Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 575, Fri Mar 10 02:55:58 2000 UTC revision 1031, Mon Jan 21 22:15:18 2002 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name: Matthias Blume  Name: Matthias Blume
17    Date: 2002/01/21
18    Tag: blume-20020121-ff
19    Description:
20    
21    Another day of fiddling with the FFI...
22    
23    1. Bug fix/workaround:  CKIT does not complain about negative array
24       dimensions, so ml-nlffigen has to guard itself against this possibility.
25       (Otherwise a negative dimension would send it into an infinite loop.)
26    
27    2. Some of the abstract types (light objects, light pointers, most "base"
28       types) in structure C are now eqtypes.
29    
30    3. Added constructors and test functions for NULL function pointers.
31    
32    ----------------------------------------------------------------------
33    Name: Matthias Blume
34    Date: 2002/01/18
35    Tag: blume-20020118-ready-for-new-release
36    Description:
37    
38    Made config/srcarchiveurl point to a new place.  (Will provide boot
39    files shortly.)
40    
41    Maybe we christen this to be 110.38?
42    
43    ----------------------------------------------------------------------
44    Name: Matthias Blume
45    Date: 2002/01/18
46    Tag: blume-20020118-more-ffifiddle
47    Description:
48    
49    Today's FFI fiddling:
50    
51      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
52        C.Set.* that use concrete (MLRep.*) arguments and results instead
53        of abstract ones.
54    
55      - Provided word-style bit operations etc. for "int" representation
56        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
57        and Long.
58    
59    ----------------------------------------------------------------------
60    Name: Matthias Blume
61    Date: 2002/01/18
62    Tag: blume-20020118-use-x86-fp
63    Description:
64    
65    Now that x86-fast-fp seems to be working, I turned it back on again
66    by default.  (Seems to work fine now, even with the FFI.)
67    
68    Other than that, I added some documentation about the FFI to
69    src/ml-nlffigen/README and updated the FFI test examples in
70    src/ml-nlffi-lib/Tests/*.
71    
72    ----------------------------------------------------------------------
73    Name: Allen Leung
74    Date: 2002/01/17
75    Tag: leunga-20020117-x86-fast-fp-call
76    Description:
77    
78       1. Fixed a problem with handling return fp values when x86's fast fp
79          mode is turned on.
80    
81       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
82          of %f32.
83    
84       3. Added a constructor INT32lit to the ast of MLRISC tools.
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2002/01/16
89    Tag: blume-20020116-ffifiddle
90    Description:
91    
92    More fiddling with the FFI interface:
93    
94     - Make constness 'c instead of rw wherever possible.  This eliminates
95       the need for certain explicit coercions.  (However, due to ML's
96       value polymorphism, there will still be many cases where explicit
97       coercions are necessary.  Phantom types are not the whole answer
98       to modeling a subtyping relationship in ML.)
99    
100     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
101    
102     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
103       clumsy workaround for varargs problem.)
104    
105    ----------------------------------------------------------------------
106    Name: Lal George
107    Date: 2002/01/15
108    Tag: <none>
109    Description:
110    
111    1. Since COPY instructions are no longer native to the architecture,
112       a generic functor can be used to implement the expandCopies function.
113    
114    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
115       TEXT segment.
116    
117    ----------------------------------------------------------------------
118    Name: Matthias Blume
119    Date: 2002/01/15
120    Tag: blume-20020115-ffiupdates
121    Description:
122    
123    1. Fix for bug resulting in single-precision float values being returned
124       incorrectly from FFI calls.
125    
126    2. Small modifications to C FFI API:
127    
128        - memory-allocation routines return straight objects (no options)
129          and raise an exception in out-of-memory situations
130        - unsafe extensions to cast between function pointers and pointers
131          from/to ints
132        - added structure C_Debug as an alternative to structure C where
133          pointer-dereferencing (|*| and |*!) always check for null-pointers
134        - added open_lib' to DynLinkage;  open_lib' works like open_lib
135          but also takes a (possibly empty) list of existing library handles
136          that the current library depends on
137    
138    ----------------------------------------------------------------------
139    Name: Matthias Blume
140    Date: 2002/01/10
141    Tag: blume-20020110-newffigen
142    Description:
143    
144    1. Updates to portable graph code.
145    
146    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
147       more scalable now so that even huge interfaces such as the one
148       for GTK compile in finite time and space. :-)
149       See src/ml-nlffigen/README for details on what's new.
150    
151    ----------------------------------------------------------------------
152    Name: Lal George
153    Date: 2001/01/09 14:31:35 EST 2002
154    Tag: george-20011206-rm-native-copy
155    Description:
156    
157            Removed the native COPY and FCOPY instructions
158            from all the architectures and replaced it with the
159            explicit COPY instruction from the previous commit.
160    
161            It is now possible to simplify many of the optimizations
162            modules that manipulate copies. This has not been
163            done in this change.
164    
165    ----------------------------------------------------------------------
166    Name: Lal George
167    Date: 2001/12/06 16:50:13 EST 2001
168    Tag: george-20011206-mlrisc-instruction
169    Description:
170    
171    Changed the representation of instructions from being fully abstract
172    to being partially concrete. That is to say:
173    
174      from
175            type instruction
176    
177      to
178            type instr                              (* machine instruction *)
179    
180            datatype instruction =
181                LIVE of {regs: C.cellset, spilled: C.cellset}
182              | KILL of {regs: C.cellset, spilled: C.cellset}
183              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
184              | ANNOTATION of {i: instruction, a: Annotations.annotation}
185              | INSTR of instr
186    
187    This makes the handling of certain special instructions that appear on
188    all architectures easier and uniform.
189    
190    LIVE and KILL say that a list of registers are live or killed at the
191    program point where they appear. No spill code is generated when an
192    element of the 'regs' field is spilled, but the register is moved to
193    the 'spilled' (which is present, more for debugging than anything else).
194    
195    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
196    We used to generate:
197    
198            DEFFREG f1
199            f1 := f2 + f3
200            trapb
201    
202    but now generate:
203    
204            f1 := f2 + f3
205            trapb
206            LIVE {regs=[f1,f2,f3], spilled=[]}
207    
208    Furthermore, the DEFFREG (hack) required that all floating point instruction
209    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
210    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
211    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
212    this architecture).
213    
214    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
215    all the architectures. This will result in further simplification of the
216    register allocator that must be aware of them for coalescing purposes, and
217    will also simplify certain aspects of the machine description that provides
218    callbacks related to parallel copies.
219    
220    ANNOTATION should be obvious, and now INSTR represents the honest to God
221    machine instruction set!
222    
223    The <arch>/instructions/<arch>Instr.sml files define certain utility
224    functions for making porting easier -- essentially converting upper case
225    to lower case. All machine instructions (of type instr) are in upper case,
226    and the lower case form generates an MLRISC instruction. For example on
227    the alpha we have:
228    
229      datatype instr =
230         LDA of {r:cell, b:cell, d:operand}
231       | ...
232    
233      val lda : {r:cell, b:cell, d:operand} -> instruction
234        ...
235    
236    where lda is just (INSTR o LDA), etc.
237    
238    ----------------------------------------------------------------------
239    Name: Matthias Blume
240    Date: 2001/11/22 21:40:00 EST
241    Tag: Release_110_37
242    Description:
243    
244    Release 110.37.  This time for real.
245    
246    ----------------------------------------------------------------------
247    Name: Matthias Blume
248    Date: 2001/11/21 16:35:00 EST
249    Tag: blume-20011121-foot-in-mouth
250    Description:
251    
252    Removed the "Release_110_37" tag because of a serious bug.
253    This will be re-tagged once the bug is fixed.
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume
257    Date: 2001/11/21 16:14:00 EST
258    Tag: blume-20011121-forgottenfile
259    Description:
260    
261    Forgot to add a file.  (Just a .tex-file -- part of
262    the CM manual source.)
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume
266    Date: 2001/11/21 16:10:00 EST
267    Tag: blume-20011121-invalid_110_37
268    Description:
269    
270    Note: I removed the original tag "Release_110_37" from this commit
271          because we found a serious bug in all non-x86 backends.
272          - Matthias
273    
274    1. Modifications to the SML/NJ code generator and to the runtime system
275       so that code object name strings are directly inserted into code
276       objects at code generation time.  The only business the runtime system
277       has with this is now to read the name strings on occasions.
278       (The encoding of the name string has also changed somewhat.)
279    
280    2. CM now implements a simple "set calculus" for specifying export lists.
281       In particular, it is now possible to refer to the export lists of
282       other libraries/groups/sources and form unions as well as differences.
283       See the latest CM manual for details.
284    
285    3. An separate notion of "proxy" libraries has again be eliminated from
286       CM's model.  (Proxy libraries are now simply a special case of using
287       the export list calculus.)
288    
289    4. Some of the existing libraries now take advantage of the new set
290       calculus.
291       (Notice that not all libraries have been converted because some
292       of the existing .cm-files are supposed to be backward compatible
293       with 110.0.x.)
294    
295    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
296       "General.exnMessage"!  The former relies on a certain hook to be
297       initialized, and that often does not happen in the stand-alone case.)
298    
299    ----------------------------------------------------------------------
300    Name: Lal George
301    Date: 2001/11/21  13:56:18 EST
302    Tag: george-2001121-pseudo-ops
303    Description:
304    
305      Implemented a complete redesign of MLRISC pseudo-ops. Now there
306      ought to never be any question of incompatabilities with
307      pseudo-op syntax expected by host assemblers.
308    
309      For now, only modules supporting GAS syntax are implemented
310      but more should follow, such as MASM, and vendor assembler
311      syntax, e.g. IBM as, Sun as, etc.
312    
313    ----------------------------------------------------------------------
314    Name: Matthias Blume
315    Date: 2001/11/14 11:52:00 EST
316    Tag: blume-20011114-srcname
317    Description:
318    
319    1. Routed the name of the current source file to mlriscgen where it
320       should be directly emitted into the code object.  (This last part
321       is yet to be done.)
322    
323    2. Some cleanup of the pgraph code to make it match the proposal that
324       I put out the other day.  (The proposal notwithstanding, things are
325       still in flux here.)
326    
327    ----------------------------------------------------------------------
328    Name: Lal George
329    Date: 2001/11/14 09:44:04 EST
330    Tag:
331    Description:
332    
333      Fix for a backpatching bug reported by Allen.
334    
335      Because the boundary between short and long span-dependent
336      instructions is +/- 128, there are an astounding number of
337      span-dependent instructions whose size is over estimated.
338    
339      Allen came up with the idea of letting the size of span
340      dependent instructions be non-monotonic, for a maxIter
341      number of times, after which the size must be monotonically
342      increasing.
343    
344      This table shows the number of span-dependent instructions
345      whose size was over-estimated as a function of maxIter, for the
346      file Parse/parse/ml.grm.sml:
347    
348         maxIter            # of instructions:
349            10                      687
350            20                      438
351            30                      198
352            40                        0
353    
354      In compiling the compiler, there is no significant difference in
355      compilation speed between maxIter=10 and maxIter=40. Actually,
356      my measurements showed that maxIter=40 was a tad faster than
357      maxIter=10! Also 96% of the  files in the compiler reach a fix
358      point within 13 iterations, so fixing maxIter at 40, while high,
359      is okay.
360    
361    ----------------------------------------------------------------------
362    Name: Matthias Blume
363    Date: 2001/10/31 15:25:00 EST
364    Tag: blume-20011031-pgraph
365    Description:
366    
367    CKIT:
368    * Changed the "Function" constructor of type Ast.ctype to carry optional
369      argument identifiers.
370    * Changed the return type of TypeUtil.getFunction accordingly.
371    * Type equality ignores the argument names.
372    * TypeUtil.composite tries to preserve argument names but gives up quickly
373      if there is a mismatch.
374    
375    installation script:
376    * attempts to use "curl" if available (unless "wget" is available as well)
377    
378    CM:
379    * has an experimental implementation of "portable graphs" which I will
380      soon propose as an implementation-independent library format
381    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
382    
383    NLFFI-LIB:
384    * some cleanup (all cosmetic)
385    
386    NLFFIGEN:
387    * temporarily disabled the mechanism that suppresses ML output for
388      C definitions whose identifiers start with an underscore character
389    * generate val bindings for enum constants
390    * user can request that only one style (light or heavy) is being used;
391      default is to use both (command-line arguments: -heavy and -light)
392    * fixed bug in handling of function types involving incomplete pointers
393    * generate ML entry points that take record arguments (i.e., using
394      named arguments) for C functions that have a prototype with named
395      arguments
396      (see changes to CKIT)
397    
398    ----------------------------------------------------------------------
399    Name: Allen Leung
400    Date: 2001/10/27 20:34:00 EDT
401    Tag: leunga-20011027-x86-fast-fp-call
402    Description:
403    
404       Fixed the bug described in blume-20010920-slowfp.
405    
406       The fix involves
407          1. generating FCOPYs in FSTP in ia32-svid
408          2. marking a CALL with the appropriate annotation
409    
410    ----------------------------------------------------------------------
411    Name: Matthias Blume
412    Date: 2001/10/16 11:32:00 EDT
413    Tag: blume-20011016-netbsd
414    Description:
415    
416    Underscore patch from Chris Richards (fixing problem with compiling
417    runtime system under recent NetBSD).
418    
419    ----------------------------------------------------------------------
420    Name: Allen Leung
421    Date: 2001/10/12 17:18:32 EDT 2001
422    Tag: leung-20011012-x86-printflowgraph
423    Description:
424    
425    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
426    
427    ----------------------------------------------------------------------
428    Name: Lal George
429    Date: 2001/10/11 23:51:34 EDT
430    Tag: george-20011011-too-many-instrs
431    Description:
432    
433    The representation of a program point never expected to see more
434    than 65536 instructions in a basic block!
435    
436    ----------------------------------------------------------------------
437    Name: Lal George
438    Date: 2001/10/09 09:41:37 EDT
439    Tag: george-20011008-mlrisc-labels
440    Description:
441    
442    Changed the machine description files to support printing of
443    local and global labels in assembly code, based on host assembler
444    conventions.
445    
446    ----------------------------------------------------------------------
447    Name: Matthias Blume
448    Date: 2001/09/25 15:25:00 EDT
449    Tag: blume-20010925-exninfo
450    Description:
451    
452    I provided a non-hook implementation of exnName (at the toplevel) and
453    made the "dummy" implementation of exnMessage (at the toplevel) more
454    useful: if nothing gets "hooked in", then at least you are going to
455    see the exception name and a message indicating why you don't see more.
456    
457    [For the time being, programs that need exnMessage and want to use
458    ml-build should either use General.exnMessage (strongly recommended) or
459    refer to structure General at some other point so that CM sees a
460    static dependency.]
461    
462    [Similar remarks go for "print" and "use":  If you want to use their
463    functionality in stand-alone programs generated by ml-build, then use
464    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
465    
466    ----------------------------------------------------------------------
467    Name: Matthias Blume
468    Date: 2001/09/20 17:28:00 EDT
469    Tag: blume-20010920-slowfp
470    Description:
471    
472    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
473    
474    ----------------------------------------------------------------------
475    Name: Matthias Blume
476    Date: 2001/09/20 17:20:00 EDT
477    Tag: blume-20010920-canonicalpaths
478    Description:
479    
480    0. Updated the BOOT file (something that I forgot to do earlier).
481    
482    1. Small internal change to CM so that it avoids "/../" in filenames
483       as much as possible (but only where it is safe).
484    
485    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
486       that arguments that contain delimiters are passed through correctly.
487       This change also means that all "special" arguments of the form
488       @SMLxxx... must come first.
489    
490    3. Changed install script to put relative anchor names for tool commands
491       into pathconfig.
492    
493    ----------------------------------------------------------------------
494    Name: Matthias Blume
495    >>>>>>> 1.169
496    Date: 2001/09/18 15:35:00 EDT
497    Tag: blume-20010918-readme11036
498    Description:
499    
500    Added README files.
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume
504    Date: 2001/09/18 11:45:00 EDT
505    Tag: Release_110_36 (retag)
506    Description:
507    
508    Fixed mistake in config/preloads. Retagged as 110.36.
509    
510    ----------------------------------------------------------------------
511    Name: Matthias Blume
512    Date: 2001/09/18 09:40:00 EDT
513    Tag: Release_110_36_orig (tag changed)
514    Description:
515    
516    New version (110.36).  New bootfiles.
517    
518    ----------------------------------------------------------------------
519    Name: Matthias Blume
520    Date: 2001/09/14 16:15:00 EDT
521    Tag: blume-20010914-x86fastfp
522    Description:
523    
524    John committed some changes that Allen made, in particular a (hopefully)
525    correctly working version of the x86-fp module.
526    
527    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
528    flag to "true".  Everything seems to compile to a fixpoint ok, and
529    "mandelbrot" speeds up by about 15%.
530    
531    ----------------------------------------------------------------------
532    Name: Matthias Blume
533    Date: 2001/09/13 11:20:00 EDT
534    Tag: blume-20010913-minimal
535    Description:
536    
537    1. Stefan Monnier's patch to fix a miscompilation problem that
538       was brought to light by John Reppy's work on Moby.
539    
540    2. Implemented a minimal "structure Compiler" that contains just
541       "version" and "architecture".  The minimal version will be
542       available when the full version is not.  This is for backward-
543       compatibility with code that wants to test Compiler.version.
544    
545    ----------------------------------------------------------------------
546    Name: Matthias Blume
547    Date: 2001/08/28 14:03:00 EDT
548    Tag: blume-20010828-ml-lex
549    Description:
550    
551    Fix for bug 1581, received from Neophytos Michael.
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume
555    Date: 2001/08/27 11:20:00 EDT
556    Tag: blume-20010827-readme11035
557    Description:
558    
559    Fleshed out the README file for 110.35.
560    
561    ----------------------------------------------------------------------
562    Name: Matthias Blume
563    Date: 2001/08/24 17:10:00 EDT
564    Tag: Release_110_35
565    Description:
566    
567    New version number (110.35).  New bootfiles.
568    
569    ----------------------------------------------------------------------
570    Name: Lal George
571    Date: 2001/08/24 13:47:18 EDT 2001
572    Tag: george-20010824-MLRISC-graphs
573    Description:
574    
575     removed clusters from MLRISC completely and replaced with graphs.
576    
577    ----------------------------------------------------------------------
578    Name: Matthias Blume
579    Date: 2001/08/23 17:50:00 EDT
580    Tag: blume-20010823-toplevel
581    Description:
582    
583    - some reorganization of the code that implements various kinds of
584      environments in the compiler (static, dynamic, symbolic, combined)
585    - re-implemented the EnvRef module so that evalStream works properly
586      (if the stream contains references to "use", "CM.make", etc.)
587    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume
591    Date: 2001/08/20 15:50 EDT
592    Tag: blume20010820-slipup
593    Description:
594    
595    I forgot to commit a few files.  Here they are...
596    
597    ----------------------------------------------------------------------
598    Name: Matthias Blume
599    Date: 2001/08/20 15:35:00 EDT
600    Tag: blume-20010820-debugprof
601    Description:
602    
603    !!!! NEW BOOTFILES !!!!
604    
605    This is another round of reorganizing the compiler sources.  This
606    time the main goal was to factor out all the "instrumentation"
607    passes (for profiling and backtracing) into their own library.
608    The difficulty was to do it in such a way that it does not depend
609    on elaborate.cm but only on elabdata.cm.
610    
611    Therefore there have been further changes to both elaborate.cm and
612    elabdata.cm -- more "generic" things have been moved from the former
613    to the latter.  As a result, I was forced to split the assignment
614    of numbers indicating "primtyc"s into two portions: SML-generic and
615    SML/NJ-specific.  Since it would have been awkward to maintain,
616    I bit the bullet and actually _changed_ the mapping between these
617    numbers and primtycs.  The bottom line of this is that you need
618    a new set of bin- and bootfiles.
619    
620    I have built new bootfiles for all architectures, so doing a fresh
621    checkout and config/install.sh should be all you need.
622    
623    The newly created library's name is
624    
625        $smlnj/viscomp/debugprof.cm
626    
627    and its sources live under
628    
629        src/compiler/DebugProf
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume
633    Date: 2001/08/15 17:15:00 EDT
634    Tag: blume-20010815-compreorg
635    Description:
636    
637    This is a first cut at reorganizing the CM libraries that make up the
638    core of the compiler.  The idea is to separate out pieces that could
639    be used independently by tools, e.g., the parser, the typechecker, etc.
640    
641    The current status is a step in this direction, but it is not quite
642    satisfactory yet.  Expect more changes in the future.
643    
644    Here is the current (new) organization...
645    
646        What used to be $smlnj/viscomp/core.cm is now divided into
647        six CM libraries:
648    
649             $smlnj/viscomp/basics.cm
650                           /parser.cm
651                           /elabdata.cm
652                           /elaborate.cm
653                           /execute.cm
654                           /core.cm
655    
656        The CM files for these libraries live under src/system/smlnj/viscomp.
657        All these libraries are proxy libraries that contain precisely
658        one CM library component.  Here are the locations of the components
659        (all within the src/compiler tree):
660    
661             Basics/basics.cm
662             Parse/parser.cm
663             ElabData/elabdata.cm
664             Elaborator/elaborate.cm
665             Execution/execute.cm
666             core.cm
667    
668         [This organization is the same that has been used already
669         for a while for the architecture-specific parts of the visible
670         compiler and for the old version of core.cm.]
671    
672         As you will notice, many source files have been moved from their
673         respective original locations to a new home in one of the above
674         subtrees.
675    
676         The division of labor between the new libraries is the following:
677    
678             basics.cm:
679                - Simple, basic definitions that pertain to many (or all) of
680                  the other libraries.
681             parser.cm:
682                - The SML parser, producing output of type Ast.dec.
683                - The type family for Ast is also defined and exported here.
684             elabdata.cm:
685                - The datatypes that describe input and output of the elaborator.
686                  This includes types, absyn, and static environments.
687             elaborator.cm:
688                - The SML/NJ type checker and elaborator.
689                  This maps an Ast.dec (with a given static environment) to
690                  an Absyn.dec (with a new static environment).
691                - This libraries implements certain modules that used to be
692                  structures as functors (to remove dependencies on FLINT).
693             execute.cm:
694                - Everything having to do with executing binary code objects.
695                - Dynamic environments.
696             core.cm:
697                - SML/NJ-specific instantiations of the elaborator and MLRISC.
698                - Top-level modules.
699                - FLINT (this should eventually become its own library)
700    
701    Notes:
702    
703    I am not 100% happy with the way I separated the elaborator (and its
704    data structures) from FLINT.  Two instances of the same problem:
705    
706        1. Data structures contain certain fields that carry FLINT-specific
707           information.  I hacked around this using exn and the property list
708           module from smlnj-lib.  But the fact that there are middle-end
709           specific fields around at all is a bit annoying.
710    
711        2. The elaborator calculates certain FLINT-related information.  I tried
712           to make this as abstract as I could using functorization, but, again,
713           the fact that the elaborator has to perform calculations on behalf
714           of the middle-end at all is not nice.
715    
716        3. Having to used exn and property lists is unfortunate because it
717           weakens type checking.  The other alternative (parameterizing
718           nearly *everything*) is not appealing, though.
719    
720    I removed the "rebinding =" warning hack because due to the new organization
721    it was awkward to maintain it.  As a result, the compiler now issues some of
722    these warnings when compiling init.cmi during bootstrap compilation. On
723    the plus side, you also get a warning when you do, for example:
724       val op = = Int32.+
725    which was not the case up to now.
726    
727    I placed "assign" and "deref" into the _Core structure so that the
728    code that deals with the "lazy" keyword can find them there.  This
729    removes the need for having access to the primitive environment
730    during elaboration.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume
734    Date: 2001/08/13
735    Tag: blume-20010813-closures
736    Description:
737    
738    This fix was sent to us by Zhong Shao.  It is supposed to improve the
739    performance of certain loops by avoiding needless closure allocation.
740    
741    ----------------------------------------------------------------------
742    Name: Lal George
743    Date: 2001/07/31 10:03:23 EDT 2001
744    Tag: george-20010731-x86-fmalloc
745    Description: Fixed bug in x86 calls
746    
747        There was a bug where call instructions would mysteriously
748        vanish. The call instruction had to be one that returned
749        a floating point value.
750    
751    ----------------------------------------------------------------------
752    Name: Lal George
753    Date: 2001/07/19 16:36:29 EDT 2001
754    Tag: george-20010719-simple-cells
755    Description:
756    
757    I have dramatically simplified the interface for CELLS in MLRISC.
758    
759    In summary, the cells interface is broken up into three parts:
760    
761      1. CellsBasis : CELLS_BASIS
762    
763            CellsBasis is a top level structure and common for all
764            architectures.  it contains the definitions of basic datatypes
765            and utility  functions over these types.
766    
767      2. functor Cells() : CELLS
768    
769            Cells generates an interface for CELLS that incorporates the
770            specific resources on the target architecture, such as the
771            presence of special register classes, their number and size,
772            and various useful substructures.
773    
774      3. <ARCH>CELLS
775    
776            e.g. SparcCells: SPARCCELLS
777    
778            <ARCH>CELLS usually contains additional bindings for special
779            registers  on the architecture, such as:
780    
781                    val r0 : cell           (* register zero *)
782                    val y : cell            (* Y register *)
783                    val psr : cell          (* processor status register *)
784                    ...
785    
786            The structure returned by applying the Cells functor is opened
787            in this interface.
788    
789    The main implication of all this is that the datatypes for cells is
790    split between CellsBasis and CELLS -- a fairly simple change for user
791    code.
792    
793    In the old scheme the CELLS interface had a definitional binding of
794    the form:
795    
796            signature CELLS = sig
797    
798               structure CellsBasis = CellsBasis
799    
800               ...
801    
802            end
803    
804    With all the sharing constraints that goes on in MLRISC, this old
805    design  quickly leads to errors such as:
806    
807            "structure definition spec inside of sharing ... "
808    
809    
810    and appears to require an unacceptable amount of sharing and where
811    constraint hackery.
812    
813    I think this error message (the interaction of definitional specs and
814    sharing) requires more explanation on our web page.
815    
816    ----------------------------------------------------------------------
817    Name: Matthias Blume
818    Date: 2001/07/19 15:00:00 EDT
819    Tag: blume-20010719-libreorg
820    Description:
821    
822    This update puts together a fairly extensive but straightforward change
823    to the way the libraries that implement the interactive system are
824    organized:
825    
826       The biggest change is the elimination of structure Compiler.  As a
827       replacement for this structure, there is now a CM library
828       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
829       that exports all the substructures of the original structure Compiler
830       directly.  So instead of saying Compiler.Foo.bar one now simply
831       says Foo.bar.  (The CM libraries actually export a collection of
832       structures that is richer than the collection of substructures of
833       structure Compiler.)
834    
835       To make the transition smooth, there is a separate library called
836       $smlnj/compiler/compiler.cm which puts together and exports the
837       original structure Compiler (or at least something very close to it).
838    
839       There are five members of the original structure Compiler
840       that are not exported directly but which instead became members
841       of a new structure Backend (described by signature BACKEND).  These are:
842       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
843       Interact (: INTERACT), structure Machine (: MACHINE), and val
844       architecture (: string).
845    
846       Structure Compiler.Version has become structure CompilerVersion.
847    
848       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
849       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
850       or x86, respectively.
851       Each of these exports the same frontend structures that
852       $smlnj/compiler.cm exports.  But they do not have a structure Backend
853       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
854       Hppa, PPC, Sparc, or X86, respectively.
855    
856       Library $smlnj/compiler/all.cm exports the union of the exports of
857       $smlnj/compiler/<arch>.cm
858    
859       There are no structures <Arch>Compiler anymore, use
860       $smlnj/compiler/<arch>.cm instead.
861    
862       Library host-compiler-0.cm is gone.  Instead, the internal library
863       that instantiates CM is now called cm0.cm.  Selection of the host
864       compiler (backend) is no longer done here but. (Responsibility for it
865       now lies with $smlnj/compiler/current.cm.  This seems to be more
866       logical.)
867    
868       Many individual files have been moved or renamed.  Some files have
869       been split into multiple files, and some "dead" files have been deleted.
870    
871    Aside from these changes to library organization, there are also changes
872    to the way the code itself is organized:
873    
874       Structure Binfile has been re-implemented in such a way that it no
875       longer needs any knowledge of the compiler.  It exclusively deals
876       with the details of binfile layout.  It no longer invokes the
877       compiler (for the purpose of creating new prospective binfile
878       content), and it no longer has any knowledge of how to interpret
879       pickles.
880    
881       Structure Compile (: COMPILE) has been stripped down to the bare
882       essentials of compilation.  It no longer deals with linking/execution.
883       The interface has been cleaned up considerably.
884    
885       Utility routines for dealing with linking and execution have been
886       moved into their own substructures.
887    
888       (The ultimate goal of these changes is to provide a light-weight
889       binfile loader/linker (at least for, e.g., stable libraries) that
890       does not require CM or the compiler to be present.)
891    
892    CM documentation has been updated to reflect the changes to library
893    organization.
894    
895    ----------------------------------------------------------------------
896    Name: Matthias Blume
897    Date: 2001/07/10 17:30:00 EDT
898    Tag: Release_110_34
899    Description:
900    
901    Minor tweak to 110.34 (re-tagged):
902    
903      - README.html file added to CVS repository
904      - runtime compiles properly under FreeBSD 3.X and 4.X
905    
906    ----------------------------------------------------------------------
907    Name: Matthias Blume
908    Date: 2001/07/10 17:30:00 EDT
909    Tag: Release_110_34
910    Description:
911    
912    New version number (110.34). New bootfiles.
913    
914    ----------------------------------------------------------------------
915    Name: Matthias Blume
916    Date: 2001/07/09 16:00:00 EDT
917    Tag: blume-20010709-more-varargs
918    Description:
919    
920    I changed the handling of varargs in ml-nlffigen again:
921    The ellipsis ... will now simply be ignored (with an accompanying warning).
922    
923    The immediate effect is that you can actually call a varargs function
924    from ML -- but you can't actually supply any arguments beyond the ones
925    specified explicitly.  (For example, you can call printf with its format
926    string, but you cannot pass additional arguments.)
927    
928    This behavior is only marginally more useful than the one before, but
929    it has the advantage that a function or, more importantly, a function
930    type never gets dropped on the floor, thus avoiding follow-up problems with
931    other types that refer to the offending one.
932    
933    ----------------------------------------------------------------------
934    Name: Matthias Blume
935    Date: 2001/07/09 11:25:00 EDT
936    Tag: blume-20010709-varargs
937    Description:
938    
939    1. ckit-lib.cm now exports structure Error
940    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
941       with a warning accompanied by a source location.  Moreover, it
942       merely skips the offending function or type and proceeds with the
943       rest of its work.u  As a result, one can safely feed C code containing
944       "..." to ml-nlffigen.
945    3. There are some internal improvements to CM, providing slightly
946       more general string substitutions in the tools subsystem.
947    
948    ----------------------------------------------------------------------
949    Name: Matthias Blume
950    Date: 2001/06/27 15:10:00 EDT
951    Tag: blume-20010627-concur
952    Description:
953    
954    Fixed a small bug in CM's handling of parallel compilation.
955    (You could observe the bug by Control-C-interrupting an ordinary
956    CMB.make or CM.stabilize and then attaching some compile servers.
957    The result was that all of a sudden the previously interrupted
958    compilation would continue on its own.  This was because of
959    an over-optimization: CM did not bother to clean out certain queues
960    when no servers were attached "anyway", resulting in the contents
961    of these queues to grab control when new servers did get attached.)
962    
963    There is also another minor update to the CM manual.
964    
965    ----------------------------------------------------------------------
966    Name: Matthias Blume
967    Date: 2001/06/26 16:15:00 EDT
968    Tag: blume-20010626-cmdoc
969    Description:
970    
971    Minor typo fixed in CM manual (syntax diagram for libraries).
972    
973    ----------------------------------------------------------------------
974    Name: Matthias Blume
975    Date: 2001/06/25 22:55:00 EDT
976    Tag: blume-20010625-x86pc
977    Description:
978    
979    Fixed a nasty bug in the X86 assembly code that caused signal
980    handlers to fail (crash) randomly.
981    
982    ----------------------------------------------------------------------
983    Name: Matthias Blume
984    Date: 2001/06/25 12:05:00 EDT
985    Tag: blume-20010625-nlffigen
986    Description:
987    
988    This update fixes a number of minor bugs in ml-nlffigen as reported by
989    Nick Carter <nbc@andrew.cmu.edu>.
990    
991      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
992      2. Default names for generated files are now derived from the name of
993         the C file *without its directory*.  In particular, this causes generated
994         files to be placed locally even if the C file is in some system directory.
995      3. Default names for generated signatures and structures are also derived
996         from the C file name without its directory.  This avoids silly things
997         like "structure GL/GL".
998         (Other silly names are still possible because ml-nlffigen does not do
999          a thorough check of whether generated names are legal ML identifiers.
1000          When in doubt, use command line arguments to force particular names.)
1001    
1002    ----------------------------------------------------------------------
1003    Name: Matthias Blume
1004    Date: 2001/06/21 12:25:00 EDT
1005    Tag: blume-20010621-eXene
1006    Description:
1007    
1008    eXene now compiles and (sort of) works again.
1009    
1010    The library name (for version > 110.33) is $/eXene.cm.
1011    
1012    I also added an new example in src/eXene/examples/nbody.  See the
1013    README file there for details.
1014    
1015    ----------------------------------------------------------------------
1016    Name: Matthias Blume
1017    Date: 2001/06/20 16:40:00 EDT
1018    Tag: blume-20010620-cml
1019    Description:
1020    
1021    CML now compiles and works again.
1022    
1023    Libraries (for version > 110.33):
1024    
1025      $cml/cml.cm            Main CML library.
1026      $cml/basis.cm          CML's version of $/basis.cm.
1027      $cml/cml-internal.cm   Internal helper library.
1028      $cml/core-cml.cm       Internal helper library.
1029      $cml-lib/trace-cml.cm  Tracing facility.
1030      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
1031    
1032    The installer (config/install.sh) has been taught how to properly
1033    install this stuff.
1034    
1035    ----------------------------------------------------------------------
1036    Name: Matthias Blume
1037    Date: 2001/06/19 17:55:00 EDT
1038    Tag: blume-20010619-instantiate
1039    Description:
1040    
1041    This un-breaks the fix for bug 1432.
1042    (The bug was originally fixed in 110.9 but I broke it again some
1043    time after that.)
1044    
1045    ----------------------------------------------------------------------
1046    Name: Matthias Blume
1047    Date: 2001/06/19 17:25:00 EDT
1048    Tag: blume-20010619-signals
1049    Description:
1050    
1051    This should (hopefully) fix the long-standing signal handling bug.
1052    (The runtime system was constructing a continuation record with an
1053    incorrect descriptor which would cause the GC to drop data on the floor...)
1054    
1055    ----------------------------------------------------------------------
1056    Name: Matthias Blume
1057    Date: 2001/06/15 15:05:00 EDT
1058    Tag: blume-20010615-moresparc
1059    Description:
1060    
1061    Here is a short late-hour update related to Sparc c-calls:
1062    
1063     -- made handling of double-word arguments a bit smarter
1064    
1065     -- instruction selection phase tries to collapse certain clumsily
1066        constructed ML-Trees; typical example:
1067    
1068            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
1069    
1070        This currently has no further impact on SML/NJ since mlriscGen does
1071        not seem to generate such patterns in the first place, and c-calls
1072        (which did generate them in the beginning) has meanwhile been fixed
1073        so as to avoid them as well.
1074    
1075    ----------------------------------------------------------------------
1076    Name: Matthias Blume
1077    Date: 2001/06/15 15:05:00 EDT
1078    Tag: blume-20010615-sparc
1079    Description:
1080    
1081    The purpose of this update is to provide an implementation of NLFFI
1082    on Sparc machines.
1083    
1084    Here are the changes in detail:
1085    
1086       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
1087       the Sparc implementation of the c-calls API.
1088       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
1089       for accessing the ML frame.  Thus, we have a real frame pointer and
1090       can freely modify %sp without need for an omit-frame-ptr phase.
1091       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
1092       this case.
1093       * ml-nlffigen has been taught to produce code for different architectures
1094       and calling conventions.
1095       * In a way similar to what was done in the x86 case, the Sparc
1096       backend uses its own specific extension to mltree.  (For example,
1097       it needs to be able to generate UNIMP instructions which are part
1098       of the calling convention.)
1099       * ml-nlffi-lib was reorganized to make it more modular (in particular,
1100       to make it easier to plug in new machine- and os-dependent parts).
1101    
1102    There are some other fairly unrelated bug fixes and cleanups as well:
1103    
1104       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
1105       that they properly share their libraries with existing SML/NJ libraries.
1106       * I fixed a minor cosmetic bug in CM, supressing certain spurious
1107       follow-up error messages.
1108       * Updates to CM/CMB documentation.
1109    
1110    TODO items:
1111    
1112       * MLRISC should use a different register as its asmTemp on the Sparc.
1113         (The current %o2 is a really bad choice because it is part of the
1114         calling conventions, so things might interfere in unexpected ways.)
1115    
1116    ----------------------------------------------------------------------
1117    Name: Matthias Blume
1118    Date: 2001/06/07
1119    Tag: blume-20010607-calls
1120    Description:
1121    
1122    A number of internal changes related to C calls and calling conventions:
1123    
1124    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
1125       number of bytes popped implicitly (by the callee).  In most cases
1126       this field is 0 but on x86/win32 it is some non-zero value.  This
1127       is information provided for the benefit of the "omit-frameptr" pass.
1128    2. The CALL instruction on the x86 carries a similar "pops" field.
1129       The instruction selection phase copies its value from the ML-Tree
1130       CALL statement.
1131    3. On all other architectures, the instruction selection phase checks
1132       whether "pops=0" and complains if not.
1133    4. The c-calls implementation for x86 now accepts two calling conventions:
1134       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
1135       up after the call and pops is set to 0.  For "stdcall", the caller
1136       does nothing, leaving the cleanup to the callee; pops is set to
1137       the number of bytes that were pushed onto the stack.
1138    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1139       distinguish between "ccall" and "stdcall".
1140    6. The UNIMP instruction has been added to the supported Sparc instruction
1141       set. (This is needed for implementing the official C calling convention
1142       on this architecture.)
1143    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1144       work with the latest CM.
1145    
1146    ----------------------------------------------------------------------
1147    Name: Matthias Blume
1148    Date: 2001/06/05 15:10:00 EDT
1149    Tag: blume-20010605-cm-index
1150    Description:
1151    
1152    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1153    
1154    1. CM can now generate "index files".  These are human-readable files
1155       that list on a per-.cm-file basis each toplevel symbol defined or
1156       imported.  The location of the index file for
1157       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1158       To enable index-file generation, set CM.Control.generate_index to true
1159       or export an environment-symbol: export CM_GENERATE_INDEX=true.
1160    
1161       The CM manual has been updated accordingly.
1162    
1163    2. I made some slight modifications to the c-calls API in MLRISC.
1164    
1165         a) There is now a callback to support saving/restoring of
1166            dedicated but caller-save registers around the actual call
1167            instruction.
1168         b) One can optionally specify a comment-annotation for the
1169            call instruction.
1170    
1171    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1172       (For example, the comment annotation shows the C prototype of
1173        the function being called.)
1174    
1175    ----------------------------------------------------------------------
1176    Name: Matthias Blume
1177    Date: 2001/06/01 13:30:00 EDT
1178    Tag: blume-20010601-nlffi-cleanup
1179    Description:
1180    
1181    This is mostly a cleanup of MLFFI stuff:
1182    
1183       - some signature files have been put into a more exposed place
1184       - the ugly 'f type parameter is gone (simplifies types tremendously!)
1185       - ml-nlffigen changed accordingly
1186       - tutorial updated
1187    
1188    Other changes:
1189    
1190       - author's affiliation in CM manual(s) updated
1191       - some more recognized keywords added to Allen's sml.sty
1192    
1193    ----------------------------------------------------------------------
1194    Name: Matthias Blume
1195    Date: 2001/05/25 15:30:00 EDT
1196    Tag: blume-20010525-iptr
1197    Description:
1198    
1199      - put the official 110.33-README (as it appears on the ftp server) under
1200        CVS
1201      - fixed a small bug related to incomplete pointer types in
1202        ml-nlffigen
1203      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1204        (it does not need the 'f type parameter)
1205    
1206    ----------------------------------------------------------------------
1207    Name: Matthias Blume
1208    Date: 2001/05/23 14:30:00 EDT
1209    Tag: Release_110_33
1210    Description:
1211    
1212    New version number (110.33).  New bootfiles.
1213    
1214    ----------------------------------------------------------------------
1215    Name: Matthias Blume
1216    Date: 2001/05/22 18:06:00 EDT
1217    Tag: blume-20010522-targets
1218    Description:
1219    
1220    Made install.sh use file config/targets.customized if it exists, falling
1221    back to config/targets if it doesn't.  This way one can have a customized
1222    version of the targets file without touching the "real thing", thus
1223    eliminating the constant fear of accidentally checking something bogus
1224    back into the CVS repository...  (File config/targets.customized must
1225    not be added to the repository!)
1226    
1227    ----------------------------------------------------------------------
1228    Name: Matthias Blume
1229    Date: 2001/05/22 16:30:00 EDT
1230    Tag: blume-20010522-minitut
1231    Description:
1232    
1233    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1234       struct returns.
1235    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
1236       incomplete, preliminary documentation for NLFFI.
1237    
1238    ----------------------------------------------------------------------
1239    Name: Matthias Blume
1240    Date: 2001/05/14 11:30:00 EDT
1241    Tag: blume-20010514-script
1242    Description:
1243    
1244    Some bugs in install script fixed.
1245    
1246    In addition to that I also made a slight change to the NLFFI API:
1247    Functors generated by ml-nlffigen now take the dynamic library as a
1248    straight functor argument, not as a suspended one.  (The original
1249    functor code used to force the suspension right away anyway, so there
1250    was nothing gained by this complication of the interface.)
1251    
1252    ----------------------------------------------------------------------
1253    Name: Matthias Blume
1254    Date: 2001/05/11 14:35:00 EDT
1255    Tag: blume-20010511-ml-nlffi
1256    Description:
1257    
1258    I finally took the plunge and added my new FFI code to the main
1259    repository.  For x86-linux it is now ready for prime-time.
1260    
1261    There are two new subdirectories of "src":
1262    
1263      - ml-nlffi-lib:
1264           The utility library for programs using the FFI interface.
1265           Here is the implementation of $/c.cm and its associated low-level
1266           partners $/c-int.cm and $/memory.cm.
1267      - ml-nlffigen:
1268           A stand-alone program for generating ML glue code from C source
1269           code.
1270    
1271    Building ml-nlffigen requires $/ckit-lib.cm.
1272    
1273    The config/install.sh script has been updates to do the Right Thing
1274    (hopefully).
1275    
1276    Notice that the source tree for the C-Kit will not be put under "src"
1277    but directly under the installation root directory.  (This is the
1278    structure that currently exists on the CVS server when you check out
1279    module "sml".)  Fortunately, config/install.sh knows about this oddity.
1280    
1281    Bugs: No documentation yet.
1282    
1283    ----------------------------------------------------------------------
1284    Name: Matthias Blume
1285    Date: 2001/05/09 16:35:00 EDT
1286    Tag: blume-20010509-cpscontract
1287    Description:
1288    
1289    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
1290    wrapper/unwrapper elimination did not decrement usage counts and some
1291    dead variables got overlooked by the dead-up logic.)
1292    
1293    ----------------------------------------------------------------------
1294    Name: Lal George
1295    Date: 2001/05/08  17:26:09 EDT
1296    Tag: george-20010508-omit-frameptr
1297    Description:
1298    
1299    Changes to implement the omit-frame-pointer optimization to support
1300    raw C calls. For now, there is only support on the Intel x86, but
1301    other architectures will follow as more experience is gained with this.
1302    
1303    
1304    ----------------------------------------------------------------------
1305    Name: Matthias Blume
1306    Date: 2001/05/07 14:40:00 EDT
1307    Tag: blume-20010507-proxies
1308    Description:
1309    
1310    I made into "proxy libraries" all libraries that qualify for such a
1311    change.  (A qualifying library is a library that has another library or
1312    groups as its sole member and repeats that member's export list
1313    verbatim.  A proxy library avoids this repetition by omitting its export
1314    list, effectively inheriting the list that its (only) member exports.
1315    See the CM manual for more explanation.)
1316    The main effect is that explicit export lists for these libraries
1317    do not have to be kepts in sync, making maintenance a bit easier.
1318    
1319    I also added copyright notices to many .cm-files.
1320    
1321    Last but not least, I made a new set of bootfiles.
1322    
1323    ----------------------------------------------------------------------
1324    Name: Matthias Blume
1325    Date: 2001/05/04 17:00:00 EDT
1326    Tag: blume-20010504-cm-lsplit
1327    Description:
1328    
1329    0. John merged pending changes to $/smlnj-lib.cm
1330    
1331    1. Allen's previous change accidentally backed out of one of Lal's
1332       earlier changes.  I undid this mistake (re-introducing Lal's change).
1333    
1334    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1335       within the compiler where applicable.  There is some code simplification
1336       because of that.
1337    
1338    3. The "split" phase (in FLINT) is now part of the default list of phases.
1339       Compiler.Control.LambdaSplitting.* can be used to globally control the
1340       lambda-splitting (cross-module-inlining) engine.  In addition to that,
1341       it can now also be controlled on a per-source basis: CM has been taught
1342       a new tool parameter applicable to ML source files.
1343    
1344       - To turn lambda-splitting off completely:
1345            local open Compiler.Control.LambdaSplitting in
1346                val _ = set Off
1347            end
1348       - To make "no lambda-splitting" the global default (but allow per-source
1349         overriding); this is the initial setting:
1350            local open Compiler.Control.LambdaSplitting in
1351                val _ = set (Default NONE)
1352            end
1353       - To make "lambda-splitting with aggressiveness a" the global default
1354         (and allow per-source overriding):
1355            local open Compiler.Control.LambdaSplitting in
1356                val _ = set (Default (SOME a))
1357            end
1358    
1359       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
1360         write (in the respective .cm-file):
1361            a.sml (lambdasplitting:off)
1362       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
1363            a.sml (lambdasplitting:on)
1364       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1365         <a> is a decimal non-negative integer):
1366            a.sml (lambdasplitting:<a>)
1367       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
1368            a.sml (lambdasplitting:infinity)
1369       - To use the global default for a.sml:
1370            a.sml (lambdasplitting:default)
1371         or simply
1372            a.sml
1373    
1374    ----------------------------------------------------------------------
1375    Name: Allen Leung
1376    Date: 2001/05/04 01:57:00 EDT
1377    Tag: leunga-20010504-sync
1378    Description:
1379    
1380      MLRISC features.
1381    
1382      1. Fix to CMPXCHG instructions.
1383      2. Changed RA interface to allow annotations in callbacks.
1384      3. Added a new method to the stream interface to allow annotations updates.
1385    
1386    ----------------------------------------------------------------------
1387    Name: Matthias Blume
1388    Date: 2001/05/01 11:45:00 EDT
1389    Tag: blume-20010501-pcedittmp
1390    Description:
1391    
1392    Changed install.sh to use the current working directory instead of
1393    /usr/tmp for a temporary file (pcedittmp).  The previous choice
1394    of /usr/tmp caused trouble with MacOS X because of file premission
1395    problems.
1396    
1397    ----------------------------------------------------------------------
1398    Name: Matthias Blume
1399    Date: 2001/04/20 11:10:00 EDT
1400    Tag: blume-20010420-inMLflag
1401    Description:
1402    
1403     - added vp_limitPtrMask to vproc-state.h
1404       (for use by the raw-C-calls mechanism to implement proper interrupt
1405        handling)
1406     - made the ML compiler aware of various data-structure offsets so it
1407       can generate code for accessing the vp_inML flag and vp_limitPtrMask
1408     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
1409       raw C-calls
1410    
1411    ----------------------------------------------------------------------
1412    Name: Lal George
1413    Date: 2001/04/20 09:15:28 EDT
1414    Tag: george-20010420-macosX
1415    Description:
1416    
1417     - Changes to port to Mac OS X; Darwin.
1418    
1419     - In the process I found that sqrt was broken on the PPC, because the
1420       fsqrt instruction is not implemented.
1421    
1422    ----------------------------------------------------------------------
1423    Name: Matthias Blume
1424    Date: 2001/04/18 12:45:00 EDT
1425    Tag: blume-20010418-ccalls
1426    Description:
1427    
1428     - fixed two off-by-4 errors in the x86-specific c-calls implementation
1429       (this bug prevented structure arguments containing pointers from being
1430        passed correctly)
1431     - changed the raw-C-call code in mlriscGen.sml in such a way that
1432       structure arguments are represented as a pointer to the beginning
1433       of the structure (instead of having a series of synthesized arguments,
1434       one for each structure member)
1435    
1436     - made makeml script's verbosity level configurable via environment
1437       variable (MAKEML_VERBOSITY)
1438    
1439     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1440       in rawmem-x86.sml; we are now using the real thing
1441    
1442    ----------------------------------------------------------------------
1443    Name: Matthias Blume
1444    Date: 2001/03/22 16:25:00 EST
1445    Tag: blume-20010322-bootfiles
1446    Description:
1447    
1448    Created a new set of bootfiles (for your automatic installation convenience).
1449    
1450    ----------------------------------------------------------------------
1451    Name: Matthias Blume
1452    Date: 2001/03/22 15:10:00 EST
1453    Tag: blume-20010322-rawmem-parcm
1454    Description:
1455    
1456    1. All "raw memory access" primitives for the new FFI are implemented now
1457       (at least on the x86).
1458    2. Some further cleanup of CM's parallel make mechanism.
1459    
1460    ----------------------------------------------------------------------
1461    Name: Matthias Blume
1462    Date: 2001/03/19 17:53:00 EST
1463    Tag: blume-20010319-parallel
1464    Description:
1465    
1466    Parallel make (using compile servers) now works again.
1467    
1468    To this end, CM.stabilize and CMB.make have been modified to work in
1469    two passes when compile servers are attached:
1470       1. Compile everything, do not perform stabilization; this pass
1471          uses compile servers
1472       2. Stabilize everything; this pass does not use compile servers
1473    If there are no compile servers, the two passes are combined into one
1474    (as before).  Splitting the passes increases the inherent parallelism
1475    in the dependency graph because the entire graph including all
1476    libraries is available at the same time.  This, in turn, improves
1477    server utilization.  The downside is that the master process will
1478    have to do some extra work after compilation is done (because for
1479    technical reasons it must re-read all the binfiles during stabilization).
1480    
1481    ----------------------------------------------------------------------
1482    Name: Matthias Blume
1483    Date: 2001/03/16 12:22:00 EST
1484    Tag: blume-20010316-bootfiles
1485    Description:
1486    
1487    Created a new set of bootfiles (for your automatic installation convenience).
1488    
1489    ----------------------------------------------------------------------
1490    Name: Matthias Blume
1491    Date: 2001/03/16 11:00:00 EST
1492    Tag: blume-20010316-MLTREE-fixup
1493    Description:
1494    
1495    This is a minor fixup for an (untagged) earlier commit by Allen.
1496    (A file was missing).
1497    
1498    ----------------------------------------------------------------------
1499    Name: Allen Leung
1500    Date: Mon Mar  5 18:54:57 EST 2001
1501    Tag: leunga-20010305-cut-support
1502    
1503    1. New support for alternative control-flow in MLTREE.
1504       Currently we support
1505    
1506          FLOW_TO(CALL ...., [k1,...,kn])
1507    
1508       This is needed for 'cuts to' in C-- and try/handle-like constructs
1509       in Moby
1510    
1511       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
1512    
1513    2. Register Allocator
1514    
1515       Changes in interface [from Fermin, John]
1516    
1517    3. Alpha 8-bit SLL support [Fermin]
1518    
1519    4. All architectures
1520    
1521       A new module (ClusterExpandCopies) for expanding parallel copies.
1522    
1523    ----------------------------------------------------------------------
1524    Name: Allen Leung
1525    Date: 2001/02/27 23:07:00 EST
1526    Tag: leunga-20010227-minor-stuff
1527    
1528    1. Alpha bug fix for CMOVNE
1529    2. Handle mltree COND(..,FCMP ...,...)
1530    3. Bug fix in simplifier
1531    
1532    ----------------------------------------------------------------------
1533    Name: Matthias Blume
1534    Date: 2001/01/30 17:50:00 EST
1535    Tag: blume-20010130-sync
1536    Description:
1537    
1538    This is just a minor update to sync my devel branch with the main brach.
1539    The only visible change is the addition of some README files.
1540    
1541    ----------------------------------------------------------------------
1542    Name: Matthias Blume
1543    Date: 2001/01/12 23:30:00 JST
1544    Tag: blume-20010112-bootfiles
1545    Description:
1546    
1547    Made a new set of bootfiles that goes with the current state of the
1548    repository.
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2001/01/12 21:20:00 JST
1553    Tag: blume-20010112-sync
1554    Description:
1555    
1556    I am just flushing out some minor changes that had accumulated in
1557    my private branch in order to sync with the main tree.  (This is
1558    mainly because I had CVS trouble when trying to merge _into_ my
1559    private branch.)
1560    
1561    Most people should be completely unaffected by this.
1562    
1563    ----------------------------------------------------------------------
1564    Name: Allen Leung
1565    Date: Thu Jan 11 21:03:00 EST 2001
1566    Tag: leunga-20010111-labexp=mltree
1567    Description:
1568    
1569    1.  Removed the type LabelExp and replace it by MLTree.
1570    2.  Rewritten mltree-simplify with the pattern matcher tool.
1571    3.  There were some bugs in alpha code generator which would break
1572        64-bit code generation.
1573    4.  Redo the tools to generate code with the
1574    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
1575        are now generated by perl scripts.
1576    
1577    ----------------------------------------------------------------------
1578    Name: Matthias Blume
1579    Date: 2001/01/10 21:55:00 JST
1580    Tag: blume-20010110-rcc
1581    Description:
1582    
1583    The RCC stuff now seems to work (but only on the x86).
1584    This required hacking of the c-calls interface (and -implementation) in
1585    MLRISC.
1586    
1587    Normal compiler users should be unaffected.
1588    
1589    ----------------------------------------------------------------------
1590    Name: Matthias Blume
1591    Date: 2001/01/09 01:20:00 JST
1592    Tag: blume-20010109-rcc
1593    Description:
1594    
1595    This is a fairly big patch, flushing out a large number of pending
1596    changes that I made to my development copy over the last couple of days.
1597    
1598    Of practical relevance at this moment is a workaround for a pickling
1599    bug that Allen ran into the other day.  The cause of the bug itself is
1600    still unknown and it might be hard to fix it properly, but the
1601    workaround has some merits of its own (namely somewhat reducing pickling
1602    overhead for certain libraries).  Therefore, I think this solution should
1603    be satisfactory at this time.
1604    
1605    The rest of the changes (i.e., the vast majority) has to do with my
1606    ongoing efforts of providing direct support for C function calls from
1607    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
1608    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
1609    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
1610    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
1611    handle RCC.
1612    
1613    The last part (the changes to mlriscGen.sml) are still known to be
1614    wrong on the x86 and not implemented on all other architectures.  But
1615    the infrastructure is in place. I had to change a few functor
1616    signatures in the backend to be able to route the CCalls interface
1617    from MLRISC there, and I had to specialize the mltree type (on the
1618    x86) to include the necessary extensions. (The extensions themselves
1619    were already there and redy to go in MLRISC/x86).
1620    
1621    Everything should be very happy as soon as someone helps me with
1622    mlriscGen.sml...
1623    
1624    In any case, nothing of this should matter to anyone as long as the
1625    new primop is not being used (which is going to be the case unless you
1626    find it where I hid it :). The rest of the compiler is completely
1627    unaffected.
1628    
1629    ----------------------------------------------------------------------
1630    Name: Matthias Blume
1631    Date: 2001/01/05 00:30:00 JST
1632    Tag: blume-20010105-primops
1633    Description:
1634    
1635    Added some experimental support for work that I am doing right now.
1636    These changes mostly concern added primops, but there is also a new
1637    experimental C library in the runtime system (but currently not enabled
1638    anywhere except on Linux/X86).
1639    
1640    In the course of adding primops (and playing with them), I discovered that
1641    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
1642    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
1643    To recover, I made type information in INL_PRIM mandatory and changed
1644    prim.sml as well as built-in.sml accordingly.  The InLine structure now
1645    has complete, correct type information (i.e., no bottom types).
1646    
1647    Since all these changes mean that we need new binfiles, I also bumped the
1648    version number to 110.32.1.
1649    
1650    ----------------------------------------------------------------------
1651    Name: Matthias Blume
1652    Date: 2000/12/30 22:10:00 JST
1653    Tag: blume-20001230-various
1654    Description:
1655    
1656    Added proxy libraries for MLRISC and let MLRISC libraries refer
1657    to each other using path anchors.  (See CM manual for explanation.)
1658    
1659    Updated CM documentation.
1660    
1661    Fixed some bugs in CM.
1662    
1663    Implemented "proxy" libraries (= syntactic sugar for CM).
1664    
1665    Added "-quiet" option to makeml and changed runtime system accordingly.
1666    
1667    Added cleanup handler for exportML to reset timers and compiler stats.
1668    
1669    ----------------------------------------------------------------------
1670    Name: Lal George
1671    Date: 2000/12/22 22:22:58 EST 2000
1672    Tag: Release_110_32
1673    Description:
1674    
1675            Infinite precision used throughout MLRISC.
1676            see MLRISC/mltree/machine-int.sig
1677    
1678    ----------------------------------------------------------------------
1679    Name: Matthias Blume
1680    Date: 2000/12/22 23:16:00 JST
1681    Tag: blume-20001222-warn
1682    Description:
1683    
1684    Corrected wording and formatting of some CM warning message which I
1685    broke in my previous patch.
1686    
1687    ----------------------------------------------------------------------
1688    Name: Matthias Blume
1689    Date: 2000/12/22 21:20:00 JST
1690    Tag: blume-20001222-anchorenv
1691    Description:
1692    
1693    Fixed CM's handling of anchor environments in connection with CMB.make.
1694    
1695    ----------------------------------------------------------------------
1696    Name: Matthias Blume
1697    Date: 2000/12/22 13:15:00 JST
1698    Tag: blume-20001222-cleanup
1699    Description:
1700    
1701    Removed src/cm/ffi which does not (and did not) belong here.
1702    
1703    ----------------------------------------------------------------------
1704    Name: Matthias Blume
1705    Date: 2000/12/21 23:55:00 JST
1706    Tag: blume-20001221-exn
1707    Description:
1708    
1709    Probably most important: CM no longer silently swallows all exceptions
1710    in the compiler.
1711    Plus: some other minor CM changes.  For example, CM now reports some
1712    sizes for generated binfiles (code, data, envpickle, lambdapickle).
1713    
1714    ----------------------------------------------------------------------
1715    Name: Matthias Blume
1716    Date: 2000/12/15 00:01:05 JST
1717    Tag: blume-20001215-dirtool
1718    Description:
1719    
1720    - "dir" tool added.
1721    - improvements and cleanup to Tools structure
1722    - documentation updates
1723    
1724    ----------------------------------------------------------------------
1725    Name: Allen Leung
1726    Date: Thu Dec 14 03:45:24 EST 2000
1727    Description:
1728    Tag:  leunga-20001214-int-inf
1729    Description:
1730    
1731       In IntInf, added these standard functions, which are missing from our
1732    implementation:
1733    
1734        andb : int * int -> int
1735        xorb : int * int -> int
1736        orb  : int * int -> int
1737        notb : int -> int
1738         <<   : int * word -> int
1739        ~>>  : int * word -> int
1740    
1741       Not tested, I hope they are correct.
1742    
1743    ----------------------------------------------------------------------
1744    Name: Allen Leung
1745    Date: Fri Dec  8 19:23:26 EST 2000
1746    Description:
1747    Tag:  leunga-20001208-nowhere
1748    Description:
1749    
1750      Slight improvements to the 'nowhere' tool to handle OR-patterns,
1751    to generate better error messages etc.  Plus a brief manual.
1752    
1753    ----------------------------------------------------------------------
1754    Name: Lal George
1755    Date: 2000/12/08 09:54:02 EST 2000
1756    Tag: Release_110_31
1757    Description:
1758    
1759    - Version 110.31
1760    ----------------------------------------------------------------------
1761    Name: Allen Leung
1762    Date: Thu Dec  7 22:01:04 EST 2000
1763    Tag:  leunga-20001207-cell-monster-hack
1764    Description:
1765    
1766    Major MLRISC internal changes.  Affect all clients.
1767    Summary:
1768    
1769    1.  Type CELLS.cell = int is now replaced by a datatype.
1770        As a result, the old regmap is now gone.  Almost all interfaces
1771        in MLRISC change as a consequence.
1772    
1773    2.  A new brand version of machine description tool (v3.0) that generates
1774        modules expecting the new interface.  The old version is removed.
1775    
1776    3.  The RA interface has been further abstracted into two new functors.
1777        RISC_RA and X86RA.  These functors have much simpler interfaces.
1778        [See also directory MLRISC/demo.]
1779    
1780    4.  Some other new source->source code generation tools are available:
1781    
1782        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1783        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1784           I use this tool to generate the peephole optimizers---with the new
1785           cell type changes, peephole rules are becoming difficult to write
1786           without conditional pattern matching.
1787    
1788    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
1789        cover the entire MLRISC source tree so many things broke.
1790    
1791    6.  CM files have been moved to the subdirectory MLRISC/cm.
1792        They are moved because there are a lot of them and they clutter up the
1793        root dir.
1794    
1795    7.  More detailed documentation to come...
1796    
1797        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1798        a makeml -rebuild first.  This is because of other other
1799        changes that Matthias has made (see below).
1800    
1801    
1802    ----------------------------------------------------------------------
1803    Name: Matthias Blume
1804    Date: 2000/11/30 23:12:00 JST
1805    Tag: blume-20001130-filereorg
1806    Description:
1807    
1808    Some manual updates and some file reorganizations in CM.
1809    
1810    ----------------------------------------------------------------------
1811    Name: Matthias Blume
1812    Date: 2000/11/24 17:45:00 JST
1813    Tag: blume-20001124-link
1814    Description:
1815    
1816    Drastically improved link traversal code for the case that the dynamic
1817    value was already loaded at bootstrap time.  As a result, CM and CMB
1818    now both load blazingly fast -- even on a very slow machine.  Also,
1819    memory consumption has been further reduced by this.
1820    
1821    Warning: The format of the PIDMAP file has changed.  THerefore, to
1822    bootstrap you have to do this:
1823    
1824    1. Run CMB.make
1825    2. Make a symbolic link for the boot directory:
1826         ln -s sml.boot.ARCH-OS xxx
1827    3. "Rebuild" the boot directory:
1828         ./makeml -boot xxx -rebuild sml ; rm xxx
1829    4. Boot normally:
1830          ./makeml
1831    
1832    ----------------------------------------------------------------------
1833    Name: Matthias Blume
1834    Date: 2000/11/21 21:20:00 JST
1835    Tag: blume-20001121-tools
1836    Description:
1837    
1838    Continued hacking on autoloading problem -- with success this time.
1839    Also changed tool-plugin mechanism.  See new CM manual.
1840    
1841    ----------------------------------------------------------------------
1842    Name: Matthias Blume
1843    Date: 2000/11/19 14:30:00 JST
1844    Tag:  blume-20001119-autoload
1845    Description:
1846    
1847    Some hacking to make autoloading faster.  Success for CMB, no success
1848    so far for CM.  There is a reduced structure CM' that autoloads faster.
1849    (This is a temporary, non-documented hack to be eliminated again when
1850    the general problem is solved.)
1851    
1852    ----------------------------------------------------------------------
1853    Name: Matthias Blume
1854    Date: 2000/11/17 14:10:00 JST
1855    Tag: blume-20001117-pickle-lib
1856    Description:
1857    
1858    1. Eliminated comp-lib.cm
1859    2. Made pickle-lib.cm
1860    3. Eliminated all uses of intset.sml (from comp-lib.cm)
1861    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1862       equivalent constructs from smlnj-lib.cm (INtHashTable).
1863    5. Point 4. also goes for those uses of intmap.* in MLRISC.
1864       Duplicated intmap modules thrown out.
1865    6. Hunted down all duplicated SCC code and replaced it with
1866       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1867    7. Rewrote Feedback module.
1868    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
1869       should be thrown out and equivalent modules from smlnj-lib.cm
1870       should be used (IntRedBlackSet, IntListSet, ...).
1871    
1872    Confirmed that compiler compiles to fixpoint.
1873    
1874    ----------------------------------------------------------------------
1875    Name: Allen Leung
1876    Date: 2000/11/10 18:00:00
1877    Tag: leunga-20001110-new-x86-fp
1878    
1879    A new x86 floating point code generator has been added.
1880    By default this is turned off.  To turn this on, do:
1881    
1882        CM.autoload "$smlnj/compiler.cm";
1883        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1884    
1885    Changes:
1886    
1887    1.  Changed FTAN to FPTAN so that the assembly output is correct.
1888    2.  Changed the extension callback for FTANGENT to generate:
1889    
1890              fptan
1891              fstp  %st(0)
1892        instead of
1893              fptan
1894              fstpl ftempmem
1895    
1896    3.  Numerous assembly fixes for x86.
1897    
1898    5.  Cleaned up the machine code output module x86/x86MC.sml and added
1899        support for a whole bunch of instructions and addressing modes:
1900    
1901          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
1902          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
1903          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
1904          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1905          fxch %st(n)
1906          fld %st(n)
1907          fst %st(n)
1908          fst mem
1909          fstp %st(n)
1910          fucom %st(n)
1911          fucomp %st(n)
1912    
1913        All these are now generated when the fast fp mode is turned on.
1914    
1915    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1916    
1917    ----------------------------------------------------------------------
1918    Name: Matthias Blume
1919    Date: 2000/11/09 11:20:00 JST
1920    Tag: blume-20001109-scc
1921    Description:
1922    
1923    Eliminated some code duplication:
1924    
1925    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1926       (Otherwise the functor is useless.)
1927    2. Used GraphSCCFn where SCCUtilFun was used previously.
1928    3. Got rid of SCCUtilFun (in comp-lib.cm).
1929    
1930    ----------------------------------------------------------------------
1931    Name: Lal George
1932    Date: 2000/11/06 09:02:21 EST 2000
1933    Tag: Release_110_30
1934    Description:
1935    
1936    - Version 110.30
1937    ----------------------------------------------------------------------
1938    Name: Matthias Blume
1939    Date: 2000/11/04 14:45:00
1940    Tag: blume-20001104-mlbuild
1941    Description:
1942    
1943    - Made ml-build faster on startup.
1944    - Documentation fixes.
1945    
1946    ----------------------------------------------------------------------
1947    Name: Matthias Blume
1948    Date: 2000/11/02 17:00:00 JST
1949    Tag: blume-20001102-condcomp
1950    Description:
1951    
1952    - Small tweaks to pickler -- new BOOTFILES!
1953    - Version bumped to 110.29.2.
1954    - Added conditional compilation facility to init.cmi (see comment there).
1955    ----------------------------------------------------------------------
1956    Name: Allen Leung
1957    Date: 2000/10/23 19:31:00
1958    Tag: leunga-20001023-demo-ra
1959    
1960    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1961    2. Test programs for the graph library updated
1962    3. Some new MLRISC demo programs added
1963    
1964    ----------------------------------------------------------------------
1965    Name: Matthias Blume
1966    Date: 2000/08/31 22:15:00 JST
1967    Tag: blume-20001017-errmsg
1968    Description:
1969    
1970    More error message grief: Where there used to be no messages, there
1971    now were some that had bogus error regions.  Fixed.
1972    
1973    ----------------------------------------------------------------------
1974    Name: Matthias Blume
1975    Date: 2000/08/31 17:30:00 JST
1976    Tag: blume-20001017-v110p29p1
1977    Description:
1978    
1979    I made a version 110.29.1 with new bootfiles.
1980    
1981    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1982              CM documentation changes and a small bugfix in CM's error reporting.
1983    
1984    ----------------------------------------------------------------------
1985    Name: Lal George
1986    Date: 2000/09/27 14:42:35 EDT
1987    Tag: george-20000927-nodestatus
1988    Description:
1989    
1990    Changed the type of the nodestatus, so that:
1991    
1992            SPILLED(~1)             is now SPILLED
1993            SPILLED(m) where m>=0   is now MEMREG(m)
1994            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1995    
1996    ----------------------------------------------------------------------
1997    Name: Matthias Blume
1998    Date: 2000/09/07 14:45:00 JST
1999    Tag: blume-20000907-cmerrmsg
2000    Description:
2001    
2002    Small tweak to CM to avoid getting ML syntax error messages twice.
2003    
2004    ----------------------------------------------------------------------
2005    Name: Matthias Blume
2006    Date: 2000/08/31 18:00:00 JST
2007    Tag: blume-20000831-cvsbootfiles
2008    Description:
2009    
2010    New URL for boot files (because the 110.29 files on the BL server do
2011    now work correctly with my updated install scripts for yacc and lex).
2012    
2013    ----------------------------------------------------------------------
2014    Name: Matthias Blume
2015    Date: 2000/08/08 12:33:00 JST
2016    Tag: blume-20000808-manual
2017    Description:
2018    
2019    Tiny update to CM manual.
2020    
2021    ----------------------------------------------------------------------
2022    Name: Allen Leung
2023    Date: 2000/08/7 19:31:00
2024    Tag: leunga-20000807-a-whole-bunch-of-stuff
2025    
2026      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
2027    and Mobdy.
2028    
2029    1.  x86
2030    
2031       a.  Fixes to peephole module by John and Dan.
2032       b.  Assembly fix to SETcc by Allen.
2033       c.  Fix to c-call by John.
2034       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
2035       e.  Instruction selection optimization to SETcc as suggested by John.
2036    
2037           For example,
2038    
2039            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
2040    
2041           should generate:
2042    
2043            MOVL a, x
2044            SUBL b, x
2045            SHRL 31, x
2046    
2047    2.  IR stuff
2048    
2049         A bunch of new DJ-graph related algorithms added.  These
2050         speed up SSA construction.
2051    
2052    3.  SSA + Scheduling
2053    
2054         Added code for SSA and scheduling to the repository
2055    
2056    ----------------------------------------------------------------------
2057    Name: Lal George
2058    Date: 2000/07/27 11:53:14 EDT
2059    
2060    Tag: lal-20000727-linux-ppc
2061    Description:
2062    
2063     Made changes to support Linux PPC.
2064     p.s. I have confirmation that the 110.29 boot files work fine.
2065    
2066    ----------------------------------------------------------------------
2067    Name: Matthias Blume
2068    Date: 2000/07/27 17:40:00 JST
2069    Tag: blume-20000727-scripts
2070    Description:
2071    
2072    !!!! WARNING !!!!
2073    You must recompile the runtime system!
2074    !!!! WARNING !!!!
2075    
2076    This is basically another round of script-enhancements:
2077    
2078    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
2079       and undefine CM preprocessor symbols.
2080    
2081    2. ml-build avoids generating a new heap image if it finds that the
2082       existing one is still ok.  (The condition is that no ML file had to
2083       be recompiled and all ML files are found to be older that the heap
2084       file.)
2085    
2086       To make this work smoothly, I also hacked the runtime system as
2087       well as SMLofNJ.SysInfo to get access to the heap image suffix
2088       (.sparc-solaris, ...) that is currently being used.
2089    
2090       Moreover, the signature of CM.mk_standalone has changed.  See the
2091       CM manual.
2092    
2093    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
2094       CM manual for details.)
2095    
2096    4. More CM manual updates:
2097        - all of the above has been documented.
2098        - there is now a section describing the (CM-related) command line
2099          arguments that are accepted by the "sml" command
2100    
2101    ----------------------------------------------------------------------
2102    Name: Matthias Blume
2103    Date: 2000/07/25 16:20:00 JST
2104    Tag: blume-20000725-makedepend
2105    Description:
2106    
2107    Added a script called ml-makedepend.  This can be used in makefiles
2108    for Unix' make in a way very similar to the "makedepend" command for
2109    C.
2110    
2111    The script internally uses function CM.sources.
2112    
2113    Synopsis:
2114    
2115        ml-makedepend [-f makefile] cmfile targetname
2116    
2117    The default for the makefile is "makefile" (or "Makefile" should
2118    "makefile" not exist).
2119    
2120    ml-makedepend adds a cmfile/targetname-specific section to this
2121    makefile (after removing the previous version of this section).  The
2122    section contains a single dependency specification with targetname on
2123    the LHS (targetname is an arbitrary name), and a list of files derived
2124    from the cmfile on the RHS.  Some of the files on the RHS are
2125    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
2126    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
2127    path names.  The makefile writer is responsible for making sure that
2128    these variables have correct at the time "make" is invoked.
2129    
2130    ----------------------------------------------------------------------
2131    Name: Matthias Blume
2132    Date: 2000/07/22 23:30:00 JST
2133    Tag: blume-20000722-urlupdate
2134    Description:
2135    
2136    Changed BOOT and config/srcarchiveurl to point to BL server:
2137    
2138        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
2139    
2140    ----------------------------------------------------------------------
2141    Name: Matthias Blume
2142    Date: 2000/07/18 18:00:00 JST
2143    Tag: blume-20000718-Version_110_29
2144    Description:
2145    
2146    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
2147    
2148    2. Updated config/version to 110.29
2149    
2150    3. Updated config/srcarchiveurl
2151    
2152    3. New boot files!
2153       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
2154    
2155    ----------------------------------------------------------------------
2156    Name: Matthias Blume
2157    Date: 2000/07/11 13:58:00 JST
2158    Tag: blume-20000711-doctypo
2159    Description:
2160    
2161    Fixed a few typos in CM manual.
2162    
2163    ----------------------------------------------------------------------
2164    Name: Allen Leung
2165    Date: 2000/06/15 00:38:00
2166    Tag: leunga-20000704-sparc-x86
2167    
2168    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
2169    2. fix to x86 RET bug [found by Dan Grossman]
2170    3. sparc assembly bug fix for ticc instructions [found by Fermin]
2171    
2172       Affects c-- and moby only
2173    
2174    ----------------------------------------------------------------------
2175    Name: Matthias Blume
2176    Date: 2000/07/04 15:26:00
2177    Tag: blume-20000704-trigger
2178    Description:
2179    
2180    1. Improvements to CM manual.
2181    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
2182       of getting a back-trace.  The function, when called, raises an
2183       internal exception which explicitly carries the full back-trace history,
2184       so it is unaffected by any intervening handle-raise pairs ("trivial"
2185       or not).  The interactive loop will print that history once it arrives
2186       at top level.
2187       Short of having all exceptions implicitly carry the full history, the
2188       recommended way of using this facility is:
2189         - compile your program with instrumentation "on"
2190         - run it, when it raises an exception, look at the history
2191         - if the history is "cut off" because of some handler, go and modify
2192           your program so that it explicitly calls BTrace.trigger
2193         - recompile (still instrumented), and rerun; look at the full history
2194    
2195    ----------------------------------------------------------------------
2196    Name: Matthias Blume
2197    Date: 2000/07/03 15:36:00 JST
2198    Tag: blume-20000702-manual
2199    Description:
2200    
2201    Small corrections and updates to CM manual.
2202    
2203    ----------------------------------------------------------------------
2204    Name: Matthias Blume
2205    Date: 2000/06/29 16:04:00 JST
2206    Tag: blume-20000629-yacctool
2207    Description:
2208    
2209    Changes:
2210    
2211    1. Class "mlyacc" now takes separate arguments to pass options to
2212       generated .sml- and .sig-files independently.
2213    2. Corresponding CM manual updates.
2214    3. BTrace module now also reports call sites.  (However, for loop clusters
2215       it only shows from where the cluster was entered.)  There are associated
2216       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
2217    
2218    ----------------------------------------------------------------------
2219    Name: Matthias Blume
2220    Date: 2000/06/27 16:51:00 JST
2221    Tag: blume-20000627-noweb
2222    Description:
2223    
2224    Changes:
2225    
2226     1. Implemented "subdir" and "witness" options for noweb tool.
2227        This caused some slight internal changes in CM's tool implementation.
2228     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
2229        some remaining issues from earlier path anchor changes.
2230     3. Updated CM manual accordingly.
2231    
2232     4. Changed implementation of back-tracing so that I now consider it
2233        ready for prime-time.
2234    
2235        In particular, you don't have to explicitly trigger the back-trace
2236        anymore.  Instead, if you are running BTrace-instrumented code and
2237        there is an uncaught exception (regardless of whether or not it was
2238        raised in instrumented code), the top-level evalloop will print
2239        the back-trace.
2240    
2241        Features:
2242    
2243          - Instrumented and uninstrumented code work together seemlessly.
2244            (Of course, uninstrumented code is never mentioned in actual
2245             back-traces.)
2246    
2247          - Asymptotic time- and space-complexity of instrumented code is
2248            equal to that of uninstrumented code.  (This means that
2249            tail-recursion is preserved by the instrumentation phase.)
2250    
2251          - Modules whose code has been instrumented in different sessions
2252            work together without problem.
2253    
2254          - There is no penalty whatsoever on uninstrumented code.
2255    
2256          - There is no penalty on "raise" expressions, even in
2257            instrumented code.
2258    
2259        A potential bug (or perhaps it is a feature, too):
2260    
2261          A back-trace reaches no further than the outermost instrumented
2262          non-trivial "raise".  Here, a "trivial" raise is one that is the
2263          sole RHS of a "handle" rule.  Thus, back-traces reach trough
2264    
2265               <exp> handle e => raise e
2266    
2267          and even
2268    
2269               <exp> handle Foo => raise Bar
2270    
2271          and, of course, through
2272    
2273               <exp> handle Foo => ...
2274    
2275         if the exception was not Foo.
2276    
2277         Back-traces always reach right through any un-instrumented code
2278         including any of its "handle" expressions, trivial or not.
2279    
2280       To try this out, do the following:
2281    
2282         - Erase all existing binfiles for your program.
2283           (You may keep binfiles for those modules where you think you
2284            definitely don't need back-tracing.)
2285         - Turn on back-trace instrumentation:
2286              SMLofNJ.Internals.BTrace.mode (SOME true);
2287         - Recompile your program.  (I.e., run "CM.make" or "use".)
2288         - You may now turn instrumentation off again (if you want):
2289              SMLofNJ.Internals.BTrace.mode (SOME false);
2290         - Run your program as usual.  If it raises an exception that
2291           reaches the interactive toplevel, then a back-trace will
2292           automatically be printed.  After that, the toplevel loop
2293           will print the exception history as usual.
2294    
2295    ----------------------------------------------------------------------
2296    Name: Matthias Blume
2297    Date: 2000/06/26 09:56:46 JST
2298    Tag: blume-20000626-setup
2299    Description:
2300    
2301    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2302          ML code before and after compiling a file (e.g., to set compiler
2303          flags)
2304    
2305    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2306              - associated changes to btrace.sml (BTrace instrumentation pass)
2307              - cleaner implementation of btimp.sml (BTrace tracing and report
2308                module)
2309    
2310    CM manual: * new path encoding documented
2311               * description of setup-parameter to "sml" added
2312    
2313    The biggest user-visible change to back-tracing is that it is no
2314    longer necessary to compile all traced modules within the same
2315    session.  (This was a real limitation.)
2316    
2317    ----------------------------------------------------------------------
2318    Name: Matthias Blume
2319    Date: 2000/06/24 12:40:00 JST
2320    Tag: blume-20000624-startup
2321    Description:
2322    
2323    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
2324    bit too often -- to put it mildly. :)
2325    
2326    ----------------------------------------------------------------------
2327    Name: Matthias Blume
2328    Date: 2000/06/23 18:20:00 JST
2329    Tag: blume-20000623-btrace
2330    Description:
2331    
2332    This updates adds a backtrace facility to aid programmers in debugging
2333    their programs.  This involves the following changes:
2334    
2335    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
2336       keeping track of the current call stack.  When programs are compiled
2337       in a special mode, the compiler will insert calls to these hooks
2338       into the user program.
2339       "Hook" means that it is possible for different implementations of
2340       back-tracing to register themselves (at different times).
2341    
2342    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
2343       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
2344       It can be turned on using this call:
2345         SMLofNJ.Internals.BTrace.mode (SOME true);
2346       Turning it off again:
2347         SMLofNJ.Internals.BTrace.mode (SOME false);
2348       Querying the current status:
2349         SMLofNJ.Internals.BTrace.mode NONE;
2350       Annotated programs are about twice as big as normal ones, and they
2351       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
2352       where all hooks do nothing).  The slowdown with a plugin that is
2353       actually useful (such as the one supplied by default) is even greater,
2354       but in the case of the default plugin it is still only an constant
2355       factor (amortized).
2356    
2357    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
2358       with a sub-structure BTrace for controlling back-tracing.  In particular,
2359       the above-mentioned function "mode" controls whether the annotation
2360       phase is invoked by the compiler.  Another important function is
2361       "trigger": when called it aborts the current execution and causes
2362       the top-level loop to print a full back-trace.
2363    
2364    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
2365       for back-tracing.  It keeps track of the dynamic call stack and in
2366       addition to that it keeps a partial history at each "level" of that
2367       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
2368       d and b (at separate times, dynamically), then the report will show:
2369    
2370       GOTO   d
2371             /c
2372       GOTO  \b
2373       CALL   a
2374    
2375       This shows that there was an initial non-tail call of a, then a
2376       tail-call to b or c, looping behavior in a cluster of functions that
2377       consist of b and c, and then a goto from that cluster (i.e., either from
2378       b or from c) to d.
2379    
2380       Note that (depending on the user program) the amount of information
2381       that the back-trace module has to keep track of at each level is bounded
2382       by a constant.  Thus, the whole implementation has the same asymptotical
2383       complexity as the original program (both in space and in time).
2384    
2385    5. compiler/TopLevel/interact/evalloop.sml has been modified to
2386       handle the special exception SMLofNJ.Internals.BTrace.BTrace
2387       which is raised by the "trigger" function mentioned above.
2388    
2389    Notes on usage:
2390    
2391    - Annotated code works well together with unannotated code:
2392    Unannotated calls simply do not show up at all in the backtrace.
2393    
2394    - It is not a good idea to let modules that were annotated during
2395    different sessions run at the same time.  This is because the compiler
2396    chooses small integers to identify individual functions, and there
2397    will be clashes if different modules were compiled in separate sessions.
2398    (Nothing will crash, and you will even be told about the clashes, but
2399    back-trace information will in general not be useful.)
2400    
2401    - Back-tracing can be confused by callcc and capture.
2402    
2403    - The only way of getting a back-trace right now is to explicitly
2404    invoke the "trigger" function from your user program.  Eventually, we
2405    should make every exception carry back-trace information (if
2406    available).  But since this creates more overhead at "raise"-time
2407    (similar to the current exnHistory overhead), I have not yet
2408    implemented this.  (The implementation will be rather easy.)  With
2409    exceptions carrying back-trace information, this facility will be even
2410    more useful because users don't need to modify their programs...
2411    
2412    - While it is possible to compile the compiler with back-trace
2413    annotations turned on (I did it to get some confidence in
2414    correctness), you must make absolutely sure that core.sml and
2415    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
2416    be compiled with annotation because there is no core access yet, but
2417    if you compile btimp.sml with annotation, then the system will go into
2418    an infinite recursion and crash.)
2419    Since CM currently does not know about BTrace, the only way to turn
2420    annotations on and off for different modules of the compiler is to
2421    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
2422    this is awkward and clumsy.
2423    
2424    Sample sessions:
2425    
2426    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2427    - SMLofNJ.Internals.BTrace.mode (SOME true);
2428    [autoloading]
2429    [autoloading done]
2430    val it = false : bool
2431    - structure X = struct
2432    -     fun main n = let
2433    -         fun a (x, 0) = d x
2434    -           | a (x, n) = b (x, n - 1)
2435    -         and b (x, n) = c (x, n)
2436    -         and c (x, n) = a (x, n)
2437    -         and d x = e (x, 3)
2438    -         and e (x, 0) = f x
2439    -           | e (x, n) = e (x, n - 1)
2440    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
2441    -           | f n = n * g (n - 1)
2442    -         and g n = a (n, 3)
2443    -     in
2444    -         f n
2445    -     end
2446    - end;
2447    structure X : sig val main : int -> int end
2448    - X.main 3;
2449    *** BACK-TRACE ***
2450    GOTO   stdIn:4.2-13.20: X.main[2].f
2451    GOTO-( stdIn:4.2-13.20: X.main[2].e
2452    GOTO   stdIn:4.2-13.20: X.main[2].d
2453         / stdIn:4.2-13.20: X.main[2].a
2454         | stdIn:4.2-13.20: X.main[2].b
2455    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2456    CALL   stdIn:4.2-13.20: X.main[2].g
2457    GOTO   stdIn:4.2-13.20: X.main[2].f
2458    GOTO-( stdIn:4.2-13.20: X.main[2].e
2459    GOTO   stdIn:4.2-13.20: X.main[2].d
2460         / stdIn:4.2-13.20: X.main[2].a
2461         | stdIn:4.2-13.20: X.main[2].b
2462    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2463    CALL   stdIn:4.2-13.20: X.main[2].g
2464    GOTO   stdIn:4.2-13.20: X.main[2].f
2465    GOTO-( stdIn:4.2-13.20: X.main[2].e
2466    GOTO   stdIn:4.2-13.20: X.main[2].d
2467         / stdIn:4.2-13.20: X.main[2].a
2468         | stdIn:4.2-13.20: X.main[2].b
2469    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2470    CALL   stdIn:4.2-13.20: X.main[2].g
2471    GOTO   stdIn:4.2-13.20: X.main[2].f
2472    CALL   stdIn:2.15-17.4: X.main[2]
2473    -
2474    
2475    (Note that because of a FLINt bug the above code currently does not
2476    compile without BTrace turned on.)
2477    
2478    Here is another example, using my modified Tiger compiler:
2479    
2480    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2481    - SMLofNJ.Internals.BTrace.mode (SOME true);
2482    [autoloading]
2483    [autoloading done]
2484    val it = false : bool
2485    - CM.make "sources.cm";
2486    [autoloading]
2487    ...
2488    [autoloading done]
2489    [scanning sources.cm]
2490    [parsing (sources.cm):parse.sml]
2491    [creating directory CM/SKEL ...]
2492    [parsing (sources.cm):tiger.lex.sml]
2493    ...
2494    [wrote CM/sparc-unix/semant.sml]
2495    [compiling (sources.cm):main.sml]
2496    [wrote CM/sparc-unix/main.sml]
2497    [New bindings added.]
2498    val it = true : bool
2499    - Main.compile ("../testcases/merge.tig", "foo.out");
2500    *** BACK-TRACE ***
2501    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
2502    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2503    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2504    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2505    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2506    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2507    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
2508         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
2509    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
2510    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2511    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
2512    CALL   main.sml:1.18-118.4: Main.compile[2]
2513    -
2514    
2515    ----------------------------------------------------------------------
2516    Name: Matthias Blumen
2517    Date: 2000/06/21 18:00:00 JST
2518    Tag: blume-20000621-manual
2519    Description:
2520    
2521    CM manual update: Path environments documented.
2522    
2523    ----------------------------------------------------------------------
2524    Name: Matthias Blume
2525    Date: 2000/06/19 13:40:00
2526    Tag: blume-20000619-manual
2527    Description:
2528    
2529    CM manual and system/README update.  This only covers the fact that
2530    there are no more implicit anchors.  (Path environments and the "bind"
2531    option to "cm" have yet to be documented.)
2532    
2533    ----------------------------------------------------------------------
2534    Name: Matthias Blume
2535    Date: 2000/06/19 11:05:00 JST
2536    Tag: blume-20000619-chdir-bugfix
2537    Description:
2538    
2539    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
2540    
2541    ----------------------------------------------------------------------
2542    Name: Matthias Blume
2543    Date: 2000/06/18 22:00:10 JST
2544    Tag: blume-20000618-implicit-anchors-really-gone
2545    Description:
2546    
2547    I updates the previous HISTORY entry where I forgot to mention that
2548    implicit anchors are no longer with us.
2549    
2550    The current update also gets rid of the (now useless) controller
2551    CM.Control.implicit_anchors.
2552    
2553    ----------------------------------------------------------------------
2554    Name: Matthias Blume
2555    Date: 2000/06/16 17:30:00 JST
2556    Tag: blume-20000616-anchorenv
2557    Description:
2558    
2559    This patch implements the long anticipated (just kidding :) "anchor
2560    environment" mechanism.  In the course of doing this, I also
2561    re-implemented CM's internal "SrcPath" module from scratch.  The new
2562    one should be more robust in certain boundary cases.  In any case, it
2563    is a lot cleaner than its predecessor (IMHO).
2564    
2565    This time, although there is yet another boot file format change, I
2566    kept the unpickler backward-compatible.  As a result, no new bootfiles
2567    are necessary and bootstrapping is straightforward.  (You cannot read
2568    new bootfiles into an old system, but the other way around is no
2569    problem.)
2570    
2571    Visible changes:
2572    
2573    ** 0. Implicit path anchors (without the leading $-symbol) are no
2574    longer recognized at all. This means that such path names are not
2575    illegal either.  For example, the name basis.cm simply refers to a
2576    local file called "basis.cm" (i.e, the name is an ordinary path
2577    relative to .cm-files directory).  Or, to put it differently, only
2578    names that start with $ are anchored paths.
2579    
2580    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
2581    vanished.
2582    
2583    John (Reppy) had critizised this as soon as I originally proposed and
2584    implemented it, but at that time I did not really deeply believe
2585    him. :) Now I came full-circle because I need the $<singlearc> syntax
2586    in another place where it cannot be seen as an abbreviation for
2587    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
2588    seems to mean (i.e., it "expands" into the corresponding anchor
2589    value).
2590    
2591    However, when paths are used as members in CM description files, it
2592    continues to be true that there must be at least another arc after the
2593    anchor.  This is now enforced separately during semantic analysis
2594    (i.e., from a lexical/syntactical point of view, the notation is ok.)
2595    
2596    ** 2. The "cm" class now accepts an option "bind".  The option's value
2597    is a sub-option list of precisely two items -- one labeled "anchor"
2598    and the other one labeled "value".  As you might expect, "anchor" is
2599    used to specify an anchor name to be bound, and "value" specifies what
2600    the anchor is being bound to.
2601    
2602    The value must be a directory name and can be given in either standard
2603    syntax (including the possibility that it is itself an anchored path)
2604    or native syntax.
2605    
2606    Examples:
2607    
2608       foo.cm (bind:(anchor:bar value:$mystuff/bar))
2609       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
2610    
2611    and so on.
2612    
2613    The meaning of this is that the .cm-file will be processed with an
2614    augmented anchor environment where the given anchor(s) is/are bound to
2615    the given values(s).
2616    
2617    The rationale for having this feature is this: Suppose you are trying
2618    to use two different (already stable) libraries a.cm and b.cm (that
2619    you perhaps didn't write yourself).  Further, suppose each of these
2620    two libraries internally uses its own auxiliary library $aux/lib.cm.
2621    Normally you would now have a problem because the anchor "lib" can not
2622    be bound to more than one value globally.  Therefore, the project that
2623    uses both a.cm and b.cm must locally redirect the anchor to some other
2624    place:
2625    
2626       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
2627       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
2628    
2629    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
2630    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
2631    
2632    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
2633    you when you do so at the time of CM.stabilize).  Therefore, you can
2634    also use an anchored path as the value:
2635    
2636      a.cm (bind:(anchor:lib value:$a-lib))
2637      b.cm (bind:(anchor:lib value:$b-lib))
2638    
2639    Now you can globally configure (using the usual CM.Anchor.anchor or
2640    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
2641    itself is always locally bound, setting it globally is no longer
2642    meaningful or necessary (but it does not hurt either).  In fact, "lib"
2643    can still be used as a global anchor for separate purposes.  As a
2644    matter of fact, one can locally define "lib" in terms of a global
2645    "lib":
2646    
2647      a.cm (bind:(anchor:lib value:$lib/a))
2648      b.cm (bind:(anchor:lib value:$lib/b))
2649    
2650    ** 3: The encoding of path names has changed.  This affects the way
2651    path names are shown in CM's progress report and also the internal
2652    protocol encoding used for parallel make.
2653    
2654    The encoding now uses one or more ':'-separated segments.  Each
2655    segments corresponds to a file that has been specified relative to the
2656    file given by its preceding segment.  The first segment is either
2657    relative to the CWD, absolute, or anchored.  Each segment itself is
2658    basically a Unix pathname; all segments but the first are relative.
2659    
2660    Example:
2661    
2662       $foo/bar/baz.cm:a/b/c.sml
2663    
2664    This path denotes the file bar/a/b/c.sml relative to the directory
2665    denoted by anchor "foo".  Notice that the encoding also includes
2666    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
2667    paths are resolved relative to the .cm-files directory, so baz.cm must
2668    be ignored to get the "real" pathname.
2669    
2670    To make this fact more obvious, CM puts the names of such "virtual
2671    arcs" into parentheses when they appear in progress reports. (No
2672    parentheses will appear in the internal protocol encoding.)  Thus,
2673    what you really see is:
2674    
2675      $foo/bar/(baz.cm):a/b/c.sml
2676    
2677    I find this notation to be much more informative than before.
2678    
2679    Another new feature of the encoding is that special characters
2680    including parentheses, colons, (back)slashes, and white space are
2681    written as \ddd (where ddd is the decimal encoding of the character).
2682    
2683    *** The CM manual still needs to be updated.
2684    
2685    ----------------------------------------------------------------------
2686    Name: Allen Leung
2687    Date: 2000/06/15 00:38:00
2688    Tag: leunga-20000615-x86-peephole
2689    
2690    x86 Peephole fix by Fermin.  Affects c-- and moby only.
2691    
2692    ----------------------------------------------------------------------
2693    Name: Matthias Blume
2694    Date: 2000/06/12 11:40:00
2695    Tag: blume-20000612-parmakefix
2696    Description:
2697    
2698    More cleanup after changing the file naming scheme: This time I
2699    repaired the parallel make mechanism for CMB.make which I broke earlier.
2700    
2701    ----------------------------------------------------------------------
2702    Name: Allen Leung
2703    Date: 2000/06/09 01:25:00
2704    Tag: leunga-20000609-various
2705    
2706    None of these things should affect normal SML/NJ operations
2707    
2708    1. Peephole improvements provided by Fermin (c--)
2709    2. New annotation DEFUSE for adding extra dependence (moby)
2710    3. New X86 LOCK instructions (moby)
2711    4. New machine description language for reservation tables (scheduling)
2712    5. Fixes to various optimization/analysis modules (branch chaining, dominator
2713       trees etc.)
2714    6. I've changed the CM files so that they can work with versions
2715       110.0.6, 110.25 and 110.28
2716    
2717    ----------------------------------------------------------------------
2718    Name: Matthias Blume
2719    Date: 2000/06/09 12:40:00
2720    Tag: blume-20000609-log
2721    Description:
2722    
2723    - Removed all(?) remaining RCS Log entries from sources.
2724    
2725    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
2726      anchored paths).
2727    
2728    ----------------------------------------------------------------------
2729    Name: Matthias Blume
2730    Date: 2000/06/07 17:00:00 JST
2731    Tag: blume-20000607-no-implicit-anchors
2732    Description:
2733    
2734    1. This update changes the default setting for
2735    CM.Control.implicit_anchors from true to false.  This means that
2736    implicit anchors are no longer permitted by default.  I also tried to
2737    make sure that nothing else still relies on implicit anchors.
2738    (This is the next step on the schedule towards a CM that does not even
2739    have the notion of implicit anchors anymore.)
2740    
2741    2. More CM manual updates.
2742    
2743    3. I managed to track down and fix the pickling bug I mentioned last
2744    time.  Because of the previously existing workaround, this entails no
2745    immediate practical changes.
2746    
2747    ----------------------------------------------------------------------
2748    Name: Matthias Blume
2749    Date: 2000/06/06 11:15:00 JST
2750    Tag: blume-20000606-lazierpickle
2751    Description:
2752    
2753    !!!! NEW BOOT FILES !!!!
2754    
2755    * The main purpose of this update is to make library pickles lazier in
2756    order to reduce the initial space penalty for autoloading a library.
2757    As a result, it is now possible to have $smlnj/compiler.cm
2758    pre-registered.  This should take care of the many complaints or
2759    inquiries about missing structure Compiler.  This required changes to
2760    CM's internal data structures and small tweaks to some algorithms.
2761    
2762    As a neat additional effect, it is no longer necessary (for the sake
2763    of lean heap image files) to distinguish between a "minimal" CM and a
2764    "full" CM.  Now, there is only one CM (i.e., the "full" version:
2765    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
2766    interactive top level. ($smlnj/cm/minimal.cm is gone.)
2767    
2768    To make the life of compiler-hackers easier, "makeml" now also
2769    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
2770    words, after you bootstrap a new sml for the first time, you will not
2771    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
2772    around you will still have to do it, though.)
2773    
2774    * A second change consists of major updates to the CM manual.  There
2775    are now several appendices with summary information and also a full
2776    specification of the CM description file syntax.
2777    
2778    * In directory src/system I added the script "allcross".  This script
2779    invokes sml and cross-compiles the compiler for all supported
2780    architectures.  (Useful when providing a new set of boot files.)
2781    
2782    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
2783    added a small tweak to pickle-util.sml to work around this problem,
2784    but it is not a proper fix yet.  I will investigate further.  (The
2785    effect of the bug was an inflation of library pickle size.)
2786    
2787    * Version number increased to 110.28.1 (to avoid compatibility problems).
2788    
2789    ----------------------------------------------------------------------
2790    Name: Allen Leung
2791    Date: 2000/05/25 17:28 EDT
2792    Tag: leunga-20000525-ra
2793    Description:
2794    
2795      Fixed a bug in freezing phase of the register allocator.
2796    
2797    ----------------------------------------------------------------------
2798    Name: Allen Leung
2799    Date: 2000/05/15 22:53 EDT
2800    Tag: leunga-20000515-alpha-x86-ra
2801    Description:
2802    
2803      1. Alpha
2804    
2805          Slight cleanup.  Removed the instruction SGNXL
2806    
2807      2. X86
2808    
2809          Added the following instructions to the instruction set:
2810    
2811            ROLx, RORx,
2812            BTx, BTSx, BTLx, BTRx,
2813            XCHGx, and variants with the LOCK prefix
2814    
2815      3. Register Allocation
2816    
2817          The module ra-rewrite-with-renaming has been improved.
2818    
2819      These have no effect on SML/NJ.
2820    
2821    ----------------------------------------------------------------------
2822    Name: Matthias Blume
2823    Date: 2000/05/15 16:20:00 JST
2824    Tag: blume-20000515-lightrebuild
2825    Description:
2826    
2827    1. I added an alternative to "-rebuild" to "makeml".  The difference is
2828       that prior to calling CMB.make' the CM-variable "LIGHT" will be
2829       defined.  In effect, the command will not build any cross-compiler
2830       backends and therefore finish more quickly.
2831    
2832       The "fixpt" script also takes a "-light" switch to be able to use
2833       this new facility while compiling for a fixpoint.
2834    
2835    2. I replaced all mentions of anchored paths in group owner specifications
2836       with simple relative paths (usually starting with "..").
2837       The rationale is that a library's internal workings should not be
2838       compromised by the lack of some anchor.  (An anchor is necessary
2839       for someone who wants to refer to the library by an anchored path,
2840       but it should not be necessary to build the same library in the first
2841       place.)
2842    
2843    3. I changed the way CM's tool mechanism determines the shell command
2844       string used for things like ml-yacc etc. so that it does not break
2845       when CM.Control.implicit_anchors is turned off.
2846    
2847    ----------------------------------------------------------------------
2848    Name: Matthias Blume
2849    Date: 2000/05/12 18:20:00 JST
2850    Tag: blume-20000512-ml-build
2851    Description:
2852    
2853    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
2854    getting installed properly (by config/install.sh).
2855    
2856    ----------------------------------------------------------------------
2857    Name: Matthias Blume
2858    Date: 2000/05/12 17:30:00 JST
2859    Tag: blume-20000512-anchors
2860    Description:
2861    
2862    !!! NEW BOOT FILES !!!
2863    
2864    This change is in preparation of fading out support for "implicitly
2865    anchored path names".  I went through all sources and used the
2866    explicit (and relatively new) $-notation.  See system/README and the
2867    CM manual for more info on this.
2868    
2869    I also modified the anchoring scheme for some things such as "smlnj",
2870    "MLRISC", "cm", etc. to take advantage of the fact that explicit
2871    anchors are more expressive: anchor name and first arc do not have to
2872    coincide.  This entails the following user-visible change:
2873    
2874    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
2875    particular, when you fire up sml with a command-line argument, say,
2876    e.g.:
2877    
2878       sml '$smlnj/cmb.cm'
2879    
2880    At the ML toplevel prompt:
2881    
2882       CM.autoload "$smlnj/cmb.cm";
2883    
2884    There is also a new controller in CM.Control that can be used to turn
2885    off all remaining support for implicit anchors by saying:
2886    
2887        CM.autoload "$smlnj/
2888        #set CM.Control.implicit_anchors false;
2889    
2890    This causes CM to reject implicitly anchored paths.  This is (for the
2891    time being) less permissive than the "final" version where there will
2892    be no more such implicit anchors and relative paths will be just that:
2893    relative.
2894    
2895    The next step (version after next version?) will be to make the
2896    default for CM.Control.implicit_anchors false.  After the dust has
2897    settled, I can then produce the "final" version of this...
2898    
2899    Note: Since bootstrapping is a bit tricky, I provided new boot files.
2900    
2901    ----------------------------------------------------------------------
2902    Name: Matthias Blume
2903    Date: 2000/05/11 16:30:00 JST
2904    Tag: blume-20000511-sources
2905    Description:
2906    
2907    The main change is that I added function CM.sources as a generalized
2908    version of the earlier CM.makedepend.  This entails the following
2909    additional changes:
2910    
2911      - CM.makedepend has been dropped.
2912    
2913      - CM manual has been updated.
2914    
2915      - TOOLS signature and API have been changed.
2916    
2917    ----------------------------------------------------------------------
2918    Name: Allen Leung
2919    Date: 2000/05/10 21:17 EDT
2920    Tag: leunga-20000510-moby-c--ssa
2921    Description:
2922    
2923      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2924    None of these affect SML/NJ.
2925    
2926    1. Register Allocation
2927    
2928        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2929           This module tries to remove local (i.e. basic block level) redundancies
2930           during spilling.
2931    
2932        b. A new framework for performing region based register allocation.
2933           Not yet entirely functional.
2934    
2935    2. X86
2936    
2937       a. DefUse for POP was missing the stack pointer [found by Lal]
2938       b. Reload for CALL was incorrect in X86Spill [found by John]
2939       c. Various fixes in X86Spill so that it can be used correctly for
2940          the new spilling module.
2941    
2942    3. SSA/IR
2943    
2944       a. New module ir/dj-dataflow.sml implements elimination based
2945          data flow analysis.
2946    
2947    4. MLRiscGen
2948    
2949       a. Fix for gc type annotation
2950    
2951    5. MDGen
2952    
2953       Various fixes for machine description -> ml code translation.  For ssa
2954       only.
2955    
2956    ----------------------------------------------------------------------
2957    Name: Allen Leung
2958    Date: 2000/05/08 22:17 EDT
2959    Tag: leunga-20000508-labexp
2960    Description:
2961    
2962      Fermin has found a few assembly problems with constant expressions
2963      generated in LabelExp.  Mostly, the problems involve extra parentheses,
2964      which choke on dumb assemblers.  This is his fix.
2965    
2966    ----------------------------------------------------------------------
2967    Name: Dave MacQueen
2968    Date: 2000/04/09 14:00 EDT
2969    Tag: dbm-20000502-Version_110_28
2970    Description:
2971    
2972    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2973    
2974    2. Updated config/version to 110.28
2975    
2976    3. Updated config/srcarchiveurl
2977    
2978    3. New boot files!
2979       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2980    
2981    ----------------------------------------------------------------------
2982    Name: Matthias Blume
2983    Date: 2000/05/01 19:05:00 JST
2984    Tag: blume-20000501-noweb
2985    Description:
2986    
2987    A new noweb tool has been added.  The existing system is entirely
2988    unaffected by this, but some CM users have asked for renewed noweb
2989    support.  Everything is documented in the CM manual.
2990    
2991    New (plugin) libraries:
2992    
2993       noweb-tool.cm
2994       nw-ext.cm
2995    
2996    ----------------------------------------------------------------------
2997    Name: Dave MacQueen
2998    Date: 2000/04/30 12:40PM EDT
2999    Tag: dbm-20000430-bug_fixes
3000    Description:
3001    
3002    1. Fix for bug 1498
3003       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
3004       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
3005         added toRealArray function
3006       smlnj/src/compiler/MiscUtil/print/ppobj.sml
3007         added check for tag Obj.RealArray to array printing case in ppObj
3008    
3009    2. Fix for bug 1510
3010       smlnj/src/compiler/Semant/types/typesutil.sml
3011         fixed definition of dummyargs (used by equalTycon) so that
3012         dummy args are distinct types
3013    
3014    ----------------------------------------------------------------------
3015    Name: Matthias Blume
3016    Date: 2000/04/30 01:00:00 JST
3017    Tag: blume-20000430-versions
3018    Description:
3019    
3020    1. CM version numbering added.  This is an implementation of Lal's
3021       proposal for adding version numbers and version checking to .cm
3022       files.  Lal said that his proposal was just that -- a proposal.
3023       For the time being I went ahead and implemented it so that people
3024       can comment on it.  Everything is completely backward-compatible
3025       (except for the stable library format, i.e., new bootfiles!).
3026    
3027       As usual, see the CM manual for details.
3028    
3029    2. An alternative syntax for anchored paths has been implemented.
3030       Dave has recently voiced the same concerns that I had when I did
3031       this, so there should be some support.  My take is that eventually
3032       I will let support for the current syntax (where anchors are
3033       "implicit") fade out in favor of the new, explicit syntax.
3034       In order to be backward-compatible, both old and new syntax are
3035       currently supported.
3036    
3037       Again, see the CM manual for details.
3038    
3039    3. Parallel make is trying to be slightly smarter:  When the master
3040       process finds a "bottleneck", i.e., when there is only one
3041       compilation unit that can be compiled and everybody else is
3042       waiting on it, then it will simply compile it directly instead
3043       of clumsily telling one of the slaves to do it.
3044    
3045    4. Support for "unsharing" added.  This is necessary in order to be
3046       able to have two different versions of the same library running
3047       at the same time (e.g., for trying out a new MLRISC while still
3048       having the old MLRISC linked into the current compiler, etc.)
3049       See the CM manual.
3050    
3051    5. Simple "makedepend" functionality added for generating Makefile
3052       dependency information.  (This is rather crude at the moment.
3053       Expect some changes here in the future.)
3054    
3055    6. ".fun" added as a recognized suffix for ML files. Also documented
3056       explicitly in the manual that the fallback behavior (unknown suffix
3057       -> ML file) is not an official feature!
3058    
3059    7. Small changes to the pickler for stable libraries.
3060    
3061    8. Several internal changes to CM (for cleanup/improvement).
3062    
3063    
3064    !!!! NEW BINFILES !!!!
3065    
3066    ----------------------------------------------------------------------
3067    Name: Matthias Blume
3068    Date: 2000/04/28 17:30:00 JST
3069    Tag: blume-20000428-pathconfig
3070    Description:
3071    
3072    1. I changed config/install.sh to remove duplicate entries from the
3073       lib/pathconfig file at the end.  Moreover, the final version of
3074       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
3075       in src/system/installml.
3076    
3077    2. The config/install.sh script now consistently uses relative
3078       pathnames in lib/pathconfig whenever the anchor is in the lib
3079       directory.  (So far this was true for the libraries that come
3080       pre-compiled and bundled as part of the bootfiles but not for
3081       libraries that are compiled by the script itself.)
3082    
3083    ----------------------------------------------------------------------
3084    Name: Matthias Blume
3085    Date: 2000/04/26 13:10:00 JST
3086    Tag: blume-20000426-fun_suffix
3087    Description:
3088    
3089    Added ".fun" as a recognized file name suffix (for ML code).
3090    
3091    ----------------------------------------------------------------------
3092    Name: Allen Leung
3093    Date: 2000/04/25 17:00:00 EST
3094    Tag: leunga-20000425-alpha-ra
3095    Description:
3096    
3097    1. Alpha
3098    
3099        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
3100    in C--.
3101    
3102    2. RA
3103    
3104       Added a flag "ra-dump-size" to print out the size of the flowgraph
3105       and the interference graph.
3106    
3107    ----------------------------------------------------------------------
3108    Name: Dave MacQueen
3109    Date: 2000/04/25/
3110    Tag: dbm-20000425-mlyacc_doc_examples
3111    Description:
3112      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
3113      Updated all three examples in src/ml-yacc/examples to run
3114      under 110.* using CM.make.
3115    
3116    ----------------------------------------------------------------------
3117    Name: Allen Leung
3118    Date: 2000/04/20 23:04:00 EST
3119    Tag: leunga-20000420-ssa-c---stuff
3120    Description:
3121    
3122      This update synchronizes my repository with Yale's.  Most of these
3123    changes, however, do not affect SML/NJ at all (the RA is an exception).
3124    
3125    1. Register Allocator
3126    
3127       a. An improvement in the interference graph construction:
3128          Given a copy
3129    
3130                s <- t
3131    
3132          no interference edge between s and t is added for this definition of s.
3133    
3134       b. I've added two new spill heuristic modules that Fermin and I developed
3135          (in the new library RA.cm). These are unused in SML/NJ but maybe
3136          useful for others (Moby?)
3137    
3138    2. X86
3139    
3140       a. Various fixes in the backend provided by Fermin [C--] and Lal.
3141    
3142    3. Alpha
3143    
3144       a. Added the BSR instruction and code generation that goes with it [C--]
3145       b. Other fixes too numerous to recount provided by Fermin [C--]
3146    
3147    4. Regmaps
3148    
3149       a. The regmaps are not initialized with the identity physical bindings
3150          at creation time.  This is unneeded.
3151    
3152    5. MLRISC Optimizations
3153    
3154       a. The DJ-Graph module can now compute the iterated dominance frontiers
3155          intersects with liveness incrementally in linear time! Woohoo!
3156          This is now used in my new SSA construction algorithm.
3157    
3158       b. THe branch reorganization module is now smarter about linear chains of
3159          basic blocks.
3160    
3161    
3162    ----------------------------------------------------------------------
3163    Name: Matthias Blume
3164    Date: 2000/04/12 13:52:00 JST
3165    Tag: blume_main_v110p27_1
3166    Description:
3167    
3168    Changed install.sh script to handle archive files without version number
3169    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
3170    name of the boot file archive.
3171    
3172    ----------------------------------------------------------------------
3173    Name: Dave MacQueen
3174    Date: 2000/04/09 14:00 EDT
3175    Tag: dbm-20000410-Version_110_27
3176    Description:
3177    
3178    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
3179    
3180    2. Updated src/config/version to 110.27
3181    
3182    3. New boot files!
3183    
3184    ----------------------------------------------------------------------
3185    Name: Allen Leung
3186    Date: 2000/04/09 19:09:00 EST
3187    Tag: leunga-20000409-misc
3188    Description:
3189    
3190    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
3191    
3192    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
3193    
3194    ----------------------------------------------------------------------
3195    Name: Stefan
3196    Date: 2000/04/07 10:00:00 EDT
3197    Tag: monnier-20000406-branch-handling
3198    Description:
3199    
3200    Improved handling of branches (mostly those generated from
3201    polymorphic equality), removed switchoff and changed the
3202    default optimization settings (more cpsopt and less flintopt).
3203    
3204    ----------------------------------------------------------------------
3205    Name: Allen Leung
3206    Date: 2000/04/06 01:30:00 EST
3207    Tag: leunga-20000406-peephole-x86-SSA-2
3208    Description:
3209    
3210       Forgot a few files.
3211    
3212    ----------------------------------------------------------------------
3213    Name: Allen Leung
3214    Date: 2000/04/06 00:36:00 EST
3215    Tag: leunga-20000406-peephole-x86-SSA
3216    Description:
3217    
3218    1.  New Peephole code
3219    
3220    2.  Minor improvement to X86 instruction selection
3221    
3222    3.  Various fixes to SSA and machine description -> code translator
3223    
3224    ----------------------------------------------------------------------
3225    Name: Matthias Blume
3226    Date: 2000/04/05 12:30:00 JST
3227    Tag: blume_main_v110p26p2_3
3228    Description:
3229    
3230    This update just merges three minor cosmetic updates to CM's sources
3231    to get ready for the 110.27 code freeze on Friday.  No functionality
3232    has changed.
3233    
3234    ----------------------------------------------------------------------
3235    Name: Allen Leung
3236    Date: 2000/04/04 19:39:00 EST
3237    Tag: leunga-20000404-x86-asm
3238    Description:
3239    
3240    1.  Fixed a problem in X86 assembly.
3241    
3242        Things like
3243    
3244           jmp %eax
3245           jmp (%eax)
3246    
3247        should be output as
3248    
3249           jmp *%eax
3250           jmp *(%eax)
3251    
3252    2.  Assembly output
3253    
3254          Added a new flag
3255    
3256              "asm-indent-copies" (default to false)
3257    
3258          When this flag is on, parallel copies will be indented an extra level.
3259    
3260    ----------------------------------------------------------------------
3261    Name: Allen Leung
3262    Date: 2000/04/04 03:18:00 EST
3263    Tag: leunga-20000404-C--Moby
3264    Description:
3265    
3266        All of these fixes are related to C--, Moby, and my own optimization
3267        stuff; so they shouldn't affect SML/NJ.
3268    
3269    1.  X86
3270    
3271        Various fixes related floating point, and extensions.
3272    
3273    2.  Alpha
3274    
3275        Some extra patterns related to loads with signed/zero extension
3276        provided by Fermin.
3277    
3278    3.  Assembly
3279    
3280        When generating assembly, resolve the value of client defined constants,
3281        instead of generating symbolic values.  This is controlled by the
3282        new flag "asm-resolve-constants", which is default to true.
3283    
3284    4.  Machine Descriptions
3285    
3286        a. The precedence parser was slightly broken when parsing infixr symbols.
3287        b. The type generalizing code had the bound variables reversed, resulting
3288           in a problem during arity raising.
3289        c. Various fixes in machine descriptions.
3290    
3291    ----------------------------------------------------------------------
3292    Name: Matthias Blume
3293    Date: 2000/04/03 16:05:00 JST
3294    Tag: blume_main_v110p26p2_2
3295    Description:
3296    
3297    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
3298    now done via the ordinary static environment that is context to each
3299    compilation unit.
3300    
3301    To this end, I arranged that instead of "structure Core" as "structure
3302    _Core" is bound in the pervasive environment.  Core access is done via
3303    _Core (which can never be accidentally rebound because _Core is not a
3304    legal surface-syntax symbol).
3305    
3306    The current solution is much cleaner because the core environment is
3307    now simply part of the pervasive environment which is part of every
3308    compilation unit's context anyway.  In particular, this eliminates all
3309    special-case handling that was necessary until now in order to deal
3310    with dynamic and symbolic parts of the core environment.
3311    
3312    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3313    compilation manager's bootstrap compiler (actually: in the "init group"
3314    handling).  See the comments in src/system/smlnj/init/init.cmi for
3315    more details.
3316    
3317    I also tried to track down all mentions of "Core" (as string argument
3318    to Symbol.strSymbol) in the compiler and replaced them with a
3319    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
3320    name appears in one place only.
3321    
3322    Binfile and bootfile format have not changed, but the switchover from
3323    the old "init.cmi" to the new one is a bit tricky, so I supplied new
3324    bootfiles anyway.
3325    
3326    ----------------------------------------------------------------------
3327    Name: Allen Leung
3328    Date: 2000/04/02 21:17:00 EST
3329    Tag: leunga-20000402-mltree
3330    Description:
3331    
3332       1. Renamed the constructor CALL in MLTREE by popular demand.
3333       2. Added a bunch of files from my repository.  These are currently
3334          used by other non-SMLNJ backends.
3335    
3336    ----------------------------------------------------------------------
3337    Name: Allen Leung
3338    Date: 2000/03/31 21:15:00 EST
3339    Tag: leunga-20000331-aliasing
3340    Description:
3341    
3342    This update contains a rewritten (and hopefully more correct) module
3343    for extracting aliasing information from CPS.
3344    
3345       To turn on this feature:
3346    
3347            Compiler.Control.CG.memDisambiguate := true
3348    
3349       To pretty print the region information with assembly
3350    
3351           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
3352    
3353       To control how many levels of aliasing information are printed, use:
3354    
3355           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
3356    
3357       The default of n is 3.
3358    
3359    ----------------------------------------------------------------------
3360    Name: David MacQueen
3361    Date: 2000/03/31 11:15:00 EST
3362    Tag: dbm-20000331-runtime_fix
3363    Description:
3364    
3365    This update contains:
3366    
3367    1. runtime/c-lib/c-libraries.c
3368       includes added in revision 1.2 caused compilation errors on hppa-hpux
3369    
3370    2. fix for bug 1556
3371       system/Basis/Implementation/NJ/internal-signals.sml
3372    
3373    ----------------------------------------------------------------------
3374    Name: Matthias Blume
3375    Date: 2000/03/31 18:00:00 JST
3376    Tag: blume_main_v110p26p2_1
3377    Description:
3378    
3379    This update contains:
3380    
3381    1. A small change to CM's handling of stable libraries:
3382       CM now maintains one "global" modmap that is used for all stable
3383       libraries.  The use of such a global modmap maximizes sharing and
3384       minimizes the need for re-traversing parts of environments during
3385       modmap construction.  (However, this has minor impact since modmap
3386       construction seems to account for just one percent or less of total
3387       compile time.)
3388    
3389    2. I added a "genmap" phase to the statistics.  This is where I got the
3390       "one percent" number (see above).
3391    
3392    3. CM's new tool parameter mechanism just became _even_ better. :)
3393       - The parser understands named parameters and recursive options.
3394       - The "make" and "shell" tools use these new features.
3395         (This makes it a lot easier to cascade these tools.)
3396       - There is a small syntax change: named parameters use a
3397    
3398           <name> : ( <option> ... )            or
3399           <name> : <string>
3400    
3401         syntax.  Previously, named parameters were implemented in an
3402         ad-hoc fashion by each tool individually (by parsing strings)
3403         and had the form
3404    
3405           <name>=<string>
3406    
3407       See the CM manual for a full description of these issues.
3408    
3409    ----------------------------------------------------------------------
3410    Name: Matthias Blume
3411    Date: 2000/03/30 18:00:00 JST
3412    Tag: blume_main_v110p26p2_0
3413    Description:
3414    
3415    !!!!! WARNING !!!!!!
3416    !!  New binfiles  !!
3417    !!!!!!!!!!!!!!!!!!!!
3418    
3419    This update contains:
3420    
3421    1. Moderate changes to CM:
3422    
3423       - Changes to CM's tools mechanism.  In particular, it is now possible
3424       to have tools that accept additional "command line" parameters
3425       (specified in the .cm file at each instance where the tool's class is
3426       used).
3427    
3428       This was done to accommodate the new "make" and "shell" tools which
3429       facilitate fairly seamless hookup to portions of code managed using
3430       Makefiles or Shell scripts.
3431    
3432       There are no classes "shared" or "private" anymore.  Instead, the
3433       sharing annotation is now a parameter to the "sml" class.
3434    
3435       There is a bit of generic machinery for implementing one's own
3436       tools that accept command-line parameters.  However, I am not yet fully
3437       satisfied with that part, so expect changes here in the future.
3438    
3439       All existing tools are described in the CM manual.
3440    
3441       - Slightly better error handling.  (CM now suppresses many followup
3442       error messages that tended to be more annoying than helpful.)
3443    
3444    2. Major changes to the compiler's static environment data structures.
3445    
3446       - no CMStaticEnv anymore.
3447            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
3448              but it is called Environment), no conversions between different
3449              kinds of static environments
3450    
3451       - There is still a notion of a "modmap", but such modmaps are generated
3452         on demand at the time when they are needed.  This sounds slow, but I
3453         sped up the code that generates modmaps enough for this not to lead to
3454         a slowdown of the compiler (at least I didn't detect any).
3455    
3456       - To facilitate rapid modmap generation, static environments now
3457         contain an (optional) "modtree" structure.  Modtree annotations are
3458         constructed by the unpickler during unpickling.  (This means that
3459         the elaborator does not have to worry about modtrees at all.)
3460         Modtrees have the advantage that they are compositional in the same
3461         way as the environment data structure itself is compositional.
3462         As a result, modtrees never hang on to parts of an environment that
3463         has already been rendered "stale" by filtering or rebinding.
3464    
3465       - I went through many, many trials and errors before arriving at the
3466         current solution.  (The initial idea of "linkpaths" did not work.)
3467         But the result of all this is that I have touched a lot of files that
3468         depend on the "modules" and "types" data structures (most of the
3469         elaborator). There were a lot of changes during my "linkpath" trials
3470         that could have been reverted to their original state but weren't.
3471         Please, don't be too harsh on me for messing with this code a bit more
3472         than what was strictly necessary...  (I _did_ resist the temptation
3473         of doing any "global reformatting" to avoid an untimely death at
3474         Dave's hands. :)
3475    
3476       - One positive aspect of the previous point:  At least I made sure that
3477         all files that I touched now compile without warnings (other than
3478         "polyEqual").
3479    
3480       - compiler now tends to run "leaner" (i.e., ties up less memory in
3481         redundant modmaps)
3482    
3483    ----------------------------------------------------------------------
3484    Name: Allen Leung
3485    Date: 2000/03/29 18:00:00
3486    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
3487    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
3488    Description:
3489    
3490       This update contains *MAJOR* changes to the way code is generated from CPS
3491    in the module mlriscGen, and in various backend modules.
3492    
3493    CHANGES
3494    =======
3495    
3496    1. MLRiscGen: forward propagation fix.
3497    
3498       There was a bug in forward propagation introduced at about the same time
3499       as the MLRISC x86 backend, which prohibits coalescing to be
3500       performed effectively in loops.
3501    
3502       Effect: speed up of loops in RISC architectures.
3503               By itself, this actually slowed down certain benchmarks on the x86.
3504    
3505    2. MLRiscGen:  forward propagating addresses from consing.
3506    
3507       I've changed the way consing code is generated.  Basically I separated
3508       out the initialization part:
3509    
3510            store tag,   offset(allocptr)
3511            store elem1, offset+4(allocptr)
3512            store elem2, offset+8(allocptr)
3513            ...
3514            store elemn, offset+4n(allocptr)
3515    
3516       and the address computation part:
3517    
3518            celladdr <- offset+4+alloctpr
3519    
3520       and move the address computation part
3521    
3522       Effect:  register pressure is generally lower as a result.  This
3523                makes compilation of certain expressions much faster, such as
3524                long lists with non-trivial elements.
3525    
3526                 [(0,0), (0,0), .... (0,0)]
3527    
3528    3. MLRiscGen: base pointer elimination.
3529    
3530        As part of the linkage mechanism, we generate the sequence:
3531    
3532         L:  ...  <- start of the code fragment
3533    
3534         L1:
3535             base pointer <- linkreg - L1 + L
3536    
3537         The base pointer was then used for computing relocatable addresses
3538       in the code fragment.  Frequently (such as in lots of continuations)
3539       this is not needed.  We now eliminate this sequence whenever possible.
3540    
3541         For compile time efficiency, I'm using a very stupid local heuristic.
3542       But in general, this should be done as a control flow analysis.
3543    
3544       Effect:  Smaller code size.  Speed up of most programs.
3545    
3546    4. Hppa back end
3547    
3548         Long jumps in span dependence resolution used to depend on the existence
3549      of the base pointer.
3550    
3551         A jump to a long label L was expanded into the following sequence:
3552    
3553          LDIL %hi(L-8192), %r29
3554          LDO  %lo(L-8192)(%r29), %r29
3555          ADD  %r29, baseptr, %r29
3556          BV,n %r0(%r29)
3557    
3558         In the presence of change (3) above, this will not work.  I've changed
3559       it so that the following sequence of instructions are generated, which
3560       doesn't mention the base pointer at all:
3561    
3562             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
3563        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
3564             BV,n  %r0(%r29)          /* Jump */
3565    
3566    5. Alpha back end
3567    
3568          New alpha instructions LDB/LDW have been added, as per Fermin's
3569       suggestions.   This is unrelated to all other changes.
3570    
3571    6. X86 back end
3572    
3573         I've changed andl to testl in the floating point test sequence
3574         whenever appropriate.  The Intel optimization guide states that
3575         testl is preferable to andl.
3576    
3577    7. RA (x86 only)
3578    
3579         I've improved the spill propagation algorithm, using an approximation
3580       of maximal weighted independent sets.   This seems to be necessary to
3581       alleviate the negative effect in light of the slow down in (1).
3582    
3583         I'll write down the algorithm one of these days.
3584    
3585    8. MLRiscGen: frequencies
3586    
3587         I've added an annotation that states that all call gc blocks have zero
3588       execution frequencies.  This improves register allocation on the x86.
3589    
3590    BENCHMARKS
3591    ==========
3592    
3593       I've only perform the comparison on 110.25.
3594    
3595       The platforms are:
3596    
3597        HPPA  A four processor HP machine (E9000) with 5G of memory.
3598        X86   A 300Hhz Pentium II with 128M of memory, and
3599        SPARC An Ultra sparc 2 with 512M of memory.
3600    
3601       I used the following parameters for the SML benchmarks:
3602    
3603                 @SMLalloc
3604         HPPA    256k
3605         SPARC   512k
3606         X86     256k
3607    
3608    COMPILATION TIME
3609    ----------------
3610       Here are the numbers comparing the compilation times of the compilers.
3611       I've only compared 110.25 compiling the new sources versus
3612       a fixpoint version of the new compiler compiling the same.
3613    
3614                     110.25                                  New
3615               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
3616         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
3617         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
3618         X86    999s    315s       94006+130691   987s    296s    108877+141957
3619    
3620                   110.25         New
3621                Code Size      Code Size
3622         HPPA   8596736         8561421
3623         SPARC  8974299         8785143
3624         X86    9029180         8716783
3625    
3626       So in summary, things are at least as good as before.   Dramatic
3627       reduction in compilation is obtained on the Sparc; I can't explain it,
3628       but it is reproducible.  Perhaps someone should try to reproduce this
3629       on their own machines.
3630    
3631    SML BENCHMARKS
3632