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

Legend:
Removed from v.586  
changed lines
  Added in v.958

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