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 1142 - (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 : leunga 1142 Name: Allen Leung
17 :     Date: 2002/03/13 17:30:00 EST
18 :     Tag: leunga-20020313-x86-fp-unary
19 :     Description:
20 :    
21 :     Bug fix for:
22 :    
23 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
24 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
25 :     > - fun f(x,(y,z)) = Real.~ y;
26 :     > [autoloading]
27 :     > [autoloading done]
28 :     > fchsl (%eax), 184(%esp)
29 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
30 :     >
31 :     > uncaught exception Error
32 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
33 :    
34 :     The problem was that the code generator did not generate any fp registers
35 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
36 :     translate the pseudo fp instruction. This only happened with unary fp
37 :     operators in certain situations.
38 :    
39 :     ----------------------------------------------------------------------
40 : blume 1137 Name: Matthias Blume
41 : blume 1140 Date: 2002/03/13 14:00:00 EST
42 :     Tag: blume-20020313-overload-etc
43 :     Description:
44 :    
45 :     1. Added _overload as a synonym for overload for backward compatibility.
46 :     (Control.overloadKW must be true for either version to be accepted.)
47 :    
48 :     2. Fixed bug in install script that caused more things to be installed
49 :     than what was requested in config/targets.
50 :    
51 :     3. Made CM aware of the (_)overload construct so that autoloading
52 :     works.
53 :    
54 :     ----------------------------------------------------------------------
55 :     Name: Matthias Blume
56 : blume 1138 Date: 2002/03/12 22:03:00 EST
57 :     Tag: blume-20020312-url
58 :     Description:
59 :    
60 :     Forgot to update BOOT and srcarchiveurl.
61 :    
62 :     ----------------------------------------------------------------------
63 :     Name: Matthias Blume
64 : blume 1137 Date: 2002/03/12 17:30:00 EST
65 :     Tag: blume-20020312-version110392
66 :     Description:
67 :    
68 :     Yet another version number bump (because of small changes to the
69 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
70 :    
71 :     Changes:
72 :    
73 :     The new pid generation scheme described a few weeks ago was overly
74 :     complicated. I implemented a new mechanism that is simpler and
75 :     provides a bit more "stability": Once CM has seen a compilation
76 :     unit, it keeps its identity constant (as long as you do not delete
77 :     those crucial CM/GUID/* files). This means that when you change
78 :     and interface, compiler, then go back to the old interface, and
79 :     compile again, you arrive at the original pid.
80 :    
81 :     There now also is a mechanism that instructs CM to use the plain
82 :     environment hash as a module's pid (effectively making its GUID
83 :     the empty string). For this, "noguid" must be specified as an
84 :     option to the .sml file in question within its .cm file.
85 :     This is most useful for code that is being generated by tools such
86 :     as ml-nlffigen (because during development programmers tend to
87 :     erase the tool's entire output directory tree including CM's cached
88 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
89 :     revert to the old, broken behavior of SML/NJ, but in specific cases
90 :     where there is no danger of interface confusion, its use is ok
91 :     (I think).
92 :    
93 :     ml-nlffigen by default generates "noguid" annotations. They can be
94 :     turned off by specifying -guid in its command line.
95 :    
96 :     ----------------------------------------------------------------------
97 : george 1136 Name: Lal George
98 :     Date: 2002/03/12 12 14:42:36 EST
99 :     Tag: george-20020312-frequency-computation
100 :     Description:
101 :    
102 :     Integrated jump chaining and static block frequency into the
103 :     compiler. More details and numbers later.
104 : leunga 1025
105 : george 1003 ----------------------------------------------------------------------
106 : george 1133 Name: Lal George
107 :     Date: 2002/03/11 11 22:38:53 EST
108 :     Tag: george-20020311-jump-chain-elim
109 :     Description:
110 :    
111 :     Tested the jump chain elimination on all architectures (except the
112 :     hppa). This is on by default right now and is profitable for the
113 :     alpha and x86, however, it may not be profitable for the sparc and ppc
114 :     when compiling the compiler.
115 :    
116 :     The gc test will typically jump to a label at the end of the cluster,
117 :     where there is another jump to an external cluster containing the actual
118 :     code to invoke gc. This is to allow factoring of common gc invocation
119 :     sequences. That is to say, we generate:
120 :    
121 :     f:
122 :     testgc
123 :     ja L1 % jump if above to L1
124 :    
125 :     L1:
126 :     jmp L2
127 :    
128 :    
129 :     After jump chain elimination the 'ja L1' instructions is converted to
130 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
131 :     up being implemented in their long form (if L2 is far away) using:
132 :    
133 :     jbe L3 % jump if below or equal to L3
134 :     jmp L2
135 :     L3:
136 :     ...
137 :    
138 :    
139 :     For large compilation units L2 may be far away.
140 :    
141 :    
142 :     ----------------------------------------------------------------------
143 : blume 1132 Name: Matthias Blume
144 :     Date: 2002/03/11 13:30:00 EST
145 :     Tag: blume-20020311-mltreeeval
146 :     Description:
147 :    
148 :     A functor parameter was missing.
149 :    
150 :     ----------------------------------------------------------------------
151 : leunga 1130 Name: Allen Leung
152 : leunga 1131 Date: 2002/03/11 10:30:00 EST
153 : leunga 1142 Tag: leunga-20020311-runtime-string0
154 : leunga 1131 Description:
155 :    
156 :     The representation of the empty string now points to a
157 :     legal null terminated C string instead of unit. It is now possible
158 :     to convert an ML string into C string with InlineT.CharVector.getData.
159 :     This compiles into one single machine instruction.
160 :    
161 :     ----------------------------------------------------------------------
162 :     Name: Allen Leung
163 : leunga 1130 Date: 2002/03/10 23:55:00 EST
164 :     Tag: leunga-20020310-x86-call
165 :     Description:
166 :    
167 :     Added machine generation for CALL instruction (relative displacement mode)
168 :    
169 :     ----------------------------------------------------------------------
170 : blume 1128 Name: Matthias Blume
171 :     Date: 2002/03/08 16:05:00
172 :     Tag: blume-20020308-entrypoints
173 :     Description:
174 :    
175 :     Version number bumped to 110.39.1. NEW BOOTFILES!
176 :    
177 :     Entrypoints: non-zero offset into a code object where execution should begin.
178 :    
179 :     - Added the notion of an entrypoint to CodeObj.
180 :     - Added reading/writing of entrypoint info to Binfile.
181 :     - Made runtime system bootloader aware of entrypoints.
182 :     - Use the address of the label of the first function given to mlriscGen
183 :     as the entrypoint. This address is currently always 0, but it will
184 :     not be 0 once we turn on block placement.
185 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
186 :     with entry points) from mlriscGen.
187 :    
188 :     ----------------------------------------------------------------------
189 : leunga 1127 Name: Allen Leung
190 :     Date: 2002/03/07 20:45:00 EST
191 :     Tag: leunga-20020307-x86-cmov
192 :     Description:
193 :    
194 :     Bug fixes for CMOVcc on x86.
195 :    
196 :     1. Added machine code generation for CMOVcc
197 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
198 :     3. CMOVcc cannot have an immediate operand as argument.
199 :    
200 :     ----------------------------------------------------------------------
201 : blume 1126 Name: Matthias Blume
202 :     Date: 2002/03/07 16:15:00 EST
203 :     Tag: blume-20020307-controls
204 :     Description:
205 :    
206 :     This is a very large but mostly boring patch which makes (almost)
207 :     every tuneable compiler knob (i.e., pretty much everything under
208 :     Control.* plus a few other things) configurable via both the command
209 :     line and environment variables in the style CM did its configuration
210 :     until now.
211 :    
212 :     Try starting sml with '-h' (or, if you are brave, '-H')
213 :    
214 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
215 :     implements the underlying generic mechanism.
216 :    
217 :     The interface to some of the existing such facilities has changed somewhat.
218 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
219 :     (The getFoo interface is still there for backward-compatibility, but its
220 :     use is deprecated.)
221 :    
222 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
223 :     that one can now twiddle the compiler settings when using this "batch"
224 :     compiler.
225 :    
226 :     TODO items:
227 :    
228 :     We should go through and throw out all controls that are no longer
229 :     connected to anything. Moreover, we should go through and provide
230 :     meaningful (and correct!) documentation strings for those controls
231 :     that still are connected.
232 :    
233 :     Currently, multiple calls to Controls.new are accepted (only the first
234 :     has any effect). Eventually we should make sure that every control
235 :     is being made (via Controls.new) exactly once. Future access can then
236 :     be done using Controls.acc.
237 :    
238 :     Finally, it would probably be a good idea to use the getter-setter
239 :     interface to controls rather than ref cells. For the time being, both
240 :     styles are provided by the Controls module, but getter-setter pairs are
241 :     better if thread-safety is of any concern because they can be wrapped.
242 :    
243 :     *****************************************
244 :    
245 :     One bug fix: The function blockPlacement in three of the MLRISC
246 :     backpatch files used to be hard-wired to one of two possibilities at
247 :     link time (according to the value of the placementFlag). But (I
248 :     think) it should rather sense the flag every time.
249 :    
250 :     *****************************************
251 :    
252 :     Other assorted changes (by other people who did not supply a HISTORY entry):
253 :    
254 :     1. the cross-module inliner now works much better (Monnier)
255 :     2. representation of weights, frequencies, and probabilities in MLRISC
256 :     changed in preparation of using those for weighted block placement
257 :     (Reppy, George)
258 :    
259 :     ----------------------------------------------------------------------
260 : george 1116 Name: Lal George
261 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
262 :     Tag: george-20020307-weighted-block-placement
263 :    
264 :     Tested the weighted block placement optimization on all architectures
265 :     (except the hppa) using AMPL to generate the block and edge frequencies.
266 :     Changes were required in the machine properties to correctly
267 :     categorize trap instructions. There is an MLRISC flag
268 :     "weighted-block-placement" that can be used to enable weighted block
269 :     placement, but this will be ineffective without block/edge
270 :     frequencies (coming soon).
271 :    
272 :    
273 :     ----------------------------------------------------------------------
274 :     Name: Lal George
275 : george 1116 Date: 2002/03/05 17:24:48 EST
276 :     Tag: george-20020305-linkage-cluster
277 :    
278 : george 1124 In order to support the block placement optimization, a new cluster
279 :     is generated as the very first cluster (called the linkage cluster).
280 :     It contains a single jump to the 'real' entry point for the compilation
281 :     unit. Block placement has no effect on the linkage cluster itself, but
282 :     all the other clusters have full freedom in the manner in which they
283 :     reorder blocks or functions.
284 : george 1116
285 :     On the x86 the typical linkage code that is generated is:
286 :     ----------------------
287 :     .align 2
288 :     L0:
289 :     addl $L1-L0, 72(%esp)
290 : george 1124 jmp L1
291 : george 1116
292 :    
293 :     .align 2
294 :     L1:
295 :     ----------------------
296 :    
297 :     72(%esp) is the memory location for the stdlink register. This
298 :     must contain the address of the CPS function being called. In the
299 : george 1124 above example, it contains the address of L0; before
300 : george 1116 calling L1 (the real entry point for the compilation unit), it
301 :     must contain the address for L1, and hence
302 :    
303 :     addl $L1-L0, 72(%esp)
304 :    
305 : george 1124 I have tested this on all architectures except the hppa.The increase
306 :     in code size is of course negligible
307 : george 1116
308 :     ----------------------------------------------------------------------
309 : leunga 1115 Name: Allen Leung
310 :     Date: 2002/03/03 13:20:00 EST
311 :     Tag: leunga-20020303-mlrisc-tools
312 :    
313 :     Added #[ ... ] expressions to mlrisc tools
314 :    
315 :     ----------------------------------------------------------------------
316 : blume 1096 Name: Matthias Blume
317 : blume 1098 Date: 2002/02/27 12:29:00 EST
318 :     Tag: blume-20020227-cdebug
319 :     Description:
320 :    
321 :     - made types in structure C and C_Debug to be equal
322 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
323 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
324 :    
325 :     ----------------------------------------------------------------------
326 :     Name: Matthias Blume
327 : blume 1096 Date: 2002/02/26 12:00:00 EST
328 :     Tag: blume-20020226-ffi
329 :     Description:
330 :    
331 :     1. Fixed a minor bug in CM's "noweb" tool:
332 :     If numbering is turned off, then truly don't number (i.e., do not
333 :     supply the -L option to noweb). The previous behavior was to supply
334 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
335 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
336 :    
337 :     2. Once again, I reworked some aspects of the FFI:
338 :    
339 :     A. The incomplete/complete type business:
340 :    
341 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
342 :     gone!
343 :     - ML types representing an incomplete type are now *equal* to
344 :     ML types representing their corresponding complete types (just like
345 :     in C). This is still safe because ml-nlffigen will not generate
346 :     RTTI for incomplete types, nor will it generate functions that
347 :     require access to such RTTI. But when ML code generated from both
348 :     incomplete and complete versions of the C type meet, the ML types
349 :     are trivially interoperable.
350 :    
351 :     NOTE: These changes restore the full generality of the translation
352 :     (which was previously lost when I eliminated functorization)!
353 :    
354 :     B. Enum types:
355 :    
356 :     - Structure C now has a type constructor "enum" that is similar to
357 :     how the "su" constructor works. However, "enum" is not a phantom
358 :     type because each "T enum" has values (and is isomorphic to
359 :     MLRep.Signed.int).
360 :     - There are generic access operations for enum objects (using
361 :     MLRep.Signed.int).
362 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
363 :     * The structure contains the definition of type "mlrep" (the ML-side
364 :     representation type of the enum). Normally, mlrep is the same
365 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
366 :     then mlrep will be defined as a datatype -- thus facilitating
367 :     pattern matching on mlrep values.
368 :     ("-ec" will be suppressed if there are duplicate values in an
369 :     enumeration.)
370 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
371 :     will be generated for each C enum constant xxx.
372 :     * Conversion functions m2i and i2m convert between mlrep and
373 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
374 :     * Coversion functions c and ml convert between mlrep and "tag enum".
375 :     * Access functions (get/set) fetch and store mlrep values.
376 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
377 :     enumerations are merged into one single enumeration represented by
378 :     structure E_'.
379 :    
380 :     ----------------------------------------------------------------------
381 : leunga 1080 Name: Allen Leung
382 : leunga 1094 Date: 2002/02/25 04:45:00 EST
383 :     Tag: leunga-20020225-cps-spill
384 :    
385 :     This is a new implementation of the CPS spill phase.
386 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
387 :     In case of problems, replace it with the old file spill.sml
388 :    
389 :     The current compiler runs into some serious performance problems when
390 :     constructing a large record. This can happen when we try to compile a
391 :     structure with many items. Even a very simple structure like the following
392 :     makes the compiler slow down.
393 :    
394 :     structure Foo = struct
395 :     val x_1 = 0w1 : Word32.int
396 :     val x_2 = 0w2 : Word32.int
397 :     val x_3 = 0w3 : Word32.int
398 :     ...
399 :     val x_N = 0wN : Word32.int
400 :     end
401 :    
402 :     The following table shows the compile time, from N=1000 to N=4000,
403 :     with the old compiler:
404 :    
405 :     N
406 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
407 :     MLRISC ra 0.06u 0.00s 0.05g
408 :     (spills = 0 reloads = 0)
409 :     TOTAL 0.63u 0.07s 0.21g
410 :    
411 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
412 :     MLRISC ra 5.68u 0.59s 3.93g
413 :     (spills = 0 reloads = 0)
414 :     TOTAL 14.71u 0.99s 4.81g
415 :    
416 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
417 :     MLRISC ra 5.54u 0.65s 3.91g
418 :     (spills = 543 reloads = 1082)
419 :     TOTAL 65.40u 3.13s 6.00g
420 :    
421 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
422 :     MLRISC ra 0.80u 0.10s 0.55g
423 :     (spills = 42 reloads = 84)
424 :     TOTAL 129.42u 5.10s 4.13g
425 :    
426 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
427 :     MLRISC ra 2.69u 0.27s 1.38g
428 :     (spills = 62 reloads = 124)
429 :     TOTAL 682.48u 19.61s 13.99g
430 :    
431 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
432 :     MLRISC ra 4.96u 0.27s 2.72g
433 :     (spills = 85 reloads = 170)
434 :     TOTAL 2375.26u 57.21s 48.00g
435 :    
436 :     As you can see the old cps spill module suffers from some serious
437 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
438 :     instead of patching the problems up, I'm reimplementing it
439 : leunga 1094 with a different algorithm. The new code is more modular,
440 :     smaller when compiled, and substantially faster
441 :     (O(n log n) time and O(n) space). Timing of the new spill module:
442 :    
443 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
444 :     MLRISC ra 0.25u 0.02s 0.15g
445 :     (spills=1 reloads=3)
446 :     TOTAL 7.74u 0.34s 1.62g
447 :    
448 :     Implementation details:
449 :    
450 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
451 :     number of live variables at any program point (the bandwidth)
452 :     does not exceed a certain limit, which is determined by the
453 :     size of the spill area.
454 :    
455 :     When the bandwidth is too large, we decrease the register pressure by
456 :     packing live variables into spill records. How we achieve this is
457 :     completely different than what we did in the old code.
458 :    
459 :     First, there is something about the MLRiscGen code generator
460 :     that we should be aware of:
461 :    
462 :     o MLRiscGen performs code motion!
463 :    
464 :     In particular, it will move floating point computations and
465 :     address computations involving only the heap pointer to
466 :     their use sites (if there is only a single use).
467 :     What this means is that if we have a CPS record construction
468 :     statement
469 :    
470 :     RECORD(k,vl,w,e)
471 :    
472 :     we should never count the new record address w as live if w
473 :     has only one use (which is often the case).
474 :    
475 :     We should do something similar to floating point, but the transformation
476 :     there is much more complex, so I won't deal with that.
477 :    
478 :     Secondly, there are now two new cps primops at our disposal:
479 :    
480 :     1. rawrecord of record_kind option
481 :     This pure operator allocates some uninitialized storage from the heap.
482 :     There are two forms:
483 :    
484 :     rawrecord NONE [INT n] allocates a tagless record of length n
485 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
486 :     and initializes the tag.
487 :    
488 :     2. rawupdate of cty
489 :     rawupdate cty (v,i,x)
490 :     Assigns to x to the ith component of record v.
491 :     The storelist is not updated.
492 :    
493 :     We use these new primops for both spilling and increment record construction.
494 :    
495 :     1. Spilling.
496 :    
497 :     This is implemented with a linear scan algorithm (but generalized
498 :     to trees). The algorithm will create a single spill record at the
499 :     beginning of the cps function and use rawupdate to spill to it,
500 :     and SELECT or SELp to reload from it. So both spills and reloads
501 :     are fine-grain operations. In contrast, in the old algorithm
502 :     "spills" have to be bundled together in records.
503 :    
504 :     Ideally, we should sink the spill record construction to where
505 :     it is needed. We can even split the spill record into multiple ones
506 :     at the places where they are needed. But CPS is not a good
507 :     representation for global code motion, so I'll keep it simple and
508 :     am not attempting this.
509 :    
510 :     2. Incremental record construction (aka record splitting).
511 :    
512 :     Long records with many component values which are simulatenously live
513 :     (recall that single use record addresses are not considered to
514 :     be live) are constructed with rawrecord and rawupdate.
515 :     We allocate space on the heap with rawrecord first, then gradually
516 :     fill it in with rawupdate. This is the technique suggested to me
517 :     by Matthias.
518 :    
519 :     Some restrictions on when this is applicable:
520 :     1. It is not a VECTOR record. The code generator currently does not handle
521 :     this case. VECTOR record uses double indirection like arrays.
522 :     2. All the record component values are defined in the same "basic block"
523 :     as the record constructor. This is to prevent speculative
524 :     record construction.
525 :    
526 :     ----------------------------------------------------------------------
527 :     Name: Allen Leung
528 : leunga 1086 Date: 2002/02/22 01:02:00 EST
529 :     Tag: leunga-20020222-mlrisc-tools
530 :    
531 :     Minor bug fixes in the parser and rewriter
532 :    
533 :     ----------------------------------------------------------------------
534 :     Name: Allen Leung
535 : leunga 1085 Date: 2002/02/21 20:20:00 EST
536 :     Tag: leunga-20020221-peephole
537 :    
538 :     Regenerated the peephole files. Some contained typos in the specification
539 :     and some didn't compile because of pretty printing bugs in the old version
540 :     of 'nowhere'.
541 :    
542 :     ----------------------------------------------------------------------
543 :     Name: Allen Leung
544 : leunga 1079 Date: 2002/02/19 20:20:00 EST
545 :     Tag: leunga-20020219-mlrisc-tools
546 :     Description:
547 :    
548 :     Minor bug fixes to the mlrisc-tools library:
549 :    
550 :     1. Fixed up parsing colon suffixed keywords
551 :     2. Added the ability to shut the error messages up
552 :     3. Reimplemented the pretty printer and fixed up/improved
553 :     the pretty printing of handle and -> types.
554 :     4. Fixed up generation of literal symbols in the nowhere tool.
555 :     5. Added some SML keywords to to sml.sty
556 :    
557 :     ----------------------------------------------------------------------
558 :     Name: Matthias Blume
559 : blume 1078 Date: 2002/02/19 16:20:00 EST
560 :     Tag: blume-20020219-cmffi
561 :     Description:
562 :    
563 :     A wild mix of changes, some minor, some major:
564 :    
565 :     * All C FFI-related libraries are now anchored under $c:
566 :     $/c.cm --> $c/c.cm
567 :     $/c-int.cm --> $c/internals/c-int.cm
568 :     $/memory.cm --> $c/memory/memory.cm
569 :    
570 :     * "make" tool (in CM) now treats its argument pathname slightly
571 :     differently:
572 :     1. If the native expansion is an absolute name, then before invoking
573 :     the "make" command on it, CM will apply OS.Path.mkRelative
574 :     (with relativeTo = OS.FileSys.getDir()) to it.
575 :     2. The argument will be passed through to subsequent phases of CM
576 :     processing without "going native". In particular, if the argument
577 :     was an anchored path, then "make" will not lose track of that anchor.
578 :    
579 :     * Compiler backends now "know" their respective C calling conventions
580 :     instead of having to be told about it by ml-nlffigen. This relieves
581 :     ml-nlffigen from one of its burdens.
582 :    
583 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
584 :    
585 :     * Export C_DEBUG and C_Debug from $c/c.cm.
586 :    
587 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
588 :     subtyping relationship between incomplete pointers and their complete
589 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
590 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
591 :     the pointer target type is complete. In the incomplete case, 'o
592 :     will be instantiated with some "'c iobj" -- a type obtained by
593 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
594 :    
595 :     Operations that work on both incomplete and complete pointer types are
596 :     typed as taking an 'o ptr while operations that require the target to
597 :     be known are typed as taking some ('t, 'c) obj ptr.
598 :    
599 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
600 :     where void is an eqtype without any values. This makes it possible
601 :     to work on voidptr values using functions meant to operate on light
602 :     incomplete pointers.
603 :    
604 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
605 :     been vastly simplified.
606 :    
607 :     ----------------------------------------------------------------------
608 :     Name: Matthias Blume
609 : blume 1076 Date: 2002/02/19 10:48:00 EST
610 :     Tag: blume-20020219-pqfix
611 :     Description:
612 :    
613 :     Applied Chris Okasaki's bug fix for priority queues.
614 :    
615 :     ----------------------------------------------------------------------
616 :     Name: Matthias Blume
617 : blume 1073 Date: 2002/02/15 17:05:00
618 : blume 1069 Tag: Release_110_39
619 :     Description:
620 :    
621 : blume 1073 Last-minute retagging is becoming a tradition... :-(
622 :    
623 :     This is the working release 110.39.
624 :    
625 :     ----------------------------------------------------------------------
626 :     Name: Matthias Blume
627 :     Date: 2002/02/15 16:00:00 EST
628 :     Tag: Release_110_39-orig
629 :     Description:
630 :    
631 : blume 1069 Working release 110.39. New bootfiles.
632 :    
633 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
634 :     with all shells. So I retagged. -Matthias)
635 :    
636 : blume 1069 ----------------------------------------------------------------------
637 :     Name: Matthias Blume
638 : blume 1068 Date: 2002/02/15 14:17:00 EST
639 :     Tag: blume-20020215-showbindings
640 :     Description:
641 :    
642 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
643 :     the latter can be useful for exploring what bindings are available at
644 :     the interactive prompt. (The first function returns only the list
645 :     of symbols that are really bound, the second prints those but also the
646 :     ones that CM's autoloading mechanism knows about.)
647 :    
648 :     ----------------------------------------------------------------------
649 :     Name: Matthias Blume
650 : blume 1067 Date: 2002/02/15 12:08:00 EST
651 :     Tag: blume-20020215-iptrs
652 :     Description:
653 :    
654 :     Two improvements to ml-nlffigen:
655 :    
656 :     1. Write files only if they do not exist or if their current contents
657 :     do not coincide with what's being written. (That is, avoid messing
658 :     with the time stamps unless absolutely necessary.)
659 :    
660 :     2. Implement a "repository" mechanism for generated files related
661 :     to "incomplete pointer types". See the README file for details.
662 :    
663 :     ----------------------------------------------------------------------
664 :     Name: Matthias Blume
665 : blume 1066 Date: 2002/02/14 11:50:00 EST
666 :     Tag: blume-20020214-quote
667 :     Description:
668 :    
669 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
670 :     because of the new and improved tag generation scheme. (Thanks to Allen
671 :     Leung for pointing it out.)
672 :    
673 :     ----------------------------------------------------------------------
674 : george 1065 Name: Lal George
675 :     Date: 2002/02/14 09:55:27 EST 2002
676 :     Tag: george-20020214-isabelle-bug
677 :     Description:
678 :    
679 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
680 :     of Isabelle on the x86.
681 :    
682 :     From Allen:
683 :     -----------
684 :     I've found the problem:
685 :    
686 :     in ra-core.sml, I use the counter "blocked" to keep track of the
687 :     true number of elements in the freeze queue. When the counter goes
688 :     to zero, I skip examining the queue. But I've messed up the
689 :     bookkeeping in combine():
690 :    
691 :     else ();
692 :     case !ucol of
693 :     PSEUDO => (if !cntv > 0 then
694 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
695 :     ^^^^^^^^^^^^^^^^^^^^^^^
696 :     moveu := mergeMoveList(!movev, !moveu)
697 :     )
698 :     else ();
699 :    
700 :     combine() is called to coalesce two nodes u and v.
701 :     I think I was thinking that if the move counts of u and v are both
702 :     greater than zero then after they are coalesced then one node is
703 :     removed from the freeze queue. Apparently I was thinking that
704 :     both u and v are of low degree, but that's clearly not necessarily true.
705 :    
706 :    
707 :     02/12/2002:
708 :     Here's the patch. HOL now compiles.
709 :    
710 :     I don't know how this impact on performance (compile
711 :     time or runtime). This bug caused the RA (especially on the x86)
712 :     to go thru the potential spill phase when there are still nodes on the
713 :     freeze queue.
714 :    
715 :    
716 :    
717 :    
718 :     ----------------------------------------------------------------------
719 : blume 1058 Name: Matthias Blume
720 : blume 1064 Date: 2002/02/13 22:40:00 EST
721 :     Tag: blume-20020213-fptr-rtti
722 :     Description:
723 :    
724 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
725 :     updates.
726 :    
727 :     ----------------------------------------------------------------------
728 :     Name: Matthias Blume
729 : blume 1063 Date: 2002/02/13 16:41:00 EST
730 :     Tag: blume-20020213-cmlpq
731 :     Description:
732 :    
733 :     Added new priority queue export symbols (which have just been added to
734 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
735 :     not compile and the installer would choke.)
736 :    
737 :     ----------------------------------------------------------------------
738 :     Name: Matthias Blume
739 : blume 1062 Date: 2002/02/13 16:15:00 EST
740 :     Tag: blume-20020213-various
741 :     Description:
742 :    
743 :     1. More tweaks to ml-nlffigen:
744 :    
745 :     - better internal datastructures (resulting in slight speedup)
746 :     - "-match" option requires exact match
747 :     - "localized" gensym counters (untagged structs/unions nested within
748 :     other structs/unions or within typedefs get a fresh counter; their
749 :     tag will be prefixed by a concatenation of their parents' tags)
750 :     - bug fixes (related to calculation of transitive closure of types
751 :     to be included in the output)
752 :    
753 :     2. Minor Basis updates:
754 :    
755 :     - added implementations for List.collate and Option.app
756 :    
757 :     ----------------------------------------------------------------------
758 :     Name: Matthias Blume
759 : blume 1060 Date: 2002/02/11 15:55:00 EST
760 :     Tag: blume-20020211-gensym
761 :     Description:
762 :    
763 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
764 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
765 :     names (ML structure names that correspond to unnamed C structs, unions,
766 :     and enums), so that separate runs of ml-nlffigen do not clash.
767 :    
768 :     ----------------------------------------------------------------------
769 :     Name: Matthias Blume
770 : blume 1059 Date: 2002/02/11 12:05:00 EST
771 :     Tag: blume-20020211-gensml
772 :     Description:
773 :    
774 :     A quick fix for a problem with GenSML (in the pgraph-util library):
775 :     Make generation of toplevel "local" optional. (Strictly speaking,
776 :     signature definitions within "local" are not legal SML.)
777 :    
778 :     Other than that: updates to INSTALL and cm/TODO.
779 :    
780 :     ----------------------------------------------------------------------
781 :     Name: Matthias Blume
782 : blume 1058 Date: 2002/02/08 15:00:00 EST
783 :     Tag: blume-20020208-uniquepid
784 :     Description:
785 :    
786 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
787 :    
788 :     1. The installer (config/install.sh) has gotten smarter:
789 :    
790 :     - Configuration options are a bit easier to specify now
791 :     (in config/targets).
792 :     - Bug in recognizing .tar.bz2 files fixed.
793 :     - Installer automatically resolves dependencies between
794 :     configuration options (e.g., if you ask for eXene, you will
795 :     also get cml -- regardless whether you asked for it or not).
796 :     - Installer can run in "quieter mode" by setting the environment
797 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
798 :     "completely silent", though.
799 :     - Build HashCons library as part of smlnj-lib.
800 :    
801 :     2. A new scheme for assigning persistent identifiers to compilation
802 :     units (and, by extension, to types etc.) has been put into place.
803 :     This fixes a long-standing bug where types and even dynamic values
804 :     can get internally confused, thereby compromising type safety
805 :     (abstraction) and dynamic correctness. See
806 :    
807 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
808 :    
809 :     for an example of how things could go wrong until now.
810 :    
811 :     The downside of the new scheme is that pids are not quite as
812 :     persistent as they used to be: CM will generate a fresh pid
813 :     for every compilation unit that it thinks it sees for the first
814 :     time. That means that if you compile starting from a clean, fresh
815 :     source tree at two different times, you end up with different
816 :     binaries.
817 :    
818 :     Cutoff recompilation, however, has not been compromised because
819 :     CM keeps pid information in special caches between runs.
820 :    
821 :     ----------------------------------------------------------------------
822 : george 1053 Name: Lal George
823 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
824 :     Tag: <none>
825 :     Description:
826 :    
827 :     Compilers that generate assembly code may produce global labels
828 :     whose value is resolved at link time. The various peephole optimization
829 :     modules did not take this in account.
830 :    
831 :     TODO. The Labels.addrOf function should really return an option
832 :     type so that clients are forced to deal with this issue, rather
833 :     than an exception being raised.
834 :    
835 :     ----------------------------------------------------------------------
836 :     Name: Lal George
837 : george 1053 Date: 2002/02/06 13:55:02 EST
838 :     Tag: george-20020206-ra-breakup
839 :     Description:
840 :    
841 :     1. A bug fix from Allen.
842 :    
843 :     A typo causes extra fstp %st(0)'s to be generated at compensation
844 :     edges, which might cause stack underflow traps at runtime. This
845 :     occurs in fft where there are extraneous fstps right before the 'into'
846 :     trap instruction (in this case they are harmless since none of the
847 :     integers overflow.)
848 :    
849 : george 1054 2. Pulled out various utility modules that were embedded in the modules
850 : george 1053 of the register allocator. I need these modules for other purposes, but
851 : george 1054 they are not complete enough to put into a library (just yet).
852 : george 1053 ----------------------------------------------------------------------
853 : blume 1049 Name: Matthias Blume
854 :     Date: 2002/01/31 16:05:00 EST
855 :     Tag: blume-20020131-sparc-ccalls
856 :     Description:
857 :    
858 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
859 :     of extra stack space by mistake. Fixed.
860 :    
861 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
862 :    
863 :     ----------------------------------------------------------------------
864 : leunga 1048 Name: Allen Leung
865 :     Date: 2002/01/30
866 :     Tag: leunga-20020130-nowhere-bug-fix
867 :     Description:
868 :    
869 :     MLRISC bug fixes:
870 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
871 :     2. MachineInt.fromString was negating its value.
872 :    
873 :     ----------------------------------------------------------------------
874 : blume 1039 Name: Matthias Blume
875 : blume 1047 Date: 2002/01/29
876 :     Tag: blume-20020129-INSTALL
877 :     Description:
878 :    
879 :     - Added somewhat detailed installation instructions (file INSTALL).
880 :     - Fixed curl-detection bug in config/install.sh.
881 :     - It is now possible to select the URL getter using the URLGETTER
882 :     environment variable:
883 :    
884 :     not set / "unknown" --> automatic detection (script tries wget,
885 :     curl, and lynx)
886 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
887 :     how to properly invoke them)
888 :     other --> use $URLGETTER directly, it must take
889 :     precisely two command-line arguments
890 :     (source URL and destination file name)
891 :    
892 :     ----------------------------------------------------------------------
893 :     Name: Matthias Blume
894 : blume 1044 Date: 2002/01/28
895 : blume 1045 Tag: blume-20020128-sparc-ccalls
896 :     Description:
897 :    
898 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
899 :     - Make use of the allocParam argument in sparc-c-calls.
900 :    
901 :     ----------------------------------------------------------------------
902 :     Name: Matthias Blume
903 :     Date: 2002/01/28
904 : blume 1044 Tag: blume-20020128-allocParam
905 :     Description:
906 :    
907 :     John Reppy: Changes c-calls API to accept client-callback for
908 :     allocating extra stack space.
909 :     me: Corresponding changes to mlriscGen (using a dummy argument that
910 :     does not change the current behavior).
911 :    
912 :     ----------------------------------------------------------------------
913 :     Name: Matthias Blume
914 : blume 1040 Date: 2002/01/28 12:00:00
915 :     Tag: Release_110_38
916 :     Description:
917 :    
918 :     This time for real!!!
919 :    
920 :     ----------------------------------------------------------------------
921 :     Name: Matthias Blume
922 : blume 1039 Date: 2002/01/28 10:56:00 EST
923 :     Tag: blume-20020128-retraction
924 :     Description:
925 :    
926 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
927 :     with blume-Release_110_38-retracted.)
928 :    
929 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
930 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
931 :    
932 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
933 :     and enums fixed. The naming of corresponding ML identifiers should
934 :     now be consistent again.
935 :    
936 :     ----------------------------------------------------------------------
937 : leunga 1038 Name: Allen Leung
938 :     Date: 2002/01/27
939 :     Tag: leunga-20020127-nowhere
940 :     Description:
941 :    
942 :     Added a target called nowhere in the configuration scripts.
943 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
944 :     programs with where-clauses into legal SML code) during installation.
945 :    
946 :     ----------------------------------------------------------------------
947 : blume 1036 Name: Matthias Blume
948 : leunga 1038 Date: 2002/01/25 21:27:00 EST
949 : blume 1039 Tag: blume-Release_110_38-retracted
950 : blume 1037 Description:
951 :    
952 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
953 :    
954 :     README will be added later.
955 :    
956 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
957 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
958 :    
959 : blume 1037 ----------------------------------------------------------------------
960 :     Name: Matthias Blume
961 : blume 1036 Date: 2002/01/25
962 :     Tag: blume-20020125-ffi
963 :     Description:
964 :    
965 :     A large number of tweaks and improvements to ml-nlffi-lib and
966 :     ml-nlffigen:
967 :    
968 :     - ML represenation types have been streamlined
969 :     - getter and setter functions work with concrete values, not abstract
970 :     ones where possible
971 :     - ml-nlffigen command line more flexible (see README file there)
972 :     - some bugs have been fixed (hopefully)
973 :    
974 :     ----------------------------------------------------------------------
975 : george 1034 Name: Lal George
976 :     Date: 2002/01/24
977 :     Tag: george-20020124-risc-ra-interface
978 :     Description:
979 :    
980 :     There is a dramatic simplification in the interface to the
981 :     register allocator for RISC architectures as a result of making
982 :     parallel copy instructions explicit.
983 :    
984 :     ----------------------------------------------------------------------
985 : blume 1028 Name: Matthias Blume
986 : blume 1032 Date: 2002/01/22
987 :     Tag: blume-20020122-x86-ccalls
988 :     Description:
989 :    
990 :     Bug fix for c-calls on x86 (having to do with how char- and
991 :     short-arguments are being handled).
992 :    
993 :     ----------------------------------------------------------------------
994 :     Name: Matthias Blume
995 : blume 1031 Date: 2002/01/21
996 :     Tag: blume-20020121-ff
997 :     Description:
998 :    
999 :     Another day of fiddling with the FFI...
1000 :    
1001 :     1. Bug fix/workaround: CKIT does not complain about negative array
1002 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
1003 :     (Otherwise a negative dimension would send it into an infinite loop.)
1004 :    
1005 :     2. Some of the abstract types (light objects, light pointers, most "base"
1006 :     types) in structure C are now eqtypes.
1007 :    
1008 :     3. Added constructors and test functions for NULL function pointers.
1009 :    
1010 :     ----------------------------------------------------------------------
1011 :     Name: Matthias Blume
1012 : blume 1028 Date: 2002/01/18
1013 : blume 1030 Tag: blume-20020118-ready-for-new-release
1014 :     Description:
1015 :    
1016 :     Made config/srcarchiveurl point to a new place. (Will provide boot
1017 :     files shortly.)
1018 :    
1019 :     Maybe we christen this to be 110.38?
1020 :    
1021 :     ----------------------------------------------------------------------
1022 :     Name: Matthias Blume
1023 :     Date: 2002/01/18
1024 : blume 1029 Tag: blume-20020118-more-ffifiddle
1025 :     Description:
1026 :    
1027 :     Today's FFI fiddling:
1028 :    
1029 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1030 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
1031 :     of abstract ones.
1032 :    
1033 :     - Provided word-style bit operations etc. for "int" representation
1034 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1035 :     and Long.
1036 :    
1037 :     ----------------------------------------------------------------------
1038 :     Name: Matthias Blume
1039 :     Date: 2002/01/18
1040 : blume 1028 Tag: blume-20020118-use-x86-fp
1041 :     Description:
1042 :    
1043 :     Now that x86-fast-fp seems to be working, I turned it back on again
1044 :     by default. (Seems to work fine now, even with the FFI.)
1045 :    
1046 :     Other than that, I added some documentation about the FFI to
1047 :     src/ml-nlffigen/README and updated the FFI test examples in
1048 :     src/ml-nlffi-lib/Tests/*.
1049 :    
1050 :     ----------------------------------------------------------------------
1051 : leunga 1025 Name: Allen Leung
1052 :     Date: 2002/01/17
1053 :     Tag: leunga-20020117-x86-fast-fp-call
1054 :     Description:
1055 :    
1056 :     1. Fixed a problem with handling return fp values when x86's fast fp
1057 :     mode is turned on.
1058 :    
1059 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
1060 :     of %f32.
1061 :    
1062 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
1063 :    
1064 :     ----------------------------------------------------------------------
1065 : blume 1021 Name: Matthias Blume
1066 :     Date: 2002/01/16
1067 :     Tag: blume-20020116-ffifiddle
1068 :     Description:
1069 :    
1070 :     More fiddling with the FFI interface:
1071 :    
1072 :     - Make constness 'c instead of rw wherever possible. This eliminates
1073 :     the need for certain explicit coercions. (However, due to ML's
1074 :     value polymorphism, there will still be many cases where explicit
1075 :     coercions are necessary. Phantom types are not the whole answer
1076 :     to modeling a subtyping relationship in ML.)
1077 :    
1078 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
1079 :    
1080 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
1081 :     clumsy workaround for varargs problem.)
1082 :    
1083 :     ----------------------------------------------------------------------
1084 : george 1016 Name: Lal George
1085 :     Date: 2002/01/15
1086 :     Tag: <none>
1087 :     Description:
1088 :    
1089 :     1. Since COPY instructions are no longer native to the architecture,
1090 :     a generic functor can be used to implement the expandCopies function.
1091 :    
1092 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1093 :     TEXT segment.
1094 :    
1095 :     ----------------------------------------------------------------------
1096 : blume 1011 Name: Matthias Blume
1097 : blume 1015 Date: 2002/01/15
1098 :     Tag: blume-20020115-ffiupdates
1099 :     Description:
1100 :    
1101 :     1. Fix for bug resulting in single-precision float values being returned
1102 :     incorrectly from FFI calls.
1103 :    
1104 :     2. Small modifications to C FFI API:
1105 :    
1106 :     - memory-allocation routines return straight objects (no options)
1107 :     and raise an exception in out-of-memory situations
1108 :     - unsafe extensions to cast between function pointers and pointers
1109 :     from/to ints
1110 :     - added structure C_Debug as an alternative to structure C where
1111 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
1112 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
1113 :     but also takes a (possibly empty) list of existing library handles
1114 :     that the current library depends on
1115 :    
1116 :     ----------------------------------------------------------------------
1117 :     Name: Matthias Blume
1118 : blume 1011 Date: 2002/01/10
1119 :     Tag: blume-20020110-newffigen
1120 :     Description:
1121 :    
1122 :     1. Updates to portable graph code.
1123 :    
1124 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
1125 :     more scalable now so that even huge interfaces such as the one
1126 :     for GTK compile in finite time and space. :-)
1127 :     See src/ml-nlffigen/README for details on what's new.
1128 :    
1129 :     ----------------------------------------------------------------------
1130 : george 1003 Name: Lal George
1131 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
1132 :     Tag: george-20011206-rm-native-copy
1133 :     Description:
1134 :    
1135 :     Removed the native COPY and FCOPY instructions
1136 :     from all the architectures and replaced it with the
1137 :     explicit COPY instruction from the previous commit.
1138 :    
1139 :     It is now possible to simplify many of the optimizations
1140 :     modules that manipulate copies. This has not been
1141 :     done in this change.
1142 :    
1143 :     ----------------------------------------------------------------------
1144 :     Name: Lal George
1145 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
1146 :     Tag: george-20011206-mlrisc-instruction
1147 :     Description:
1148 : george 824
1149 : george 1003 Changed the representation of instructions from being fully abstract
1150 :     to being partially concrete. That is to say:
1151 :    
1152 :     from
1153 :     type instruction
1154 :    
1155 :     to
1156 :     type instr (* machine instruction *)
1157 :    
1158 :     datatype instruction =
1159 :     LIVE of {regs: C.cellset, spilled: C.cellset}
1160 :     | KILL of {regs: C.cellset, spilled: C.cellset}
1161 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1162 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
1163 :     | INSTR of instr
1164 :    
1165 :     This makes the handling of certain special instructions that appear on
1166 :     all architectures easier and uniform.
1167 :    
1168 :     LIVE and KILL say that a list of registers are live or killed at the
1169 :     program point where they appear. No spill code is generated when an
1170 :     element of the 'regs' field is spilled, but the register is moved to
1171 :     the 'spilled' (which is present, more for debugging than anything else).
1172 :    
1173 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1174 :     We used to generate:
1175 :    
1176 :     DEFFREG f1
1177 :     f1 := f2 + f3
1178 :     trapb
1179 :    
1180 :     but now generate:
1181 :    
1182 :     f1 := f2 + f3
1183 :     trapb
1184 :     LIVE {regs=[f1,f2,f3], spilled=[]}
1185 :    
1186 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
1187 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1188 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1189 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1190 :     this architecture).
1191 :    
1192 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
1193 :     all the architectures. This will result in further simplification of the
1194 :     register allocator that must be aware of them for coalescing purposes, and
1195 :     will also simplify certain aspects of the machine description that provides
1196 :     callbacks related to parallel copies.
1197 :    
1198 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
1199 :     machine instruction set!
1200 :    
1201 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
1202 :     functions for making porting easier -- essentially converting upper case
1203 :     to lower case. All machine instructions (of type instr) are in upper case,
1204 :     and the lower case form generates an MLRISC instruction. For example on
1205 :     the alpha we have:
1206 :    
1207 :     datatype instr =
1208 :     LDA of {r:cell, b:cell, d:operand}
1209 :     | ...
1210 :    
1211 :     val lda : {r:cell, b:cell, d:operand} -> instruction
1212 :     ...
1213 :    
1214 :     where lda is just (INSTR o LDA), etc.
1215 :    
1216 : leunga 815 ----------------------------------------------------------------------
1217 : blume 986 Name: Matthias Blume
1218 : blume 993 Date: 2001/11/22 21:40:00 EST
1219 :     Tag: Release_110_37
1220 :     Description:
1221 :    
1222 :     Release 110.37. This time for real.
1223 :    
1224 :     ----------------------------------------------------------------------
1225 :     Name: Matthias Blume
1226 : blume 988 Date: 2001/11/21 16:35:00 EST
1227 :     Tag: blume-20011121-foot-in-mouth
1228 :     Description:
1229 :    
1230 :     Removed the "Release_110_37" tag because of a serious bug.
1231 :     This will be re-tagged once the bug is fixed.
1232 :    
1233 :     ----------------------------------------------------------------------
1234 :     Name: Matthias Blume
1235 : blume 987 Date: 2001/11/21 16:14:00 EST
1236 :     Tag: blume-20011121-forgottenfile
1237 :     Description:
1238 :    
1239 :     Forgot to add a file. (Just a .tex-file -- part of
1240 :     the CM manual source.)
1241 :    
1242 :     ----------------------------------------------------------------------
1243 :     Name: Matthias Blume
1244 : blume 986 Date: 2001/11/21 16:10:00 EST
1245 : blume 988 Tag: blume-20011121-invalid_110_37
1246 : blume 986 Description:
1247 :    
1248 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
1249 :     because we found a serious bug in all non-x86 backends.
1250 :     - Matthias
1251 :    
1252 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
1253 :     so that code object name strings are directly inserted into code
1254 :     objects at code generation time. The only business the runtime system
1255 :     has with this is now to read the name strings on occasions.
1256 :     (The encoding of the name string has also changed somewhat.)
1257 :    
1258 :     2. CM now implements a simple "set calculus" for specifying export lists.
1259 :     In particular, it is now possible to refer to the export lists of
1260 :     other libraries/groups/sources and form unions as well as differences.
1261 :     See the latest CM manual for details.
1262 :    
1263 :     3. An separate notion of "proxy" libraries has again be eliminated from
1264 :     CM's model. (Proxy libraries are now simply a special case of using
1265 :     the export list calculus.)
1266 :    
1267 :     4. Some of the existing libraries now take advantage of the new set
1268 :     calculus.
1269 :     (Notice that not all libraries have been converted because some
1270 :     of the existing .cm-files are supposed to be backward compatible
1271 :     with 110.0.x.)
1272 :    
1273 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
1274 :     "General.exnMessage"! The former relies on a certain hook to be
1275 :     initialized, and that often does not happen in the stand-alone case.)
1276 :    
1277 :     ----------------------------------------------------------------------
1278 : george 984 Name: Lal George
1279 :     Date: 2001/11/21 13:56:18 EST
1280 :     Tag: george-2001121-pseudo-ops
1281 :     Description:
1282 :    
1283 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
1284 :     ought to never be any question of incompatabilities with
1285 :     pseudo-op syntax expected by host assemblers.
1286 :    
1287 :     For now, only modules supporting GAS syntax are implemented
1288 :     but more should follow, such as MASM, and vendor assembler
1289 :     syntax, e.g. IBM as, Sun as, etc.
1290 :    
1291 :     ----------------------------------------------------------------------
1292 : blume 977 Name: Matthias Blume
1293 :     Date: 2001/11/14 11:52:00 EST
1294 :     Tag: blume-20011114-srcname
1295 :     Description:
1296 :    
1297 :     1. Routed the name of the current source file to mlriscgen where it
1298 :     should be directly emitted into the code object. (This last part
1299 :     is yet to be done.)
1300 :    
1301 :     2. Some cleanup of the pgraph code to make it match the proposal that
1302 :     I put out the other day. (The proposal notwithstanding, things are
1303 :     still in flux here.)
1304 :    
1305 :     ----------------------------------------------------------------------
1306 : george 976 Name: Lal George
1307 :     Date: 2001/11/14 09:44:04 EST
1308 :     Tag:
1309 :     Description:
1310 :    
1311 :     Fix for a backpatching bug reported by Allen.
1312 :    
1313 :     Because the boundary between short and long span-dependent
1314 :     instructions is +/- 128, there are an astounding number of
1315 :     span-dependent instructions whose size is over estimated.
1316 :    
1317 :     Allen came up with the idea of letting the size of span
1318 :     dependent instructions be non-monotonic, for a maxIter
1319 :     number of times, after which the size must be monotonically
1320 :     increasing.
1321 :    
1322 :     This table shows the number of span-dependent instructions
1323 :     whose size was over-estimated as a function of maxIter, for the
1324 :     file Parse/parse/ml.grm.sml:
1325 :    
1326 :     maxIter # of instructions:
1327 :     10 687
1328 :     20 438
1329 :     30 198
1330 :     40 0
1331 :    
1332 :     In compiling the compiler, there is no significant difference in
1333 :     compilation speed between maxIter=10 and maxIter=40. Actually,
1334 :     my measurements showed that maxIter=40 was a tad faster than
1335 :     maxIter=10! Also 96% of the files in the compiler reach a fix
1336 :     point within 13 iterations, so fixing maxIter at 40, while high,
1337 :     is okay.
1338 :    
1339 :     ----------------------------------------------------------------------
1340 : blume 975 Name: Matthias Blume
1341 :     Date: 2001/10/31 15:25:00 EST
1342 :     Tag: blume-20011031-pgraph
1343 :     Description:
1344 :    
1345 :     CKIT:
1346 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
1347 :     argument identifiers.
1348 :     * Changed the return type of TypeUtil.getFunction accordingly.
1349 :     * Type equality ignores the argument names.
1350 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
1351 :     if there is a mismatch.
1352 :    
1353 :     installation script:
1354 :     * attempts to use "curl" if available (unless "wget" is available as well)
1355 :    
1356 :     CM:
1357 :     * has an experimental implementation of "portable graphs" which I will
1358 :     soon propose as an implementation-independent library format
1359 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1360 :    
1361 :     NLFFI-LIB:
1362 :     * some cleanup (all cosmetic)
1363 :    
1364 :     NLFFIGEN:
1365 :     * temporarily disabled the mechanism that suppresses ML output for
1366 :     C definitions whose identifiers start with an underscore character
1367 :     * generate val bindings for enum constants
1368 :     * user can request that only one style (light or heavy) is being used;
1369 :     default is to use both (command-line arguments: -heavy and -light)
1370 :     * fixed bug in handling of function types involving incomplete pointers
1371 :     * generate ML entry points that take record arguments (i.e., using
1372 :     named arguments) for C functions that have a prototype with named
1373 :     arguments
1374 :     (see changes to CKIT)
1375 :    
1376 :     ----------------------------------------------------------------------
1377 : leunga 972 Name: Allen Leung
1378 :     Date: 2001/10/27 20:34:00 EDT
1379 :     Tag: leunga-20011027-x86-fast-fp-call
1380 :     Description:
1381 :    
1382 :     Fixed the bug described in blume-20010920-slowfp.
1383 :    
1384 :     The fix involves
1385 :     1. generating FCOPYs in FSTP in ia32-svid
1386 :     2. marking a CALL with the appropriate annotation
1387 :    
1388 :     ----------------------------------------------------------------------
1389 : blume 961 Name: Matthias Blume
1390 :     Date: 2001/10/16 11:32:00 EDT
1391 :     Tag: blume-20011016-netbsd
1392 :     Description:
1393 :    
1394 :     Underscore patch from Chris Richards (fixing problem with compiling
1395 :     runtime system under recent NetBSD).
1396 :    
1397 :     ----------------------------------------------------------------------
1398 : george 959 Name: Allen Leung
1399 :     Date: 2001/10/12 17:18:32 EDT 2001
1400 :     Tag: leung-20011012-x86-printflowgraph
1401 :     Description:
1402 :    
1403 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1404 :    
1405 :     ----------------------------------------------------------------------
1406 : george 952 Name: Lal George
1407 : george 958 Date: 2001/10/11 23:51:34 EDT
1408 :     Tag: george-20011011-too-many-instrs
1409 :     Description:
1410 :    
1411 :     The representation of a program point never expected to see more
1412 :     than 65536 instructions in a basic block!
1413 :    
1414 :     ----------------------------------------------------------------------
1415 :     Name: Lal George
1416 : george 952 Date: 2001/10/09 09:41:37 EDT
1417 :     Tag: george-20011008-mlrisc-labels
1418 :     Description:
1419 :    
1420 :     Changed the machine description files to support printing of
1421 :     local and global labels in assembly code, based on host assembler
1422 :     conventions.
1423 :    
1424 :     ----------------------------------------------------------------------
1425 : blume 916 Name: Matthias Blume
1426 : blume 939 Date: 2001/09/25 15:25:00 EDT
1427 :     Tag: blume-20010925-exninfo
1428 :     Description:
1429 :    
1430 :     I provided a non-hook implementation of exnName (at the toplevel) and
1431 :     made the "dummy" implementation of exnMessage (at the toplevel) more
1432 :     useful: if nothing gets "hooked in", then at least you are going to
1433 :     see the exception name and a message indicating why you don't see more.
1434 :    
1435 :     [For the time being, programs that need exnMessage and want to use
1436 :     ml-build should either use General.exnMessage (strongly recommended) or
1437 :     refer to structure General at some other point so that CM sees a
1438 :     static dependency.]
1439 :    
1440 :     [Similar remarks go for "print" and "use": If you want to use their
1441 :     functionality in stand-alone programs generated by ml-build, then use
1442 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1443 :    
1444 :     ----------------------------------------------------------------------
1445 :     Name: Matthias Blume
1446 : blume 938 Date: 2001/09/20 17:28:00 EDT
1447 :     Tag: blume-20010920-slowfp
1448 :     Description:
1449 :    
1450 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1451 :    
1452 :     ----------------------------------------------------------------------
1453 :     Name: Matthias Blume
1454 : blume 937 Date: 2001/09/20 17:20:00 EDT
1455 :     Tag: blume-20010920-canonicalpaths
1456 :     Description:
1457 :    
1458 :     0. Updated the BOOT file (something that I forgot to do earlier).
1459 :    
1460 :     1. Small internal change to CM so that it avoids "/../" in filenames
1461 :     as much as possible (but only where it is safe).
1462 :    
1463 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1464 :     that arguments that contain delimiters are passed through correctly.
1465 :     This change also means that all "special" arguments of the form
1466 :     @SMLxxx... must come first.
1467 :    
1468 :     3. Changed install script to put relative anchor names for tool commands
1469 :     into pathconfig.
1470 :    
1471 :     ----------------------------------------------------------------------
1472 :     Name: Matthias Blume
1473 : blume 932 Date: 2001/09/18 15:35:00 EDT
1474 :     Tag: blume-20010918-readme11036
1475 :     Description:
1476 :    
1477 :     Added README files.
1478 :    
1479 :     ----------------------------------------------------------------------
1480 :     Name: Matthias Blume
1481 : blume 929 Date: 2001/09/18 11:45:00 EDT
1482 :     Tag: Release_110_36 (retag)
1483 :     Description:
1484 :    
1485 :     Fixed mistake in config/preloads. Retagged as 110.36.
1486 :    
1487 :     ----------------------------------------------------------------------
1488 :     Name: Matthias Blume
1489 : blume 928 Date: 2001/09/18 09:40:00 EDT
1490 : blume 929 Tag: Release_110_36_orig (tag changed)
1491 : blume 928 Description:
1492 :    
1493 :     New version (110.36). New bootfiles.
1494 :    
1495 :     ----------------------------------------------------------------------
1496 :     Name: Matthias Blume
1497 : blume 927 Date: 2001/09/14 16:15:00 EDT
1498 :     Tag: blume-20010914-x86fastfp
1499 :     Description:
1500 :    
1501 :     John committed some changes that Allen made, in particular a (hopefully)
1502 :     correctly working version of the x86-fp module.
1503 :    
1504 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1505 :     flag to "true". Everything seems to compile to a fixpoint ok, and
1506 :     "mandelbrot" speeds up by about 15%.
1507 :    
1508 :     ----------------------------------------------------------------------
1509 :     Name: Matthias Blume
1510 : blume 923 Date: 2001/09/13 11:20:00 EDT
1511 :     Tag: blume-20010913-minimal
1512 :     Description:
1513 :    
1514 :     1. Stefan Monnier's patch to fix a miscompilation problem that
1515 :     was brought to light by John Reppy's work on Moby.
1516 :    
1517 :     2. Implemented a minimal "structure Compiler" that contains just
1518 :     "version" and "architecture". The minimal version will be
1519 :     available when the full version is not. This is for backward-
1520 :     compatibility with code that wants to test Compiler.version.
1521 :    
1522 :     ----------------------------------------------------------------------
1523 :     Name: Matthias Blume
1524 : blume 920 Date: 2001/08/28 14:03:00 EDT
1525 :     Tag: blume-20010828-ml-lex
1526 :     Description:
1527 :    
1528 :     Fix for bug 1581, received from Neophytos Michael.
1529 :    
1530 :     ----------------------------------------------------------------------
1531 :     Name: Matthias Blume
1532 : blume 918 Date: 2001/08/27 11:20:00 EDT
1533 :     Tag: blume-20010827-readme11035
1534 :     Description:
1535 :    
1536 :     Fleshed out the README file for 110.35.
1537 :    
1538 :     ----------------------------------------------------------------------
1539 :     Name: Matthias Blume
1540 : blume 916 Date: 2001/08/24 17:10:00 EDT
1541 :     Tag: Release_110_35
1542 :     Description:
1543 :    
1544 :     New version number (110.35). New bootfiles.
1545 :    
1546 :     ----------------------------------------------------------------------
1547 : george 915 Name: Lal George
1548 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1549 :     Tag: george-20010824-MLRISC-graphs
1550 :     Description:
1551 :    
1552 :     removed clusters from MLRISC completely and replaced with graphs.
1553 :    
1554 :     ----------------------------------------------------------------------
1555 :     Name: Matthias Blume
1556 : blume 905 Date: 2001/08/23 17:50:00 EDT
1557 :     Tag: blume-20010823-toplevel
1558 :     Description:
1559 :    
1560 :     - some reorganization of the code that implements various kinds of
1561 :     environments in the compiler (static, dynamic, symbolic, combined)
1562 :     - re-implemented the EnvRef module so that evalStream works properly
1563 :     (if the stream contains references to "use", "CM.make", etc.)
1564 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1565 :    
1566 :     ----------------------------------------------------------------------
1567 :     Name: Matthias Blume
1568 : blume 904 Date: 2001/08/20 15:50 EDT
1569 :     Tag: blume20010820-slipup
1570 :     Description:
1571 :    
1572 :     I forgot to commit a few files. Here they are...
1573 :    
1574 :     ----------------------------------------------------------------------
1575 :     Name: Matthias Blume
1576 : blume 903 Date: 2001/08/20 15:35:00 EDT
1577 :     Tag: blume-20010820-debugprof
1578 :     Description:
1579 :    
1580 :     !!!! NEW BOOTFILES !!!!
1581 :    
1582 :     This is another round of reorganizing the compiler sources. This
1583 :     time the main goal was to factor out all the "instrumentation"
1584 :     passes (for profiling and backtracing) into their own library.
1585 :     The difficulty was to do it in such a way that it does not depend
1586 :     on elaborate.cm but only on elabdata.cm.
1587 :    
1588 :     Therefore there have been further changes to both elaborate.cm and
1589 :     elabdata.cm -- more "generic" things have been moved from the former
1590 :     to the latter. As a result, I was forced to split the assignment
1591 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1592 :     SML/NJ-specific. Since it would have been awkward to maintain,
1593 :     I bit the bullet and actually _changed_ the mapping between these
1594 :     numbers and primtycs. The bottom line of this is that you need
1595 :     a new set of bin- and bootfiles.
1596 :    
1597 :     I have built new bootfiles for all architectures, so doing a fresh
1598 :     checkout and config/install.sh should be all you need.
1599 :    
1600 :     The newly created library's name is
1601 :    
1602 :     $smlnj/viscomp/debugprof.cm
1603 :    
1604 :     and its sources live under
1605 :    
1606 :     src/compiler/DebugProf
1607 :    
1608 :     ----------------------------------------------------------------------
1609 :     Name: Matthias Blume
1610 : blume 902 Date: 2001/08/15 17:15:00 EDT
1611 :     Tag: blume-20010815-compreorg
1612 :     Description:
1613 :    
1614 :     This is a first cut at reorganizing the CM libraries that make up the
1615 :     core of the compiler. The idea is to separate out pieces that could
1616 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1617 :    
1618 :     The current status is a step in this direction, but it is not quite
1619 :     satisfactory yet. Expect more changes in the future.
1620 :    
1621 :     Here is the current (new) organization...
1622 :    
1623 :     What used to be $smlnj/viscomp/core.cm is now divided into
1624 :     six CM libraries:
1625 :    
1626 :     $smlnj/viscomp/basics.cm
1627 :     /parser.cm
1628 :     /elabdata.cm
1629 :     /elaborate.cm
1630 :     /execute.cm
1631 :     /core.cm
1632 :    
1633 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1634 :     All these libraries are proxy libraries that contain precisely
1635 :     one CM library component. Here are the locations of the components
1636 :     (all within the src/compiler tree):
1637 :    
1638 :     Basics/basics.cm
1639 :     Parse/parser.cm
1640 :     ElabData/elabdata.cm
1641 :     Elaborator/elaborate.cm
1642 :     Execution/execute.cm
1643 :     core.cm
1644 :    
1645 :     [This organization is the same that has been used already
1646 :     for a while for the architecture-specific parts of the visible
1647 :     compiler and for the old version of core.cm.]
1648 :    
1649 :     As you will notice, many source files have been moved from their
1650 :     respective original locations to a new home in one of the above
1651 :     subtrees.
1652 :    
1653 :     The division of labor between the new libraries is the following:
1654 :    
1655 :     basics.cm:
1656 :     - Simple, basic definitions that pertain to many (or all) of
1657 :     the other libraries.
1658 :     parser.cm:
1659 :     - The SML parser, producing output of type Ast.dec.
1660 :     - The type family for Ast is also defined and exported here.
1661 :     elabdata.cm:
1662 :     - The datatypes that describe input and output of the elaborator.
1663 :     This includes types, absyn, and static environments.
1664 :     elaborator.cm:
1665 :     - The SML/NJ type checker and elaborator.
1666 :     This maps an Ast.dec (with a given static environment) to
1667 :     an Absyn.dec (with a new static environment).
1668 :     - This libraries implements certain modules that used to be
1669 :     structures as functors (to remove dependencies on FLINT).
1670 :     execute.cm:
1671 :     - Everything having to do with executing binary code objects.
1672 :     - Dynamic environments.
1673 :     core.cm:
1674 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1675 :     - Top-level modules.
1676 :     - FLINT (this should eventually become its own library)
1677 :    
1678 :     Notes:
1679 :    
1680 :     I am not 100% happy with the way I separated the elaborator (and its
1681 :     data structures) from FLINT. Two instances of the same problem:
1682 :    
1683 :     1. Data structures contain certain fields that carry FLINT-specific
1684 :     information. I hacked around this using exn and the property list
1685 :     module from smlnj-lib. But the fact that there are middle-end
1686 :     specific fields around at all is a bit annoying.
1687 :    
1688 :     2. The elaborator calculates certain FLINT-related information. I tried
1689 :     to make this as abstract as I could using functorization, but, again,
1690 :     the fact that the elaborator has to perform calculations on behalf
1691 :     of the middle-end at all is not nice.
1692 :    
1693 :     3. Having to used exn and property lists is unfortunate because it
1694 :     weakens type checking. The other alternative (parameterizing
1695 :     nearly *everything*) is not appealing, though.
1696 :    
1697 :     I removed the "rebinding =" warning hack because due to the new organization
1698 :     it was awkward to maintain it. As a result, the compiler now issues some of
1699 :     these warnings when compiling init.cmi during bootstrap compilation. On
1700 :     the plus side, you also get a warning when you do, for example:
1701 :     val op = = Int32.+
1702 :     which was not the case up to now.
1703 :    
1704 :     I placed "assign" and "deref" into the _Core structure so that the
1705 :     code that deals with the "lazy" keyword can find them there. This
1706 :     removes the need for having access to the primitive environment
1707 :     during elaboration.
1708 :    
1709 :     ----------------------------------------------------------------------
1710 :     Name: Matthias Blume
1711 : blume 898 Date: 2001/08/13
1712 :     Tag: blume-20010813-closures
1713 :     Description:
1714 :    
1715 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1716 :     performance of certain loops by avoiding needless closure allocation.
1717 :    
1718 :     ----------------------------------------------------------------------
1719 : george 890 Name: Lal George
1720 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1721 :     Tag: george-20010731-x86-fmalloc
1722 :     Description: Fixed bug in x86 calls
1723 :    
1724 :     There was a bug where call instructions would mysteriously
1725 :     vanish. The call instruction had to be one that returned
1726 :     a floating point value.
1727 :    
1728 :     ----------------------------------------------------------------------
1729 :     Name: Lal George
1730 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1731 :     Tag: george-20010719-simple-cells
1732 :     Description:
1733 :    
1734 :     I have dramatically simplified the interface for CELLS in MLRISC.
1735 :    
1736 :     In summary, the cells interface is broken up into three parts:
1737 :    
1738 :     1. CellsBasis : CELLS_BASIS
1739 :    
1740 :     CellsBasis is a top level structure and common for all
1741 :     architectures. it contains the definitions of basic datatypes
1742 :     and utility functions over these types.
1743 :    
1744 :     2. functor Cells() : CELLS
1745 :    
1746 :     Cells generates an interface for CELLS that incorporates the
1747 :     specific resources on the target architecture, such as the
1748 :     presence of special register classes, their number and size,
1749 :     and various useful substructures.
1750 :    
1751 :     3. <ARCH>CELLS
1752 :    
1753 :     e.g. SparcCells: SPARCCELLS
1754 :    
1755 :     <ARCH>CELLS usually contains additional bindings for special
1756 :     registers on the architecture, such as:
1757 :    
1758 :     val r0 : cell (* register zero *)
1759 :     val y : cell (* Y register *)
1760 :     val psr : cell (* processor status register *)
1761 :     ...
1762 :    
1763 :     The structure returned by applying the Cells functor is opened
1764 :     in this interface.
1765 :    
1766 :     The main implication of all this is that the datatypes for cells is
1767 :     split between CellsBasis and CELLS -- a fairly simple change for user
1768 :     code.
1769 :    
1770 :     In the old scheme the CELLS interface had a definitional binding of
1771 :     the form:
1772 :    
1773 :     signature CELLS = sig
1774 :    
1775 :     structure CellsBasis = CellsBasis
1776 :    
1777 :     ...
1778 :    
1779 :     end
1780 :    
1781 :     With all the sharing constraints that goes on in MLRISC, this old
1782 :     design quickly leads to errors such as:
1783 :    
1784 :     "structure definition spec inside of sharing ... "
1785 :    
1786 :    
1787 :     and appears to require an unacceptable amount of sharing and where
1788 :     constraint hackery.
1789 :    
1790 :     I think this error message (the interaction of definitional specs and
1791 :     sharing) requires more explanation on our web page.
1792 :    
1793 :     ----------------------------------------------------------------------
1794 : blume 826 Name: Matthias Blume
1795 : blume 879 Date: 2001/07/19 15:00:00 EDT
1796 :     Tag: blume-20010719-libreorg
1797 :     Description:
1798 :    
1799 :     This update puts together a fairly extensive but straightforward change
1800 :     to the way the libraries that implement the interactive system are
1801 :     organized:
1802 :    
1803 :     The biggest change is the elimination of structure Compiler. As a
1804 :     replacement for this structure, there is now a CM library
1805 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1806 :     that exports all the substructures of the original structure Compiler
1807 :     directly. So instead of saying Compiler.Foo.bar one now simply
1808 :     says Foo.bar. (The CM libraries actually export a collection of
1809 :     structures that is richer than the collection of substructures of
1810 :     structure Compiler.)
1811 :    
1812 :     To make the transition smooth, there is a separate library called
1813 :     $smlnj/compiler/compiler.cm which puts together and exports the
1814 :     original structure Compiler (or at least something very close to it).
1815 :    
1816 :     There are five members of the original structure Compiler
1817 :     that are not exported directly but which instead became members
1818 :     of a new structure Backend (described by signature BACKEND). These are:
1819 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1820 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
1821 :     architecture (: string).
1822 :    
1823 :     Structure Compiler.Version has become structure CompilerVersion.
1824 :    
1825 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1826 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1827 :     or x86, respectively.
1828 :     Each of these exports the same frontend structures that
1829 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
1830 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1831 :     Hppa, PPC, Sparc, or X86, respectively.
1832 :    
1833 :     Library $smlnj/compiler/all.cm exports the union of the exports of
1834 :     $smlnj/compiler/<arch>.cm
1835 :    
1836 :     There are no structures <Arch>Compiler anymore, use
1837 :     $smlnj/compiler/<arch>.cm instead.
1838 :    
1839 :     Library host-compiler-0.cm is gone. Instead, the internal library
1840 :     that instantiates CM is now called cm0.cm. Selection of the host
1841 :     compiler (backend) is no longer done here but. (Responsibility for it
1842 :     now lies with $smlnj/compiler/current.cm. This seems to be more
1843 :     logical.)
1844 :    
1845 :     Many individual files have been moved or renamed. Some files have
1846 :     been split into multiple files, and some "dead" files have been deleted.
1847 :    
1848 :     Aside from these changes to library organization, there are also changes
1849 :     to the way the code itself is organized:
1850 :    
1851 :     Structure Binfile has been re-implemented in such a way that it no
1852 :     longer needs any knowledge of the compiler. It exclusively deals
1853 :     with the details of binfile layout. It no longer invokes the
1854 :     compiler (for the purpose of creating new prospective binfile
1855 :     content), and it no longer has any knowledge of how to interpret
1856 :     pickles.
1857 :    
1858 :     Structure Compile (: COMPILE) has been stripped down to the bare
1859 :     essentials of compilation. It no longer deals with linking/execution.
1860 :     The interface has been cleaned up considerably.
1861 :    
1862 :     Utility routines for dealing with linking and execution have been
1863 :     moved into their own substructures.
1864 :    
1865 :     (The ultimate goal of these changes is to provide a light-weight
1866 :     binfile loader/linker (at least for, e.g., stable libraries) that
1867 :     does not require CM or the compiler to be present.)
1868 :    
1869 :     CM documentation has been updated to reflect the changes to library
1870 :     organization.
1871 :    
1872 :     ----------------------------------------------------------------------
1873 :     Name: Matthias Blume
1874 : blume 875 Date: 2001/07/10 17:30:00 EDT
1875 :     Tag: Release_110_34
1876 :     Description:
1877 :    
1878 : blume 876 Minor tweak to 110.34 (re-tagged):
1879 :    
1880 :     - README.html file added to CVS repository
1881 :     - runtime compiles properly under FreeBSD 3.X and 4.X
1882 :    
1883 :     ----------------------------------------------------------------------
1884 :     Name: Matthias Blume
1885 :     Date: 2001/07/10 17:30:00 EDT
1886 :     Tag: Release_110_34
1887 :     Description:
1888 :    
1889 : blume 875 New version number (110.34). New bootfiles.
1890 :    
1891 :     ----------------------------------------------------------------------
1892 :     Name: Matthias Blume
1893 : blume 874 Date: 2001/07/09 16:00:00 EDT
1894 :     Tag: blume-20010709-more-varargs
1895 :     Description:
1896 :    
1897 :     I changed the handling of varargs in ml-nlffigen again:
1898 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
1899 :    
1900 :     The immediate effect is that you can actually call a varargs function
1901 :     from ML -- but you can't actually supply any arguments beyond the ones
1902 :     specified explicitly. (For example, you can call printf with its format
1903 :     string, but you cannot pass additional arguments.)
1904 :    
1905 :     This behavior is only marginally more useful than the one before, but
1906 :     it has the advantage that a function or, more importantly, a function
1907 :     type never gets dropped on the floor, thus avoiding follow-up problems with
1908 :     other types that refer to the offending one.
1909 :    
1910 :     ----------------------------------------------------------------------
1911 :     Name: Matthias Blume
1912 : blume 873 Date: 2001/07/09 11:25:00 EDT
1913 :     Tag: blume-20010709-varargs
1914 :     Description:
1915 :    
1916 :     1. ckit-lib.cm now exports structure Error
1917 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1918 :     with a warning accompanied by a source location. Moreover, it
1919 :     merely skips the offending function or type and proceeds with the
1920 :     rest of its work.u As a result, one can safely feed C code containing
1921 :     "..." to ml-nlffigen.
1922 :     3. There are some internal improvements to CM, providing slightly
1923 :     more general string substitutions in the tools subsystem.
1924 :    
1925 :     ----------------------------------------------------------------------
1926 :     Name: Matthias Blume
1927 : blume 854 Date: 2001/06/27 15:10:00 EDT
1928 :     Tag: blume-20010627-concur
1929 :     Description:
1930 :    
1931 :     Fixed a small bug in CM's handling of parallel compilation.
1932 :     (You could observe the bug by Control-C-interrupting an ordinary
1933 :     CMB.make or CM.stabilize and then attaching some compile servers.
1934 :     The result was that all of a sudden the previously interrupted
1935 :     compilation would continue on its own. This was because of
1936 :     an over-optimization: CM did not bother to clean out certain queues
1937 :     when no servers were attached "anyway", resulting in the contents
1938 :     of these queues to grab control when new servers did get attached.)
1939 :    
1940 :     There is also another minor update to the CM manual.
1941 :    
1942 :     ----------------------------------------------------------------------
1943 :     Name: Matthias Blume
1944 : blume 853 Date: 2001/06/26 16:15:00 EDT
1945 :     Tag: blume-20010626-cmdoc
1946 :     Description:
1947 :    
1948 :     Minor typo fixed in CM manual (syntax diagram for libraries).
1949 :    
1950 :     ----------------------------------------------------------------------
1951 :     Name: Matthias Blume
1952 : blume 852 Date: 2001/06/25 22:55:00 EDT
1953 :     Tag: blume-20010625-x86pc
1954 :     Description:
1955 :    
1956 :     Fixed a nasty bug in the X86 assembly code that caused signal
1957 :     handlers to fail (crash) randomly.
1958 :    
1959 :     ----------------------------------------------------------------------
1960 :     Name: Matthias Blume
1961 : blume 846 Date: 2001/06/25 12:05:00 EDT
1962 :     Tag: blume-20010625-nlffigen
1963 :     Description:
1964 :    
1965 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
1966 :     Nick Carter <nbc@andrew.cmu.edu>.
1967 :    
1968 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1969 :     2. Default names for generated files are now derived from the name of
1970 :     the C file *without its directory*. In particular, this causes generated
1971 :     files to be placed locally even if the C file is in some system directory.
1972 :     3. Default names for generated signatures and structures are also derived
1973 :     from the C file name without its directory. This avoids silly things
1974 :     like "structure GL/GL".
1975 :     (Other silly names are still possible because ml-nlffigen does not do
1976 : blume 847 a thorough check of whether generated names are legal ML identifiers.
1977 :     When in doubt, use command line arguments to force particular names.)
1978 : blume 846
1979 :     ----------------------------------------------------------------------
1980 :     Name: Matthias Blume
1981 : blume 845 Date: 2001/06/21 12:25:00 EDT
1982 :     Tag: blume-20010621-eXene
1983 :     Description:
1984 :    
1985 :     eXene now compiles and (sort of) works again.
1986 :    
1987 :     The library name (for version > 110.33) is $/eXene.cm.
1988 :    
1989 :     I also added an new example in src/eXene/examples/nbody. See the
1990 :     README file there for details.
1991 :    
1992 :     ----------------------------------------------------------------------
1993 :     Name: Matthias Blume
1994 : blume 844 Date: 2001/06/20 16:40:00 EDT
1995 :     Tag: blume-20010620-cml
1996 :     Description:
1997 :    
1998 :     CML now compiles and works again.
1999 :    
2000 :     Libraries (for version > 110.33):
2001 :    
2002 :     $cml/cml.cm Main CML library.
2003 :     $cml/basis.cm CML's version of $/basis.cm.
2004 :     $cml/cml-internal.cm Internal helper library.
2005 :     $cml/core-cml.cm Internal helper library.
2006 :     $cml-lib/trace-cml.cm Tracing facility.
2007 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
2008 :    
2009 :     The installer (config/install.sh) has been taught how to properly
2010 :     install this stuff.
2011 :    
2012 :     ----------------------------------------------------------------------
2013 :     Name: Matthias Blume
2014 : blume 843 Date: 2001/06/19 17:55:00 EDT
2015 :     Tag: blume-20010619-instantiate
2016 :     Description:
2017 :    
2018 :     This un-breaks the fix for bug 1432.
2019 :     (The bug was originally fixed in 110.9 but I broke it again some
2020 :     time after that.)
2021 :    
2022 :     ----------------------------------------------------------------------
2023 :     Name: Matthias Blume
2024 : blume 842 Date: 2001/06/19 17:25:00 EDT
2025 :     Tag: blume-20010619-signals
2026 :     Description:
2027 :    
2028 :     This should (hopefully) fix the long-standing signal handling bug.
2029 :     (The runtime system was constructing a continuation record with an
2030 :     incorrect descriptor which would cause the GC to drop data on the floor...)
2031 :    
2032 :     ----------------------------------------------------------------------
2033 :     Name: Matthias Blume
2034 : blume 840 Date: 2001/06/15 15:05:00 EDT
2035 : blume 841 Tag: blume-20010615-moresparc
2036 :     Description:
2037 :    
2038 :     Here is a short late-hour update related to Sparc c-calls:
2039 :    
2040 :     -- made handling of double-word arguments a bit smarter
2041 :    
2042 :     -- instruction selection phase tries to collapse certain clumsily
2043 :     constructed ML-Trees; typical example:
2044 :    
2045 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
2046 :    
2047 :     This currently has no further impact on SML/NJ since mlriscGen does
2048 :     not seem to generate such patterns in the first place, and c-calls
2049 :     (which did generate them in the beginning) has meanwhile been fixed
2050 :     so as to avoid them as well.
2051 :    
2052 :     ----------------------------------------------------------------------
2053 :     Name: Matthias Blume
2054 :     Date: 2001/06/15 15:05:00 EDT
2055 : blume 840 Tag: blume-20010615-sparc
2056 :     Description:
2057 :    
2058 :     The purpose of this update is to provide an implementation of NLFFI
2059 :     on Sparc machines.
2060 :    
2061 :     Here are the changes in detail:
2062 :    
2063 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2064 :     the Sparc implementation of the c-calls API.
2065 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2066 :     for accessing the ML frame. Thus, we have a real frame pointer and
2067 :     can freely modify %sp without need for an omit-frame-ptr phase.
2068 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2069 :     this case.
2070 :     * ml-nlffigen has been taught to produce code for different architectures
2071 :     and calling conventions.
2072 :     * In a way similar to what was done in the x86 case, the Sparc
2073 :     backend uses its own specific extension to mltree. (For example,
2074 :     it needs to be able to generate UNIMP instructions which are part
2075 :     of the calling convention.)
2076 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
2077 :     to make it easier to plug in new machine- and os-dependent parts).
2078 :    
2079 :     There are some other fairly unrelated bug fixes and cleanups as well:
2080 :    
2081 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2082 :     that they properly share their libraries with existing SML/NJ libraries.
2083 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
2084 :     follow-up error messages.
2085 :     * Updates to CM/CMB documentation.
2086 :    
2087 :     TODO items:
2088 :    
2089 :     * MLRISC should use a different register as its asmTemp on the Sparc.
2090 :     (The current %o2 is a really bad choice because it is part of the
2091 :     calling conventions, so things might interfere in unexpected ways.)
2092 :    
2093 :     ----------------------------------------------------------------------
2094 :     Name: Matthias Blume
2095 : blume 839 Date: 2001/06/07
2096 :     Tag: blume-20010607-calls
2097 :     Description:
2098 :    
2099 :     A number of internal changes related to C calls and calling conventions:
2100 :    
2101 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
2102 :     number of bytes popped implicitly (by the callee). In most cases
2103 :     this field is 0 but on x86/win32 it is some non-zero value. This
2104 :     is information provided for the benefit of the "omit-frameptr" pass.
2105 :     2. The CALL instruction on the x86 carries a similar "pops" field.
2106 :     The instruction selection phase copies its value from the ML-Tree
2107 :     CALL statement.
2108 :     3. On all other architectures, the instruction selection phase checks
2109 :     whether "pops=0" and complains if not.
2110 :     4. The c-calls implementation for x86 now accepts two calling conventions:
2111 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
2112 :     up after the call and pops is set to 0. For "stdcall", the caller
2113 :     does nothing, leaving the cleanup to the callee; pops is set to
2114 :     the number of bytes that were pushed onto the stack.
2115 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2116 :     distinguish between "ccall" and "stdcall".
2117 :     6. The UNIMP instruction has been added to the supported Sparc instruction
2118 :     set. (This is needed for implementing the official C calling convention
2119 :     on this architecture.)
2120 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2121 :     work with the latest CM.
2122 :    
2123 :     ----------------------------------------------------------------------
2124 :     Name: Matthias Blume
2125 : blume 838 Date: 2001/06/05 15:10:00 EDT
2126 :     Tag: blume-20010605-cm-index
2127 :     Description:
2128 :    
2129 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2130 :    
2131 :     1. CM can now generate "index files". These are human-readable files
2132 :     that list on a per-.cm-file basis each toplevel symbol defined or
2133 :     imported. The location of the index file for
2134 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2135 :     To enable index-file generation, set CM.Control.generate_index to true
2136 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
2137 :    
2138 :     The CM manual has been updated accordingly.
2139 :    
2140 :     2. I made some slight modifications to the c-calls API in MLRISC.
2141 :    
2142 :     a) There is now a callback to support saving/restoring of
2143 :     dedicated but caller-save registers around the actual call
2144 :     instruction.
2145 :     b) One can optionally specify a comment-annotation for the
2146 :     call instruction.
2147 :    
2148 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2149 :     (For example, the comment annotation shows the C prototype of
2150 :     the function being called.)
2151 :    
2152 :     ----------------------------------------------------------------------
2153 :     Name: Matthias Blume
2154 : blume 837 Date: 2001/06/01 13:30:00 EDT
2155 :     Tag: blume-20010601-nlffi-cleanup
2156 :     Description:
2157 :    
2158 :     This is mostly a cleanup of MLFFI stuff:
2159 :    
2160 :     - some signature files have been put into a more exposed place
2161 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
2162 :     - ml-nlffigen changed accordingly
2163 :     - tutorial updated
2164 :    
2165 :     Other changes:
2166 :    
2167 :     - author's affiliation in CM manual(s) updated
2168 :     - some more recognized keywords added to Allen's sml.sty
2169 :    
2170 :     ----------------------------------------------------------------------
2171 :     Name: Matthias Blume
2172 : blume 836 Date: 2001/05/25 15:30:00 EDT
2173 :     Tag: blume-20010525-iptr
2174 :     Description:
2175 :    
2176 :     - put the official 110.33-README (as it appears on the ftp server) under
2177 :     CVS
2178 :     - fixed a small bug related to incomplete pointer types in
2179 :     ml-nlffigen
2180 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2181 :     (it does not need the 'f type parameter)
2182 :    
2183 :     ----------------------------------------------------------------------
2184 :     Name: Matthias Blume
2185 : blume 834 Date: 2001/05/23 14:30:00 EDT
2186 :     Tag: Release_110_33
2187 :     Description:
2188 :    
2189 :     New version number (110.33). New bootfiles.
2190 :    
2191 :     ----------------------------------------------------------------------
2192 :     Name: Matthias Blume
2193 : blume 833 Date: 2001/05/22 18:06:00 EDT
2194 :     Tag: blume-20010522-targets
2195 :     Description:
2196 :    
2197 :     Made install.sh use file config/targets.customized if it exists, falling
2198 :     back to config/targets if it doesn't. This way one can have a customized
2199 :     version of the targets file without touching the "real thing", thus
2200 :     eliminating the constant fear of accidentally checking something bogus
2201 :     back into the CVS repository... (File config/targets.customized must
2202 :     not be added to the repository!)
2203 :    
2204 :     ----------------------------------------------------------------------
2205 :     Name: Matthias Blume
2206 : blume 832 Date: 2001/05/22 16:30:00 EDT
2207 :     Tag: blume-20010522-minitut
2208 :     Description:
2209 :    
2210 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2211 :     struct returns.
2212 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
2213 :     incomplete, preliminary documentation for NLFFI.
2214 :    
2215 :     ----------------------------------------------------------------------
2216 :     Name: Matthias Blume
2217 : blume 829 Date: 2001/05/14 11:30:00 EDT
2218 :     Tag: blume-20010514-script
2219 :     Description:
2220 :    
2221 :     Some bugs in install script fixed.
2222 :    
2223 :     In addition to that I also made a slight change to the NLFFI API:
2224 :     Functors generated by ml-nlffigen now take the dynamic library as a
2225 :     straight functor argument, not as a suspended one. (The original
2226 :     functor code used to force the suspension right away anyway, so there
2227 :     was nothing gained by this complication of the interface.)
2228 :    
2229 :     ----------------------------------------------------------------------
2230 :     Name: Matthias Blume
2231 : blume 828 Date: 2001/05/11 14:35:00 EDT
2232 :     Tag: blume-20010511-ml-nlffi
2233 :     Description:
2234 :    
2235 :     I finally took the plunge and added my new FFI code to the main
2236 :     repository. For x86-linux it is now ready for prime-time.
2237 :    
2238 :     There are two new subdirectories of "src":
2239 :    
2240 :     - ml-nlffi-lib:
2241 :     The utility library for programs using the FFI interface.
2242 :     Here is the implementation of $/c.cm and its associated low-level
2243 :     partners $/c-int.cm and $/memory.cm.
2244 :     - ml-nlffigen:
2245 :     A stand-alone program for generating ML glue code from C source
2246 :     code.
2247 :    
2248 :     Building ml-nlffigen requires $/ckit-lib.cm.
2249 :    
2250 :     The config/install.sh script has been updates to do the Right Thing
2251 :     (hopefully).
2252 :    
2253 :     Notice that the source tree for the C-Kit will not be put under "src"
2254 :     but directly under the installation root directory. (This is the
2255 :     structure that currently exists on the CVS server when you check out
2256 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
2257 :    
2258 :     Bugs: No documentation yet.
2259 :    
2260 :     ----------------------------------------------------------------------
2261 :     Name: Matthias Blume
2262 : blume 826 Date: 2001/05/09 16:35:00 EDT
2263 :     Tag: blume-20010509-cpscontract
2264 :     Description:
2265 :    
2266 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
2267 :     wrapper/unwrapper elimination did not decrement usage counts and some
2268 :     dead variables got overlooked by the dead-up logic.)
2269 :    
2270 :     ----------------------------------------------------------------------
2271 : george 824 Name: Lal George
2272 :     Date: 2001/05/08 17:26:09 EDT
2273 :     Tag: george-20010508-omit-frameptr
2274 :     Description:
2275 :    
2276 :     Changes to implement the omit-frame-pointer optimization to support
2277 :     raw C calls. For now, there is only support on the Intel x86, but
2278 :     other architectures will follow as more experience is gained with this.
2279 :    
2280 :    
2281 :     ----------------------------------------------------------------------
2282 : blume 818 Name: Matthias Blume
2283 : blume 819 Date: 2001/05/07 14:40:00 EDT
2284 :     Tag: blume-20010507-proxies
2285 :     Description:
2286 :    
2287 :     I made into "proxy libraries" all libraries that qualify for such a
2288 :     change. (A qualifying library is a library that has another library or
2289 :     groups as its sole member and repeats that member's export list
2290 :     verbatim. A proxy library avoids this repetition by omitting its export
2291 :     list, effectively inheriting the list that its (only) member exports.
2292 :     See the CM manual for more explanation.)
2293 :     The main effect is that explicit export lists for these libraries
2294 :     do not have to be kepts in sync, making maintenance a bit easier.
2295 :    
2296 :     I also added copyright notices to many .cm-files.
2297 :    
2298 :     Last but not least, I made a new set of bootfiles.
2299 :    
2300 :     ----------------------------------------------------------------------
2301 :     Name: Matthias Blume
2302 : blume 818 Date: 2001/05/04 17:00:00 EDT
2303 :     Tag: blume-20010504-cm-lsplit
2304 :     Description:
2305 :    
2306 :     0. John merged pending changes to $/smlnj-lib.cm
2307 :    
2308 :     1. Allen's previous change accidentally backed out of one of Lal's
2309 :     earlier changes. I undid this mistake (re-introducing Lal's change).
2310 :    
2311 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
2312 :     within the compiler where applicable. There is some code simplification
2313 :     because of that.
2314 :    
2315 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
2316 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
2317 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
2318 :     it can now also be controlled on a per-source basis: CM has been taught
2319 :     a new tool parameter applicable to ML source files.
2320 :    
2321 :     - To turn lambda-splitting off completely:
2322 :     local open Compiler.Control.LambdaSplitting in
2323 :     val _ = set Off
2324 :     end
2325 :     - To make "no lambda-splitting" the global default (but allow per-source
2326 :     overriding); this is the initial setting:
2327 :     local open Compiler.Control.LambdaSplitting in
2328 :     val _ = set (Default NONE)
2329 :     end
2330 :     - To make "lambda-splitting with aggressiveness a" the global default
2331 :     (and allow per-source overriding):
2332 :     local open Compiler.Control.LambdaSplitting in
2333 :     val _ = set (Default (SOME a))
2334 :     end
2335 :    
2336 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2337 :     write (in the respective .cm-file):
2338 :     a.sml (lambdasplitting:off)
2339 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2340 :     a.sml (lambdasplitting:on)
2341 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2342 :     <a> is a decimal non-negative integer):
2343 :     a.sml (lambdasplitting:<a>)
2344 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2345 :     a.sml (lambdasplitting:infinity)
2346 :     - To use the global default for a.sml:
2347 :     a.sml (lambdasplitting:default)
2348 :     or simply
2349 :     a.sml
2350 :    
2351 :     ----------------------------------------------------------------------
2352 : leunga 815 Name: Allen Leung
2353 :     Date: 2001/05/04 01:57:00 EDT
2354 :     Tag: leunga-20010504-sync
2355 :     Description:
2356 : leunga 788
2357 : leunga 815 MLRISC features.
2358 :    
2359 :     1. Fix to CMPXCHG instructions.
2360 :     2. Changed RA interface to allow annotations in callbacks.
2361 :     3. Added a new method to the stream interface to allow annotations updates.
2362 :    
2363 : george 749 ----------------------------------------------------------------------
2364 : blume 812 Name: Matthias Blume
2365 : blume 814 Date: 2001/05/01 11:45:00 EDT
2366 :     Tag: blume-20010501-pcedittmp
2367 :     Description:
2368 :    
2369 :     Changed install.sh to use the current working directory instead of
2370 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
2371 :     of /usr/tmp caused trouble with MacOS X because of file premission
2372 :     problems.
2373 :    
2374 :     ----------------------------------------------------------------------
2375 :     Name: Matthias Blume
2376 : blume 812 Date: 2001/04/20 11:10:00 EDT
2377 :     Tag: blume-20010420-inMLflag
2378 :     Description:
2379 :    
2380 :     - added vp_limitPtrMask to vproc-state.h
2381 :     (for use by the raw-C-calls mechanism to implement proper interrupt
2382 :     handling)
2383 :     - made the ML compiler aware of various data-structure offsets so it
2384 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
2385 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2386 :     raw C-calls
2387 :    
2388 :     ----------------------------------------------------------------------
2389 : george 810 Name: Lal George
2390 :     Date: 2001/04/20 09:15:28 EDT
2391 :     Tag: george-20010420-macosX
2392 :     Description:
2393 :    
2394 :     - Changes to port to Mac OS X; Darwin.
2395 :    
2396 :     - In the process I found that sqrt was broken on the PPC, because the
2397 :     fsqrt instruction is not implemented.
2398 :    
2399 :     ----------------------------------------------------------------------
2400 : blume 799 Name: Matthias Blume
2401 : blume 808 Date: 2001/04/18 12:45:00 EDT
2402 :     Tag: blume-20010418-ccalls
2403 :     Description:
2404 :    
2405 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2406 :     (this bug prevented structure arguments containing pointers from being
2407 :     passed correctly)
2408 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
2409 :     structure arguments are represented as a pointer to the beginning
2410 :     of the structure (instead of having a series of synthesized arguments,
2411 :     one for each structure member)
2412 :    
2413 :     - made makeml script's verbosity level configurable via environment
2414 :     variable (MAKEML_VERBOSITY)
2415 :    
2416 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2417 : blume 808 in rawmem-x86.sml; we are now using the real thing
2418 :    
2419 :     ----------------------------------------------------------------------
2420 :     Name: Matthias Blume
2421 : blume 806 Date: 2001/03/22 16:25:00 EST
2422 :     Tag: blume-20010322-bootfiles
2423 :     Description:
2424 :    
2425 :     Created a new set of bootfiles (for your automatic installation convenience).
2426 :    
2427 :     ----------------------------------------------------------------------
2428 :     Name: Matthias Blume
2429 : blume 805 Date: 2001/03/22 15:10:00 EST
2430 :     Tag: blume-20010322-rawmem-parcm
2431 :     Description:
2432 :    
2433 :     1. All "raw memory access" primitives for the new FFI are implemented now
2434 :     (at least on the x86).
2435 :     2. Some further cleanup of CM's parallel make mechanism.
2436 :    
2437 :     ----------------------------------------------------------------------
2438 :     Name: Matthias Blume
2439 : blume 801 Date: 2001/03/19 17:53:00 EST
2440 :     Tag: blume-20010319-parallel
2441 :     Description:
2442 :    
2443 :     Parallel make (using compile servers) now works again.
2444 :    
2445 :     To this end, CM.stabilize and CMB.make have been modified to work in
2446 :     two passes when compile servers are attached:
2447 :     1. Compile everything, do not perform stabilization; this pass
2448 :     uses compile servers
2449 :     2. Stabilize everything; this pass does not use compile servers
2450 :     If there are no compile servers, the two passes are combined into one
2451 :     (as before). Splitting the passes increases the inherent parallelism
2452 :     in the dependency graph because the entire graph including all
2453 :     libraries is available at the same time. This, in turn, improves
2454 :     server utilization. The downside is that the master process will
2455 :     have to do some extra work after compilation is done (because for
2456 :     technical reasons it must re-read all the binfiles during stabilization).
2457 :    
2458 :     ----------------------------------------------------------------------
2459 :     Name: Matthias Blume
2460 : blume 800 Date: 2001/03/16 12:22:00 EST
2461 :     Tag: blume-20010316-bootfiles
2462 :     Description:
2463 :    
2464 :     Created a new set of bootfiles (for your automatic installation convenience).
2465 :    
2466 :     ----------------------------------------------------------------------
2467 :     Name: Matthias Blume
2468 : blume 799 Date: 2001/03/16 11:00:00 EST
2469 :     Tag: blume-20010316-MLTREE-fixup
2470 :     Description:
2471 :    
2472 :     This is a minor fixup for an (untagged) earlier commit by Allen.
2473 :     (A file was missing).
2474 :    
2475 :     ----------------------------------------------------------------------
2476 : leunga 788 Name: Allen Leung
2477 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
2478 :     Tag: leunga-20010305-cut-support
2479 :    
2480 :     1. New support for alternative control-flow in MLTREE.
2481 :     Currently we support
2482 :    
2483 :     FLOW_TO(CALL ...., [k1,...,kn])
2484 :    
2485 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
2486 :     in Moby
2487 :    
2488 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2489 :    
2490 :     2. Register Allocator
2491 :    
2492 :     Changes in interface [from Fermin, John]
2493 :    
2494 :     3. Alpha 8-bit SLL support [Fermin]
2495 :    
2496 :     4. All architectures
2497 :    
2498 :     A new module (ClusterExpandCopies) for expanding parallel copies.
2499 :    
2500 :     ----------------------------------------------------------------------
2501 :     Name: Allen Leung
2502 : leunga 788 Date: 2001/02/27 23:07:00 EST
2503 :     Tag: leunga-20010227-minor-stuff
2504 :    
2505 :     1. Alpha bug fix for CMOVNE
2506 :     2. Handle mltree COND(..,FCMP ...,...)
2507 :     3. Bug fix in simplifier
2508 :    
2509 :     ----------------------------------------------------------------------
2510 : blume 777 Name: Matthias Blume
2511 : blume 781 Date: 2001/01/30 17:50:00 EST
2512 :     Tag: blume-20010130-sync
2513 :     Description:
2514 :    
2515 :     This is just a minor update to sync my devel branch with the main brach.
2516 :     The only visible change is the addition of some README files.
2517 :    
2518 :     ----------------------------------------------------------------------
2519 :     Name: Matthias Blume
2520 : blume 778 Date: 2001/01/12 23:30:00 JST
2521 :     Tag: blume-20010112-bootfiles
2522 :     Description:
2523 :    
2524 :     Made a new set of bootfiles that goes with the current state of the
2525 :     repository.
2526 :    
2527 :     ----------------------------------------------------------------------
2528 :     Name: Matthias Blume
2529 : blume 777 Date: 2001/01/12 21:20:00 JST
2530 :     Tag: blume-20010112-sync
2531 :     Description:
2532 :    
2533 :     I am just flushing out some minor changes that had accumulated in
2534 :     my private branch in order to sync with the main tree. (This is
2535 :     mainly because I had CVS trouble when trying to merge _into_ my
2536 :     private branch.)
2537 :    
2538 :     Most people should be completely unaffected by this.
2539 :    
2540 :     ----------------------------------------------------------------------
2541 : leunga 775 Name: Allen Leung
2542 :     Date: Thu Jan 11 21:03:00 EST 2001
2543 :     Tag: leunga-20010111-labexp=mltree
2544 :     Description:
2545 :    
2546 :     1. Removed the type LabelExp and replace it by MLTree.
2547 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2548 :     3. There were some bugs in alpha code generator which would break
2549 :     64-bit code generation.
2550 :     4. Redo the tools to generate code with the
2551 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2552 :     are now generated by perl scripts.
2553 :    
2554 :     ----------------------------------------------------------------------
2555 : blume 771 Name: Matthias Blume
2556 : blume 774 Date: 2001/01/10 21:55:00 JST
2557 :     Tag: blume-20010110-rcc
2558 :     Description:
2559 :    
2560 :     The RCC stuff now seems to work (but only on the x86).
2561 :     This required hacking of the c-calls interface (and -implementation) in
2562 :     MLRISC.
2563 :    
2564 :     Normal compiler users should be unaffected.
2565 :    
2566 :     ----------------------------------------------------------------------
2567 :     Name: Matthias Blume
2568 : blume 773 Date: 2001/01/09 01:20:00 JST
2569 :     Tag: blume-20010109-rcc
2570 :     Description:
2571 :    
2572 :     This is a fairly big patch, flushing out a large number of pending
2573 :     changes that I made to my development copy over the last couple of days.
2574 :    
2575 :     Of practical relevance at this moment is a workaround for a pickling
2576 :     bug that Allen ran into the other day. The cause of the bug itself is
2577 :     still unknown and it might be hard to fix it properly, but the
2578 :     workaround has some merits of its own (namely somewhat reducing pickling
2579 :     overhead for certain libraries). Therefore, I think this solution should
2580 :     be satisfactory at this time.
2581 :    
2582 :     The rest of the changes (i.e., the vast majority) has to do with my
2583 :     ongoing efforts of providing direct support for C function calls from
2584 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2585 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2586 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2587 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2588 :     handle RCC.
2589 :    
2590 :     The last part (the changes to mlriscGen.sml) are still known to be
2591 :     wrong on the x86 and not implemented on all other architectures. But
2592 :     the infrastructure is in place. I had to change a few functor
2593 :     signatures in the backend to be able to route the CCalls interface
2594 :     from MLRISC there, and I had to specialize the mltree type (on the
2595 :     x86) to include the necessary extensions. (The extensions themselves
2596 :     were already there and redy to go in MLRISC/x86).
2597 :    
2598 :     Everything should be very happy as soon as someone helps me with
2599 :     mlriscGen.sml...
2600 :    
2601 :     In any case, nothing of this should matter to anyone as long as the
2602 :     new primop is not being used (which is going to be the case unless you
2603 :     find it where I hid it :). The rest of the compiler is completely
2604 :     unaffected.
2605 :    
2606 :     ----------------------------------------------------------------------
2607 :     Name: Matthias Blume
2608 : blume 772 Date: 2001/01/05 00:30:00 JST
2609 :     Tag: blume-20010105-primops
2610 :     Description:
2611 :    
2612 :     Added some experimental support for work that I am doing right now.
2613 :     These changes mostly concern added primops, but there is also a new
2614 :     experimental C library in the runtime system (but currently not enabled
2615 :     anywhere except on Linux/X86).
2616 :    
2617 :     In the course of adding primops (and playing with them), I discovered that
2618 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2619 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2620 :     To recover, I made type information in INL_PRIM mandatory and changed
2621 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2622 :     has complete, correct type information (i.e., no bottom types).
2623 :    
2624 :     Since all these changes mean that we need new binfiles, I also bumped the
2625 :     version number to 110.32.1.
2626 :    
2627 :     ----------------------------------------------------------------------
2628 :     Name: Matthias Blume
2629 : blume 771 Date: 2000/12/30 22:10:00 JST
2630 :     Tag: blume-20001230-various
2631 :     Description:
2632 :    
2633 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2634 :     to each other using path anchors. (See CM manual for explanation.)
2635 :    
2636 :     Updated CM documentation.
2637 :    
2638 :     Fixed some bugs in CM.
2639 :    
2640 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2641 :    
2642 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2643 :    
2644 :     Added cleanup handler for exportML to reset timers and compiler stats.
2645 :    
2646 :     ----------------------------------------------------------------------
2647 : george 761 Name: Lal George
2648 :     Date: 2000/12/22 22:22:58 EST 2000
2649 :     Tag: Release_110_32
2650 :     Description:
2651 :    
2652 :     Infinite precision used throughout MLRISC.
2653 :     see MLRISC/mltree/machine-int.sig
2654 :    
2655 :     ----------------------------------------------------------------------
2656 : blume 756 Name: Matthias Blume
2657 : blume 760 Date: 2000/12/22 23:16:00 JST
2658 :     Tag: blume-20001222-warn
2659 :     Description:
2660 : blume 759
2661 : blume 760 Corrected wording and formatting of some CM warning message which I
2662 :     broke in my previous patch.
2663 : blume 759
2664 :     ----------------------------------------------------------------------
2665 :     Name: Matthias Blume
2666 : blume 763 Date: 2000/12/22 21:20:00 JST
2667 :     Tag: blume-20001222-anchorenv
2668 :     Description:
2669 :    
2670 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2671 :    
2672 :     ----------------------------------------------------------------------
2673 :     Name: Matthias Blume
2674 : blume 758 Date: 2000/12/22 13:15:00 JST
2675 :     Tag: blume-20001222-cleanup
2676 :     Description:
2677 :    
2678 :     Removed src/cm/ffi which does not (and did not) belong here.
2679 :    
2680 :     ----------------------------------------------------------------------
2681 :     Name: Matthias Blume
2682 : blume 757 Date: 2000/12/21 23:55:00 JST
2683 :     Tag: blume-20001221-exn
2684 :     Description:
2685 :    
2686 :     Probably most important: CM no longer silently swallows all exceptions
2687 :     in the compiler.
2688 :     Plus: some other minor CM changes. For example, CM now reports some
2689 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2690 :    
2691 :     ----------------------------------------------------------------------
2692 :     Name: Matthias Blume
2693 : blume 756 Date: 2000/12/15 00:01:05 JST
2694 :     Tag: blume-20001215-dirtool
2695 :     Description:
2696 :    
2697 :     - "dir" tool added.
2698 :     - improvements and cleanup to Tools structure
2699 :     - documentation updates
2700 :    
2701 :     ----------------------------------------------------------------------
2702 : leunga 752 Name: Allen Leung
2703 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2704 :     Description:
2705 :     Tag: leunga-20001214-int-inf
2706 :     Description:
2707 :    
2708 :     In IntInf, added these standard functions, which are missing from our
2709 :     implementation:
2710 :    
2711 :     andb : int * int -> int
2712 :     xorb : int * int -> int
2713 :     orb : int * int -> int
2714 :     notb : int -> int
2715 :     << : int * word -> int
2716 :     ~>> : int * word -> int
2717 :    
2718 :     Not tested, I hope they are correct.
2719 :    
2720 :     ----------------------------------------------------------------------
2721 :     Name: Allen Leung
2722 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2723 :     Description:
2724 :     Tag: leunga-20001208-nowhere
2725 :     Description:
2726 :    
2727 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2728 :     to generate better error messages etc. Plus a brief manual.
2729 :    
2730 :     ----------------------------------------------------------------------
2731 : george 749 Name: Lal George
2732 :     Date: 2000/12/08 09:54:02 EST 2000
2733 :     Tag: Release_110_31
2734 :     Description:
2735 : george 705
2736 : george 749 - Version 110.31
2737 : leunga 591 ----------------------------------------------------------------------
2738 : leunga 744 Name: Allen Leung
2739 :     Date: Thu Dec 7 22:01:04 EST 2000
2740 :     Tag: leunga-20001207-cell-monster-hack
2741 :     Description:
2742 :    
2743 :     Major MLRISC internal changes. Affect all clients.
2744 :     Summary:
2745 :    
2746 :     1. Type CELLS.cell = int is now replaced by a datatype.
2747 :     As a result, the old regmap is now gone. Almost all interfaces
2748 :     in MLRISC change as a consequence.
2749 :    
2750 :     2. A new brand version of machine description tool (v3.0) that generates
2751 :     modules expecting the new interface. The old version is removed.
2752 :    
2753 :     3. The RA interface has been further abstracted into two new functors.
2754 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2755 :     [See also directory MLRISC/demo.]
2756 :    
2757 :     4. Some other new source->source code generation tools are available:
2758 :    
2759 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2760 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2761 :     I use this tool to generate the peephole optimizers---with the new
2762 :     cell type changes, peephole rules are becoming difficult to write
2763 :     without conditional pattern matching.
2764 :    
2765 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2766 :     cover the entire MLRISC source tree so many things broke.
2767 :    
2768 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2769 :     They are moved because there are a lot of them and they clutter up the
2770 :     root dir.
2771 :    
2772 :     7. More detailed documentation to come...
2773 :    
2774 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2775 :     a makeml -rebuild first. This is because of other other
2776 :     changes that Matthias has made (see below).
2777 :    
2778 :    
2779 :     ----------------------------------------------------------------------
2780 : blume 733 Name: Matthias Blume
2781 : blume 742 Date: 2000/11/30 23:12:00 JST
2782 :     Tag: blume-20001130-filereorg
2783 :     Description:
2784 :    
2785 :     Some manual updates and some file reorganizations in CM.
2786 :    
2787 :     ----------------------------------------------------------------------
2788 :     Name: Matthias Blume
2789 : blume 737 Date: 2000/11/24 17:45:00 JST
2790 :     Tag: blume-20001124-link
2791 :     Description:
2792 :    
2793 :     Drastically improved link traversal code for the case that the dynamic
2794 :     value was already loaded at bootstrap time. As a result, CM and CMB
2795 :     now both load blazingly fast -- even on a very slow machine. Also,
2796 :     memory consumption has been further reduced by this.
2797 :    
2798 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2799 :     bootstrap you have to do this:
2800 :    
2801 :     1. Run CMB.make
2802 :     2. Make a symbolic link for the boot directory:
2803 :     ln -s sml.boot.ARCH-OS xxx
2804 :     3. "Rebuild" the boot directory:
2805 :     ./makeml -boot xxx -rebuild sml ; rm xxx
2806 :     4. Boot normally:
2807 :     ./makeml
2808 :    
2809 :     ----------------------------------------------------------------------
2810 :     Name: Matthias Blume
2811 : blume 735 Date: 2000/11/21 21:20:00 JST
2812 :     Tag: blume-20001121-tools
2813 :     Description:
2814 :    
2815 :     Continued hacking on autoloading problem -- with success this time.
2816 :     Also changed tool-plugin mechanism. See new CM manual.
2817 :    
2818 :     ----------------------------------------------------------------------
2819 :     Name: Matthias Blume
2820 : blume 734 Date: 2000/11/19 14:30:00 JST
2821 :     Tag: blume-20001119-autoload
2822 :     Description:
2823 :    
2824 :     Some hacking to make autoloading faster. Success for CMB, no success
2825 :     so far for CM. There is a reduced structure CM' that autoloads faster.
2826 :     (This is a temporary, non-documented hack to be eliminated again when
2827 :     the general problem is solved.)
2828 :    
2829 :     ----------------------------------------------------------------------
2830 :     Name: Matthias Blume
2831 : blume 733 Date: 2000/11/17 14:10:00 JST
2832 :     Tag: blume-20001117-pickle-lib
2833 :     Description:
2834 :    
2835 :     1. Eliminated comp-lib.cm
2836 :     2. Made pickle-lib.cm
2837 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
2838 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
2839 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
2840 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
2841 :     Duplicated intmap modules thrown out.
2842 :     6. Hunted down all duplicated SCC code and replaced it with
2843 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
2844 :     7. Rewrote Feedback module.
2845 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
2846 :     should be thrown out and equivalent modules from smlnj-lib.cm
2847 :     should be used (IntRedBlackSet, IntListSet, ...).
2848 :    
2849 :     Confirmed that compiler compiles to fixpoint.
2850 :    
2851 :     ----------------------------------------------------------------------
2852 : leunga 731 Name: Allen Leung
2853 :     Date: 2000/11/10 18:00:00
2854 :     Tag: leunga-20001110-new-x86-fp
2855 :    
2856 :     A new x86 floating point code generator has been added.
2857 :     By default this is turned off. To turn this on, do:
2858 :    
2859 :     CM.autoload "$smlnj/compiler.cm";
2860 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
2861 :    
2862 :     Changes:
2863 :    
2864 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
2865 :     2. Changed the extension callback for FTANGENT to generate:
2866 :    
2867 :     fptan
2868 :     fstp %st(0)
2869 :     instead of
2870 :     fptan
2871 :     fstpl ftempmem
2872 :    
2873 :     3. Numerous assembly fixes for x86.
2874 :    
2875 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
2876 :     support for a whole bunch of instructions and addressing modes:
2877 :    
2878 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
2879 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
2880 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
2881 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
2882 :     fxch %st(n)
2883 :     fld %st(n)
2884 :     fst %st(n)
2885 :     fst mem
2886 :     fstp %st(n)
2887 :     fucom %st(n)
2888 :     fucomp %st(n)
2889 :    
2890 :     All these are now generated when the fast fp mode is turned on.
2891 :    
2892 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
2893 :    
2894 :     ----------------------------------------------------------------------
2895 : blume 729 Name: Matthias Blume
2896 :     Date: 2000/11/09 11:20:00 JST
2897 :     Tag: blume-20001109-scc
2898 :     Description:
2899 :    
2900 :     Eliminated some code duplication:
2901 :    
2902 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
2903 :     (Otherwise the functor is useless.)
2904 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
2905 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
2906 :    
2907 :     ----------------------------------------------------------------------
2908 : george 721 Name: Lal George
2909 :     Date: 2000/11/06 09:02:21 EST 2000
2910 :     Tag: Release_110_30
2911 :     Description:
2912 :    
2913 :     - Version 110.30
2914 :     ----------------------------------------------------------------------
2915 : blume 715 Name: Matthias Blume
2916 : blume 716 Date: 2000/11/04 14:45:00
2917 :     Tag: blume-20001104-mlbuild
2918 :     Description:
2919 :    
2920 :     - Made ml-build faster on startup.
2921 :     - Documentation fixes.
2922 :    
2923 :     ----------------------------------------------------------------------
2924 :     Name: Matthias Blume
2925 : blume 715 Date: 2000/11/02 17:00:00 JST
2926 :     Tag: blume-20001102-condcomp
2927 :     Description:
2928 :    
2929 :     - Small tweaks to pickler -- new BOOTFILES!
2930 :     - Version bumped to 110.29.2.
2931 :     - Added conditional compilation facility to init.cmi (see comment there).
2932 :     ----------------------------------------------------------------------
2933 : leunga 713 Name: Allen Leung
2934 :     Date: 2000/10/23 19:31:00
2935 :     Tag: leunga-20001023-demo-ra
2936 :    
2937 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
2938 :     2. Test programs for the graph library updated
2939 :     3. Some new MLRISC demo programs added
2940 :    
2941 :     ----------------------------------------------------------------------
2942 : blume 710 Name: Matthias Blume
2943 : blume 711 Date: 2000/08/31 22:15:00 JST
2944 :     Tag: blume-20001017-errmsg
2945 :     Description:
2946 :    
2947 :     More error message grief: Where there used to be no messages, there
2948 :     now were some that had bogus error regions. Fixed.
2949 :    
2950 :     ----------------------------------------------------------------------
2951 :     Name: Matthias Blume
2952 : blume 710 Date: 2000/08/31 17:30:00 JST
2953 :     Tag: blume-20001017-v110p29p1
2954 :     Description:
2955 :    
2956 :     I made a version 110.29.1 with new bootfiles.
2957 :    
2958 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
2959 :     CM documentation changes and a small bugfix in CM's error reporting.
2960 :    
2961 :     ----------------------------------------------------------------------
2962 : george 705 Name: Lal George
2963 :     Date: 2000/09/27 14:42:35 EDT
2964 :     Tag: george-20000927-nodestatus
2965 :     Description:
2966 :    
2967 :     Changed the type of the nodestatus, so that:
2968 :    
2969 :     SPILLED(~1) is now SPILLED
2970 :     SPILLED(m) where m>=0 is now MEMREG(m)
2971 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
2972 :    
2973 :     ----------------------------------------------------------------------
2974 : blume 697 Name: Matthias Blume
2975 : blume 703 Date: 2000/09/07 14:45:00 JST
2976 :     Tag: blume-20000907-cmerrmsg
2977 :     Description:
2978 :    
2979 :     Small tweak to CM to avoid getting ML syntax error messages twice.
2980 :    
2981 :     ----------------------------------------------------------------------
2982 :     Name: Matthias Blume
2983 : blume 701 Date: 2000/08/31 18:00:00 JST
2984 :     Tag: blume-20000831-cvsbootfiles
2985 :     Description:
2986 :    
2987 :     New URL for boot files (because the 110.29 files on the BL server do
2988 :     now work correctly with my updated install scripts for yacc and lex).
2989 :    
2990 :     ----------------------------------------------------------------------
2991 :     Name: Matthias Blume
2992 : blume 697 Date: 2000/08/08 12:33:00 JST
2993 :     Tag: blume-20000808-manual
2994 :     Description:
2995 :    
2996 :     Tiny update to CM manual.
2997 :    
2998 :     ----------------------------------------------------------------------
2999 : leunga 695 Name: Allen Leung
3000 :     Date: 2000/08/7 19:31:00
3001 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
3002 :    
3003 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
3004 :     and Mobdy.
3005 :    
3006 :     1. x86
3007 :    
3008 :     a. Fixes to peephole module by John and Dan.
3009 :     b. Assembly fix to SETcc by Allen.
3010 :     c. Fix to c-call by John.
3011 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
3012 :     e. Instruction selection optimization to SETcc as suggested by John.
3013 :    
3014 :     For example,
3015 :    
3016 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
3017 :    
3018 :     should generate:
3019 :    
3020 :     MOVL a, x
3021 :     SUBL b, x
3022 :     SHRL 31, x
3023 :    
3024 :     2. IR stuff
3025 :    
3026 :     A bunch of new DJ-graph related algorithms added. These
3027 :     speed up SSA construction.
3028 :    
3029 :     3. SSA + Scheduling
3030 :    
3031 :     Added code for SSA and scheduling to the repository
3032 :    
3033 :     ----------------------------------------------------------------------
3034 : george 694 Name: Lal George
3035 :     Date: 2000/07/27 11:53:14 EDT
3036 :    
3037 :     Tag: lal-20000727-linux-ppc
3038 :     Description:
3039 :    
3040 :     Made changes to support Linux PPC.
3041 :     p.s. I have confirmation that the 110.29 boot files work fine.
3042 :    
3043 :     ----------------------------------------------------------------------
3044 : blume 684 Name: Matthias Blume
3045 : blume 692 Date: 2000/07/27 17:40:00 JST
3046 :     Tag: blume-20000727-scripts
3047 :     Description:
3048 :    
3049 :     !!!! WARNING !!!!
3050 :     You must recompile the runtime system!
3051 :     !!!! WARNING !!!!
3052 :    
3053 :     This is basically another round of script-enhancements:
3054 :    
3055 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
3056 :     and undefine CM preprocessor symbols.
3057 :    
3058 :     2. ml-build avoids generating a new heap image if it finds that the
3059 :     existing one is still ok. (The condition is that no ML file had to
3060 :     be recompiled and all ML files are found to be older that the heap
3061 :     file.)
3062 :    
3063 :     To make this work smoothly, I also hacked the runtime system as
3064 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
3065 :     (.sparc-solaris, ...) that is currently being used.
3066 :    
3067 :     Moreover, the signature of CM.mk_standalone has changed. See the
3068 :     CM manual.
3069 :    
3070 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
3071 :     CM manual for details.)
3072 :    
3073 :     4. More CM manual updates:
3074 :     - all of the above has been documented.
3075 :     - there is now a section describing the (CM-related) command line
3076 :     arguments that are accepted by the "sml" command
3077 :    
3078 :     ----------------------------------------------------------------------
3079 :     Name: Matthias Blume
3080 : blume 691 Date: 2000/07/25 16:20:00 JST
3081 :     Tag: blume-20000725-makedepend
3082 :     Description:
3083 :    
3084 :     Added a script called ml-makedepend. This can be used in makefiles
3085 :     for Unix' make in a way very similar to the "makedepend" command for
3086 :     C.
3087 :    
3088 :     The script internally uses function CM.sources.
3089 :    
3090 :     Synopsis:
3091 :    
3092 :     ml-makedepend [-f makefile] cmfile targetname
3093 :    
3094 :     The default for the makefile is "makefile" (or "Makefile" should
3095 :     "makefile" not exist).
3096 :    
3097 :     ml-makedepend adds a cmfile/targetname-specific section to this
3098 :     makefile (after removing the previous version of this section). The
3099 :     section contains a single dependency specification with targetname on
3100 :     the LHS (targetname is an arbitrary name), and a list of files derived
3101 :     from the cmfile on the RHS. Some of the files on the RHS are
3102 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
3103 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
3104 :     path names. The makefile writer is responsible for making sure that
3105 :     these variables have correct at the time "make" is invoked.
3106 :    
3107 :     ----------------------------------------------------------------------
3108 :     Name: Matthias Blume
3109 : blume 690 Date: 2000/07/22 23:30:00 JST
3110 :     Tag: blume-20000722-urlupdate
3111 :     Description:
3112 :    
3113 :     Changed BOOT and config/srcarchiveurl to point to BL server:
3114 :    
3115 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
3116 :    
3117 :     ----------------------------------------------------------------------
3118 :     Name: Matthias Blume
3119 : blume 686 Date: 2000/07/18 18:00:00 JST
3120 :     Tag: blume-20000718-Version_110_29
3121 :     Description:
3122 :    
3123 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
3124 :    
3125 :     2. Updated config/version to 110.29
3126 :    
3127 :     3. Updated config/srcarchiveurl
3128 :    
3129 :     3. New boot files!
3130 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
3131 :    
3132 :     ----------------------------------------------------------------------
3133 :     Name: Matthias Blume
3134 : blume 684 Date: 2000/07/11 13:58:00 JST
3135 :     Tag: blume-20000711-doctypo
3136 :     Description:
3137 :    
3138 :     Fixed a few typos in CM manual.
3139 :    
3140 :     ----------------------------------------------------------------------
3141 : leunga 683 Name: Allen Leung
3142 :     Date: 2000/06/15 00:38:00
3143 :     Tag: leunga-20000704-sparc-x86
3144 :    
3145 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
3146 :     2. fix to x86 RET bug [found by Dan Grossman]
3147 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
3148 :    
3149 :     Affects c-- and moby only
3150 :    
3151 :     ----------------------------------------------------------------------
3152 : blume 680 Name: Matthias Blume
3153 : blume 682 Date: 2000/07/04 15:26:00
3154 :     Tag: blume-20000704-trigger
3155 :     Description:
3156 :    
3157 :     1. Improvements to CM manual.
3158 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
3159 :     of getting a back-trace. The function, when called, raises an
3160 :     internal exception which explicitly carries the full back-trace history,
3161 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
3162 :     or not). The interactive loop will print that history once it arrives
3163 :     at top level.
3164 :     Short of having all exceptions implicitly carry the full history, the
3165 :     recommended way of using this facility is:
3166 :     - compile your program with instrumentation "on"
3167 :     - run it, when it raises an exception, look at the history
3168 :     - if the history is "cut off" because of some handler, go and modify
3169 :     your program so that it explicitly calls BTrace.trigger
3170 :     - recompile (still instrumented), and rerun; look at the full history
3171 :    
3172 :     ----------------------------------------------------------------------
3173 :     Name: Matthias Blume
3174 :     Date: 2000/07/03 15:36:00 JST
3175 : blume 680 Tag: blume-20000702-manual
3176 :     Description:
3177 :    
3178 :     Small corrections and updates to CM manual.
3179 :    
3180 :     ----------------------------------------------------------------------
3181 : blume 682 Name: Matthias Blume
3182 : blume 679 Date: 2000/06/29 16:04:00 JST
3183 :     Tag: blume-20000629-yacctool
3184 :     Description:
3185 :    
3186 :     Changes:
3187 :    
3188 :     1. Class "mlyacc" now takes separate arguments to pass options to
3189 :     generated .sml- and .sig-files independently.
3190 :     2. Corresponding CM manual updates.
3191 :     3. BTrace module now also reports call sites. (However, for loop clusters
3192 :     it only shows from where the cluster was entered.) There are associated
3193 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
3194 :    
3195 :     ----------------------------------------------------------------------
3196 : blume 675 Name: Matthias Blume
3197 : blume 678 Date: 2000/06/27 16:51:00 JST
3198 :     Tag: blume-20000627-noweb
3199 :     Description:
3200 :    
3201 :     Changes:
3202 :    
3203 :     1. Implemented "subdir" and "witness" options for noweb tool.
3204 :     This caused some slight internal changes in CM's tool implementation.
3205 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
3206 :     some remaining issues from earlier path anchor changes.
3207 :     3. Updated CM manual accordingly.
3208 :    
3209 :     4. Changed implementation of back-tracing so that I now consider it
3210 :     ready for prime-time.
3211 :    
3212 :     In particular, you don't have to explicitly trigger the back-trace
3213 :     anymore. Instead, if you are running BTrace-instrumented code and
3214 :     there is an uncaught exception (regardless of whether or not it was
3215 :     raised in instrumented code), the top-level evalloop will print
3216 :     the back-trace.
3217 :    
3218 :     Features:
3219 :    
3220 :     - Instrumented and uninstrumented code work together seemlessly.
3221 :     (Of course, uninstrumented code is never mentioned in actual
3222 :     back-traces.)
3223 :    
3224 :     - Asymptotic time- and space-complexity of instrumented code is
3225 :     equal to that of uninstrumented code. (This means that
3226 :     tail-recursion is preserved by the instrumentation phase.)
3227 :    
3228 :     - Modules whose code has been instrumented in different sessions
3229 :     work together without problem.
3230 :    
3231 :     - There is no penalty whatsoever on uninstrumented code.
3232 :    
3233 :     - There is no penalty on "raise" expressions, even in
3234 :     instrumented code.
3235 :    
3236 :     A potential bug (or perhaps it is a feature, too):
3237 :    
3238 :     A back-trace reaches no further than the outermost instrumented
3239 :     non-trivial "raise". Here, a "trivial" raise is one that is the
3240 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
3241 :    
3242 :     <exp> handle e => raise e
3243 :    
3244 :     and even
3245 :    
3246 :     <exp> handle Foo => raise Bar
3247 :    
3248 :     and, of course, through
3249 :    
3250 :     <exp> handle Foo => ...
3251 :    
3252 :     if the exception was not Foo.
3253 :    
3254 :     Back-traces always reach right through any un-instrumented code
3255 :     including any of its "handle" expressions, trivial or not.
3256 :    
3257 :     To try this out, do the following:
3258 :    
3259 :     - Erase all existing binfiles for your program.
3260 :     (You may keep binfiles for those modules where you think you
3261 :     definitely don't need back-tracing.)
3262 :     - Turn on back-trace instrumentation:
3263 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3264 :     - Recompile your program. (I.e., run "CM.make" or "use".)
3265 :     - You may now turn instrumentation off again (if you want):
3266 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3267 :     - Run your program as usual. If it raises an exception that
3268 :     reaches the interactive toplevel, then a back-trace will
3269 :     automatically be printed. After that, the toplevel loop
3270 :     will print the exception history as usual.
3271 :    
3272 :     ----------------------------------------------------------------------
3273 :     Name: Matthias Blume
3274 : blume 677 Date: 2000/06/26 09:56:46 JST
3275 :     Tag: blume-20000626-setup
3276 :     Description:
3277 :    
3278 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
3279 :     ML code before and after compiling a file (e.g., to set compiler
3280 :     flags)
3281 :    
3282 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
3283 :     - associated changes to btrace.sml (BTrace instrumentation pass)
3284 :     - cleaner implementation of btimp.sml (BTrace tracing and report
3285 :     module)
3286 :    
3287 :     CM manual: * new path encoding documented
3288 :     * description of setup-parameter to "sml" added
3289 :    
3290 :     The biggest user-visible change to back-tracing is that it is no
3291 :     longer necessary to compile all traced modules within the same
3292 :     session. (This was a real limitation.)
3293 :    
3294 :     ----------------------------------------------------------------------
3295 :     Name: Matthias Blume
3296 : blume 676 Date: 2000/06/24 12:40:00 JST
3297 :     Tag: blume-20000624-startup
3298 :     Description:
3299 :    
3300 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
3301 :     bit too often -- to put it mildly. :)
3302 :    
3303 :     ----------------------------------------------------------------------
3304 :     Name: Matthias Blume
3305 : blume 675 Date: 2000/06/23 18:20:00 JST
3306 :     Tag: blume-20000623-btrace
3307 :     Description:
3308 :    
3309 :     This updates adds a backtrace facility to aid programmers in debugging
3310 :     their programs. This involves the following changes:
3311 :    
3312 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
3313 :     keeping track of the current call stack. When programs are compiled
3314 :     in a special mode, the compiler will insert calls to these hooks
3315 :     into the user program.
3316 :     "Hook" means that it is possible for different implementations of
3317 :     back-tracing to register themselves (at different times).
3318 :    
3319 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
3320 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
3321 :     It can be turned on using this call:
3322 :     SMLofNJ.Internals.BTrace.mode (SOME true);
3323 :     Turning it off again:
3324 :     SMLofNJ.Internals.BTrace.mode (SOME false);
3325 :     Querying the current status:
3326 :     SMLofNJ.Internals.BTrace.mode NONE;
3327 :     Annotated programs are about twice as big as normal ones, and they
3328 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
3329 :     where all hooks do nothing). The slowdown with a plugin that is
3330 :     actually useful (such as the one supplied by default) is even greater,
3331 :     but in the case of the default plugin it is still only an constant
3332 :     factor (amortized).
3333 :    
3334 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
3335 :     with a sub-structure BTrace for controlling back-tracing. In particular,
3336 :     the above-mentioned function "mode" controls whether the annotation
3337 :     phase is invoked by the compiler. Another important function is
3338 :     "trigger": when called it aborts the current execution and causes
3339 :     the top-level loop to print a full back-trace.
3340 :    
3341 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
3342 :     for back-tracing. It keeps track of the dynamic call stack and in
3343 :     addition to that it keeps a partial history at each "level" of that
3344 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
3345 :     d and b (at separate times, dynamically), then the report will show:
3346 :    
3347 :     GOTO d
3348 :     /c
3349 :     GOTO \b
3350 :     CALL a
3351 :    
3352 :     This shows that there was an initial non-tail call of a, then a
3353 :     tail-call to b or c, looping behavior in a cluster of functions that
3354 :     consist of b and c, and then a goto from that cluster (i.e., either from
3355 :     b or from c) to d.
3356 :    
3357 :     Note that (depending on the user program) the amount of information
3358 :     that the back-trace module has to keep track of at each level is bounded
3359 :     by a constant. Thus, the whole implementation has the same asymptotical
3360 :     complexity as the original program (both in space and in time).
3361 :    
3362 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
3363 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
3364 :     which is raised by the "trigger" function mentioned above.
3365 :    
3366 :     Notes on usage:
3367 :    
3368 :     - Annotated code works well together with unannotated code:
3369 :     Unannotated calls simply do not show up at all in the backtrace.
3370 :    
3371 :     - It is not a good idea to let modules that were annotated during
3372 :     different sessions run at the same time. This is because the compiler
3373 :     chooses small integers to identify individual functions, and there
3374 :     will be clashes if different modules were compiled in separate sessions.
3375 :     (Nothing will crash, and you will even be told about the clashes, but
3376 :     back-trace information will in general not be useful.)
3377 :    
3378 :     - Back-tracing can be confused by callcc and capture.
3379 :    
3380 :     - The only way of getting a back-trace right now is to explicitly
3381 :     invoke the "trigger" function from your user program. Eventually, we
3382 :     should make every exception carry back-trace information (if
3383 :     available). But since this creates more overhead at "raise"-time
3384 :     (similar to the current exnHistory overhead), I have not yet
3385 :     implemented this. (The implementation will be rather easy.) With
3386 :     exceptions carrying back-trace information, this facility will be even
3387 :     more useful because users don't need to modify their programs...
3388 :    
3389 :     - While it is possible to compile the compiler with back-trace
3390 :     annotations turned on (I did it to get some confidence in
3391 :     correctness), you must make absolutely sure that core.sml and
3392 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
3393 :     be compiled with annotation because there is no core access yet, but
3394 :     if you compile btimp.sml with annotation, then the system will go into
3395 :     an infinite recursion and crash.)
3396 :     Since CM currently does not know about BTrace, the only way to turn
3397 :     annotations on and off for different modules of the compiler is to
3398 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
3399 :     this is awkward and clumsy.
3400 :    
3401 :     Sample sessions:
3402 :    
3403 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3404 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3405 :     [autoloading]
3406 :     [autoloading done]
3407 :     val it = false : bool
3408 :     - structure X = struct
3409 :     - fun main n = let
3410 :     - fun a (x, 0) = d x
3411 :     - | a (x, n) = b (x, n - 1)
3412 :     - and b (x, n) = c (x, n)
3413 :     - and c (x, n) = a (x, n)
3414 :     - and d x = e (x, 3)
3415 :     - and e (x, 0) = f x
3416 :     - | e (x, n) = e (x, n - 1)
3417 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
3418 :     - | f n = n * g (n - 1)
3419 :     - and g n = a (n, 3)
3420 :     - in
3421 :     - f n
3422 :     - end
3423 :     - end;
3424 :     structure X : sig val main : int -> int end
3425 :     - X.main 3;
3426 :     *** BACK-TRACE ***
3427 :     GOTO stdIn:4.2-13.20: X.main[2].f
3428 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3429 :     GOTO stdIn:4.2-13.20: X.main[2].d
3430 :     / stdIn:4.2-13.20: X.main[2].a
3431 :     | stdIn:4.2-13.20: X.main[2].b
3432 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3433 :     CALL stdIn:4.2-13.20: X.main[2].g
3434 :     GOTO stdIn:4.2-13.20: X.main[2].f
3435 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3436 :     GOTO stdIn:4.2-13.20: X.main[2].d
3437 :     / stdIn:4.2-13.20: X.main[2].a
3438 :     | stdIn:4.2-13.20: X.main[2].b
3439 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3440 :     CALL stdIn:4.2-13.20: X.main[2].g
3441 :     GOTO stdIn:4.2-13.20: X.main[2].f
3442 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
3443 :     GOTO stdIn:4.2-13.20: X.main[2].d
3444 :     / stdIn:4.2-13.20: X.main[2].a
3445 :     | stdIn:4.2-13.20: X.main[2].b
3446 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
3447 :     CALL stdIn:4.2-13.20: X.main[2].g
3448 :     GOTO stdIn:4.2-13.20: X.main[2].f
3449 :     CALL stdIn:2.15-17.4: X.main[2]
3450 :     -
3451 :    
3452 :     (Note that because of a FLINt bug the above code currently does not
3453 :     compile without BTrace turned on.)
3454 :    
3455 :     Here is another example, using my modified Tiger compiler:
3456 :    
3457 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
3458 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
3459 :     [autoloading]
3460 :     [autoloading done]
3461 :     val it = false : bool
3462 :     - CM.make "sources.cm";
3463 :     [autoloading]
3464 :     ...
3465 :     [autoloading done]
3466 :     [scanning sources.cm]
3467 :     [parsing (sources.cm):parse.sml]
3468 :     [creating directory CM/SKEL ...]
3469 :     [parsing (sources.cm):tiger.lex.sml]
3470 :     ...
3471 :     [wrote CM/sparc-unix/semant.sml]
3472 :     [compiling (sources.cm):main.sml]
3473 :     [wrote CM/sparc-unix/main.sml]
3474 :     [New bindings added.]
3475 :     val it = true : bool
3476 :     - Main.compile ("../testcases/merge.tig", "foo.out");
3477 :     *** BACK-TRACE ***
3478 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
3479 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3480 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3481 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
3482 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3483 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3484 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
3485 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
3486 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
3487 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
3488 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
3489 :     CALL main.sml:1.18-118.4: Main.compile[2]
3490 :     -
3491 :    
3492 :     ----------------------------------------------------------------------
3493 : blume 673 Name: Matthias Blumen
3494 :     Date: 2000/06/21 18:00:00 JST
3495 :     Tag: blume-20000621-manual
3496 :     Description:
3497 :    
3498 :     CM manual update: Path environments documented.
3499 :    
3500 :     ----------------------------------------------------------------------
3501 : blume 666 Name: Matthias Blume
3502 : blume 672 Date: 2000/06/19 13:40:00
3503 :     Tag: blume-20000619-manual
3504 :     Description:
3505 :    
3506 :     CM manual and system/README update. This only covers the fact that
3507 :     there are no more implicit anchors. (Path environments and the "bind"
3508 :     option to "cm" have yet to be documented.)
3509 :    
3510 :     ----------------------------------------------------------------------
3511 :     Name: Matthias Blume
3512 : blume 671 Date: 2000/06/19 11:05:00 JST
3513 :     Tag: blume-20000619-chdir-bugfix
3514 :     Description:
3515 :    
3516 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
3517 :    
3518 :     ----------------------------------------------------------------------
3519 :     Name: Matthias Blume
3520 : blume 670 Date: 2000/06/18 22:00:10 JST
3521 :     Tag: blume-20000618-implicit-anchors-really-gone
3522 :     Description:
3523 :    
3524 :     I updates the previous HISTORY entry where I forgot to mention that
3525 :     implicit anchors are no longer with us.
3526 :    
3527 :     The current update also gets rid of the (now useless) controller
3528 :     CM.Control.implicit_anchors.
3529 :    
3530 :     ----------------------------------------------------------------------
3531 :     Name: Matthias Blume
3532 : blume 666 Date: 2000/06/16 17:30:00 JST
3533 :     Tag: blume-20000616-anchorenv
3534 :     Description:
3535 :    
3536 :     This patch implements the long anticipated (just kidding :) "anchor
3537 :     environment" mechanism. In the course of doing this, I also
3538 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3539 :     one should be more robust in certain boundary cases. In any case, it
3540 :     is a lot cleaner than its predecessor (IMHO).
3541 :    
3542 :     This time, although there is yet another boot file format change, I
3543 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3544 :     are necessary and bootstrapping is straightforward. (You cannot read
3545 :     new bootfiles into an old system, but the other way around is no
3546 :     problem.)
3547 :    
3548 :     Visible changes:
3549 :    
3550 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3551 :     longer recognized at all. This means that such path names are not
3552 :     illegal either. For example, the name basis.cm simply refers to a
3553 :     local file called "basis.cm" (i.e, the name is an ordinary path
3554 :     relative to .cm-files directory). Or, to put it differently, only
3555 :     names that start with $ are anchored paths.
3556 : blume 666
3557 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3558 :     vanished.
3559 :    
3560 :     John (Reppy) had critizised this as soon as I originally proposed and
3561 : blume 666 implemented it, but at that time I did not really deeply believe
3562 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3563 :     in another place where it cannot be seen as an abbreviation for
3564 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3565 :     seems to mean (i.e., it "expands" into the corresponding anchor
3566 :     value).
3567 :    
3568 :     However, when paths are used as members in CM description files, it
3569 :     continues to be true that there must be at least another arc after the
3570 :     anchor. This is now enforced separately during semantic analysis
3571 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3572 : blume 666
3573 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3574 :     is a sub-option list of precisely two items -- one labeled "anchor"
3575 :     and the other one labeled "value". As you might expect, "anchor" is
3576 :     used to specify an anchor name to be bound, and "value" specifies what
3577 :     the anchor is being bound to.
3578 :    
3579 :     The value must be a directory name and can be given in either standard
3580 :     syntax (including the possibility that it is itself an anchored path)
3581 :     or native syntax.
3582 :    
3583 :     Examples:
3584 :    
3585 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3586 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3587 :    
3588 :     and so on.
3589 :    
3590 :     The meaning of this is that the .cm-file will be processed with an
3591 :     augmented anchor environment where the given anchor(s) is/are bound to
3592 :     the given values(s).
3593 :    
3594 :     The rationale for having this feature is this: Suppose you are trying
3595 :     to use two different (already stable) libraries a.cm and b.cm (that
3596 :     you perhaps didn't write yourself). Further, suppose each of these
3597 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3598 :     Normally you would now have a problem because the anchor "lib" can not
3599 :     be bound to more than one value globally. Therefore, the project that
3600 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3601 :     place:
3602 :    
3603 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3604 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3605 :    
3606 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3607 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3608 :    
3609 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3610 :     you when you do so at the time of CM.stabilize). Therefore, you can
3611 :     also use an anchored path as the value:
3612 :    
3613 :     a.cm (bind:(anchor:lib value:$a-lib))
3614 :     b.cm (bind:(anchor:lib value:$b-lib))
3615 :    
3616 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3617 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3618 :     itself is always locally bound, setting it globally is no longer
3619 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3620 :     can still be used as a global anchor for separate purposes. As a
3621 :     matter of fact, one can locally define "lib" in terms of a global
3622 :     "lib":
3623 :    
3624 :     a.cm (bind:(anchor:lib value:$lib/a))
3625 :     b.cm (bind:(anchor:lib value:$lib/b))
3626 :    
3627 :     ** 3: The encoding of path names has changed. This affects the way
3628 :     path names are shown in CM's progress report and also the internal
3629 :     protocol encoding used for parallel make.
3630 :    
3631 :     The encoding now uses one or more ':'-separated segments. Each
3632 :     segments corresponds to a file that has been specified relative to the
3633 :     file given by its preceding segment. The first segment is either
3634 :     relative to the CWD, absolute, or anchored. Each segment itself is
3635 :     basically a Unix pathname; all segments but the first are relative.
3636 :    
3637 :     Example:
3638 :    
3639 :     $foo/bar/baz.cm:a/b/c.sml
3640 :    
3641 :     This path denotes the file bar/a/b/c.sml relative to the directory
3642 :     denoted by anchor "foo". Notice that the encoding also includes
3643 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3644 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3645 :     be ignored to get the "real" pathname.
3646 :    
3647 :     To make this fact more obvious, CM puts the names of such "virtual
3648 :     arcs" into parentheses when they appear in progress reports. (No
3649 :     parentheses will appear in the internal protocol encoding.) Thus,
3650 :     what you really see is:
3651 :    
3652 :     $foo/bar/(baz.cm):a/b/c.sml
3653 :    
3654 :     I find this notation to be much more informative than before.
3655 :    
3656 :     Another new feature of the encoding is that special characters
3657 :     including parentheses, colons, (back)slashes, and white space are
3658 :     written as \ddd (where ddd is the decimal encoding of the character).
3659 :    
3660 : blume 670 *** The CM manual still needs to be updated.
3661 :    
3662 : blume 666 ----------------------------------------------------------------------
3663 : leunga 660 Name: Allen Leung
3664 :     Date: 2000/06/15 00:38:00
3665 :     Tag: leunga-20000615-x86-peephole
3666 :    
3667 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3668 :    
3669 :     ----------------------------------------------------------------------
3670 : blume 658 Name: Matthias Blume
3671 :     Date: 2000/06/12 11:40:00
3672 :     Tag: blume-20000612-parmakefix
3673 :     Description:
3674 :    
3675 :     More cleanup after changing the file naming scheme: This time I
3676 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3677 :    
3678 :     ----------------------------------------------------------------------
3679 : leunga 657 Name: Allen Leung
3680 :     Date: 2000/06/09 01:25:00
3681 :     Tag: leunga-20000609-various
3682 :    
3683 :     None of these things should affect normal SML/NJ operations
3684 :    
3685 :     1. Peephole improvements provided by Fermin (c--)
3686 :     2. New annotation DEFUSE for adding extra dependence (moby)
3687 :     3. New X86 LOCK instructions (moby)
3688 :     4. New machine description language for reservation tables (scheduling)
3689 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3690 :     trees etc.)
3691 :     6. I've changed the CM files so that they can work with versions
3692 :     110.0.6, 110.25 and 110.28
3693 :    
3694 :     ----------------------------------------------------------------------
3695 : blume 652 Name: Matthias Blume
3696 : blume 656 Date: 2000/06/09 12:40:00
3697 :     Tag: blume-20000609-log
3698 :     Description:
3699 :    
3700 :     - Removed all(?) remaining RCS Log entries from sources.
3701 :    
3702 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3703 :     anchored paths).
3704 :    
3705 :     ----------------------------------------------------------------------
3706 :     Name: Matthias Blume
3707 : blume 653 Date: 2000/06/07 17:00:00 JST
3708 :     Tag: blume-20000607-no-implicit-anchors
3709 :     Description:
3710 :    
3711 :     1. This update changes the default setting for
3712 :     CM.Control.implicit_anchors from true to false. This means that
3713 :     implicit anchors are no longer permitted by default. I also tried to
3714 :     make sure that nothing else still relies on implicit anchors.
3715 :     (This is the next step on the schedule towards a CM that does not even
3716 :     have the notion of implicit anchors anymore.)
3717 :    
3718 :     2. More CM manual updates.
3719 :    
3720 :     3. I managed to track down and fix the pickling bug I mentioned last
3721 :     time. Because of the previously existing workaround, this entails no
3722 :     immediate practical changes.
3723 :    
3724 :     ----------------------------------------------------------------------
3725 :     Name: Matthias Blume
3726 : blume 652 Date: 2000/06/06 11:15:00 JST
3727 :     Tag: blume-20000606-lazierpickle
3728 :     Description:
3729 :    
3730 :     !!!! NEW BOOT FILES !!!!
3731 :    
3732 :     * The main purpose of this update is to make library pickles lazier in
3733 :     order to reduce the initial space penalty for autoloading a library.
3734 :     As a result, it is now possible to have $smlnj/compiler.cm
3735 :     pre-registered. This should take care of the many complaints or
3736 :     inquiries about missing structure Compiler. This required changes to
3737 :     CM's internal data structures and small tweaks to some algorithms.
3738 :    
3739 :     As a neat additional effect, it is no longer necessary (for the sake
3740 :     of lean heap image files) to distinguish between a "minimal" CM and a
3741 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3742 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3743 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3744 :    
3745 :     To make the life of compiler-hackers easier, "makeml" now also
3746 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3747 :     words, after you bootstrap a new sml for the first time, you will not
3748 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3749 :     around you will still have to do it, though.)
3750 :    
3751 :     * A second change consists of major updates to the CM manual. There
3752 :     are now several appendices with summary information and also a full
3753 :     specification of the CM description file syntax.
3754 :    
3755 :     * In directory src/system I added the script "allcross". This script
3756 :     invokes sml and cross-compiles the compiler for all supported
3757 :     architectures. (Useful when providing a new set of boot files.)
3758 :    
3759 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3760 :     added a small tweak to pickle-util.sml to work around this problem,
3761 :     but it is not a proper fix yet. I will investigate further. (The
3762 :     effect of the bug was an inflation of library pickle size.)
3763 :    
3764 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3765 :    
3766 :     ----------------------------------------------------------------------
3767 : leunga 646 Name: Allen Leung
3768 : leunga 648 Date: 2000/05/25 17:28 EDT
3769 :     Tag: leunga-20000525-ra
3770 :     Description:
3771 :    
3772 :     Fixed a bug in freezing phase of the register allocator.
3773 :    
3774 :     ----------------------------------------------------------------------
3775 :     Name: Allen Leung
3776 : leunga 647 Date: 2000/05/15 22:53 EDT
3777 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3778 :     Description:
3779 :    
3780 :     1. Alpha
3781 :    
3782 :     Slight cleanup. Removed the instruction SGNXL
3783 :    
3784 :     2. X86
3785 :    
3786 :     Added the following instructions to the instruction set:
3787 :    
3788 :     ROLx, RORx,
3789 :     BTx, BTSx, BTLx, BTRx,
3790 :     XCHGx, and variants with the LOCK prefix
3791 :    
3792 :     3. Register Allocation
3793 :    
3794 :     The module ra-rewrite-with-renaming has been improved.
3795 :    
3796 : leunga 647 These have no effect on SML/NJ.
3797 :    
3798 : leunga 646 ----------------------------------------------------------------------
3799 : blume 642 Name: Matthias Blume
3800 : blume 645 Date: 2000/05/15 16:20:00 JST
3801 :     Tag: blume-20000515-lightrebuild
3802 :     Description:
3803 :    
3804 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
3805 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
3806 :     defined. In effect, the command will not build any cross-compiler
3807 :     backends and therefore finish more quickly.
3808 :    
3809 :     The "fixpt" script also takes a "-light" switch to be able to use
3810 :     this new facility while compiling for a fixpoint.
3811 :    
3812 :     2. I replaced all mentions of anchored paths in group owner specifications
3813 :     with simple relative paths (usually starting with "..").
3814 :     The rationale is that a library's internal workings should not be
3815 :     compromised by the lack of some anchor. (An anchor is necessary
3816 :     for someone who wants to refer to the library by an anchored path,
3817 :     but it should not be necessary to build the same library in the first
3818 :     place.)
3819 :    
3820 :     3. I changed the way CM's tool mechanism determines the shell command
3821 :     string used for things like ml-yacc etc. so that it does not break
3822 :     when CM.Control.implicit_anchors is turned off.
3823 :    
3824 :     ----------------------------------------------------------------------
3825 :     Name: Matthias Blume
3826 : blume 644 Date: 2000/05/12 18:20:00 JST
3827 :     Tag: blume-20000512-ml-build
3828 :     Description:
3829 :    
3830 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
3831 :     getting installed properly (by config/install.sh).
3832 :    
3833 :     ----------------------------------------------------------------------
3834 :     Name: Matthias Blume
3835 : blume 643 Date: 2000/05/12 17:30:00 JST
3836 :     Tag: blume-20000512-anchors
3837 :     Description:
3838 :    
3839 :     !!! NEW BOOT FILES !!!
3840 :    
3841 :     This change is in preparation of fading out support for "implicitly
3842 :     anchored path names". I went through all sources and used the
3843 :     explicit (and relatively new) $-notation. See system/README and the
3844 :     CM manual for more info on this.
3845 :    
3846 :     I also modified the anchoring scheme for some things such as "smlnj",
3847 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
3848 :     anchors are more expressive: anchor name and first arc do not have to
3849 :     coincide. This entails the following user-visible change:
3850 :    
3851 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
3852 :     particular, when you fire up sml with a command-line argument, say,
3853 :     e.g.:
3854 :    
3855 :     sml '$smlnj/cmb.cm'
3856 :    
3857 :     At the ML toplevel prompt:
3858 :    
3859 :     CM.autoload "$smlnj/cmb.cm";
3860 :    
3861 :     There is also a new controller in CM.Control that can be used to turn
3862 :     off all remaining support for implicit anchors by saying:
3863 :    
3864 :     CM.autoload "$smlnj/
3865 :     #set CM.Control.implicit_anchors false;
3866 :    
3867 :     This causes CM to reject implicitly anchored paths. This is (for the
3868 :     time being) less permissive than the "final" version where there will
3869 :     be no more such implicit anchors and relative paths will be just that:
3870 :     relative.
3871 :    
3872 :     The next step (version after next version?) will be to make the
3873 :     default for CM.Control.implicit_anchors false. After the dust has
3874 :     settled, I can then produce the "final" version of this...
3875 :    
3876 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
3877 :    
3878 :     ----------------------------------------------------------------------
3879 :     Name: Matthias Blume
3880 : blume 642 Date: 2000/05/11 16:30:00 JST
3881 :     Tag: blume-20000511-sources
3882 :     Description:
3883 :    
3884 :     The main change is that I added function CM.sources as a generalized
3885 :     version of the earlier CM.makedepend. This entails the following
3886 :     additional changes:
3887 :    
3888 :     - CM.makedepend has been dropped.
3889 :    
3890 :     - CM manual has been updated.
3891 :    
3892 :     - TOOLS signature and API have been changed.
3893 :    
3894 :     ----------------------------------------------------------------------
3895 : leunga 640 Name: Allen Leung
3896 : leunga 641 Date: 2000/05/10 21:17 EDT
3897 :     Tag: leunga-20000510-moby-c--ssa
3898 :     Description:
3899 :    
3900 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
3901 :     None of these affect SML/NJ.
3902 :    
3903 :     1. Register Allocation
3904 :    
3905 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
3906 :     This module tries to remove local (i.e. basic block level) redundancies
3907 :     during spilling.
3908 :    
3909 :     b. A new framework for performing region based register allocation.
3910 :     Not yet entirely functional.
3911 :    
3912 :     2. X86
3913 :    
3914 :     a. DefUse for POP was missing the stack pointer [found by Lal]
3915 :     b. Reload for CALL was incorrect in X86Spill [found by John]
3916 :     c. Various fixes in X86Spill so that it can be used correctly for
3917 :     the new spilling module.
3918 :    
3919 :     3. SSA/IR
3920 :    
3921 :     a. New module ir/dj-dataflow.sml implements elimination based
3922 :     data flow analysis.
3923 :    
3924 :     4. MLRiscGen
3925 :    
3926 :     a. Fix for gc type annotation
3927 :    
3928 :     5. MDGen
3929 :    
3930 :     Various fixes for machine description -> ml code translation. For ssa
3931 :     only.
3932 :    
3933 :     ----------------------------------------------------------------------
3934 :     Name: Allen Leung
3935 : leunga 640 Date: 2000/05/08 22:17 EDT
3936 :     Tag: leunga-20000508-labexp
3937 :     Description:
3938 :    
3939 :     Fermin has found a few assembly problems with constant expressions
3940 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
3941 :     which choke on dumb assemblers. This is his fix.
3942 :    
3943 :     ----------------------------------------------------------------------
3944 : dbm 635 Name: Dave MacQueen
3945 :     Date: 2000/04/09 14:00 EDT
3946 :     Tag: dbm-20000502-Version_110_28
3947 :     Description:
3948 :    
3949 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
3950 :    
3951 :     2. Updated config/version to 110.28
3952 :    
3953 :     3. Updated config/srcarchiveurl
3954 :    
3955 :     3. New boot files!
3956 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
3957 :    
3958 :     ----------------------------------------------------------------------
3959 : blume 634 Name: Matthias Blume
3960 :     Date: 2000/05/01 19:05:00 JST
3961 :     Tag: blume-20000501-noweb
3962 :     Description:
3963 :    
3964 :     A new noweb tool has been added. The existing system is entirely
3965 :     unaffected by this, but some CM users have asked for renewed noweb
3966 :     support. Everything is documented in the CM manual.
3967 :    
3968 :     New (plugin) libraries:
3969 :    
3970 :     noweb-tool.cm
3971 :     nw-ext.cm
3972 :    
3973 :     ----------------------------------------------------------------------
3974 : dbm 633 Name: Dave MacQueen
3975 :     Date: 2000/04/30 12:40PM EDT
3976 :     Tag: dbm-20000430-bug_fixes
3977 :     Description:
3978 :    
3979 :     1. Fix for bug 1498
3980 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
3981 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
3982 :     added toRealArray function
3983 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
3984 :     added check for tag Obj.RealArray to array printing case in ppObj
3985 :    
3986 :     2. Fix for bug 1510
3987 :     smlnj/src/compiler/Semant/types/typesutil.sml
3988 :     fixed definition of dummyargs (used by equalTycon) so that
3989 :     dummy args are distinct types
3990 :    
3991 :     ----------------------------------------------------------------------
3992 : blume 629 Name: Matthias Blume
3993 : blume 632 Date: 2000/04/30 01:00:00 JST
3994 :     Tag: blume-20000430-versions
3995 :     Description:
3996 :    
3997 :     1. CM version numbering added. This is an implementation of Lal's
3998 :     proposal for adding version numbers and version checking to .cm
3999 :     files. Lal said that his proposal was just that -- a proposal.
4000 :     For the time being I went ahead and implemented it so that people
4001 :     can comment on it. Everything is completely backward-compatible
4002 :     (except for the stable library format, i.e., new bootfiles!).
4003 :    
4004 :     As usual, see the CM manual for details.
4005 :    
4006 :     2. An alternative syntax for anchored paths has been implemented.
4007 :     Dave has recently voiced the same concerns that I had when I did
4008 :     this, so there should be some support. My take is that eventually
4009 :     I will let support for the current syntax (where anchors are
4010 :     "implicit") fade out in favor of the new, explicit syntax.
4011 :     In order to be backward-compatible, both old and new syntax are
4012 :     currently supported.
4013 :    
4014 :     Again, see the CM manual for details.
4015 :    
4016 :     3. Parallel make is trying to be slightly smarter: When the master
4017 :     process finds a "bottleneck", i.e., when there is only one
4018 :     compilation unit that can be compiled and everybody else is
4019 :     waiting on it, then it will simply compile it directly instead
4020 :     of clumsily telling one of the slaves to do it.
4021 :    
4022 :     4. Support for "unsharing" added. This is necessary in order to be
4023 :     able to have two different versions of the same library running
4024 :     at the same time (e.g., for trying out a new MLRISC while still
4025 :     having the old MLRISC linked into the current compiler, etc.)
4026 :     See the CM manual.
4027 :    
4028 :     5. Simple "makedepend" functionality added for generating Makefile
4029 :     dependency information. (This is rather crude at the moment.
4030 :     Expect some changes here in the future.)
4031 :    
4032 :     6. ".fun" added as a recognized suffix for ML files. Also documented
4033 :     explicitly in the manual that the fallback behavior (unknown suffix
4034 :     -> ML file) is not an official feature!
4035 :    
4036 :     7. Small changes to the pickler for stable libraries.
4037 :    
4038 :     8. Several internal changes to CM (for cleanup/improvement).
4039 :    
4040 :    
4041 :     !!!! NEW BINFILES !!!!
4042 :    
4043 :     ----------------------------------------------------------------------
4044 :     Name: Matthias Blume
4045 : blume 631 Date: 2000/04/28 17:30:00 JST
4046 :     Tag: blume-20000428-pathconfig
4047 :     Description:
4048 :    
4049 :     1. I changed config/install.sh to remove duplicate entries from the
4050 :     lib/pathconfig file at the end. Moreover, the final version of
4051 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
4052 :     in src/system/installml.
4053 :    
4054 :     2. The config/install.sh script now consistently uses relative
4055 :     pathnames in lib/pathconfig whenever the anchor is in the lib
4056 :     directory. (So far this was true for the libraries that come
4057 :     pre-compiled and bundled as part of the bootfiles but not for
4058 :     libraries that are compiled by the script itself.)
4059 :    
4060 :     ----------------------------------------------------------------------
4061 :     Name: Matthias Blume
4062 : blume 629 Date: 2000/04/26 13:10:00 JST
4063 :     Tag: blume-20000426-fun_suffix
4064 :     Description:
4065 :    
4066 :     Added ".fun" as a recognized file name suffix (for ML code).
4067 :    
4068 :     ----------------------------------------------------------------------
4069 : leunga 628 Name: Allen Leung
4070 :     Date: 2000/04/25 17:00:00 EST
4071 :     Tag: leunga-20000425-alpha-ra
4072 :     Description:
4073 :    
4074 :     1. Alpha
4075 :    
4076 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
4077 :     in C--.
4078 :    
4079 :     2. RA
4080 :    
4081 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
4082 :     and the interference graph.
4083 :    
4084 :     ----------------------------------------------------------------------
4085 : dbm 627 Name: Dave MacQueen
4086 :     Date: 2000/04/25/
4087 :     Tag: dbm-20000425-mlyacc_doc_examples
4088 :     Description:
4089 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
4090 :     Updated all three examples in src/ml-yacc/examples to run
4091 :     under 110.* using CM.make.
4092 :    
4093 :     ----------------------------------------------------------------------
4094 : leunga 624 Name: Allen Leung
4095 :     Date: 2000/04/20 23:04:00 EST
4096 :     Tag: leunga-20000420-ssa-c---stuff
4097 :     Description:
4098 :    
4099 :     This update synchronizes my repository with Yale's. Most of these
4100 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
4101 :    
4102 :     1. Register Allocator
4103 :    
4104 :     a. An improvement in the interference graph construction:
4105 :     Given a copy
4106 :    
4107 :     s <- t
4108 :    
4109 : leunga 628 no interference edge between s and t is added for this definition of s.
4110 : leunga 624
4111 :     b. I've added two new spill heuristic modules that Fermin and I developed
4112 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
4113 :     useful for others (Moby?)
4114 :    
4115 :     2. X86
4116 :    
4117 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
4118 :    
4119 :     3. Alpha
4120 :    
4121 :     a. Added the BSR instruction and code generation that goes with it [C--]
4122 :     b. Other fixes too numerous to recount provided by Fermin [C--]
4123 :    
4124 :     4. Regmaps
4125 :    
4126 :     a. The regmaps are not initialized with the identity physical bindings
4127 :     at creation time. This is unneeded.
4128 :    
4129 :     5. MLRISC Optimizations
4130 :    
4131 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
4132 :     intersects with liveness incrementally in linear time! Woohoo!
4133 :     This is now used in my new SSA construction algorithm.
4134 :    
4135 :     b. THe branch reorganization module is now smarter about linear chains of
4136 :     basic blocks.
4137 :    
4138 :    
4139 :     ----------------------------------------------------------------------
4140 : blume 615 Name: Matthias Blume
4141 :     Date: 2000/04/12 13:52:00 JST
4142 :     Tag: blume_main_v110p27_1
4143 :     Description:
4144 :    
4145 :     Changed install.sh script to handle archive files without version number
4146 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
4147 :     name of the boot file archive.
4148 :    
4149 :     ----------------------------------------------------------------------
4150 : dbm 607 Name: Dave MacQueen
4151 :     Date: 2000/04/09 14:00 EDT
4152 :     Tag: dbm-20000410-Version_110_27
4153 :     Description:
4154 :    
4155 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
4156 :    
4157 :     2. Updated src/config/version to 110.27
4158 :    
4159 :     3. New boot files!
4160 :    
4161 :     ----------------------------------------------------------------------
4162 : leunga 606 Name: Allen Leung
4163 :     Date: 2000/04/09 19:09:00 EST
4164 :     Tag: leunga-20000409-misc
4165 :     Description:
4166 :    
4167 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
4168 :    
4169 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
4170 :    
4171 :     ----------------------------------------------------------------------
4172 : monnier 605 Name: Stefan
4173 :     Date: 2000/04/07 10:00:00 EDT
4174 :     Tag: monnier-20000406-branch-handling
4175 :     Description:
4176 :    
4177 :     Improved handling of branches (mostly those generated from
4178 :     polymorphic equality), removed switchoff and changed the
4179 :     default optimization settings (more cpsopt and less flintopt).
4180 :    
4181 :     ----------------------------------------------------------------------
4182 : leunga 601 Name: Allen Leung
4183 : leunga 602 Date: 2000/04/06 01:30:00 EST
4184 :     Tag: leunga-20000406-peephole-x86-SSA-2
4185 :     Description:
4186 :    
4187 :     Forgot a few files.
4188 :    
4189 :     ----------------------------------------------------------------------
4190 :     Name: Allen Leung
4191 : leunga 601 Date: 2000/04/06 00:36:00 EST
4192 :     Tag: leunga-20000406-peephole-x86-SSA
4193 :     Description:
4194 :    
4195 :     1. New Peephole code
4196 :    
4197 :     2. Minor improvement to X86 instruction selection
4198 :    
4199 :     3. Various fixes to SSA and machine description -> code translator
4200 :    
4201 :     ----------------------------------------------------------------------
4202 : blume 595 Name: Matthias Blume
4203 :     Date: 2000/04/05 12:30:00 JST
4204 :     Tag: blume_main_v110p26p2_3
4205 :     Description:
4206 :    
4207 :     This update just merges three minor cosmetic updates to CM's sources
4208 :     to get ready for the 110.27 code freeze on Friday. No functionality
4209 :     has changed.
4210 :    
4211 :     ----------------------------------------------------------------------
4212 : leunga 593 Name: Allen Leung
4213 : leunga 594 Date: 2000/04/04 19:39:00 EST
4214 :     Tag: leunga-20000404-x86-asm
4215 :     Description:
4216 :    
4217 :     1. Fixed a problem in X86 assembly.
4218 :    
4219 :     Things like
4220 :    
4221 :     jmp %eax
4222 :     jmp (%eax)
4223 :    
4224 :     should be output as
4225 :    
4226 :     jmp *%eax
4227 :     jmp *(%eax)
4228 :    
4229 :     2. Assembly output
4230 :    
4231 :     Added a new flag
4232 :    
4233 :     "asm-indent-copies" (default to false)
4234 :    
4235 :     When this flag is on, parallel copies will be indented an extra level.
4236 :    
4237 :     ----------------------------------------------------------------------
4238 :     Name: Allen Leung
4239 : leunga 593 Date: 2000/04/04 03:18:00 EST
4240 :     Tag: leunga-20000404-C--Moby
4241 :     Description:
4242 :    
4243 :     All of these fixes are related to C--, Moby, and my own optimization
4244 :     stuff; so they shouldn't affect SML/NJ.
4245 :    
4246 :     1. X86
4247 :    
4248 :     Various fixes related floating point, and extensions.
4249 :    
4250 :     2. Alpha
4251 :    
4252 :     Some extra patterns related to loads with signed/zero extension
4253 :     provided by Fermin.
4254 :    
4255 :     3. Assembly
4256 :    
4257 : monnier 659 When generating assembly, resolve the value of client defined constants,
4258 : leunga 593 instead of generating symbolic values. This is controlled by the
4259 :     new flag "asm-resolve-constants", which is default to true.
4260 :    
4261 :     4. Machine Descriptions
4262 :    
4263 :     a. The precedence parser was slightly broken when parsing infixr symbols.
4264 :     b. The type generalizing code had the bound variables reversed, resulting
4265 :     in a problem during arity raising.
4266 :     c. Various fixes in machine descriptions.
4267 :    
4268 :     ----------------------------------------------------------------------
4269 : blume 592 Name: Matthias Blume
4270 :     Date: 2000/04/03 16:05:00 JST
4271 :     Tag: blume_main_v110p26p2_2
4272 :     Description:
4273 :    
4274 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
4275 :     now done via the ordinary static environment that is context to each
4276 :     compilation unit.
4277 :    
4278 :     To this end, I arranged that instead of "structure Core" as "structure
4279 :     _Core" is bound in the pervasive environment. Core access is done via
4280 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
4281 : blume 592 legal surface-syntax symbol).
4282 :    
4283 :     The current solution is much cleaner because the core environment is
4284 :     now simply part of the pervasive environment which is part of every
4285 :     compilation unit's context anyway. In particular, this eliminates all
4286 :     special-case handling that was necessary until now in order to deal
4287 :     with dynamic and symbolic parts of the core environment.
4288 :    
4289 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
4290 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
4291 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
4292 :     more details.
4293 :    
4294 :     I also tried to track down all mentions of "Core" (as string argument
4295 :     to Symbol.strSymbol) in the compiler and replaced them with a
4296 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
4297 :     name appears in one place only.
4298 :    
4299 :     Binfile and bootfile format have not changed, but the switchover from
4300 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
4301 :     bootfiles anyway.
4302 :    
4303 :     ----------------------------------------------------------------------
4304 : leunga 591 Name: Allen Leung
4305 :     Date: 2000/04/02 21:17:00 EST
4306 :     Tag: leunga-20000402-mltree
4307 :     Description:
4308 : leunga 585
4309 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
4310 :     2. Added a bunch of files from my repository. These are currently
4311 :     used by other non-SMLNJ backends.
4312 :    
4313 : leunga 576 ----------------------------------------------------------------------
4314 : leunga 590 Name: Allen Leung
4315 :     Date: 2000/03/31 21:15:00 EST
4316 :     Tag: leunga-20000331-aliasing
4317 :     Description:
4318 :    
4319 :     This update contains a rewritten (and hopefully more correct) module
4320 :     for extracting aliasing information from CPS.
4321 :