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 572, Thu Mar 9 02:43:06 2000 UTC revision 755, Thu Dec 14 07:57:55 2000 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    ----------------------------------------------------------------------
15    Name: Allen Leung
16    Date: Thu Dec 14 03:45:24 EST 2000
17    Description:
18    Tag:  leunga-20001214-int-inf
19    Description:
20    
21       In IntInf, added these standard functions, which are missing from our
22    implementation:
23    
24        andb : int * int -> int
25        xorb : int * int -> int
26        orb  : int * int -> int
27        notb : int -> int
28         <<   : int * word -> int
29        ~>>  : int * word -> int
30    
31       Not tested, I hope they are correct.
32    
33    ----------------------------------------------------------------------
34    Name: Allen Leung
35    Date: Fri Dec  8 19:23:26 EST 2000
36    Description:
37    Tag:  leunga-20001208-nowhere
38    Description:
39    
40      Slight improvements to the 'nowhere' tool to handle OR-patterns,
41    to generate better error messages etc.  Plus a brief manual.
42    
43    ----------------------------------------------------------------------
44    Name: Lal George
45    Date: 2000/12/08 09:54:02 EST 2000
46    Tag: Release_110_31
47    Description:
48    
49    - Version 110.31
50    ----------------------------------------------------------------------
51    Name: Allen Leung
52    Date: Thu Dec  7 22:01:04 EST 2000
53    Tag:  leunga-20001207-cell-monster-hack
54    Description:
55    
56    Major MLRISC internal changes.  Affect all clients.
57    Summary:
58    
59    1.  Type CELLS.cell = int is now replaced by a datatype.
60        As a result, the old regmap is now gone.  Almost all interfaces
61        in MLRISC change as a consequence.
62    
63    2.  A new brand version of machine description tool (v3.0) that generates
64        modules expecting the new interface.  The old version is removed.
65    
66    3.  The RA interface has been further abstracted into two new functors.
67        RISC_RA and X86RA.  These functors have much simpler interfaces.
68        [See also directory MLRISC/demo.]
69    
70    4.  Some other new source->source code generation tools are available:
71    
72        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
73        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
74           I use this tool to generate the peephole optimizers---with the new
75           cell type changes, peephole rules are becoming difficult to write
76           without conditional pattern matching.
77    
78    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
79        cover the entire MLRISC source tree so many things broke.
80    
81    6.  CM files have been moved to the subdirectory MLRISC/cm.
82        They are moved because there are a lot of them and they clutter up the
83        root dir.
84    
85    7.  More detailed documentation to come...
86    
87        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
88        a makeml -rebuild first.  This is because of other other
89        changes that Matthias has made (see below).
90    
91    
92    ----------------------------------------------------------------------
93    Name: Matthias Blume
94    Date: 2000/11/30 23:12:00 JST
95    Tag: blume-20001130-filereorg
96    Description:
97    
98    Some manual updates and some file reorganizations in CM.
99    
100    ----------------------------------------------------------------------
101    Name: Matthias Blume
102    Date: 2000/11/24 17:45:00 JST
103    Tag: blume-20001124-link
104    Description:
105    
106    Drastically improved link traversal code for the case that the dynamic
107    value was already loaded at bootstrap time.  As a result, CM and CMB
108    now both load blazingly fast -- even on a very slow machine.  Also,
109    memory consumption has been further reduced by this.
110    
111    Warning: The format of the PIDMAP file has changed.  THerefore, to
112    bootstrap you have to do this:
113    
114    1. Run CMB.make
115    2. Make a symbolic link for the boot directory:
116         ln -s sml.boot.ARCH-OS xxx
117    3. "Rebuild" the boot directory:
118         ./makeml -boot xxx -rebuild sml ; rm xxx
119    4. Boot normally:
120          ./makeml
121    
122    ----------------------------------------------------------------------
123    Name: Matthias Blume
124    Date: 2000/11/21 21:20:00 JST
125    Tag: blume-20001121-tools
126    Description:
127    
128    Continued hacking on autoloading problem -- with success this time.
129    Also changed tool-plugin mechanism.  See new CM manual.
130    
131    ----------------------------------------------------------------------
132    Name: Matthias Blume
133    Date: 2000/11/19 14:30:00 JST
134    Tag:  blume-20001119-autoload
135    Description:
136    
137    Some hacking to make autoloading faster.  Success for CMB, no success
138    so far for CM.  There is a reduced structure CM' that autoloads faster.
139    (This is a temporary, non-documented hack to be eliminated again when
140    the general problem is solved.)
141    
142    ----------------------------------------------------------------------
143    Name: Matthias Blume
144    Date: 2000/11/17 14:10:00 JST
145    Tag: blume-20001117-pickle-lib
146    Description:
147    
148    1. Eliminated comp-lib.cm
149    2. Made pickle-lib.cm
150    3. Eliminated all uses of intset.sml (from comp-lib.cm)
151    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
152       equivalent constructs from smlnj-lib.cm (INtHashTable).
153    5. Point 4. also goes for those uses of intmap.* in MLRISC.
154       Duplicated intmap modules thrown out.
155    6. Hunted down all duplicated SCC code and replaced it with
156       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
157    7. Rewrote Feedback module.
158    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
159       should be thrown out and equivalent modules from smlnj-lib.cm
160       should be used (IntRedBlackSet, IntListSet, ...).
161    
162    Confirmed that compiler compiles to fixpoint.
163    
164    ----------------------------------------------------------------------
165    Name: Allen Leung
166    Date: 2000/11/10 18:00:00
167    Tag: leunga-20001110-new-x86-fp
168    
169    A new x86 floating point code generator has been added.
170    By default this is turned off.  To turn this on, do:
171    
172        CM.autoload "$smlnj/compiler.cm";
173        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
174    
175    Changes:
176    
177    1.  Changed FTAN to FPTAN so that the assembly output is correct.
178    2.  Changed the extension callback for FTANGENT to generate:
179    
180              fptan
181              fstp  %st(0)
182        instead of
183              fptan
184              fstpl ftempmem
185    
186    3.  Numerous assembly fixes for x86.
187    
188    5.  Cleaned up the machine code output module x86/x86MC.sml and added
189        support for a whole bunch of instructions and addressing modes:
190    
191          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
192          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
193          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
194          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
195          fxch %st(n)
196          fld %st(n)
197          fst %st(n)
198          fst mem
199          fstp %st(n)
200          fucom %st(n)
201          fucomp %st(n)
202    
203        All these are now generated when the fast fp mode is turned on.
204    
205    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
206    
207    ----------------------------------------------------------------------
208    Name: Matthias Blume
209    Date: 2000/11/09 11:20:00 JST
210    Tag: blume-20001109-scc
211    Description:
212    
213    Eliminated some code duplication:
214    
215    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
216       (Otherwise the functor is useless.)
217    2. Used GraphSCCFn where SCCUtilFun was used previously.
218    3. Got rid of SCCUtilFun (in comp-lib.cm).
219    
220    ----------------------------------------------------------------------
221    Name: Lal George
222    Date: 2000/11/06 09:02:21 EST 2000
223    Tag: Release_110_30
224    Description:
225    
226    - Version 110.30
227    ----------------------------------------------------------------------
228    Name: Matthias Blume
229    Date: 2000/11/04 14:45:00
230    Tag: blume-20001104-mlbuild
231    Description:
232    
233    - Made ml-build faster on startup.
234    - Documentation fixes.
235    
236    ----------------------------------------------------------------------
237    Name: Matthias Blume
238    Date: 2000/11/02 17:00:00 JST
239    Tag: blume-20001102-condcomp
240    Description:
241    
242    - Small tweaks to pickler -- new BOOTFILES!
243    - Version bumped to 110.29.2.
244    - Added conditional compilation facility to init.cmi (see comment there).
245    ----------------------------------------------------------------------
246    Name: Allen Leung
247    Date: 2000/10/23 19:31:00
248    Tag: leunga-20001023-demo-ra
249    
250    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
251    2. Test programs for the graph library updated
252    3. Some new MLRISC demo programs added
253    
254    ----------------------------------------------------------------------
255    Name: Matthias Blume
256    Date: 2000/08/31 22:15:00 JST
257    Tag: blume-20001017-errmsg
258    Description:
259    
260    More error message grief: Where there used to be no messages, there
261    now were some that had bogus error regions.  Fixed.
262    
263    ----------------------------------------------------------------------
264    Name: Matthias Blume
265    Date: 2000/08/31 17:30:00 JST
266    Tag: blume-20001017-v110p29p1
267    Description:
268    
269    I made a version 110.29.1 with new bootfiles.
270    
271    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
272              CM documentation changes and a small bugfix in CM's error reporting.
273    
274    ----------------------------------------------------------------------
275    Name: Lal George
276    Date: 2000/09/27 14:42:35 EDT
277    Tag: george-20000927-nodestatus
278    Description:
279    
280    Changed the type of the nodestatus, so that:
281    
282            SPILLED(~1)             is now SPILLED
283            SPILLED(m) where m>=0   is now MEMREG(m)
284            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
285    
286    ----------------------------------------------------------------------
287    Name: Matthias Blume
288    Date: 2000/09/07 14:45:00 JST
289    Tag: blume-20000907-cmerrmsg
290    Description:
291    
292    Small tweak to CM to avoid getting ML syntax error messages twice.
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume
296    Date: 2000/08/31 18:00:00 JST
297    Tag: blume-20000831-cvsbootfiles
298    Description:
299    
300    New URL for boot files (because the 110.29 files on the BL server do
301    now work correctly with my updated install scripts for yacc and lex).
302    
303    ----------------------------------------------------------------------
304    Name: Matthias Blume
305    Date: 2000/08/08 12:33:00 JST
306    Tag: blume-20000808-manual
307    Description:
308    
309    Tiny update to CM manual.
310    
311    ----------------------------------------------------------------------
312    Name: Allen Leung
313    Date: 2000/08/7 19:31:00
314    Tag: leunga-20000807-a-whole-bunch-of-stuff
315    
316      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
317    and Mobdy.
318    
319    1.  x86
320    
321       a.  Fixes to peephole module by John and Dan.
322       b.  Assembly fix to SETcc by Allen.
323       c.  Fix to c-call by John.
324       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
325       e.  Instruction selection optimization to SETcc as suggested by John.
326    
327           For example,
328    
329            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
330    
331           should generate:
332    
333            MOVL a, x
334            SUBL b, x
335            SHRL 31, x
336    
337    2.  IR stuff
338    
339         A bunch of new DJ-graph related algorithms added.  These
340         speed up SSA construction.
341    
342    3.  SSA + Scheduling
343    
344         Added code for SSA and scheduling to the repository
345    
346    ----------------------------------------------------------------------
347    Name: Lal George
348    Date: 2000/07/27 11:53:14 EDT
349    
350    Tag: lal-20000727-linux-ppc
351    Description:
352    
353     Made changes to support Linux PPC.
354     p.s. I have confirmation that the 110.29 boot files work fine.
355    
356    ----------------------------------------------------------------------
357    Name: Matthias Blume
358    Date: 2000/07/27 17:40:00 JST
359    Tag: blume-20000727-scripts
360    Description:
361    
362    !!!! WARNING !!!!
363    You must recompile the runtime system!
364    !!!! WARNING !!!!
365    
366    This is basically another round of script-enhancements:
367    
368    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
369       and undefine CM preprocessor symbols.
370    
371    2. ml-build avoids generating a new heap image if it finds that the
372       existing one is still ok.  (The condition is that no ML file had to
373       be recompiled and all ML files are found to be older that the heap
374       file.)
375    
376       To make this work smoothly, I also hacked the runtime system as
377       well as SMLofNJ.SysInfo to get access to the heap image suffix
378       (.sparc-solaris, ...) that is currently being used.
379    
380       Moreover, the signature of CM.mk_standalone has changed.  See the
381       CM manual.
382    
383    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
384       CM manual for details.)
385    
386    4. More CM manual updates:
387        - all of the above has been documented.
388        - there is now a section describing the (CM-related) command line
389          arguments that are accepted by the "sml" command
390    
391    ----------------------------------------------------------------------
392    Name: Matthias Blume
393    Date: 2000/07/25 16:20:00 JST
394    Tag: blume-20000725-makedepend
395    Description:
396    
397    Added a script called ml-makedepend.  This can be used in makefiles
398    for Unix' make in a way very similar to the "makedepend" command for
399    C.
400    
401    The script internally uses function CM.sources.
402    
403    Synopsis:
404    
405        ml-makedepend [-f makefile] cmfile targetname
406    
407    The default for the makefile is "makefile" (or "Makefile" should
408    "makefile" not exist).
409    
410    ml-makedepend adds a cmfile/targetname-specific section to this
411    makefile (after removing the previous version of this section).  The
412    section contains a single dependency specification with targetname on
413    the LHS (targetname is an arbitrary name), and a list of files derived
414    from the cmfile on the RHS.  Some of the files on the RHS are
415    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
416    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
417    path names.  The makefile writer is responsible for making sure that
418    these variables have correct at the time "make" is invoked.
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume
422    Date: 2000/07/22 23:30:00 JST
423    Tag: blume-20000722-urlupdate
424    Description:
425    
426    Changed BOOT and config/srcarchiveurl to point to BL server:
427    
428        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
429    
430    ----------------------------------------------------------------------
431    Name: Matthias Blume
432    Date: 2000/07/18 18:00:00 JST
433    Tag: blume-20000718-Version_110_29
434    Description:
435    
436    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
437    
438    2. Updated config/version to 110.29
439    
440    3. Updated config/srcarchiveurl
441    
442    3. New boot files!
443       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
444    
445    ----------------------------------------------------------------------
446    Name: Matthias Blume
447    Date: 2000/07/11 13:58:00 JST
448    Tag: blume-20000711-doctypo
449    Description:
450    
451    Fixed a few typos in CM manual.
452    
453    ----------------------------------------------------------------------
454    Name: Allen Leung
455    Date: 2000/06/15 00:38:00
456    Tag: leunga-20000704-sparc-x86
457    
458    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
459    2. fix to x86 RET bug [found by Dan Grossman]
460    3. sparc assembly bug fix for ticc instructions [found by Fermin]
461    
462       Affects c-- and moby only
463    
464    ----------------------------------------------------------------------
465    Name: Matthias Blume
466    Date: 2000/07/04 15:26:00
467    Tag: blume-20000704-trigger
468    Description:
469    
470    1. Improvements to CM manual.
471    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
472       of getting a back-trace.  The function, when called, raises an
473       internal exception which explicitly carries the full back-trace history,
474       so it is unaffected by any intervening handle-raise pairs ("trivial"
475       or not).  The interactive loop will print that history once it arrives
476       at top level.
477       Short of having all exceptions implicitly carry the full history, the
478       recommended way of using this facility is:
479         - compile your program with instrumentation "on"
480         - run it, when it raises an exception, look at the history
481         - if the history is "cut off" because of some handler, go and modify
482           your program so that it explicitly calls BTrace.trigger
483         - recompile (still instrumented), and rerun; look at the full history
484    
485    ----------------------------------------------------------------------
486    Name: Matthias Blume
487    Date: 2000/07/03 15:36:00 JST
488    Tag: blume-20000702-manual
489    Description:
490    
491    Small corrections and updates to CM manual.
492    
493    ----------------------------------------------------------------------
494    Name: Matthias Blume
495    Date: 2000/06/29 16:04:00 JST
496    Tag: blume-20000629-yacctool
497    Description:
498    
499    Changes:
500    
501    1. Class "mlyacc" now takes separate arguments to pass options to
502       generated .sml- and .sig-files independently.
503    2. Corresponding CM manual updates.
504    3. BTrace module now also reports call sites.  (However, for loop clusters
505       it only shows from where the cluster was entered.)  There are associated
506       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
507    
508    ----------------------------------------------------------------------
509    Name: Matthias Blume
510    Date: 2000/06/27 16:51:00 JST
511    Tag: blume-20000627-noweb
512    Description:
513    
514    Changes:
515    
516     1. Implemented "subdir" and "witness" options for noweb tool.
517        This caused some slight internal changes in CM's tool implementation.
518     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
519        some remaining issues from earlier path anchor changes.
520     3. Updated CM manual accordingly.
521    
522     4. Changed implementation of back-tracing so that I now consider it
523        ready for prime-time.
524    
525        In particular, you don't have to explicitly trigger the back-trace
526        anymore.  Instead, if you are running BTrace-instrumented code and
527        there is an uncaught exception (regardless of whether or not it was
528        raised in instrumented code), the top-level evalloop will print
529        the back-trace.
530    
531        Features:
532    
533          - Instrumented and uninstrumented code work together seemlessly.
534            (Of course, uninstrumented code is never mentioned in actual
535             back-traces.)
536    
537          - Asymptotic time- and space-complexity of instrumented code is
538            equal to that of uninstrumented code.  (This means that
539            tail-recursion is preserved by the instrumentation phase.)
540    
541          - Modules whose code has been instrumented in different sessions
542            work together without problem.
543    
544          - There is no penalty whatsoever on uninstrumented code.
545    
546          - There is no penalty on "raise" expressions, even in
547            instrumented code.
548    
549        A potential bug (or perhaps it is a feature, too):
550    
551          A back-trace reaches no further than the outermost instrumented
552          non-trivial "raise".  Here, a "trivial" raise is one that is the
553          sole RHS of a "handle" rule.  Thus, back-traces reach trough
554    
555               <exp> handle e => raise e
556    
557          and even
558    
559               <exp> handle Foo => raise Bar
560    
561          and, of course, through
562    
563               <exp> handle Foo => ...
564    
565         if the exception was not Foo.
566    
567         Back-traces always reach right through any un-instrumented code
568         including any of its "handle" expressions, trivial or not.
569    
570       To try this out, do the following:
571    
572         - Erase all existing binfiles for your program.
573           (You may keep binfiles for those modules where you think you
574            definitely don't need back-tracing.)
575         - Turn on back-trace instrumentation:
576              SMLofNJ.Internals.BTrace.mode (SOME true);
577         - Recompile your program.  (I.e., run "CM.make" or "use".)
578         - You may now turn instrumentation off again (if you want):
579              SMLofNJ.Internals.BTrace.mode (SOME false);
580         - Run your program as usual.  If it raises an exception that
581           reaches the interactive toplevel, then a back-trace will
582           automatically be printed.  After that, the toplevel loop
583           will print the exception history as usual.
584    
585    ----------------------------------------------------------------------
586    Name: Matthias Blume
587    Date: 2000/06/26 09:56:46 JST
588    Tag: blume-20000626-setup
589    Description:
590    
591    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
592          ML code before and after compiling a file (e.g., to set compiler
593          flags)
594    
595    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
596              - associated changes to btrace.sml (BTrace instrumentation pass)
597              - cleaner implementation of btimp.sml (BTrace tracing and report
598                module)
599    
600    CM manual: * new path encoding documented
601               * description of setup-parameter to "sml" added
602    
603    The biggest user-visible change to back-tracing is that it is no
604    longer necessary to compile all traced modules within the same
605    session.  (This was a real limitation.)
606    
607    ----------------------------------------------------------------------
608    Name: Matthias Blume
609    Date: 2000/06/24 12:40:00 JST
610    Tag: blume-20000624-startup
611    Description:
612    
613    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
614    bit too often -- to put it mildly. :)
615    
616    ----------------------------------------------------------------------
617    Name: Matthias Blume
618    Date: 2000/06/23 18:20:00 JST
619    Tag: blume-20000623-btrace
620    Description:
621    
622    This updates adds a backtrace facility to aid programmers in debugging
623    their programs.  This involves the following changes:
624    
625    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
626       keeping track of the current call stack.  When programs are compiled
627       in a special mode, the compiler will insert calls to these hooks
628       into the user program.
629       "Hook" means that it is possible for different implementations of
630       back-tracing to register themselves (at different times).
631    
632    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
633       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
634       It can be turned on using this call:
635         SMLofNJ.Internals.BTrace.mode (SOME true);
636       Turning it off again:
637         SMLofNJ.Internals.BTrace.mode (SOME false);
638       Querying the current status:
639         SMLofNJ.Internals.BTrace.mode NONE;
640       Annotated programs are about twice as big as normal ones, and they
641       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
642       where all hooks do nothing).  The slowdown with a plugin that is
643       actually useful (such as the one supplied by default) is even greater,
644       but in the case of the default plugin it is still only an constant
645       factor (amortized).
646    
647    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
648       with a sub-structure BTrace for controlling back-tracing.  In particular,
649       the above-mentioned function "mode" controls whether the annotation
650       phase is invoked by the compiler.  Another important function is
651       "trigger": when called it aborts the current execution and causes
652       the top-level loop to print a full back-trace.
653    
654    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
655       for back-tracing.  It keeps track of the dynamic call stack and in
656       addition to that it keeps a partial history at each "level" of that
657       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
658       d and b (at separate times, dynamically), then the report will show:
659    
660       GOTO   d
661             /c
662       GOTO  \b
663       CALL   a
664    
665       This shows that there was an initial non-tail call of a, then a
666       tail-call to b or c, looping behavior in a cluster of functions that
667       consist of b and c, and then a goto from that cluster (i.e., either from
668       b or from c) to d.
669    
670       Note that (depending on the user program) the amount of information
671       that the back-trace module has to keep track of at each level is bounded
672       by a constant.  Thus, the whole implementation has the same asymptotical
673       complexity as the original program (both in space and in time).
674    
675    5. compiler/TopLevel/interact/evalloop.sml has been modified to
676       handle the special exception SMLofNJ.Internals.BTrace.BTrace
677       which is raised by the "trigger" function mentioned above.
678    
679    Notes on usage:
680    
681    - Annotated code works well together with unannotated code:
682    Unannotated calls simply do not show up at all in the backtrace.
683    
684    - It is not a good idea to let modules that were annotated during
685    different sessions run at the same time.  This is because the compiler
686    chooses small integers to identify individual functions, and there
687    will be clashes if different modules were compiled in separate sessions.
688    (Nothing will crash, and you will even be told about the clashes, but
689    back-trace information will in general not be useful.)
690    
691    - Back-tracing can be confused by callcc and capture.
692    
693    - The only way of getting a back-trace right now is to explicitly
694    invoke the "trigger" function from your user program.  Eventually, we
695    should make every exception carry back-trace information (if
696    available).  But since this creates more overhead at "raise"-time
697    (similar to the current exnHistory overhead), I have not yet
698    implemented this.  (The implementation will be rather easy.)  With
699    exceptions carrying back-trace information, this facility will be even
700    more useful because users don't need to modify their programs...
701    
702    - While it is possible to compile the compiler with back-trace
703    annotations turned on (I did it to get some confidence in
704    correctness), you must make absolutely sure that core.sml and
705    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
706    be compiled with annotation because there is no core access yet, but
707    if you compile btimp.sml with annotation, then the system will go into
708    an infinite recursion and crash.)
709    Since CM currently does not know about BTrace, the only way to turn
710    annotations on and off for different modules of the compiler is to
711    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
712    this is awkward and clumsy.
713    
714    Sample sessions:
715    
716    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
717    - SMLofNJ.Internals.BTrace.mode (SOME true);
718    [autoloading]
719    [autoloading done]
720    val it = false : bool
721    - structure X = struct
722    -     fun main n = let
723    -         fun a (x, 0) = d x
724    -           | a (x, n) = b (x, n - 1)
725    -         and b (x, n) = c (x, n)
726    -         and c (x, n) = a (x, n)
727    -         and d x = e (x, 3)
728    -         and e (x, 0) = f x
729    -           | e (x, n) = e (x, n - 1)
730    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
731    -           | f n = n * g (n - 1)
732    -         and g n = a (n, 3)
733    -     in
734    -         f n
735    -     end
736    - end;
737    structure X : sig val main : int -> int end
738    - X.main 3;
739    *** BACK-TRACE ***
740    GOTO   stdIn:4.2-13.20: X.main[2].f
741    GOTO-( stdIn:4.2-13.20: X.main[2].e
742    GOTO   stdIn:4.2-13.20: X.main[2].d
743         / stdIn:4.2-13.20: X.main[2].a
744         | stdIn:4.2-13.20: X.main[2].b
745    GOTO-\ stdIn:4.2-13.20: X.main[2].c
746    CALL   stdIn:4.2-13.20: X.main[2].g
747    GOTO   stdIn:4.2-13.20: X.main[2].f
748    GOTO-( stdIn:4.2-13.20: X.main[2].e
749    GOTO   stdIn:4.2-13.20: X.main[2].d
750         / stdIn:4.2-13.20: X.main[2].a
751         | stdIn:4.2-13.20: X.main[2].b
752    GOTO-\ stdIn:4.2-13.20: X.main[2].c
753    CALL   stdIn:4.2-13.20: X.main[2].g
754    GOTO   stdIn:4.2-13.20: X.main[2].f
755    GOTO-( stdIn:4.2-13.20: X.main[2].e
756    GOTO   stdIn:4.2-13.20: X.main[2].d
757         / stdIn:4.2-13.20: X.main[2].a
758         | stdIn:4.2-13.20: X.main[2].b
759    GOTO-\ stdIn:4.2-13.20: X.main[2].c
760    CALL   stdIn:4.2-13.20: X.main[2].g
761    GOTO   stdIn:4.2-13.20: X.main[2].f
762    CALL   stdIn:2.15-17.4: X.main[2]
763    -
764    
765    (Note that because of a FLINt bug the above code currently does not
766    compile without BTrace turned on.)
767    
768    Here is another example, using my modified Tiger compiler:
769    
770    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
771    - SMLofNJ.Internals.BTrace.mode (SOME true);
772    [autoloading]
773    [autoloading done]
774    val it = false : bool
775    - CM.make "sources.cm";
776    [autoloading]
777    ...
778    [autoloading done]
779    [scanning sources.cm]
780    [parsing (sources.cm):parse.sml]
781    [creating directory CM/SKEL ...]
782    [parsing (sources.cm):tiger.lex.sml]
783    ...
784    [wrote CM/sparc-unix/semant.sml]
785    [compiling (sources.cm):main.sml]
786    [wrote CM/sparc-unix/main.sml]
787    [New bindings added.]
788    val it = true : bool
789    - Main.compile ("../testcases/merge.tig", "foo.out");
790    *** BACK-TRACE ***
791    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
792    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
793    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
794    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
795    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
796    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
797    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
798         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
799    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
800    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
801    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
802    CALL   main.sml:1.18-118.4: Main.compile[2]
803    -
804    
805    ----------------------------------------------------------------------
806    Name: Matthias Blumen
807    Date: 2000/06/21 18:00:00 JST
808    Tag: blume-20000621-manual
809    Description:
810    
811    CM manual update: Path environments documented.
812    
813    ----------------------------------------------------------------------
814    Name: Matthias Blume
815    Date: 2000/06/19 13:40:00
816    Tag: blume-20000619-manual
817    Description:
818    
819    CM manual and system/README update.  This only covers the fact that
820    there are no more implicit anchors.  (Path environments and the "bind"
821    option to "cm" have yet to be documented.)
822    
823    ----------------------------------------------------------------------
824    Name: Matthias Blume
825    Date: 2000/06/19 11:05:00 JST
826    Tag: blume-20000619-chdir-bugfix
827    Description:
828    
829    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
830    
831    ----------------------------------------------------------------------
832    Name: Matthias Blume
833    Date: 2000/06/18 22:00:10 JST
834    Tag: blume-20000618-implicit-anchors-really-gone
835    Description:
836    
837    I updates the previous HISTORY entry where I forgot to mention that
838    implicit anchors are no longer with us.
839    
840    The current update also gets rid of the (now useless) controller
841    CM.Control.implicit_anchors.
842    
843    ----------------------------------------------------------------------
844    Name: Matthias Blume
845    Date: 2000/06/16 17:30:00 JST
846    Tag: blume-20000616-anchorenv
847    Description:
848    
849    This patch implements the long anticipated (just kidding :) "anchor
850    environment" mechanism.  In the course of doing this, I also
851    re-implemented CM's internal "SrcPath" module from scratch.  The new
852    one should be more robust in certain boundary cases.  In any case, it
853    is a lot cleaner than its predecessor (IMHO).
854    
855    This time, although there is yet another boot file format change, I
856    kept the unpickler backward-compatible.  As a result, no new bootfiles
857    are necessary and bootstrapping is straightforward.  (You cannot read
858    new bootfiles into an old system, but the other way around is no
859    problem.)
860    
861    Visible changes:
862    
863    ** 0. Implicit path anchors (without the leading $-symbol) are no
864    longer recognized at all. This means that such path names are not
865    illegal either.  For example, the name basis.cm simply refers to a
866    local file called "basis.cm" (i.e, the name is an ordinary path
867    relative to .cm-files directory).  Or, to put it differently, only
868    names that start with $ are anchored paths.
869    
870    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
871    vanished.
872    
873    John (Reppy) had critizised this as soon as I originally proposed and
874    implemented it, but at that time I did not really deeply believe
875    him. :) Now I came full-circle because I need the $<singlearc> syntax
876    in another place where it cannot be seen as an abbreviation for
877    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
878    seems to mean (i.e., it "expands" into the corresponding anchor
879    value).
880    
881    However, when paths are used as members in CM description files, it
882    continues to be true that there must be at least another arc after the
883    anchor.  This is now enforced separately during semantic analysis
884    (i.e., from a lexical/syntactical point of view, the notation is ok.)
885    
886    ** 2. The "cm" class now accepts an option "bind".  The option's value
887    is a sub-option list of precisely two items -- one labeled "anchor"
888    and the other one labeled "value".  As you might expect, "anchor" is
889    used to specify an anchor name to be bound, and "value" specifies what
890    the anchor is being bound to.
891    
892    The value must be a directory name and can be given in either standard
893    syntax (including the possibility that it is itself an anchored path)
894    or native syntax.
895    
896    Examples:
897    
898       foo.cm (bind:(anchor:bar value:$mystuff/bar))
899       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
900    
901    and so on.
902    
903    The meaning of this is that the .cm-file will be processed with an
904    augmented anchor environment where the given anchor(s) is/are bound to
905    the given values(s).
906    
907    The rationale for having this feature is this: Suppose you are trying
908    to use two different (already stable) libraries a.cm and b.cm (that
909    you perhaps didn't write yourself).  Further, suppose each of these
910    two libraries internally uses its own auxiliary library $aux/lib.cm.
911    Normally you would now have a problem because the anchor "lib" can not
912    be bound to more than one value globally.  Therefore, the project that
913    uses both a.cm and b.cm must locally redirect the anchor to some other
914    place:
915    
916       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
917       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
918    
919    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
920    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
921    
922    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
923    you when you do so at the time of CM.stabilize).  Therefore, you can
924    also use an anchored path as the value:
925    
926      a.cm (bind:(anchor:lib value:$a-lib))
927      b.cm (bind:(anchor:lib value:$b-lib))
928    
929    Now you can globally configure (using the usual CM.Anchor.anchor or
930    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
931    itself is always locally bound, setting it globally is no longer
932    meaningful or necessary (but it does not hurt either).  In fact, "lib"
933    can still be used as a global anchor for separate purposes.  As a
934    matter of fact, one can locally define "lib" in terms of a global
935    "lib":
936    
937      a.cm (bind:(anchor:lib value:$lib/a))
938      b.cm (bind:(anchor:lib value:$lib/b))
939    
940    ** 3: The encoding of path names has changed.  This affects the way
941    path names are shown in CM's progress report and also the internal
942    protocol encoding used for parallel make.
943    
944    The encoding now uses one or more ':'-separated segments.  Each
945    segments corresponds to a file that has been specified relative to the
946    file given by its preceding segment.  The first segment is either
947    relative to the CWD, absolute, or anchored.  Each segment itself is
948    basically a Unix pathname; all segments but the first are relative.
949    
950    Example:
951    
952       $foo/bar/baz.cm:a/b/c.sml
953    
954    This path denotes the file bar/a/b/c.sml relative to the directory
955    denoted by anchor "foo".  Notice that the encoding also includes
956    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
957    paths are resolved relative to the .cm-files directory, so baz.cm must
958    be ignored to get the "real" pathname.
959    
960    To make this fact more obvious, CM puts the names of such "virtual
961    arcs" into parentheses when they appear in progress reports. (No
962    parentheses will appear in the internal protocol encoding.)  Thus,
963    what you really see is:
964    
965      $foo/bar/(baz.cm):a/b/c.sml
966    
967    I find this notation to be much more informative than before.
968    
969    Another new feature of the encoding is that special characters
970    including parentheses, colons, (back)slashes, and white space are
971    written as \ddd (where ddd is the decimal encoding of the character).
972    
973    *** The CM manual still needs to be updated.
974    
975    ----------------------------------------------------------------------
976    Name: Allen Leung
977    Date: 2000/06/15 00:38:00
978    Tag: leunga-20000615-x86-peephole
979    
980    x86 Peephole fix by Fermin.  Affects c-- and moby only.
981    
982    ----------------------------------------------------------------------
983    Name: Matthias Blume
984    Date: 2000/06/12 11:40:00
985    Tag: blume-20000612-parmakefix
986    Description:
987    
988    More cleanup after changing the file naming scheme: This time I
989    repaired the parallel make mechanism for CMB.make which I broke earlier.
990    
991    ----------------------------------------------------------------------
992    Name: Allen Leung
993    Date: 2000/06/09 01:25:00
994    Tag: leunga-20000609-various
995    
996    None of these things should affect normal SML/NJ operations
997    
998    1. Peephole improvements provided by Fermin (c--)
999    2. New annotation DEFUSE for adding extra dependence (moby)
1000    3. New X86 LOCK instructions (moby)
1001    4. New machine description language for reservation tables (scheduling)
1002    5. Fixes to various optimization/analysis modules (branch chaining, dominator
1003       trees etc.)
1004    6. I've changed the CM files so that they can work with versions
1005       110.0.6, 110.25 and 110.28
1006    
1007    ----------------------------------------------------------------------
1008    Name: Matthias Blume
1009    Date: 2000/06/09 12:40:00
1010    Tag: blume-20000609-log
1011    Description:
1012    
1013    - Removed all(?) remaining RCS Log entries from sources.
1014    
1015    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1016      anchored paths).
1017    
1018    ----------------------------------------------------------------------
1019    Name: Matthias Blume
1020    Date: 2000/06/07 17:00:00 JST
1021    Tag: blume-20000607-no-implicit-anchors
1022    Description:
1023    
1024    1. This update changes the default setting for
1025    CM.Control.implicit_anchors from true to false.  This means that
1026    implicit anchors are no longer permitted by default.  I also tried to
1027    make sure that nothing else still relies on implicit anchors.
1028    (This is the next step on the schedule towards a CM that does not even
1029    have the notion of implicit anchors anymore.)
1030    
1031    2. More CM manual updates.
1032    
1033    3. I managed to track down and fix the pickling bug I mentioned last
1034    time.  Because of the previously existing workaround, this entails no
1035    immediate practical changes.
1036    
1037    ----------------------------------------------------------------------
1038    Name: Matthias Blume
1039    Date: 2000/06/06 11:15:00 JST
1040    Tag: blume-20000606-lazierpickle
1041    Description:
1042    
1043    !!!! NEW BOOT FILES !!!!
1044    
1045    * The main purpose of this update is to make library pickles lazier in
1046    order to reduce the initial space penalty for autoloading a library.
1047    As a result, it is now possible to have $smlnj/compiler.cm
1048    pre-registered.  This should take care of the many complaints or
1049    inquiries about missing structure Compiler.  This required changes to
1050    CM's internal data structures and small tweaks to some algorithms.
1051    
1052    As a neat additional effect, it is no longer necessary (for the sake
1053    of lean heap image files) to distinguish between a "minimal" CM and a
1054    "full" CM.  Now, there is only one CM (i.e., the "full" version:
1055    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1056    interactive top level. ($smlnj/cm/minimal.cm is gone.)
1057    
1058    To make the life of compiler-hackers easier, "makeml" now also
1059    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1060    words, after you bootstrap a new sml for the first time, you will not
1061    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1062    around you will still have to do it, though.)
1063    
1064    * A second change consists of major updates to the CM manual.  There
1065    are now several appendices with summary information and also a full
1066    specification of the CM description file syntax.
1067    
1068    * In directory src/system I added the script "allcross".  This script
1069    invokes sml and cross-compiles the compiler for all supported
1070    architectures.  (Useful when providing a new set of boot files.)
1071    
1072    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
1073    added a small tweak to pickle-util.sml to work around this problem,
1074    but it is not a proper fix yet.  I will investigate further.  (The
1075    effect of the bug was an inflation of library pickle size.)
1076    
1077    * Version number increased to 110.28.1 (to avoid compatibility problems).
1078    
1079    ----------------------------------------------------------------------
1080    Name: Allen Leung
1081    Date: 2000/05/25 17:28 EDT
1082    Tag: leunga-20000525-ra
1083    Description:
1084    
1085      Fixed a bug in freezing phase of the register allocator.
1086    
1087    ----------------------------------------------------------------------
1088    Name: Allen Leung
1089    Date: 2000/05/15 22:53 EDT
1090    Tag: leunga-20000515-alpha-x86-ra
1091    Description:
1092    
1093      1. Alpha
1094    
1095          Slight cleanup.  Removed the instruction SGNXL
1096    
1097      2. X86
1098    
1099          Added the following instructions to the instruction set:
1100    
1101            ROLx, RORx,
1102            BTx, BTSx, BTLx, BTRx,
1103            XCHGx, and variants with the LOCK prefix
1104    
1105      3. Register Allocation
1106    
1107          The module ra-rewrite-with-renaming has been improved.
1108    
1109      These have no effect on SML/NJ.
1110    
1111    ----------------------------------------------------------------------
1112    Name: Matthias Blume
1113    Date: 2000/05/15 16:20:00 JST
1114    Tag: blume-20000515-lightrebuild
1115    Description:
1116    
1117    1. I added an alternative to "-rebuild" to "makeml".  The difference is
1118       that prior to calling CMB.make' the CM-variable "LIGHT" will be
1119       defined.  In effect, the command will not build any cross-compiler
1120       backends and therefore finish more quickly.
1121    
1122       The "fixpt" script also takes a "-light" switch to be able to use
1123       this new facility while compiling for a fixpoint.
1124    
1125    2. I replaced all mentions of anchored paths in group owner specifications
1126       with simple relative paths (usually starting with "..").
1127       The rationale is that a library's internal workings should not be
1128       compromised by the lack of some anchor.  (An anchor is necessary
1129       for someone who wants to refer to the library by an anchored path,
1130       but it should not be necessary to build the same library in the first
1131       place.)
1132    
1133    3. I changed the way CM's tool mechanism determines the shell command
1134       string used for things like ml-yacc etc. so that it does not break
1135       when CM.Control.implicit_anchors is turned off.
1136    
1137    ----------------------------------------------------------------------
1138    Name: Matthias Blume
1139    Date: 2000/05/12 18:20:00 JST
1140    Tag: blume-20000512-ml-build
1141    Description:
1142    
1143    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1144    getting installed properly (by config/install.sh).
1145    
1146    ----------------------------------------------------------------------
1147    Name: Matthias Blume
1148    Date: 2000/05/12 17:30:00 JST
1149    Tag: blume-20000512-anchors
1150    Description:
1151    
1152    !!! NEW BOOT FILES !!!
1153    
1154    This change is in preparation of fading out support for "implicitly
1155    anchored path names".  I went through all sources and used the
1156    explicit (and relatively new) $-notation.  See system/README and the
1157    CM manual for more info on this.
1158    
1159    I also modified the anchoring scheme for some things such as "smlnj",
1160    "MLRISC", "cm", etc. to take advantage of the fact that explicit
1161    anchors are more expressive: anchor name and first arc do not have to
1162    coincide.  This entails the following user-visible change:
1163    
1164    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1165    particular, when you fire up sml with a command-line argument, say,
1166    e.g.:
1167    
1168       sml '$smlnj/cmb.cm'
1169    
1170    At the ML toplevel prompt:
1171    
1172       CM.autoload "$smlnj/cmb.cm";
1173    
1174    There is also a new controller in CM.Control that can be used to turn
1175    off all remaining support for implicit anchors by saying:
1176    
1177        CM.autoload "$smlnj/
1178        #set CM.Control.implicit_anchors false;
1179    
1180    This causes CM to reject implicitly anchored paths.  This is (for the
1181    time being) less permissive than the "final" version where there will
1182    be no more such implicit anchors and relative paths will be just that:
1183    relative.
1184    
1185    The next step (version after next version?) will be to make the
1186    default for CM.Control.implicit_anchors false.  After the dust has
1187    settled, I can then produce the "final" version of this...
1188    
1189    Note: Since bootstrapping is a bit tricky, I provided new boot files.
1190    
1191    ----------------------------------------------------------------------
1192    Name: Matthias Blume
1193    Date: 2000/05/11 16:30:00 JST
1194    Tag: blume-20000511-sources
1195    Description:
1196    
1197    The main change is that I added function CM.sources as a generalized
1198    version of the earlier CM.makedepend.  This entails the following
1199    additional changes:
1200    
1201      - CM.makedepend has been dropped.
1202    
1203      - CM manual has been updated.
1204    
1205      - TOOLS signature and API have been changed.
1206    
1207    ----------------------------------------------------------------------
1208    Name: Allen Leung
1209    Date: 2000/05/10 21:17 EDT
1210    Tag: leunga-20000510-moby-c--ssa
1211    Description:
1212    
1213      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1214    None of these affect SML/NJ.
1215    
1216    1. Register Allocation
1217    
1218        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1219           This module tries to remove local (i.e. basic block level) redundancies
1220           during spilling.
1221    
1222        b. A new framework for performing region based register allocation.
1223           Not yet entirely functional.
1224    
1225    2. X86
1226    
1227       a. DefUse for POP was missing the stack pointer [found by Lal]
1228       b. Reload for CALL was incorrect in X86Spill [found by John]
1229       c. Various fixes in X86Spill so that it can be used correctly for
1230          the new spilling module.
1231    
1232    3. SSA/IR
1233    
1234       a. New module ir/dj-dataflow.sml implements elimination based
1235          data flow analysis.
1236    
1237    4. MLRiscGen
1238    
1239       a. Fix for gc type annotation
1240    
1241    5. MDGen
1242    
1243       Various fixes for machine description -> ml code translation.  For ssa
1244       only.
1245    
1246    ----------------------------------------------------------------------
1247    Name: Allen Leung
1248    Date: 2000/05/08 22:17 EDT
1249    Tag: leunga-20000508-labexp
1250    Description:
1251    
1252      Fermin has found a few assembly problems with constant expressions
1253      generated in LabelExp.  Mostly, the problems involve extra parentheses,
1254      which choke on dumb assemblers.  This is his fix.
1255    
1256    ----------------------------------------------------------------------
1257    Name: Dave MacQueen
1258    Date: 2000/04/09 14:00 EDT
1259    Tag: dbm-20000502-Version_110_28
1260    Description:
1261    
1262    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1263    
1264    2. Updated config/version to 110.28
1265    
1266    3. Updated config/srcarchiveurl
1267    
1268    3. New boot files!
1269       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1270    
1271    ----------------------------------------------------------------------
1272    Name: Matthias Blume
1273    Date: 2000/05/01 19:05:00 JST
1274    Tag: blume-20000501-noweb
1275    Description:
1276    
1277    A new noweb tool has been added.  The existing system is entirely
1278    unaffected by this, but some CM users have asked for renewed noweb
1279    support.  Everything is documented in the CM manual.
1280    
1281    New (plugin) libraries:
1282    
1283       noweb-tool.cm
1284       nw-ext.cm
1285    
1286    ----------------------------------------------------------------------
1287    Name: Dave MacQueen
1288    Date: 2000/04/30 12:40PM EDT
1289    Tag: dbm-20000430-bug_fixes
1290    Description:
1291    
1292    1. Fix for bug 1498
1293       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1294       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1295         added toRealArray function
1296       smlnj/src/compiler/MiscUtil/print/ppobj.sml
1297         added check for tag Obj.RealArray to array printing case in ppObj
1298    
1299    2. Fix for bug 1510
1300       smlnj/src/compiler/Semant/types/typesutil.sml
1301         fixed definition of dummyargs (used by equalTycon) so that
1302         dummy args are distinct types
1303    
1304    ----------------------------------------------------------------------
1305    Name: Matthias Blume
1306    Date: 2000/04/30 01:00:00 JST
1307    Tag: blume-20000430-versions
1308    Description:
1309    
1310    1. CM version numbering added.  This is an implementation of Lal's
1311       proposal for adding version numbers and version checking to .cm
1312       files.  Lal said that his proposal was just that -- a proposal.
1313       For the time being I went ahead and implemented it so that people
1314       can comment on it.  Everything is completely backward-compatible
1315       (except for the stable library format, i.e., new bootfiles!).
1316    
1317       As usual, see the CM manual for details.
1318    
1319    2. An alternative syntax for anchored paths has been implemented.
1320       Dave has recently voiced the same concerns that I had when I did
1321       this, so there should be some support.  My take is that eventually
1322       I will let support for the current syntax (where anchors are
1323       "implicit") fade out in favor of the new, explicit syntax.
1324       In order to be backward-compatible, both old and new syntax are
1325       currently supported.
1326    
1327       Again, see the CM manual for details.
1328    
1329    3. Parallel make is trying to be slightly smarter:  When the master
1330       process finds a "bottleneck", i.e., when there is only one
1331       compilation unit that can be compiled and everybody else is
1332       waiting on it, then it will simply compile it directly instead
1333       of clumsily telling one of the slaves to do it.
1334    
1335    4. Support for "unsharing" added.  This is necessary in order to be
1336       able to have two different versions of the same library running
1337       at the same time (e.g., for trying out a new MLRISC while still
1338       having the old MLRISC linked into the current compiler, etc.)
1339       See the CM manual.
1340    
1341    5. Simple "makedepend" functionality added for generating Makefile
1342       dependency information.  (This is rather crude at the moment.
1343       Expect some changes here in the future.)
1344    
1345    6. ".fun" added as a recognized suffix for ML files. Also documented
1346       explicitly in the manual that the fallback behavior (unknown suffix
1347       -> ML file) is not an official feature!
1348    
1349    7. Small changes to the pickler for stable libraries.
1350    
1351    8. Several internal changes to CM (for cleanup/improvement).
1352    
1353    
1354    !!!! NEW BINFILES !!!!
1355    
1356    ----------------------------------------------------------------------
1357    Name: Matthias Blume
1358    Date: 2000/04/28 17:30:00 JST
1359    Tag: blume-20000428-pathconfig
1360    Description:
1361    
1362    1. I changed config/install.sh to remove duplicate entries from the
1363       lib/pathconfig file at the end.  Moreover, the final version of
1364       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
1365       in src/system/installml.
1366    
1367    2. The config/install.sh script now consistently uses relative
1368       pathnames in lib/pathconfig whenever the anchor is in the lib
1369       directory.  (So far this was true for the libraries that come
1370       pre-compiled and bundled as part of the bootfiles but not for
1371       libraries that are compiled by the script itself.)
1372    
1373    ----------------------------------------------------------------------
1374    Name: Matthias Blume
1375    Date: 2000/04/26 13:10:00 JST
1376    Tag: blume-20000426-fun_suffix
1377    Description:
1378    
1379    Added ".fun" as a recognized file name suffix (for ML code).
1380    
1381    ----------------------------------------------------------------------
1382    Name: Allen Leung
1383    Date: 2000/04/25 17:00:00 EST
1384    Tag: leunga-20000425-alpha-ra
1385    Description:
1386    
1387    1. Alpha
1388    
1389        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
1390    in C--.
1391    
1392    2. RA
1393    
1394       Added a flag "ra-dump-size" to print out the size of the flowgraph
1395       and the interference graph.
1396    
1397    ----------------------------------------------------------------------
1398    Name: Dave MacQueen
1399    Date: 2000/04/25/
1400    Tag: dbm-20000425-mlyacc_doc_examples
1401    Description:
1402      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
1403      Updated all three examples in src/ml-yacc/examples to run
1404      under 110.* using CM.make.
1405    
1406    ----------------------------------------------------------------------
1407    Name: Allen Leung
1408    Date: 2000/04/20 23:04:00 EST
1409    Tag: leunga-20000420-ssa-c---stuff
1410    Description:
1411    
1412      This update synchronizes my repository with Yale's.  Most of these
1413    changes, however, do not affect SML/NJ at all (the RA is an exception).
1414    
1415    1. Register Allocator
1416    
1417       a. An improvement in the interference graph construction:
1418          Given a copy
1419    
1420                s <- t
1421    
1422          no interference edge between s and t is added for this definition of s.
1423    
1424       b. I've added two new spill heuristic modules that Fermin and I developed
1425          (in the new library RA.cm). These are unused in SML/NJ but maybe
1426          useful for others (Moby?)
1427    
1428    2. X86
1429    
1430       a. Various fixes in the backend provided by Fermin [C--] and Lal.
1431    
1432    3. Alpha
1433    
1434       a. Added the BSR instruction and code generation that goes with it [C--]
1435       b. Other fixes too numerous to recount provided by Fermin [C--]
1436    
1437    4. Regmaps
1438    
1439       a. The regmaps are not initialized with the identity physical bindings
1440          at creation time.  This is unneeded.
1441    
1442    5. MLRISC Optimizations
1443    
1444       a. The DJ-Graph module can now compute the iterated dominance frontiers
1445          intersects with liveness incrementally in linear time! Woohoo!
1446          This is now used in my new SSA construction algorithm.
1447    
1448       b. THe branch reorganization module is now smarter about linear chains of
1449          basic blocks.
1450    
1451    
1452    ----------------------------------------------------------------------
1453    Name: Matthias Blume
1454    Date: 2000/04/12 13:52:00 JST
1455    Tag: blume_main_v110p27_1
1456    Description:
1457    
1458    Changed install.sh script to handle archive files without version number
1459    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
1460    name of the boot file archive.
1461    
1462    ----------------------------------------------------------------------
1463    Name: Dave MacQueen
1464    Date: 2000/04/09 14:00 EDT
1465    Tag: dbm-20000410-Version_110_27
1466    Description:
1467    
1468    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
1469    
1470    2. Updated src/config/version to 110.27
1471    
1472    3. New boot files!
1473    
1474    ----------------------------------------------------------------------
1475    Name: Allen Leung
1476    Date: 2000/04/09 19:09:00 EST
1477    Tag: leunga-20000409-misc
1478    Description:
1479    
1480    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
1481    
1482    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
1483    
1484    ----------------------------------------------------------------------
1485    Name: Stefan
1486    Date: 2000/04/07 10:00:00 EDT
1487    Tag: monnier-20000406-branch-handling
1488    Description:
1489    
1490    Improved handling of branches (mostly those generated from
1491    polymorphic equality), removed switchoff and changed the
1492    default optimization settings (more cpsopt and less flintopt).
1493    
1494    ----------------------------------------------------------------------
1495    Name: Allen Leung
1496    Date: 2000/04/06 01:30:00 EST
1497    Tag: leunga-20000406-peephole-x86-SSA-2
1498    Description:
1499    
1500       Forgot a few files.
1501    
1502    ----------------------------------------------------------------------
1503    Name: Allen Leung
1504    Date: 2000/04/06 00:36:00 EST
1505    Tag: leunga-20000406-peephole-x86-SSA
1506    Description:
1507    
1508    1.  New Peephole code
1509    
1510    2.  Minor improvement to X86 instruction selection
1511    
1512    3.  Various fixes to SSA and machine description -> code translator
1513    
1514    ----------------------------------------------------------------------
1515    Name: Matthias Blume
1516    Date: 2000/04/05 12:30:00 JST
1517    Tag: blume_main_v110p26p2_3
1518    Description:
1519    
1520    This update just merges three minor cosmetic updates to CM's sources
1521    to get ready for the 110.27 code freeze on Friday.  No functionality
1522    has changed.
1523    
1524    ----------------------------------------------------------------------
1525    Name: Allen Leung
1526    Date: 2000/04/04 19:39:00 EST
1527    Tag: leunga-20000404-x86-asm
1528    Description:
1529    
1530    1.  Fixed a problem in X86 assembly.
1531    
1532        Things like
1533    
1534           jmp %eax
1535           jmp (%eax)
1536    
1537        should be output as
1538    
1539           jmp *%eax
1540           jmp *(%eax)
1541    
1542    2.  Assembly output
1543    
1544          Added a new flag
1545    
1546              "asm-indent-copies" (default to false)
1547    
1548          When this flag is on, parallel copies will be indented an extra level.
1549    
1550    ----------------------------------------------------------------------
1551    Name: Allen Leung
1552    Date: 2000/04/04 03:18:00 EST
1553    Tag: leunga-20000404-C--Moby
1554    Description:
1555    
1556        All of these fixes are related to C--, Moby, and my own optimization
1557        stuff; so they shouldn't affect SML/NJ.
1558    
1559    1.  X86
1560    
1561        Various fixes related floating point, and extensions.
1562    
1563    2.  Alpha
1564    
1565        Some extra patterns related to loads with signed/zero extension
1566        provided by Fermin.
1567    
1568    3.  Assembly
1569    
1570        When generating assembly, resolve the value of client defined constants,
1571        instead of generating symbolic values.  This is controlled by the
1572        new flag "asm-resolve-constants", which is default to true.
1573    
1574    4.  Machine Descriptions
1575    
1576        a. The precedence parser was slightly broken when parsing infixr symbols.
1577        b. The type generalizing code had the bound variables reversed, resulting
1578           in a problem during arity raising.
1579        c. Various fixes in machine descriptions.
1580    
1581    ----------------------------------------------------------------------
1582    Name: Matthias Blume
1583    Date: 2000/04/03 16:05:00 JST
1584    Tag: blume_main_v110p26p2_2
1585    Description:
1586    
1587    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
1588    now done via the ordinary static environment that is context to each
1589    compilation unit.
1590    
1591    To this end, I arranged that instead of "structure Core" as "structure
1592    _Core" is bound in the pervasive environment.  Core access is done via
1593    _Core (which can never be accidentally rebound because _Core is not a
1594    legal surface-syntax symbol).
1595    
1596    The current solution is much cleaner because the core environment is
1597    now simply part of the pervasive environment which is part of every
1598    compilation unit's context anyway.  In particular, this eliminates all
1599    special-case handling that was necessary until now in order to deal
1600    with dynamic and symbolic parts of the core environment.
1601    
1602    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1603    compilation manager's bootstrap compiler (actually: in the "init group"
1604    handling).  See the comments in src/system/smlnj/init/init.cmi for
1605    more details.
1606    
1607    I also tried to track down all mentions of "Core" (as string argument
1608    to Symbol.strSymbol) in the compiler and replaced them with a
1609    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
1610    name appears in one place only.
1611    
1612    Binfile and bootfile format have not changed, but the switchover from
1613    the old "init.cmi" to the new one is a bit tricky, so I supplied new
1614    bootfiles anyway.
1615    
1616    ----------------------------------------------------------------------
1617    Name: Allen Leung
1618    Date: 2000/04/02 21:17:00 EST
1619    Tag: leunga-20000402-mltree
1620    Description:
1621    
1622       1. Renamed the constructor CALL in MLTREE by popular demand.
1623       2. Added a bunch of files from my repository.  These are currently
1624          used by other non-SMLNJ backends.
1625    
1626    ----------------------------------------------------------------------
1627    Name: Allen Leung
1628    Date: 2000/03/31 21:15:00 EST
1629    Tag: leunga-20000331-aliasing
1630    Description:
1631    
1632    This update contains a rewritten (and hopefully more correct) module
1633    for extracting aliasing information from CPS.
1634    
1635       To turn on this feature:
1636    
1637            Compiler.Control.CG.memDisambiguate := true
1638    
1639       To pretty print the region information with assembly
1640    
1641           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
1642    
1643       To control how many levels of aliasing information are printed, use:
1644    
1645           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
1646    
1647       The default of n is 3.
1648    
1649    ----------------------------------------------------------------------
1650    Name: David MacQueen
1651    Date: 2000/03/31 11:15:00 EST
1652    Tag: dbm-20000331-runtime_fix
1653    Description:
1654    
1655    This update contains:
1656    
1657    1. runtime/c-lib/c-libraries.c
1658       includes added in revision 1.2 caused compilation errors on hppa-hpux
1659    
1660    2. fix for bug 1556
1661       system/Basis/Implementation/NJ/internal-signals.sml
1662    
1663    ----------------------------------------------------------------------
1664    Name: Matthias Blume
1665    Date: 2000/03/31 18:00:00 JST
1666    Tag: blume_main_v110p26p2_1
1667    Description:
1668    
1669    This update contains:
1670    
1671    1. A small change to CM's handling of stable libraries:
1672       CM now maintains one "global" modmap that is used for all stable
1673       libraries.  The use of such a global modmap maximizes sharing and
1674       minimizes the need for re-traversing parts of environments during
1675       modmap construction.  (However, this has minor impact since modmap
1676       construction seems to account for just one percent or less of total
1677       compile time.)
1678    
1679    2. I added a "genmap" phase to the statistics.  This is where I got the
1680       "one percent" number (see above).
1681    
1682    3. CM's new tool parameter mechanism just became _even_ better. :)
1683       - The parser understands named parameters and recursive options.
1684       - The "make" and "shell" tools use these new features.
1685         (This makes it a lot easier to cascade these tools.)
1686       - There is a small syntax change: named parameters use a
1687    
1688           <name> : ( <option> ... )            or
1689           <name> : <string>
1690    
1691         syntax.  Previously, named parameters were implemented in an
1692         ad-hoc fashion by each tool individually (by parsing strings)
1693         and had the form
1694    
1695           <name>=<string>
1696    
1697       See the CM manual for a full description of these issues.
1698    
1699    ----------------------------------------------------------------------
1700    Name: Matthias Blume
1701    Date: 2000/03/30 18:00:00 JST
1702    Tag: blume_main_v110p26p2_0
1703    Description:
1704    
1705    !!!!! WARNING !!!!!!
1706    !!  New binfiles  !!
1707    !!!!!!!!!!!!!!!!!!!!
1708    
1709    This update contains:
1710    
1711    1. Moderate changes to CM:
1712    
1713       - Changes to CM's tools mechanism.  In particular, it is now possible
1714       to have tools that accept additional "command line" parameters
1715       (specified in the .cm file at each instance where the tool's class is
1716       used).
1717    
1718       This was done to accommodate the new "make" and "shell" tools which
1719       facilitate fairly seamless hookup to portions of code managed using
1720       Makefiles or Shell scripts.
1721    
1722       There are no classes "shared" or "private" anymore.  Instead, the
1723       sharing annotation is now a parameter to the "sml" class.
1724    
1725       There is a bit of generic machinery for implementing one's own
1726       tools that accept command-line parameters.  However, I am not yet fully
1727       satisfied with that part, so expect changes here in the future.
1728    
1729       All existing tools are described in the CM manual.
1730    
1731       - Slightly better error handling.  (CM now suppresses many followup
1732       error messages that tended to be more annoying than helpful.)
1733    
1734    2. Major changes to the compiler's static environment data structures.
1735    
1736       - no CMStaticEnv anymore.
1737            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
1738              but it is called Environment), no conversions between different
1739              kinds of static environments
1740    
1741       - There is still a notion of a "modmap", but such modmaps are generated
1742         on demand at the time when they are needed.  This sounds slow, but I
1743         sped up the code that generates modmaps enough for this not to lead to
1744         a slowdown of the compiler (at least I didn't detect any).
1745    
1746       - To facilitate rapid modmap generation, static environments now
1747         contain an (optional) "modtree" structure.  Modtree annotations are
1748         constructed by the unpickler during unpickling.  (This means that
1749         the elaborator does not have to worry about modtrees at all.)
1750         Modtrees have the advantage that they are compositional in the same
1751         way as the environment data structure itself is compositional.
1752         As a result, modtrees never hang on to parts of an environment that
1753         has already been rendered "stale" by filtering or rebinding.
1754    
1755       - I went through many, many trials and errors before arriving at the
1756         current solution.  (The initial idea of "linkpaths" did not work.)
1757         But the result of all this is that I have touched a lot of files that
1758         depend on the "modules" and "types" data structures (most of the
1759         elaborator). There were a lot of changes during my "linkpath" trials
1760         that could have been reverted to their original state but weren't.
1761         Please, don't be too harsh on me for messing with this code a bit more
1762         than what was strictly necessary...  (I _did_ resist the tempation
1763         of doing any "global reformatting" to avoid an untimely death at
1764         Dave's hands. :)
1765    
1766       - One positive aspect of the previous point:  At least I made sure that
1767         all files that I touched now compile without warnings (other than
1768         "polyEqual").
1769    
1770       - compiler now tends to run "leaner" (i.e., ties up less memory in
1771         redundant modmaps)
1772    
1773    ----------------------------------------------------------------------
1774    Name: Allen Leung
1775    Date: 2000/03/29 18:00:00
1776    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
1777    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
1778    Description:
1779    
1780       This update contains *MAJOR* changes to the way code is generated from CPS
1781    in the module mlriscGen, and in various backend modules.
1782    
1783    CHANGES
1784    =======
1785    
1786    1. MLRiscGen: forward propagation fix.
1787    
1788       There was a bug in forward propagation introduced at about the same time
1789       as the MLRISC x86 backend, which prohibits coalescing to be
1790       performed effectively in loops.
1791    
1792       Effect: speed up of loops in RISC architectures.
1793               By itself, this actually slowed down certain benchmarks on the x86.
1794    
1795    2. MLRiscGen:  forward propagating addresses from consing.
1796    
1797       I've changed the way consing code is generated.  Basically I separated
1798       out the initialization part:
1799    
1800            store tag,   offset(allocptr)
1801            store elem1, offset+4(allocptr)
1802            store elem2, offset+8(allocptr)
1803            ...
1804            store elemn, offset+4n(allocptr)
1805    
1806       and the address computation part:
1807    
1808            celladdr <- offset+4+alloctpr
1809    
1810       and move the address computation part
1811    
1812       Effect:  register pressure is generally lower as a result.  This
1813                makes compilation of certain expressions much faster, such as
1814                long lists with non-trivial elements.
1815    
1816                 [(0,0), (0,0), .... (0,0)]
1817    
1818    3. MLRiscGen: base pointer elimination.
1819    
1820        As part of the linkage mechanism, we generate the sequence:
1821    
1822         L:  ...  <- start of the code fragment
1823    
1824         L1:
1825             base pointer <- linkreg - L1 + L
1826    
1827         The base pointer was then used for computing relocatable addresses
1828       in the code fragment.  Frequently (such as in lots of continuations)
1829       this is not needed.  We now eliminate this sequence whenever possible.
1830    
1831         For compile time efficiency, I'm using a very stupid local heuristic.
1832       But in general, this should be done as a control flow analysis.
1833    
1834       Effect:  Smaller code size.  Speed up of most programs.
1835    
1836    4. Hppa back end
1837    
1838         Long jumps in span dependence resolution used to depend on the existence
1839      of the base pointer.
1840    
1841         A jump to a long label L was expanded into the following sequence:
1842    
1843          LDIL %hi(L-8192), %r29
1844          LDO  %lo(L-8192)(%r29), %r29
1845          ADD  %r29, baseptr, %r29
1846          BV,n %r0(%r29)
1847    
1848         In the presence of change (3) above, this will not work.  I've changed
1849       it so that the following sequence of instructions are generated, which
1850       doesn't mention the base pointer at all:
1851    
1852             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
1853        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
1854             BV,n  %r0(%r29)          /* Jump */
1855    
1856    5. Alpha back end
1857    
1858          New alpha instructions LDB/LDW have been added, as per Fermin's
1859       suggestions.   This is unrelated to all other changes.
1860    
1861    6. X86 back end
1862    
1863         I've changed andl to testl in the floating point test sequence
1864         whenever appropriate.  The Intel optimization guide states that
1865         testl is preferable to andl.
1866    
1867    7. RA (x86 only)
1868    
1869         I've improved the spill propagation algorithm, using an approximation
1870       of maximal weighted independent sets.   This seems to be necessary to
1871       alleviate the negative effect in light of the slow down in (1).
1872    
1873         I'll write down the algorithm one of these days.
1874    
1875    8. MLRiscGen: frequencies
1876    
1877         I've added an annotation that states that all call gc blocks have zero
1878       execution frequencies.  This improves register allocation on the x86.
1879    
1880    BENCHMARKS
1881    ==========
1882    
1883       I've only perform the comparison on 110.25.
1884    
1885       The platforms are:
1886    
1887        HPPA  A four processor HP machine (E9000) with 5G of memory.
1888        X86   A 300Hhz Pentium II with 128M of memory, and
1889        SPARC An Ultra sparc 2 with 512M of memory.
1890    
1891       I used the following parameters for the SML benchmarks:
1892    
1893                 @SMLalloc
1894         HPPA    256k
1895         SPARC   512k
1896         X86     256k
1897    
1898    COMPILATION TIME
1899    ----------------
1900       Here are the numbers comparing the compilation times of the compilers.
1901       I've only compared 110.25 compiling the new sources versus
1902       a fixpoint version of the new compiler compiling the same.
1903    
1904                     110.25                                  New
1905               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
1906         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
1907         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
1908         X86    999s    315s       94006+130691   987s    296s    108877+141957
1909    
1910                   110.25         New
1911                Code Size      Code Size
1912         HPPA   8596736         8561421
1913         SPARC  8974299         8785143
1914         X86    9029180         8716783
1915    
1916       So in summary, things are at least as good as before.   Dramatic
1917       reduction in compilation is obtained on the Sparc; I can't explain it,
1918       but it is reproducible.  Perhaps someone should try to reproduce this
1919       on their own machines.
1920    
1921    SML BENCHMARKS
1922    --------------
1923    
1924        On the average, all benchmarks perform at least as well as before.
1925    
1926          HPPA         Compilation Time     Spill+Reload      Run Time
1927                     110.25  New            110.25    New   110.25  New
1928    
1929          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
1930              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
1931       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
1932                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
1933        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
1934             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
1935               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
1936              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
1937         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
1938             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
1939            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
1940      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
1941                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
1942             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
1943                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
1944               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
1945      --------------------------------------------------------------------------
1946       Average                     6.12%                                   4.09%
1947    
1948          SPARC        Compilation Time     Spill+Reload      Run Time
1949                     110.25  New            110.25    New   110.25  New
1950    
1951          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
1952              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
1953       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
1954                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
1955        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
1956             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
1957               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
1958              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
1959         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
1960             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1961            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
1962      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
1963                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
1964             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
1965                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
1966               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
1967      ----------------------------------------------------------------------------
1968       Average                     6.94%                                    2.64%
1969    
1970          X86          Compilation Time     Spill+Reload      Run Time
1971                     110.25  New            110.25    New   110.25  New
1972    
1973          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
1974              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
1975       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
1976                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1977        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1978             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1979               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1980              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1981         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1982             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1983            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1984      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1985                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1986             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1987                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1988               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1989      ----------------------------------------------------------------------------
1990       Average                     1.22%                                     3.36%
1991    
1992    ----------------------------------------------------------------------
1993    Name: Allen Leung
1994    Date: 2000/03/23 16:25:00
1995    Tag: leunga-20000323-fix_x86_alpha
1996    Description:
1997    
1998    1. X86 fixes/changes
1999    
2000       a.  The old code generated for SETcc was completely wrong.
2001           The Intel optimization guide is VERY misleading.
2002    
2003    2. ALPHA fixes/changes
2004    
2005       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2006       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2007       c.  Added reassociation code for address computation.
2008    
2009    ----------------------------------------------------------------------
2010    Name: Allen Leung
2011    Date: 2000/03/22 01:23:00
2012    Tag: leunga-20000322-fix_x86_hppa_ra
2013    Description:
2014    
2015    1. X86 fixes/changes
2016    
2017       a.  x86Rewrite bug with MUL3 (found by Lal)
2018       b.  Added the instructions FSTS, FSTL
2019    
2020    2. PA-RISC fixes/changes
2021    
2022       a.  B label should not be a delay slot candidate!  Why did this work?
2023       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
2024           as it should be.
2025       c.  The assembly syntax for fstds and fstdd was wrong.
2026       d.  Added the composite instruction COMICLR/LDO, which is the immediate
2027           operand variant of COMCLR/LDO.
2028    
2029    3. Generic MLRISC
2030    
2031       a.  shuffle.sml rewritten to be slightly more efficient
2032       b.  DIV bug in mltree-simplify fixed (found by Fermin)
2033    
2034    4. Register Allocator
2035    
2036       a.  I now release the interference graph earlier during spilling.
2037           May improve memory usage.
2038    
2039    ----------------------------------------------------------------------
2040    Name: Matthias Blume
2041    Date: 2000/03/14 14:15:32
2042    Tag: blume_main_v110p26p1_2
2043    Description:
2044    
2045    1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2046    additional argument called "template" which is an optional string that
2047    specifies the layout of the tool command line.  See the CM manual for
2048    explanation.
2049    
2050    2. A special-purpose tool can be "registered" by simply dropping the
2051    corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2052    directory where the .cm file lives that uses this tool.  (The
2053    behavior/misfeature until now was to look for the tool description
2054    files in the current working directory.)  As before, tool description
2055    files could also be anchored -- in which case they can live anywhere
2056    they like.  Following the recent e-mail discussion, this change should
2057    make it easier to have special-purpose tools that are shipped together
2058    with the sources of the program that uses them.
2059    
2060    ----------------------------------------------------------------------
2061    Name: Matthias Blume
2062    Date: 2000/03/10 07:48:34
2063    Tag: blume_main_v110p26p1_1
2064    Description:
2065    
2066    I added a re-written version of Dave's fixpt script to src/system.
2067    Changes relative to the original version:
2068      - sh-ified (not everybody has ksh)
2069      - automatically figures out which architecture it runs on
2070      - uses ./makeml a bit more cleverly
2071      - never invokes ./installml (and, thus, does not clobber your
2072        good and working installation of sml in case something goes wrong)
2073      - accepts max iteration count using option "-iter <n>"
2074      - accepts a "base" name using option "-base <base>"
2075    
2076    It does not build any extraneous heap images but directly rebuilds
2077    bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
2078    it can incorporate existing bin- and boot- hierarchies.  For example,
2079    suppose the base is set to "sml" (which is the default).  Then it
2080    successively builds
2081    
2082            sml.bin.<arch>-unix and sml.boot.<arch>-unix
2083    then    sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
2084    then    sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
2085    ...
2086    then    sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
2087    
2088    and so on.  If any of these already exist, it will just use what's
2089    there.  In particular, many people will have the initial set of bin
2090    and boot files around, so this saves time for at least one full
2091    rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2092    <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
2093    they might be the result of an earlier partial run of fixpt (which
2094    perhaps got accidentally killed).  In this case, fixpt will quickly
2095    move through what exists before continuing where it left off earlier,
2096    and, thus, saves a lot of time.
2097    
2098    ----------------------------------------------------------------------
2099    Name: Allen Leung
2100    Date: 00/03/10 02:20:00
2101    Tag: leunga-20000310-fix_x86_asm_ra
2102    Description:
2103    
2104    More assembly output problems involving the indexed addressing mode
2105    on the x86 have been found and corrected. Thanks to Fermin Reig for the
2106    fix.
2107    
2108    The interface and implementation of the register allocator have been changed
2109    slightly to accommodate the possibility to skip the register allocation
2110    phases completely and go directly to memory allocation.  This is needed
2111    for C-- use.
2112    
2113    ----------------------------------------------------------------------
2114    Name: Matthias Blume
2115    Date: 00/03/09 10:23:53
2116    Tag: blume_main_v110p26p1_0
2117    Description:
2118    
2119    * Complete re-organization of library names.  Many libraries have been
2120    consolidated so that they share the same path anchor.  For example,
2121    all MLRISC-related libraries are anchored at MLRISC, most libraries that
2122    are SML/NJ-specific are under "smlnj".  Notice that names like
2123    host-cmb.cm or host-compiler.cm no longer exist.  See system/README
2124    for a complete description of the new naming scheme.  Quick reference:
2125    
2126       host-cmb.cm        -> smlnj/cmb.cm
2127       host-compiler.cm   -> smlnj/compiler.cm
2128       full-cm.cm         -> smlnj/cm.cm
2129       <arch>-<os>.cm     -> smlnj/cmb/<arch>-<os>.cm
2130       <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
2131    
2132    * Bug fixes in CM.
2133        - exceptions in user code are being passed through (i.e., reach top level)
2134        - more bugs in paranoia mode fixed
2135        - bug related to checking group owners fixed
2136    
2137    * New install.sh script that automagically fetches archive files:
2138      The new file config/srcarchiveurl must contain the URL of the
2139      (remote) directory that contains bin files (or other source archives).
2140      If install.sh does not find the archive locally, it tries to get
2141      it from that remote directory.
2142      This should simplify installation further:  For machines that have
2143      access to the internet, just fetch <version>-config.tgz, unpack it,
2144      edit config/targets, and go (run config/install.sh).  The script will
2145      fetch everything else that it might need all by itself.
2146    
2147      For CVS users, this mechanism is not relevant for source archives, but
2148      it is convenient for getting new sets of binfiles.
2149    
2150      Archives should be tar files compressed with either gzip, compress, or
2151      bzip2.  The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
2152    
2153  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2154  Name: Matthias Blume  Name: Matthias Blume

Legend:
Removed from v.572  
changed lines
  Added in v.755

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