
3858 :     1. This update changes the default setting for
3859 :     CM.Control.implicit_anchors from true to false. This means that
3860 :     implicit anchors are no longer permitted by default. I also tried to
3861 :     make sure that nothing else still relies on implicit anchors.
3862 :     (This is the next step on the schedule towards a CM that does not even
3863 :     have the notion of implicit anchors anymore.)
3864 :    
3865 :     2. More CM manual updates.
3866 :    
3867 :     3. I managed to track down and fix the pickling bug I mentioned last
3868 :     time. Because of the previously existing workaround, this entails no
3869 :     immediate practical changes.
3870 :    
3871 :     ----------------------------------------------------------------------
3872 :     Name: Matthias Blume
3873 : blume 652 Date: 2000/06/06 11:15:00 JST
3874 :     Tag: blume-20000606-lazierpickle
3875 :     Description:
3876 :    
3877 :     !!!! NEW BOOT FILES !!!!
3878 :    
3879 :     * The main purpose of this update is to make library pickles lazier in
3880 :     order to reduce the initial space penalty for autoloading a library.
3881 :     As a result, it is now possible to have $smlnj/compiler.cm
3882 :     pre-registered. This should take care of the many complaints or
3883 :     inquiries about missing structure Compiler. This required changes to
3884 :     CM's internal data structures and small tweaks to some algorithms.
3885 :    
3886 :     As a neat additional effect, it is no longer necessary (for the sake
3887 :     of lean heap image files) to distinguish between a "minimal" CM and a
3888 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3889 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3890 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3891 :    
3892 :     To make the life of compiler-hackers easier, "makeml" now also
3893 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3894 :     words, after you bootstrap a new sml for the first time, you will not
3895 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3896 :     around you will still have to do it, though.)
3897 :    
3898 :     * A second change consists of major updates to the CM manual. There
3899 :     are now several appendices with summary information and also a full
3900 :     specification of the CM description file syntax.
3901 :    
3902 :     * In directory src/system I added the script "allcross". This script
3903 :     invokes sml and cross-compiles the compiler for all supported
3904 :     architectures. (Useful when providing a new set of boot files.)
3905 :    
3906 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3907 :     added a small tweak to pickle-util.sml to work around this problem,
3908 :     but it is not a proper fix yet. I will investigate further. (The
3909 :     effect of the bug was an inflation of library pickle size.)
3910 :    
3911 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3912 :    
3913 :     ----------------------------------------------------------------------
3914 : leunga 646 Name: Allen Leung
3915 : leunga 648 Date: 2000/05/25 17:28 EDT
3916 :     Tag: leunga-20000525-ra
3917 :     Description:
3918 :    
3919 :     Fixed a bug in freezing phase of the register allocator.
3920 :    
3921 :     ----------------------------------------------------------------------
3922 :     Name: Allen Leung
3923 : leunga 647 Date: 2000/05/15 22:53 EDT
3924 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3925 :     Description:
3926 :    
3927 :     1. Alpha
3928 :    
3929 :     Slight cleanup. Removed the instruction SGNXL
3930 :    
3931 :     2. X86
3932 :    
3933 :     Added the following instructions to the instruction set:
3934 :    
3935 :     ROLx, RORx,
3936 :     BTx, BTSx, BTLx, BTRx,
3937 :     XCHGx, and variants with the LOCK prefix
3938 :    
3939 :     3. Register Allocation
3940 :    
3941 :     The module ra-rewrite-with-renaming has been improved.
3942 :    
3943 : leunga 647 These have no effect on SML/NJ.
3944 :    
3945 : leunga 646 ----------------------------------------------------------------------
3946 : blume 642 Name: Matthias Blume
3947 : blume 645 Date: 2000/05/15 16:20:00 JST
3948 :     Tag: blume-20000515-lightrebuild
3949 :     Description:
3950 :    
3951 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
3952 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
3953 :     defined. In effect, the command will not build any cross-compiler
3954 :     backends and therefore finish more quickly.
3955 :    
3956 :     The "fixpt" script also takes a "-light" switch to be able to use
3957 :     this new facility while compiling for a fixpoint.
3958 :    
3959 :     2. I replaced all mentions of anchored paths in group owner specifications
3960 :     with simple relative paths (usually starting with "..").
3961 :     The rationale is that a library's internal workings should not be
3962 :     compromised by the lack of some anchor. (An anchor is necessary
3963 :     for someone who wants to refer to the library by an anchored path,
3964 :     but it should not be necessary to build the same library in the first
3965 :     place.)
3966 :    
3967 :     3. I changed the way CM's tool mechanism determines the shell command
3968 :     string used for things like ml-yacc etc. so that it does not break
3969 :     when CM.Control.implicit_anchors is turned off.
3970 :    
3971 :     ----------------------------------------------------------------------
3972 :     Name: Matthias Blume
3973 : blume 644 Date: 2000/05/12 18:20:00 JST
3974 :     Tag: blume-20000512-ml-build
3975 :     Description:
3976 :    
3977 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
3978 :     getting installed properly (by config/install.sh).
3979 :    
3980 :     ----------------------------------------------------------------------
3981 :     Name: Matthias Blume
3982 : blume 643 Date: 2000/05/12 17:30:00 JST
3983 :     Tag: blume-20000512-anchors
3984 :     Description:
3985 :    
3986 :     !!! NEW BOOT FILES !!!
3987 :    
3988 :     This change is in preparation of fading out support for "implicitly
3989 :     anchored path names". I went through all sources and used the
3990 :     explicit (and relatively new) $-notation. See system/README and the
3991 :     CM manual for more info on this.
3992 :    
3993 :     I also modified the anchoring scheme for some things such as "smlnj",
3994 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
3995 :     anchors are more expressive: anchor name and first arc do not have to
3996 :     coincide. This entails the following user-visible change:
3997 :    
3998 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
3999 :     particular, when you fire up sml with a command-line argument, say,
4000 :     e.g.:
4001 :    
4002 :     sml '$smlnj/cmb.cm'
4003 :    
4004 :     At the ML toplevel prompt:
4005 :    
4006 :     CM.autoload "$smlnj/cmb.cm";
4007 :    
4008 :     There is also a new controller in CM.Control that can be used to turn
4009 :     off all remaining support for implicit anchors by saying:
4010 :    
4011 :     CM.autoload "$smlnj/
4012 :     #set CM.Control.implicit_anchors false;
4013 :    
4014 :     This causes CM to reject implicitly anchored paths. This is (for the
4015 :     time being) less permissive than the "final" version where there will
4016 :     be no more such implicit anchors and relative paths will be just that:
4017 :     relative.
4018 :    
4019 :     The next step (version after next version?) will be to make the
4020 :     default for CM.Control.implicit_anchors false. After the dust has
4021 :     settled, I can then produce the "final" version of this...
4022 :    
4023 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
4024 :    
4025 :     ----------------------------------------------------------------------
4026 :     Name: Matthias Blume
4027 : blume 642 Date: 2000/05/11 16:30:00 JST
4028 :     Tag: blume-20000511-sources
4029 :     Description:
4030 :    
4031 :     The main change is that I added function CM.sources as a generalized
4032 :     version of the earlier CM.makedepend. This entails the following
4033 :     additional changes:
4034 :    
4035 :     - CM.makedepend has been dropped.
4036 :    
4037 :     - CM manual has been updated.
4038 :    
4039 :     - TOOLS signature and API have been changed.
4040 :    
4041 :     ----------------------------------------------------------------------
4042 : leunga 640 Name: Allen Leung
4043 : leunga 641 Date: 2000/05/10 21:17 EDT
4044 :     Tag: leunga-20000510-moby-c--ssa
4045 :     Description:
4046 :    
4047 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
4048 :     None of these affect SML/NJ.
4049 :    
4050 :     1. Register Allocation
4051 :    
4052 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
4053 :     This module tries to remove local (i.e. basic block level) redundancies
4054 :     during spilling.
4055 :    
4056 :     b. A new framework for performing region based register allocation.
4057 :     Not yet entirely functional.
4058 :    
4059 :     2. X86
4060 :    
4061 :     a. DefUse for POP was missing the stack pointer [found by Lal]
4062 :     b. Reload for CALL was incorrect in X86Spill [found by John]
4063 :     c. Various fixes in X86Spill so that it can be used correctly for
4064 :     the new spilling module.
4065 :    
4066 :     3. SSA/IR
4067 :    
4068 :     a. New module ir/dj-dataflow.sml implements elimination based
4069 :     data flow analysis.
4070 :    
4071 :     4. MLRiscGen
4072 :    
4073 :     a. Fix for gc type annotation
4074 :    
4075 :     5. MDGen
4076 :    
4077 :     Various fixes for machine description -> ml code translation. For ssa
4078 :     only.
4079 :    
4080 :     ----------------------------------------------------------------------
4081 :     Name: Allen Leung
4082 : leunga 640 Date: 2000/05/08 22:17 EDT
4083 :     Tag: leunga-20000508-labexp
4084 :     Description:
4085 :    
4086 :     Fermin has found a few assembly problems with constant expressions
4087 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
4088 :     which choke on dumb assemblers. This is his fix.
4089 :    
4090 :     ----------------------------------------------------------------------
4091 : dbm 635 Name: Dave MacQueen
4092 :     Date: 2000/04/09 14:00 EDT
4093 :     Tag: dbm-20000502-Version_110_28
4094 :     Description:
4095 :    
4096 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
4097 :    
4098 :     2. Updated config/version to 110.28
4099 :    
4100 :     3. Updated config/srcarchiveurl
4101 :    
4102 :     3. New boot files!
4103 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
4104 :    
4105 :     ----------------------------------------------------------------------
4106 : blume 634 Name: Matthias Blume
4107 :     Date: 2000/05/01 19:05:00 JST
4108 :     Tag: blume-20000501-noweb
4109 :     Description:
4110 :    
4111 :     A new noweb tool has been added. The existing system is entirely
4112 :     unaffected by this, but some CM users have asked for renewed noweb
4113 :     support. Everything is documented in the CM manual.
4114 :    
4115 :     New (plugin) libraries:
4116 :    
4117 :     noweb-tool.cm
4118 :     nw-ext.cm
4119 :    
4120 :     ----------------------------------------------------------------------
4121 : dbm 633 Name: Dave MacQueen
4122 :     Date: 2000/04/30 12:40PM EDT
4123 :     Tag: dbm-20000430-bug_fixes
4124 :     Description:
4125 :    
4126 :     1. Fix for bug 1498
4127 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
4128 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
4129 :     added toRealArray function
4130 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
4131 :     added check for tag Obj.RealArray to array printing case in ppObj
4132 :    
4133 :     2. Fix for bug 1510
4134 :     smlnj/src/compiler/Semant/types/typesutil.sml
4135 :     fixed definition of dummyargs (used by equalTycon) so that
4136 :     dummy args are distinct types
4137 :    
4138 :     ----------------------------------------------------------------------
4139 : blume 629 Name: Matthias Blume
4140 : blume 632 Date: 2000/04/30 01:00:00 JST
4141 :     Tag: blume-20000430-versions
4142 :     Description:
4143 :    
4144 :     1. CM version numbering added. This is an implementation of Lal's
4145 :     proposal for adding version numbers and version checking to .cm
4146 :     files. Lal said that his proposal was just that -- a proposal.
4147 :     For the time being I went ahead and implemented it so that people
4148 :     can comment on it. Everything is completely backward-compatible
4149 :     (except for the stable library format, i.e., new bootfiles!).
4150 :    
4151 :     As usual, see the CM manual for details.
4152 :    
4153 :     2. An alternative syntax for anchored paths has been implemented.
4154 :     Dave has recently voiced the same concerns that I had when I did
4155 :     this, so there should be some support. My take is that eventually
4156 :     I will let support for the current syntax (where anchors are
4157 :     "implicit") fade out in favor of the new, explicit syntax.
4158 :     In order to be backward-compatible, both old and new syntax are
4159 :     currently supported.
4160 :    
4161 :     Again, see the CM manual for details.
4162 :    
4163 :     3. Parallel make is trying to be slightly smarter: When the master
4164 :     process finds a "bottleneck", i.e., when there is only one
4165 :     compilation unit that can be compiled and everybody else is
4166 :     waiting on it, then it will simply compile it directly instead
4167 :     of clumsily telling one of the slaves to do it.
4168 :    
4169 :     4. Support for "unsharing" added. This is necessary in order to be
4170 :     able to have two different versions of the same library running
4171 :     at the same time (e.g., for trying out a new MLRISC while still
4172 :     having the old MLRISC linked into the current compiler, etc.)
4173 :     See the CM manual.
4174 :    
4175 :     5. Simple "makedepend" functionality added for generating Makefile
4176 :     dependency information. (This is rather crude at the moment.
4177 :     Expect some changes here in the future.)
4178 :    
4179 :     6. ".fun" added as a recognized suffix for ML files. Also documented
4180 :     explicitly in the manual that the fallback behavior (unknown suffix
4181 :     -> ML file) is not an official feature!
4182 :    
4183 :     7. Small changes to the pickler for stable libraries.
4184 :    
4185 :     8. Several internal changes to CM (for cleanup/improvement).
4186 :    
4187 :    
4188 :     !!!! NEW BINFILES !!!!
4189 :    
4190 :     ----------------------------------------------------------------------
4191 :     Name: Matthias Blume
4192 : blume 631 Date: 2000/04/28 17:30:00 JST
4193 :     Tag: blume-20000428-pathconfig
4194 :     Description:
4195 :    
4196 :     1. I changed config/install.sh to remove duplicate entries from the
4197 :     lib/pathconfig file at the end. Moreover, the final version of
4198 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
4199 :     in src/system/installml.
4200 :    
4201 :     2. The config/install.sh script now consistently uses relative
4202 :     pathnames in lib/pathconfig whenever the anchor is in the lib
4203 :     directory. (So far this was true for the libraries that come
4204 :     pre-compiled and bundled as part of the bootfiles but not for
4205 :     libraries that are compiled by the script itself.)
4206 :    
4207 :     ----------------------------------------------------------------------
4208 :     Name: Matthias Blume
4209 : blume 629 Date: 2000/04/26 13:10:00 JST
4210 :     Tag: blume-20000426-fun_suffix
4211 :     Description:
4212 :    
4213 :     Added ".fun" as a recognized file name suffix (for ML code).
4214 :    
4215 :     ----------------------------------------------------------------------
4216 : leunga 628 Name: Allen Leung
4217 :     Date: 2000/04/25 17:00:00 EST
4218 :     Tag: leunga-20000425-alpha-ra
4219 :     Description:
4220 :    
4221 :     1. Alpha
4222 :    
4223 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
4224 :     in C--.
4225 :    
4226 :     2. RA
4227 :    
4228 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
4229 :     and the interference graph.
4230 :    
4231 :     ----------------------------------------------------------------------
4232 : dbm 627 Name: Dave MacQueen
4233 :     Date: 2000/04/25/
4234 :     Tag: dbm-20000425-mlyacc_doc_examples
4235 :     Description:
4236 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
4237 :     Updated all three examples in src/ml-yacc/examples to run
4238 :     under 110.* using CM.make.
4239 :    
4240 :     ----------------------------------------------------------------------
4241 : leunga 624 Name: Allen Leung
4242 :     Date: 2000/04/20 23:04:00 EST
4243 :     Tag: leunga-20000420-ssa-c---stuff
4244 :     Description:
4245 :    
4246 :     This update synchronizes my repository with Yale's. Most of these
4247 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
4248 :    
4249 :     1. Register Allocator
4250 :    
4251 :     a. An improvement in the interference graph construction:
4252 :     Given a copy
4253 :    
4254 :     s <- t
4255 :    
4256 : leunga 628 no interference edge between s and t is added for this definition of s.
4257 : leunga 624
4258 :     b. I've added two new spill heuristic modules that Fermin and I developed
4259 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
4260 :     useful for others (Moby?)
4261 :    
4262 :     2. X86
4263 :    
4264 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
4265 :    
4266 :     3. Alpha
4267 :    
4268 :     a. Added the BSR instruction and code generation that goes with it [C--]
4269 :     b. Other fixes too numerous to recount provided by Fermin [C--]
4270 :    
4271 :     4. Regmaps
4272 :    
4273 :     a. The regmaps are not initialized with the identity physical bindings
4274 :     at creation time. This is unneeded.
4275 :    
4276 :     5. MLRISC Optimizations
4277 :    
4278 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
4279 :     intersects with liveness incrementally in linear time! Woohoo!
4280 :     This is now used in my new SSA construction algorithm.
4281 :    
4282 :     b. THe branch reorganization module is now smarter about linear chains of
4283 :     basic blocks.
4284 :    
4285 :    
4286 :     ----------------------------------------------------------------------
4287 : blume 615 Name: Matthias Blume
4288 :     Date: 2000/04/12 13:52:00 JST
4289 :     Tag: blume_main_v110p27_1
4290 :     Description:
4291 :    
4292 :     Changed install.sh script to handle archive files without version number
4293 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
4294 :     name of the boot file archive.
4295 :    
4296 :     ----------------------------------------------------------------------
4297 : dbm 607 Name: Dave MacQueen
4298 :     Date: 2000/04/09 14:00 EDT
4299 :     Tag: dbm-20000410-Version_110_27
4300 :     Description:
4301 :    
4302 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
4303 :    
4304 :     2. Updated src/config/version to 110.27
4305 :    
4306 :     3. New boot files!
4307 :    
4308 :     ----------------------------------------------------------------------
4309 : leunga 606 Name: Allen Leung
4310 :     Date: 2000/04/09 19:09:00 EST
4311 :     Tag: leunga-20000409-misc
4312 :     Description:
4313 :    
4314 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
4315 :    
4316 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
4317 :    
4318 :     ----------------------------------------------------------------------
4319 : monnier 605 Name: Stefan
4320 :     Date: 2000/04/07 10:00:00 EDT
4321 :     Tag: monnier-20000406-branch-handling
4322 :     Description:
4323 :    
4324 :     Improved handling of branches (mostly those generated from
4325 :     polymorphic equality), removed switchoff and changed the
4326 :     default optimization settings (more cpsopt and less flintopt).
4327 :    
4328 :     ----------------------------------------------------------------------
4329 : leunga 601 Name: Allen Leung
4330 : leunga 602 Date: 2000/04/06 01:30:00 EST
4331 :     Tag: leunga-20000406-peephole-x86-SSA-2
4332 :     Description:
4333 :    
4334 :     Forgot a few files.
4335 :    
4336 :     ----------------------------------------------------------------------
4337 :     Name: Allen Leung
4338 : leunga 601 Date: 2000/04/06 00:36:00 EST
4339 :     Tag: leunga-20000406-peephole-x86-SSA
4340 :     Description:
4341 :    
4342 :     1. New Peephole code
4343 :    
4344 :     2. Minor improvement to X86 instruction selection
4345 :    
4346 :     3. Various fixes to SSA and machine description -> code translator
4347 :    
4348 :     ----------------------------------------------------------------------
4349 : blume 595 Name: Matthias Blume
4350 :     Date: 2000/04/05 12:30:00 JST
4351 :     Tag: blume_main_v110p26p2_3
4352 :     Description: