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 578, Tue Mar 14 05:16:29 2000 UTC revision 759, Fri Dec 22 12:29:26 2000 UTC
# Line 13  Line 13 
13  Description:  Description:
14  ----------------------------------------------------------------------  ----------------------------------------------------------------------
15  Name: Matthias Blume  Name: Matthias Blume
16    Date: 2000/12/22 21:20:00 JST
17    Tag: blume-20001222-anchorenv
18    Description:
19    
20    Fixed a CM problem related to anchor environments and their interaction
21    with CMB.make.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume
25    Date: 2000/12/22 13:15:00 JST
26    Tag: blume-20001222-cleanup
27    Description:
28    
29    Removed src/cm/ffi which does not (and did not) belong here.
30    
31    ----------------------------------------------------------------------
32    Name: Matthias Blume
33    Date: 2000/12/21 23:55:00 JST
34    Tag: blume-20001221-exn
35    Description:
36    
37    Probably most important: CM no longer silently swallows all exceptions
38    in the compiler.
39    Plus: some other minor CM changes.  For example, CM now reports some
40    sizes for generated binfiles (code, data, envpickle, lambdapickle).
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume
44    Date: 2000/12/15 00:01:05 JST
45    Tag: blume-20001215-dirtool
46    Description:
47    
48    - "dir" tool added.
49    - improvements and cleanup to Tools structure
50    - documentation updates
51    
52    ----------------------------------------------------------------------
53    Name: Allen Leung
54    Date: Thu Dec 14 03:45:24 EST 2000
55    Description:
56    Tag:  leunga-20001214-int-inf
57    Description:
58    
59       In IntInf, added these standard functions, which are missing from our
60    implementation:
61    
62        andb : int * int -> int
63        xorb : int * int -> int
64        orb  : int * int -> int
65        notb : int -> int
66         <<   : int * word -> int
67        ~>>  : int * word -> int
68    
69       Not tested, I hope they are correct.
70    
71    ----------------------------------------------------------------------
72    Name: Allen Leung
73    Date: Fri Dec  8 19:23:26 EST 2000
74    Description:
75    Tag:  leunga-20001208-nowhere
76    Description:
77    
78      Slight improvements to the 'nowhere' tool to handle OR-patterns,
79    to generate better error messages etc.  Plus a brief manual.
80    
81    ----------------------------------------------------------------------
82    Name: Lal George
83    Date: 2000/12/08 09:54:02 EST 2000
84    Tag: Release_110_31
85    Description:
86    
87    - Version 110.31
88    ----------------------------------------------------------------------
89    Name: Allen Leung
90    Date: Thu Dec  7 22:01:04 EST 2000
91    Tag:  leunga-20001207-cell-monster-hack
92    Description:
93    
94    Major MLRISC internal changes.  Affect all clients.
95    Summary:
96    
97    1.  Type CELLS.cell = int is now replaced by a datatype.
98        As a result, the old regmap is now gone.  Almost all interfaces
99        in MLRISC change as a consequence.
100    
101    2.  A new brand version of machine description tool (v3.0) that generates
102        modules expecting the new interface.  The old version is removed.
103    
104    3.  The RA interface has been further abstracted into two new functors.
105        RISC_RA and X86RA.  These functors have much simpler interfaces.
106        [See also directory MLRISC/demo.]
107    
108    4.  Some other new source->source code generation tools are available:
109    
110        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
111        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
112           I use this tool to generate the peephole optimizers---with the new
113           cell type changes, peephole rules are becoming difficult to write
114           without conditional pattern matching.
115    
116    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
117        cover the entire MLRISC source tree so many things broke.
118    
119    6.  CM files have been moved to the subdirectory MLRISC/cm.
120        They are moved because there are a lot of them and they clutter up the
121        root dir.
122    
123    7.  More detailed documentation to come...
124    
125        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
126        a makeml -rebuild first.  This is because of other other
127        changes that Matthias has made (see below).
128    
129    
130    ----------------------------------------------------------------------
131    Name: Matthias Blume
132    Date: 2000/11/30 23:12:00 JST
133    Tag: blume-20001130-filereorg
134    Description:
135    
136    Some manual updates and some file reorganizations in CM.
137    
138    ----------------------------------------------------------------------
139    Name: Matthias Blume
140    Date: 2000/11/24 17:45:00 JST
141    Tag: blume-20001124-link
142    Description:
143    
144    Drastically improved link traversal code for the case that the dynamic
145    value was already loaded at bootstrap time.  As a result, CM and CMB
146    now both load blazingly fast -- even on a very slow machine.  Also,
147    memory consumption has been further reduced by this.
148    
149    Warning: The format of the PIDMAP file has changed.  THerefore, to
150    bootstrap you have to do this:
151    
152    1. Run CMB.make
153    2. Make a symbolic link for the boot directory:
154         ln -s sml.boot.ARCH-OS xxx
155    3. "Rebuild" the boot directory:
156         ./makeml -boot xxx -rebuild sml ; rm xxx
157    4. Boot normally:
158          ./makeml
159    
160    ----------------------------------------------------------------------
161    Name: Matthias Blume
162    Date: 2000/11/21 21:20:00 JST
163    Tag: blume-20001121-tools
164    Description:
165    
166    Continued hacking on autoloading problem -- with success this time.
167    Also changed tool-plugin mechanism.  See new CM manual.
168    
169    ----------------------------------------------------------------------
170    Name: Matthias Blume
171    Date: 2000/11/19 14:30:00 JST
172    Tag:  blume-20001119-autoload
173    Description:
174    
175    Some hacking to make autoloading faster.  Success for CMB, no success
176    so far for CM.  There is a reduced structure CM' that autoloads faster.
177    (This is a temporary, non-documented hack to be eliminated again when
178    the general problem is solved.)
179    
180    ----------------------------------------------------------------------
181    Name: Matthias Blume
182    Date: 2000/11/17 14:10:00 JST
183    Tag: blume-20001117-pickle-lib
184    Description:
185    
186    1. Eliminated comp-lib.cm
187    2. Made pickle-lib.cm
188    3. Eliminated all uses of intset.sml (from comp-lib.cm)
189    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
190       equivalent constructs from smlnj-lib.cm (INtHashTable).
191    5. Point 4. also goes for those uses of intmap.* in MLRISC.
192       Duplicated intmap modules thrown out.
193    6. Hunted down all duplicated SCC code and replaced it with
194       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
195    7. Rewrote Feedback module.
196    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
197       should be thrown out and equivalent modules from smlnj-lib.cm
198       should be used (IntRedBlackSet, IntListSet, ...).
199    
200    Confirmed that compiler compiles to fixpoint.
201    
202    ----------------------------------------------------------------------
203    Name: Allen Leung
204    Date: 2000/11/10 18:00:00
205    Tag: leunga-20001110-new-x86-fp
206    
207    A new x86 floating point code generator has been added.
208    By default this is turned off.  To turn this on, do:
209    
210        CM.autoload "$smlnj/compiler.cm";
211        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
212    
213    Changes:
214    
215    1.  Changed FTAN to FPTAN so that the assembly output is correct.
216    2.  Changed the extension callback for FTANGENT to generate:
217    
218              fptan
219              fstp  %st(0)
220        instead of
221              fptan
222              fstpl ftempmem
223    
224    3.  Numerous assembly fixes for x86.
225    
226    5.  Cleaned up the machine code output module x86/x86MC.sml and added
227        support for a whole bunch of instructions and addressing modes:
228    
229          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
230          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
231          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
232          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
233          fxch %st(n)
234          fld %st(n)
235          fst %st(n)
236          fst mem
237          fstp %st(n)
238          fucom %st(n)
239          fucomp %st(n)
240    
241        All these are now generated when the fast fp mode is turned on.
242    
243    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
244    
245    ----------------------------------------------------------------------
246    Name: Matthias Blume
247    Date: 2000/11/09 11:20:00 JST
248    Tag: blume-20001109-scc
249    Description:
250    
251    Eliminated some code duplication:
252    
253    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
254       (Otherwise the functor is useless.)
255    2. Used GraphSCCFn where SCCUtilFun was used previously.
256    3. Got rid of SCCUtilFun (in comp-lib.cm).
257    
258    ----------------------------------------------------------------------
259    Name: Lal George
260    Date: 2000/11/06 09:02:21 EST 2000
261    Tag: Release_110_30
262    Description:
263    
264    - Version 110.30
265    ----------------------------------------------------------------------
266    Name: Matthias Blume
267    Date: 2000/11/04 14:45:00
268    Tag: blume-20001104-mlbuild
269    Description:
270    
271    - Made ml-build faster on startup.
272    - Documentation fixes.
273    
274    ----------------------------------------------------------------------
275    Name: Matthias Blume
276    Date: 2000/11/02 17:00:00 JST
277    Tag: blume-20001102-condcomp
278    Description:
279    
280    - Small tweaks to pickler -- new BOOTFILES!
281    - Version bumped to 110.29.2.
282    - Added conditional compilation facility to init.cmi (see comment there).
283    ----------------------------------------------------------------------
284    Name: Allen Leung
285    Date: 2000/10/23 19:31:00
286    Tag: leunga-20001023-demo-ra
287    
288    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
289    2. Test programs for the graph library updated
290    3. Some new MLRISC demo programs added
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume
294    Date: 2000/08/31 22:15:00 JST
295    Tag: blume-20001017-errmsg
296    Description:
297    
298    More error message grief: Where there used to be no messages, there
299    now were some that had bogus error regions.  Fixed.
300    
301    ----------------------------------------------------------------------
302    Name: Matthias Blume
303    Date: 2000/08/31 17:30:00 JST
304    Tag: blume-20001017-v110p29p1
305    Description:
306    
307    I made a version 110.29.1 with new bootfiles.
308    
309    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
310              CM documentation changes and a small bugfix in CM's error reporting.
311    
312    ----------------------------------------------------------------------
313    Name: Lal George
314    Date: 2000/09/27 14:42:35 EDT
315    Tag: george-20000927-nodestatus
316    Description:
317    
318    Changed the type of the nodestatus, so that:
319    
320            SPILLED(~1)             is now SPILLED
321            SPILLED(m) where m>=0   is now MEMREG(m)
322            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
323    
324    ----------------------------------------------------------------------
325    Name: Matthias Blume
326    Date: 2000/09/07 14:45:00 JST
327    Tag: blume-20000907-cmerrmsg
328    Description:
329    
330    Small tweak to CM to avoid getting ML syntax error messages twice.
331    
332    ----------------------------------------------------------------------
333    Name: Matthias Blume
334    Date: 2000/08/31 18:00:00 JST
335    Tag: blume-20000831-cvsbootfiles
336    Description:
337    
338    New URL for boot files (because the 110.29 files on the BL server do
339    now work correctly with my updated install scripts for yacc and lex).
340    
341    ----------------------------------------------------------------------
342    Name: Matthias Blume
343    Date: 2000/08/08 12:33:00 JST
344    Tag: blume-20000808-manual
345    Description:
346    
347    Tiny update to CM manual.
348    
349    ----------------------------------------------------------------------
350    Name: Allen Leung
351    Date: 2000/08/7 19:31:00
352    Tag: leunga-20000807-a-whole-bunch-of-stuff
353    
354      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
355    and Mobdy.
356    
357    1.  x86
358    
359       a.  Fixes to peephole module by John and Dan.
360       b.  Assembly fix to SETcc by Allen.
361       c.  Fix to c-call by John.
362       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
363       e.  Instruction selection optimization to SETcc as suggested by John.
364    
365           For example,
366    
367            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
368    
369           should generate:
370    
371            MOVL a, x
372            SUBL b, x
373            SHRL 31, x
374    
375    2.  IR stuff
376    
377         A bunch of new DJ-graph related algorithms added.  These
378         speed up SSA construction.
379    
380    3.  SSA + Scheduling
381    
382         Added code for SSA and scheduling to the repository
383    
384    ----------------------------------------------------------------------
385    Name: Lal George
386    Date: 2000/07/27 11:53:14 EDT
387    
388    Tag: lal-20000727-linux-ppc
389    Description:
390    
391     Made changes to support Linux PPC.
392     p.s. I have confirmation that the 110.29 boot files work fine.
393    
394    ----------------------------------------------------------------------
395    Name: Matthias Blume
396    Date: 2000/07/27 17:40:00 JST
397    Tag: blume-20000727-scripts
398    Description:
399    
400    !!!! WARNING !!!!
401    You must recompile the runtime system!
402    !!!! WARNING !!!!
403    
404    This is basically another round of script-enhancements:
405    
406    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
407       and undefine CM preprocessor symbols.
408    
409    2. ml-build avoids generating a new heap image if it finds that the
410       existing one is still ok.  (The condition is that no ML file had to
411       be recompiled and all ML files are found to be older that the heap
412       file.)
413    
414       To make this work smoothly, I also hacked the runtime system as
415       well as SMLofNJ.SysInfo to get access to the heap image suffix
416       (.sparc-solaris, ...) that is currently being used.
417    
418       Moreover, the signature of CM.mk_standalone has changed.  See the
419       CM manual.
420    
421    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
422       CM manual for details.)
423    
424    4. More CM manual updates:
425        - all of the above has been documented.
426        - there is now a section describing the (CM-related) command line
427          arguments that are accepted by the "sml" command
428    
429    ----------------------------------------------------------------------
430    Name: Matthias Blume
431    Date: 2000/07/25 16:20:00 JST
432    Tag: blume-20000725-makedepend
433    Description:
434    
435    Added a script called ml-makedepend.  This can be used in makefiles
436    for Unix' make in a way very similar to the "makedepend" command for
437    C.
438    
439    The script internally uses function CM.sources.
440    
441    Synopsis:
442    
443        ml-makedepend [-f makefile] cmfile targetname
444    
445    The default for the makefile is "makefile" (or "Makefile" should
446    "makefile" not exist).
447    
448    ml-makedepend adds a cmfile/targetname-specific section to this
449    makefile (after removing the previous version of this section).  The
450    section contains a single dependency specification with targetname on
451    the LHS (targetname is an arbitrary name), and a list of files derived
452    from the cmfile on the RHS.  Some of the files on the RHS are
453    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
454    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
455    path names.  The makefile writer is responsible for making sure that
456    these variables have correct at the time "make" is invoked.
457    
458    ----------------------------------------------------------------------
459    Name: Matthias Blume
460    Date: 2000/07/22 23:30:00 JST
461    Tag: blume-20000722-urlupdate
462    Description:
463    
464    Changed BOOT and config/srcarchiveurl to point to BL server:
465    
466        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
467    
468    ----------------------------------------------------------------------
469    Name: Matthias Blume
470    Date: 2000/07/18 18:00:00 JST
471    Tag: blume-20000718-Version_110_29
472    Description:
473    
474    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
475    
476    2. Updated config/version to 110.29
477    
478    3. Updated config/srcarchiveurl
479    
480    3. New boot files!
481       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume
485    Date: 2000/07/11 13:58:00 JST
486    Tag: blume-20000711-doctypo
487    Description:
488    
489    Fixed a few typos in CM manual.
490    
491    ----------------------------------------------------------------------
492    Name: Allen Leung
493    Date: 2000/06/15 00:38:00
494    Tag: leunga-20000704-sparc-x86
495    
496    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
497    2. fix to x86 RET bug [found by Dan Grossman]
498    3. sparc assembly bug fix for ticc instructions [found by Fermin]
499    
500       Affects c-- and moby only
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume
504    Date: 2000/07/04 15:26:00
505    Tag: blume-20000704-trigger
506    Description:
507    
508    1. Improvements to CM manual.
509    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
510       of getting a back-trace.  The function, when called, raises an
511       internal exception which explicitly carries the full back-trace history,
512       so it is unaffected by any intervening handle-raise pairs ("trivial"
513       or not).  The interactive loop will print that history once it arrives
514       at top level.
515       Short of having all exceptions implicitly carry the full history, the
516       recommended way of using this facility is:
517         - compile your program with instrumentation "on"
518         - run it, when it raises an exception, look at the history
519         - if the history is "cut off" because of some handler, go and modify
520           your program so that it explicitly calls BTrace.trigger
521         - recompile (still instrumented), and rerun; look at the full history
522    
523    ----------------------------------------------------------------------
524    Name: Matthias Blume
525    Date: 2000/07/03 15:36:00 JST
526    Tag: blume-20000702-manual
527    Description:
528    
529    Small corrections and updates to CM manual.
530    
531    ----------------------------------------------------------------------
532    Name: Matthias Blume
533    Date: 2000/06/29 16:04:00 JST
534    Tag: blume-20000629-yacctool
535    Description:
536    
537    Changes:
538    
539    1. Class "mlyacc" now takes separate arguments to pass options to
540       generated .sml- and .sig-files independently.
541    2. Corresponding CM manual updates.
542    3. BTrace module now also reports call sites.  (However, for loop clusters
543       it only shows from where the cluster was entered.)  There are associated
544       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blume
548    Date: 2000/06/27 16:51:00 JST
549    Tag: blume-20000627-noweb
550    Description:
551    
552    Changes:
553    
554     1. Implemented "subdir" and "witness" options for noweb tool.
555        This caused some slight internal changes in CM's tool implementation.
556     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
557        some remaining issues from earlier path anchor changes.
558     3. Updated CM manual accordingly.
559    
560     4. Changed implementation of back-tracing so that I now consider it
561        ready for prime-time.
562    
563        In particular, you don't have to explicitly trigger the back-trace
564        anymore.  Instead, if you are running BTrace-instrumented code and
565        there is an uncaught exception (regardless of whether or not it was
566        raised in instrumented code), the top-level evalloop will print
567        the back-trace.
568    
569        Features:
570    
571          - Instrumented and uninstrumented code work together seemlessly.
572            (Of course, uninstrumented code is never mentioned in actual
573             back-traces.)
574    
575          - Asymptotic time- and space-complexity of instrumented code is
576            equal to that of uninstrumented code.  (This means that
577            tail-recursion is preserved by the instrumentation phase.)
578    
579          - Modules whose code has been instrumented in different sessions
580            work together without problem.
581    
582          - There is no penalty whatsoever on uninstrumented code.
583    
584          - There is no penalty on "raise" expressions, even in
585            instrumented code.
586    
587        A potential bug (or perhaps it is a feature, too):
588    
589          A back-trace reaches no further than the outermost instrumented
590          non-trivial "raise".  Here, a "trivial" raise is one that is the
591          sole RHS of a "handle" rule.  Thus, back-traces reach trough
592    
593               <exp> handle e => raise e
594    
595          and even
596    
597               <exp> handle Foo => raise Bar
598    
599          and, of course, through
600    
601               <exp> handle Foo => ...
602    
603         if the exception was not Foo.
604    
605         Back-traces always reach right through any un-instrumented code
606         including any of its "handle" expressions, trivial or not.
607    
608       To try this out, do the following:
609    
610         - Erase all existing binfiles for your program.
611           (You may keep binfiles for those modules where you think you
612            definitely don't need back-tracing.)
613         - Turn on back-trace instrumentation:
614              SMLofNJ.Internals.BTrace.mode (SOME true);
615         - Recompile your program.  (I.e., run "CM.make" or "use".)
616         - You may now turn instrumentation off again (if you want):
617              SMLofNJ.Internals.BTrace.mode (SOME false);
618         - Run your program as usual.  If it raises an exception that
619           reaches the interactive toplevel, then a back-trace will
620           automatically be printed.  After that, the toplevel loop
621           will print the exception history as usual.
622    
623    ----------------------------------------------------------------------
624    Name: Matthias Blume
625    Date: 2000/06/26 09:56:46 JST
626    Tag: blume-20000626-setup
627    Description:
628    
629    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
630          ML code before and after compiling a file (e.g., to set compiler
631          flags)
632    
633    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
634              - associated changes to btrace.sml (BTrace instrumentation pass)
635              - cleaner implementation of btimp.sml (BTrace tracing and report
636                module)
637    
638    CM manual: * new path encoding documented
639               * description of setup-parameter to "sml" added
640    
641    The biggest user-visible change to back-tracing is that it is no
642    longer necessary to compile all traced modules within the same
643    session.  (This was a real limitation.)
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume
647    Date: 2000/06/24 12:40:00 JST
648    Tag: blume-20000624-startup
649    Description:
650    
651    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
652    bit too often -- to put it mildly. :)
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2000/06/23 18:20:00 JST
657    Tag: blume-20000623-btrace
658    Description:
659    
660    This updates adds a backtrace facility to aid programmers in debugging
661    their programs.  This involves the following changes:
662    
663    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
664       keeping track of the current call stack.  When programs are compiled
665       in a special mode, the compiler will insert calls to these hooks
666       into the user program.
667       "Hook" means that it is possible for different implementations of
668       back-tracing to register themselves (at different times).
669    
670    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
671       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
672       It can be turned on using this call:
673         SMLofNJ.Internals.BTrace.mode (SOME true);
674       Turning it off again:
675         SMLofNJ.Internals.BTrace.mode (SOME false);
676       Querying the current status:
677         SMLofNJ.Internals.BTrace.mode NONE;
678       Annotated programs are about twice as big as normal ones, and they
679       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
680       where all hooks do nothing).  The slowdown with a plugin that is
681       actually useful (such as the one supplied by default) is even greater,
682       but in the case of the default plugin it is still only an constant
683       factor (amortized).
684    
685    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
686       with a sub-structure BTrace for controlling back-tracing.  In particular,
687       the above-mentioned function "mode" controls whether the annotation
688       phase is invoked by the compiler.  Another important function is
689       "trigger": when called it aborts the current execution and causes
690       the top-level loop to print a full back-trace.
691    
692    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
693       for back-tracing.  It keeps track of the dynamic call stack and in
694       addition to that it keeps a partial history at each "level" of that
695       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
696       d and b (at separate times, dynamically), then the report will show:
697    
698       GOTO   d
699             /c
700       GOTO  \b
701       CALL   a
702    
703       This shows that there was an initial non-tail call of a, then a
704       tail-call to b or c, looping behavior in a cluster of functions that
705       consist of b and c, and then a goto from that cluster (i.e., either from
706       b or from c) to d.
707    
708       Note that (depending on the user program) the amount of information
709       that the back-trace module has to keep track of at each level is bounded
710       by a constant.  Thus, the whole implementation has the same asymptotical
711       complexity as the original program (both in space and in time).
712    
713    5. compiler/TopLevel/interact/evalloop.sml has been modified to
714       handle the special exception SMLofNJ.Internals.BTrace.BTrace
715       which is raised by the "trigger" function mentioned above.
716    
717    Notes on usage:
718    
719    - Annotated code works well together with unannotated code:
720    Unannotated calls simply do not show up at all in the backtrace.
721    
722    - It is not a good idea to let modules that were annotated during
723    different sessions run at the same time.  This is because the compiler
724    chooses small integers to identify individual functions, and there
725    will be clashes if different modules were compiled in separate sessions.
726    (Nothing will crash, and you will even be told about the clashes, but
727    back-trace information will in general not be useful.)
728    
729    - Back-tracing can be confused by callcc and capture.
730    
731    - The only way of getting a back-trace right now is to explicitly
732    invoke the "trigger" function from your user program.  Eventually, we
733    should make every exception carry back-trace information (if
734    available).  But since this creates more overhead at "raise"-time
735    (similar to the current exnHistory overhead), I have not yet
736    implemented this.  (The implementation will be rather easy.)  With
737    exceptions carrying back-trace information, this facility will be even
738    more useful because users don't need to modify their programs...
739    
740    - While it is possible to compile the compiler with back-trace
741    annotations turned on (I did it to get some confidence in
742    correctness), you must make absolutely sure that core.sml and
743    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
744    be compiled with annotation because there is no core access yet, but
745    if you compile btimp.sml with annotation, then the system will go into
746    an infinite recursion and crash.)
747    Since CM currently does not know about BTrace, the only way to turn
748    annotations on and off for different modules of the compiler is to
749    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
750    this is awkward and clumsy.
751    
752    Sample sessions:
753    
754    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
755    - SMLofNJ.Internals.BTrace.mode (SOME true);
756    [autoloading]
757    [autoloading done]
758    val it = false : bool
759    - structure X = struct
760    -     fun main n = let
761    -         fun a (x, 0) = d x
762    -           | a (x, n) = b (x, n - 1)
763    -         and b (x, n) = c (x, n)
764    -         and c (x, n) = a (x, n)
765    -         and d x = e (x, 3)
766    -         and e (x, 0) = f x
767    -           | e (x, n) = e (x, n - 1)
768    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
769    -           | f n = n * g (n - 1)
770    -         and g n = a (n, 3)
771    -     in
772    -         f n
773    -     end
774    - end;
775    structure X : sig val main : int -> int end
776    - X.main 3;
777    *** BACK-TRACE ***
778    GOTO   stdIn:4.2-13.20: X.main[2].f
779    GOTO-( stdIn:4.2-13.20: X.main[2].e
780    GOTO   stdIn:4.2-13.20: X.main[2].d
781         / stdIn:4.2-13.20: X.main[2].a
782         | stdIn:4.2-13.20: X.main[2].b
783    GOTO-\ stdIn:4.2-13.20: X.main[2].c
784    CALL   stdIn:4.2-13.20: X.main[2].g
785    GOTO   stdIn:4.2-13.20: X.main[2].f
786    GOTO-( stdIn:4.2-13.20: X.main[2].e
787    GOTO   stdIn:4.2-13.20: X.main[2].d
788         / stdIn:4.2-13.20: X.main[2].a
789         | stdIn:4.2-13.20: X.main[2].b
790    GOTO-\ stdIn:4.2-13.20: X.main[2].c
791    CALL   stdIn:4.2-13.20: X.main[2].g
792    GOTO   stdIn:4.2-13.20: X.main[2].f
793    GOTO-( stdIn:4.2-13.20: X.main[2].e
794    GOTO   stdIn:4.2-13.20: X.main[2].d
795         / stdIn:4.2-13.20: X.main[2].a
796         | stdIn:4.2-13.20: X.main[2].b
797    GOTO-\ stdIn:4.2-13.20: X.main[2].c
798    CALL   stdIn:4.2-13.20: X.main[2].g
799    GOTO   stdIn:4.2-13.20: X.main[2].f
800    CALL   stdIn:2.15-17.4: X.main[2]
801    -
802    
803    (Note that because of a FLINt bug the above code currently does not
804    compile without BTrace turned on.)
805    
806    Here is another example, using my modified Tiger compiler:
807    
808    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
809    - SMLofNJ.Internals.BTrace.mode (SOME true);
810    [autoloading]
811    [autoloading done]
812    val it = false : bool
813    - CM.make "sources.cm";
814    [autoloading]
815    ...
816    [autoloading done]
817    [scanning sources.cm]
818    [parsing (sources.cm):parse.sml]
819    [creating directory CM/SKEL ...]
820    [parsing (sources.cm):tiger.lex.sml]
821    ...
822    [wrote CM/sparc-unix/semant.sml]
823    [compiling (sources.cm):main.sml]
824    [wrote CM/sparc-unix/main.sml]
825    [New bindings added.]
826    val it = true : bool
827    - Main.compile ("../testcases/merge.tig", "foo.out");
828    *** BACK-TRACE ***
829    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
830    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
831    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
832    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
833    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
834    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
835    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
836         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
837    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
838    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
839    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
840    CALL   main.sml:1.18-118.4: Main.compile[2]
841    -
842    
843    ----------------------------------------------------------------------
844    Name: Matthias Blumen
845    Date: 2000/06/21 18:00:00 JST
846    Tag: blume-20000621-manual
847    Description:
848    
849    CM manual update: Path environments documented.
850    
851    ----------------------------------------------------------------------
852    Name: Matthias Blume
853    Date: 2000/06/19 13:40:00
854    Tag: blume-20000619-manual
855    Description:
856    
857    CM manual and system/README update.  This only covers the fact that
858    there are no more implicit anchors.  (Path environments and the "bind"
859    option to "cm" have yet to be documented.)
860    
861    ----------------------------------------------------------------------
862    Name: Matthias Blume
863    Date: 2000/06/19 11:05:00 JST
864    Tag: blume-20000619-chdir-bugfix
865    Description:
866    
867    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
868    
869    ----------------------------------------------------------------------
870    Name: Matthias Blume
871    Date: 2000/06/18 22:00:10 JST
872    Tag: blume-20000618-implicit-anchors-really-gone
873    Description:
874    
875    I updates the previous HISTORY entry where I forgot to mention that
876    implicit anchors are no longer with us.
877    
878    The current update also gets rid of the (now useless) controller
879    CM.Control.implicit_anchors.
880    
881    ----------------------------------------------------------------------
882    Name: Matthias Blume
883    Date: 2000/06/16 17:30:00 JST
884    Tag: blume-20000616-anchorenv
885    Description:
886    
887    This patch implements the long anticipated (just kidding :) "anchor
888    environment" mechanism.  In the course of doing this, I also
889    re-implemented CM's internal "SrcPath" module from scratch.  The new
890    one should be more robust in certain boundary cases.  In any case, it
891    is a lot cleaner than its predecessor (IMHO).
892    
893    This time, although there is yet another boot file format change, I
894    kept the unpickler backward-compatible.  As a result, no new bootfiles
895    are necessary and bootstrapping is straightforward.  (You cannot read
896    new bootfiles into an old system, but the other way around is no
897    problem.)
898    
899    Visible changes:
900    
901    ** 0. Implicit path anchors (without the leading $-symbol) are no
902    longer recognized at all. This means that such path names are not
903    illegal either.  For example, the name basis.cm simply refers to a
904    local file called "basis.cm" (i.e, the name is an ordinary path
905    relative to .cm-files directory).  Or, to put it differently, only
906    names that start with $ are anchored paths.
907    
908    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
909    vanished.
910    
911    John (Reppy) had critizised this as soon as I originally proposed and
912    implemented it, but at that time I did not really deeply believe
913    him. :) Now I came full-circle because I need the $<singlearc> syntax
914    in another place where it cannot be seen as an abbreviation for
915    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
916    seems to mean (i.e., it "expands" into the corresponding anchor
917    value).
918    
919    However, when paths are used as members in CM description files, it
920    continues to be true that there must be at least another arc after the
921    anchor.  This is now enforced separately during semantic analysis
922    (i.e., from a lexical/syntactical point of view, the notation is ok.)
923    
924    ** 2. The "cm" class now accepts an option "bind".  The option's value
925    is a sub-option list of precisely two items -- one labeled "anchor"
926    and the other one labeled "value".  As you might expect, "anchor" is
927    used to specify an anchor name to be bound, and "value" specifies what
928    the anchor is being bound to.
929    
930    The value must be a directory name and can be given in either standard
931    syntax (including the possibility that it is itself an anchored path)
932    or native syntax.
933    
934    Examples:
935    
936       foo.cm (bind:(anchor:bar value:$mystuff/bar))
937       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
938    
939    and so on.
940    
941    The meaning of this is that the .cm-file will be processed with an
942    augmented anchor environment where the given anchor(s) is/are bound to
943    the given values(s).
944    
945    The rationale for having this feature is this: Suppose you are trying
946    to use two different (already stable) libraries a.cm and b.cm (that
947    you perhaps didn't write yourself).  Further, suppose each of these
948    two libraries internally uses its own auxiliary library $aux/lib.cm.
949    Normally you would now have a problem because the anchor "lib" can not
950    be bound to more than one value globally.  Therefore, the project that
951    uses both a.cm and b.cm must locally redirect the anchor to some other
952    place:
953    
954       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
955       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
956    
957    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
958    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
959    
960    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
961    you when you do so at the time of CM.stabilize).  Therefore, you can
962    also use an anchored path as the value:
963    
964      a.cm (bind:(anchor:lib value:$a-lib))
965      b.cm (bind:(anchor:lib value:$b-lib))
966    
967    Now you can globally configure (using the usual CM.Anchor.anchor or
968    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
969    itself is always locally bound, setting it globally is no longer
970    meaningful or necessary (but it does not hurt either).  In fact, "lib"
971    can still be used as a global anchor for separate purposes.  As a
972    matter of fact, one can locally define "lib" in terms of a global
973    "lib":
974    
975      a.cm (bind:(anchor:lib value:$lib/a))
976      b.cm (bind:(anchor:lib value:$lib/b))
977    
978    ** 3: The encoding of path names has changed.  This affects the way
979    path names are shown in CM's progress report and also the internal
980    protocol encoding used for parallel make.
981    
982    The encoding now uses one or more ':'-separated segments.  Each
983    segments corresponds to a file that has been specified relative to the
984    file given by its preceding segment.  The first segment is either
985    relative to the CWD, absolute, or anchored.  Each segment itself is
986    basically a Unix pathname; all segments but the first are relative.
987    
988    Example:
989    
990       $foo/bar/baz.cm:a/b/c.sml
991    
992    This path denotes the file bar/a/b/c.sml relative to the directory
993    denoted by anchor "foo".  Notice that the encoding also includes
994    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
995    paths are resolved relative to the .cm-files directory, so baz.cm must
996    be ignored to get the "real" pathname.
997    
998    To make this fact more obvious, CM puts the names of such "virtual
999    arcs" into parentheses when they appear in progress reports. (No
1000    parentheses will appear in the internal protocol encoding.)  Thus,
1001    what you really see is:
1002    
1003      $foo/bar/(baz.cm):a/b/c.sml
1004    
1005    I find this notation to be much more informative than before.
1006    
1007    Another new feature of the encoding is that special characters
1008    including parentheses, colons, (back)slashes, and white space are
1009    written as \ddd (where ddd is the decimal encoding of the character).
1010    
1011    *** The CM manual still needs to be updated.
1012    
1013    ----------------------------------------------------------------------
1014    Name: Allen Leung
1015    Date: 2000/06/15 00:38:00
1016    Tag: leunga-20000615-x86-peephole
1017    
1018    x86 Peephole fix by Fermin.  Affects c-- and moby only.
1019    
1020    ----------------------------------------------------------------------
1021    Name: Matthias Blume
1022    Date: 2000/06/12 11:40:00
1023    Tag: blume-20000612-parmakefix
1024    Description:
1025    
1026    More cleanup after changing the file naming scheme: This time I
1027    repaired the parallel make mechanism for CMB.make which I broke earlier.
1028    
1029    ----------------------------------------------------------------------
1030    Name: Allen Leung
1031    Date: 2000/06/09 01:25:00
1032    Tag: leunga-20000609-various
1033    
1034    None of these things should affect normal SML/NJ operations
1035    
1036    1. Peephole improvements provided by Fermin (c--)
1037    2. New annotation DEFUSE for adding extra dependence (moby)
1038    3. New X86 LOCK instructions (moby)
1039    4. New machine description language for reservation tables (scheduling)
1040    5. Fixes to various optimization/analysis modules (branch chaining, dominator
1041       trees etc.)
1042    6. I've changed the CM files so that they can work with versions
1043       110.0.6, 110.25 and 110.28
1044    
1045    ----------------------------------------------------------------------
1046    Name: Matthias Blume
1047    Date: 2000/06/09 12:40:00
1048    Tag: blume-20000609-log
1049    Description:
1050    
1051    - Removed all(?) remaining RCS Log entries from sources.
1052    
1053    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1054      anchored paths).
1055    
1056    ----------------------------------------------------------------------
1057    Name: Matthias Blume
1058    Date: 2000/06/07 17:00:00 JST
1059    Tag: blume-20000607-no-implicit-anchors
1060    Description:
1061    
1062    1. This update changes the default setting for
1063    CM.Control.implicit_anchors from true to false.  This means that
1064    implicit anchors are no longer permitted by default.  I also tried to
1065    make sure that nothing else still relies on implicit anchors.
1066    (This is the next step on the schedule towards a CM that does not even
1067    have the notion of implicit anchors anymore.)
1068    
1069    2. More CM manual updates.
1070    
1071    3. I managed to track down and fix the pickling bug I mentioned last
1072    time.  Because of the previously existing workaround, this entails no
1073    immediate practical changes.
1074    
1075    ----------------------------------------------------------------------
1076    Name: Matthias Blume
1077    Date: 2000/06/06 11:15:00 JST
1078    Tag: blume-20000606-lazierpickle
1079    Description:
1080    
1081    !!!! NEW BOOT FILES !!!!
1082    
1083    * The main purpose of this update is to make library pickles lazier in
1084    order to reduce the initial space penalty for autoloading a library.
1085    As a result, it is now possible to have $smlnj/compiler.cm
1086    pre-registered.  This should take care of the many complaints or
1087    inquiries about missing structure Compiler.  This required changes to
1088    CM's internal data structures and small tweaks to some algorithms.
1089    
1090    As a neat additional effect, it is no longer necessary (for the sake
1091    of lean heap image files) to distinguish between a "minimal" CM and a
1092    "full" CM.  Now, there is only one CM (i.e., the "full" version:
1093    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1094    interactive top level. ($smlnj/cm/minimal.cm is gone.)
1095    
1096    To make the life of compiler-hackers easier, "makeml" now also
1097    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1098    words, after you bootstrap a new sml for the first time, you will not
1099    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1100    around you will still have to do it, though.)
1101    
1102    * A second change consists of major updates to the CM manual.  There
1103    are now several appendices with summary information and also a full
1104    specification of the CM description file syntax.
1105    
1106    * In directory src/system I added the script "allcross".  This script
1107    invokes sml and cross-compiles the compiler for all supported
1108    architectures.  (Useful when providing a new set of boot files.)
1109    
1110    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
1111    added a small tweak to pickle-util.sml to work around this problem,
1112    but it is not a proper fix yet.  I will investigate further.  (The
1113    effect of the bug was an inflation of library pickle size.)
1114    
1115    * Version number increased to 110.28.1 (to avoid compatibility problems).
1116    
1117    ----------------------------------------------------------------------
1118    Name: Allen Leung
1119    Date: 2000/05/25 17:28 EDT
1120    Tag: leunga-20000525-ra
1121    Description:
1122    
1123      Fixed a bug in freezing phase of the register allocator.
1124    
1125    ----------------------------------------------------------------------
1126    Name: Allen Leung
1127    Date: 2000/05/15 22:53 EDT
1128    Tag: leunga-20000515-alpha-x86-ra
1129    Description:
1130    
1131      1. Alpha
1132    
1133          Slight cleanup.  Removed the instruction SGNXL
1134    
1135      2. X86
1136    
1137          Added the following instructions to the instruction set:
1138    
1139            ROLx, RORx,
1140            BTx, BTSx, BTLx, BTRx,
1141            XCHGx, and variants with the LOCK prefix
1142    
1143      3. Register Allocation
1144    
1145          The module ra-rewrite-with-renaming has been improved.
1146    
1147      These have no effect on SML/NJ.
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume
1151    Date: 2000/05/15 16:20:00 JST
1152    Tag: blume-20000515-lightrebuild
1153    Description:
1154    
1155    1. I added an alternative to "-rebuild" to "makeml".  The difference is
1156       that prior to calling CMB.make' the CM-variable "LIGHT" will be
1157       defined.  In effect, the command will not build any cross-compiler
1158       backends and therefore finish more quickly.
1159    
1160       The "fixpt" script also takes a "-light" switch to be able to use
1161       this new facility while compiling for a fixpoint.
1162    
1163    2. I replaced all mentions of anchored paths in group owner specifications
1164       with simple relative paths (usually starting with "..").
1165       The rationale is that a library's internal workings should not be
1166       compromised by the lack of some anchor.  (An anchor is necessary
1167       for someone who wants to refer to the library by an anchored path,
1168       but it should not be necessary to build the same library in the first
1169       place.)
1170    
1171    3. I changed the way CM's tool mechanism determines the shell command
1172       string used for things like ml-yacc etc. so that it does not break
1173       when CM.Control.implicit_anchors is turned off.
1174    
1175    ----------------------------------------------------------------------
1176    Name: Matthias Blume
1177    Date: 2000/05/12 18:20:00 JST
1178    Tag: blume-20000512-ml-build
1179    Description:
1180    
1181    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1182    getting installed properly (by config/install.sh).
1183    
1184    ----------------------------------------------------------------------
1185    Name: Matthias Blume
1186    Date: 2000/05/12 17:30:00 JST
1187    Tag: blume-20000512-anchors
1188    Description:
1189    
1190    !!! NEW BOOT FILES !!!
1191    
1192    This change is in preparation of fading out support for "implicitly
1193    anchored path names".  I went through all sources and used the
1194    explicit (and relatively new) $-notation.  See system/README and the
1195    CM manual for more info on this.
1196    
1197    I also modified the anchoring scheme for some things such as "smlnj",
1198    "MLRISC", "cm", etc. to take advantage of the fact that explicit
1199    anchors are more expressive: anchor name and first arc do not have to
1200    coincide.  This entails the following user-visible change:
1201    
1202    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1203    particular, when you fire up sml with a command-line argument, say,
1204    e.g.:
1205    
1206       sml '$smlnj/cmb.cm'
1207    
1208    At the ML toplevel prompt:
1209    
1210       CM.autoload "$smlnj/cmb.cm";
1211    
1212    There is also a new controller in CM.Control that can be used to turn
1213    off all remaining support for implicit anchors by saying:
1214    
1215        CM.autoload "$smlnj/
1216        #set CM.Control.implicit_anchors false;
1217    
1218    This causes CM to reject implicitly anchored paths.  This is (for the
1219    time being) less permissive than the "final" version where there will
1220    be no more such implicit anchors and relative paths will be just that:
1221    relative.
1222    
1223    The next step (version after next version?) will be to make the
1224    default for CM.Control.implicit_anchors false.  After the dust has
1225    settled, I can then produce the "final" version of this...
1226    
1227    Note: Since bootstrapping is a bit tricky, I provided new boot files.
1228    
1229    ----------------------------------------------------------------------
1230    Name: Matthias Blume
1231    Date: 2000/05/11 16:30:00 JST
1232    Tag: blume-20000511-sources
1233    Description:
1234    
1235    The main change is that I added function CM.sources as a generalized
1236    version of the earlier CM.makedepend.  This entails the following
1237    additional changes:
1238    
1239      - CM.makedepend has been dropped.
1240    
1241      - CM manual has been updated.
1242    
1243      - TOOLS signature and API have been changed.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Allen Leung
1247    Date: 2000/05/10 21:17 EDT
1248    Tag: leunga-20000510-moby-c--ssa
1249    Description:
1250    
1251      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1252    None of these affect SML/NJ.
1253    
1254    1. Register Allocation
1255    
1256        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1257           This module tries to remove local (i.e. basic block level) redundancies
1258           during spilling.
1259    
1260        b. A new framework for performing region based register allocation.
1261           Not yet entirely functional.
1262    
1263    2. X86
1264    
1265       a. DefUse for POP was missing the stack pointer [found by Lal]
1266       b. Reload for CALL was incorrect in X86Spill [found by John]
1267       c. Various fixes in X86Spill so that it can be used correctly for
1268          the new spilling module.
1269    
1270    3. SSA/IR
1271    
1272       a. New module ir/dj-dataflow.sml implements elimination based
1273          data flow analysis.
1274    
1275    4. MLRiscGen
1276    
1277       a. Fix for gc type annotation
1278    
1279    5. MDGen
1280    
1281       Various fixes for machine description -> ml code translation.  For ssa
1282       only.
1283    
1284    ----------------------------------------------------------------------
1285    Name: Allen Leung
1286    Date: 2000/05/08 22:17 EDT
1287    Tag: leunga-20000508-labexp
1288    Description:
1289    
1290      Fermin has found a few assembly problems with constant expressions
1291      generated in LabelExp.  Mostly, the problems involve extra parentheses,
1292      which choke on dumb assemblers.  This is his fix.
1293    
1294    ----------------------------------------------------------------------
1295    Name: Dave MacQueen
1296    Date: 2000/04/09 14:00 EDT
1297    Tag: dbm-20000502-Version_110_28
1298    Description:
1299    
1300    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1301    
1302    2. Updated config/version to 110.28
1303    
1304    3. Updated config/srcarchiveurl
1305    
1306    3. New boot files!
1307       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1308    
1309    ----------------------------------------------------------------------
1310    Name: Matthias Blume
1311    Date: 2000/05/01 19:05:00 JST
1312    Tag: blume-20000501-noweb
1313    Description:
1314    
1315    A new noweb tool has been added.  The existing system is entirely
1316    unaffected by this, but some CM users have asked for renewed noweb
1317    support.  Everything is documented in the CM manual.
1318    
1319    New (plugin) libraries:
1320    
1321       noweb-tool.cm
1322       nw-ext.cm
1323    
1324    ----------------------------------------------------------------------
1325    Name: Dave MacQueen
1326    Date: 2000/04/30 12:40PM EDT
1327    Tag: dbm-20000430-bug_fixes
1328    Description:
1329    
1330    1. Fix for bug 1498
1331       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1332       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1333         added toRealArray function
1334       smlnj/src/compiler/MiscUtil/print/ppobj.sml
1335         added check for tag Obj.RealArray to array printing case in ppObj
1336    
1337    2. Fix for bug 1510
1338       smlnj/src/compiler/Semant/types/typesutil.sml
1339         fixed definition of dummyargs (used by equalTycon) so that
1340         dummy args are distinct types
1341    
1342    ----------------------------------------------------------------------
1343    Name: Matthias Blume
1344    Date: 2000/04/30 01:00:00 JST
1345    Tag: blume-20000430-versions
1346    Description:
1347    
1348    1. CM version numbering added.  This is an implementation of Lal's
1349       proposal for adding version numbers and version checking to .cm
1350       files.  Lal said that his proposal was just that -- a proposal.
1351       For the time being I went ahead and implemented it so that people
1352       can comment on it.  Everything is completely backward-compatible
1353       (except for the stable library format, i.e., new bootfiles!).
1354    
1355       As usual, see the CM manual for details.
1356    
1357    2. An alternative syntax for anchored paths has been implemented.
1358       Dave has recently voiced the same concerns that I had when I did
1359       this, so there should be some support.  My take is that eventually
1360       I will let support for the current syntax (where anchors are
1361       "implicit") fade out in favor of the new, explicit syntax.
1362       In order to be backward-compatible, both old and new syntax are
1363       currently supported.
1364    
1365       Again, see the CM manual for details.
1366    
1367    3. Parallel make is trying to be slightly smarter:  When the master
1368       process finds a "bottleneck", i.e., when there is only one
1369       compilation unit that can be compiled and everybody else is
1370       waiting on it, then it will simply compile it directly instead
1371       of clumsily telling one of the slaves to do it.
1372    
1373    4. Support for "unsharing" added.  This is necessary in order to be
1374       able to have two different versions of the same library running
1375       at the same time (e.g., for trying out a new MLRISC while still
1376       having the old MLRISC linked into the current compiler, etc.)
1377       See the CM manual.
1378    
1379    5. Simple "makedepend" functionality added for generating Makefile
1380       dependency information.  (This is rather crude at the moment.
1381       Expect some changes here in the future.)
1382    
1383    6. ".fun" added as a recognized suffix for ML files. Also documented
1384       explicitly in the manual that the fallback behavior (unknown suffix
1385       -> ML file) is not an official feature!
1386    
1387    7. Small changes to the pickler for stable libraries.
1388    
1389    8. Several internal changes to CM (for cleanup/improvement).
1390    
1391    
1392    !!!! NEW BINFILES !!!!
1393    
1394    ----------------------------------------------------------------------
1395    Name: Matthias Blume
1396    Date: 2000/04/28 17:30:00 JST
1397    Tag: blume-20000428-pathconfig
1398    Description:
1399    
1400    1. I changed config/install.sh to remove duplicate entries from the
1401       lib/pathconfig file at the end.  Moreover, the final version of
1402       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
1403       in src/system/installml.
1404    
1405    2. The config/install.sh script now consistently uses relative
1406       pathnames in lib/pathconfig whenever the anchor is in the lib
1407       directory.  (So far this was true for the libraries that come
1408       pre-compiled and bundled as part of the bootfiles but not for
1409       libraries that are compiled by the script itself.)
1410    
1411    ----------------------------------------------------------------------
1412    Name: Matthias Blume
1413    Date: 2000/04/26 13:10:00 JST
1414    Tag: blume-20000426-fun_suffix
1415    Description:
1416    
1417    Added ".fun" as a recognized file name suffix (for ML code).
1418    
1419    ----------------------------------------------------------------------
1420    Name: Allen Leung
1421    Date: 2000/04/25 17:00:00 EST
1422    Tag: leunga-20000425-alpha-ra
1423    Description:
1424    
1425    1. Alpha
1426    
1427        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
1428    in C--.
1429    
1430    2. RA
1431    
1432       Added a flag "ra-dump-size" to print out the size of the flowgraph
1433       and the interference graph.
1434    
1435    ----------------------------------------------------------------------
1436    Name: Dave MacQueen
1437    Date: 2000/04/25/
1438    Tag: dbm-20000425-mlyacc_doc_examples
1439    Description:
1440      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
1441      Updated all three examples in src/ml-yacc/examples to run
1442      under 110.* using CM.make.
1443    
1444    ----------------------------------------------------------------------
1445    Name: Allen Leung
1446    Date: 2000/04/20 23:04:00 EST
1447    Tag: leunga-20000420-ssa-c---stuff
1448    Description:
1449    
1450      This update synchronizes my repository with Yale's.  Most of these
1451    changes, however, do not affect SML/NJ at all (the RA is an exception).
1452    
1453    1. Register Allocator
1454    
1455       a. An improvement in the interference graph construction:
1456          Given a copy
1457    
1458                s <- t
1459    
1460          no interference edge between s and t is added for this definition of s.
1461    
1462       b. I've added two new spill heuristic modules that Fermin and I developed
1463          (in the new library RA.cm). These are unused in SML/NJ but maybe
1464          useful for others (Moby?)
1465    
1466    2. X86
1467    
1468       a. Various fixes in the backend provided by Fermin [C--] and Lal.
1469    
1470    3. Alpha
1471    
1472       a. Added the BSR instruction and code generation that goes with it [C--]
1473       b. Other fixes too numerous to recount provided by Fermin [C--]
1474    
1475    4. Regmaps
1476    
1477       a. The regmaps are not initialized with the identity physical bindings
1478          at creation time.  This is unneeded.
1479    
1480    5. MLRISC Optimizations
1481    
1482       a. The DJ-Graph module can now compute the iterated dominance frontiers
1483          intersects with liveness incrementally in linear time! Woohoo!
1484          This is now used in my new SSA construction algorithm.
1485    
1486       b. THe branch reorganization module is now smarter about linear chains of
1487          basic blocks.
1488    
1489    
1490    ----------------------------------------------------------------------
1491    Name: Matthias Blume
1492    Date: 2000/04/12 13:52:00 JST
1493    Tag: blume_main_v110p27_1
1494    Description:
1495    
1496    Changed install.sh script to handle archive files without version number
1497    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
1498    name of the boot file archive.
1499    
1500    ----------------------------------------------------------------------
1501    Name: Dave MacQueen
1502    Date: 2000/04/09 14:00 EDT
1503    Tag: dbm-20000410-Version_110_27
1504    Description:
1505    
1506    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
1507    
1508    2. Updated src/config/version to 110.27
1509    
1510    3. New boot files!
1511    
1512    ----------------------------------------------------------------------
1513    Name: Allen Leung
1514    Date: 2000/04/09 19:09:00 EST
1515    Tag: leunga-20000409-misc
1516    Description:
1517    
1518    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
1519    
1520    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
1521    
1522    ----------------------------------------------------------------------
1523    Name: Stefan
1524    Date: 2000/04/07 10:00:00 EDT
1525    Tag: monnier-20000406-branch-handling
1526    Description:
1527    
1528    Improved handling of branches (mostly those generated from
1529    polymorphic equality), removed switchoff and changed the
1530    default optimization settings (more cpsopt and less flintopt).
1531    
1532    ----------------------------------------------------------------------
1533    Name: Allen Leung
1534    Date: 2000/04/06 01:30:00 EST
1535    Tag: leunga-20000406-peephole-x86-SSA-2
1536    Description:
1537    
1538       Forgot a few files.
1539    
1540    ----------------------------------------------------------------------
1541    Name: Allen Leung
1542    Date: 2000/04/06 00:36:00 EST
1543    Tag: leunga-20000406-peephole-x86-SSA
1544    Description:
1545    
1546    1.  New Peephole code
1547    
1548    2.  Minor improvement to X86 instruction selection
1549    
1550    3.  Various fixes to SSA and machine description -> code translator
1551    
1552    ----------------------------------------------------------------------
1553    Name: Matthias Blume
1554    Date: 2000/04/05 12:30:00 JST
1555    Tag: blume_main_v110p26p2_3
1556    Description:
1557    
1558    This update just merges three minor cosmetic updates to CM's sources
1559    to get ready for the 110.27 code freeze on Friday.  No functionality
1560    has changed.
1561    
1562    ----------------------------------------------------------------------
1563    Name: Allen Leung
1564    Date: 2000/04/04 19:39:00 EST
1565    Tag: leunga-20000404-x86-asm
1566    Description:
1567    
1568    1.  Fixed a problem in X86 assembly.
1569    
1570        Things like
1571    
1572           jmp %eax
1573           jmp (%eax)
1574    
1575        should be output as
1576    
1577           jmp *%eax
1578           jmp *(%eax)
1579    
1580    2.  Assembly output
1581    
1582          Added a new flag
1583    
1584              "asm-indent-copies" (default to false)
1585    
1586          When this flag is on, parallel copies will be indented an extra level.
1587    
1588    ----------------------------------------------------------------------
1589    Name: Allen Leung
1590    Date: 2000/04/04 03:18:00 EST
1591    Tag: leunga-20000404-C--Moby
1592    Description:
1593    
1594        All of these fixes are related to C--, Moby, and my own optimization
1595        stuff; so they shouldn't affect SML/NJ.
1596    
1597    1.  X86
1598    
1599        Various fixes related floating point, and extensions.
1600    
1601    2.  Alpha
1602    
1603        Some extra patterns related to loads with signed/zero extension
1604        provided by Fermin.
1605    
1606    3.  Assembly
1607    
1608        When generating assembly, resolve the value of client defined constants,
1609        instead of generating symbolic values.  This is controlled by the
1610        new flag "asm-resolve-constants", which is default to true.
1611    
1612    4.  Machine Descriptions
1613    
1614        a. The precedence parser was slightly broken when parsing infixr symbols.
1615        b. The type generalizing code had the bound variables reversed, resulting
1616           in a problem during arity raising.
1617        c. Various fixes in machine descriptions.
1618    
1619    ----------------------------------------------------------------------
1620    Name: Matthias Blume
1621    Date: 2000/04/03 16:05:00 JST
1622    Tag: blume_main_v110p26p2_2
1623    Description:
1624    
1625    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
1626    now done via the ordinary static environment that is context to each
1627    compilation unit.
1628    
1629    To this end, I arranged that instead of "structure Core" as "structure
1630    _Core" is bound in the pervasive environment.  Core access is done via
1631    _Core (which can never be accidentally rebound because _Core is not a
1632    legal surface-syntax symbol).
1633    
1634    The current solution is much cleaner because the core environment is
1635    now simply part of the pervasive environment which is part of every
1636    compilation unit's context anyway.  In particular, this eliminates all
1637    special-case handling that was necessary until now in order to deal
1638    with dynamic and symbolic parts of the core environment.
1639    
1640    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1641    compilation manager's bootstrap compiler (actually: in the "init group"
1642    handling).  See the comments in src/system/smlnj/init/init.cmi for
1643    more details.
1644    
1645    I also tried to track down all mentions of "Core" (as string argument
1646    to Symbol.strSymbol) in the compiler and replaced them with a
1647    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
1648    name appears in one place only.
1649    
1650    Binfile and bootfile format have not changed, but the switchover from
1651    the old "init.cmi" to the new one is a bit tricky, so I supplied new
1652    bootfiles anyway.
1653    
1654    ----------------------------------------------------------------------
1655    Name: Allen Leung
1656    Date: 2000/04/02 21:17:00 EST
1657    Tag: leunga-20000402-mltree
1658    Description:
1659    
1660       1. Renamed the constructor CALL in MLTREE by popular demand.
1661       2. Added a bunch of files from my repository.  These are currently
1662          used by other non-SMLNJ backends.
1663    
1664    ----------------------------------------------------------------------
1665    Name: Allen Leung
1666    Date: 2000/03/31 21:15:00 EST
1667    Tag: leunga-20000331-aliasing
1668    Description:
1669    
1670    This update contains a rewritten (and hopefully more correct) module
1671    for extracting aliasing information from CPS.
1672    
1673       To turn on this feature:
1674    
1675            Compiler.Control.CG.memDisambiguate := true
1676    
1677       To pretty print the region information with assembly
1678    
1679           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
1680    
1681       To control how many levels of aliasing information are printed, use:
1682    
1683           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
1684    
1685       The default of n is 3.
1686    
1687    ----------------------------------------------------------------------
1688    Name: David MacQueen
1689    Date: 2000/03/31 11:15:00 EST
1690    Tag: dbm-20000331-runtime_fix
1691    Description:
1692    
1693    This update contains:
1694    
1695    1. runtime/c-lib/c-libraries.c
1696       includes added in revision 1.2 caused compilation errors on hppa-hpux
1697    
1698    2. fix for bug 1556
1699       system/Basis/Implementation/NJ/internal-signals.sml
1700    
1701    ----------------------------------------------------------------------
1702    Name: Matthias Blume
1703    Date: 2000/03/31 18:00:00 JST
1704    Tag: blume_main_v110p26p2_1
1705    Description:
1706    
1707    This update contains:
1708    
1709    1. A small change to CM's handling of stable libraries:
1710       CM now maintains one "global" modmap that is used for all stable
1711       libraries.  The use of such a global modmap maximizes sharing and
1712       minimizes the need for re-traversing parts of environments during
1713       modmap construction.  (However, this has minor impact since modmap
1714       construction seems to account for just one percent or less of total
1715       compile time.)
1716    
1717    2. I added a "genmap" phase to the statistics.  This is where I got the
1718       "one percent" number (see above).
1719    
1720    3. CM's new tool parameter mechanism just became _even_ better. :)
1721       - The parser understands named parameters and recursive options.
1722       - The "make" and "shell" tools use these new features.
1723         (This makes it a lot easier to cascade these tools.)
1724       - There is a small syntax change: named parameters use a
1725    
1726           <name> : ( <option> ... )            or
1727           <name> : <string>
1728    
1729         syntax.  Previously, named parameters were implemented in an
1730         ad-hoc fashion by each tool individually (by parsing strings)
1731         and had the form
1732    
1733           <name>=<string>
1734    
1735       See the CM manual for a full description of these issues.
1736    
1737    ----------------------------------------------------------------------
1738    Name: Matthias Blume
1739    Date: 2000/03/30 18:00:00 JST
1740    Tag: blume_main_v110p26p2_0
1741    Description:
1742    
1743    !!!!! WARNING !!!!!!
1744    !!  New binfiles  !!
1745    !!!!!!!!!!!!!!!!!!!!
1746    
1747    This update contains:
1748    
1749    1. Moderate changes to CM:
1750    
1751       - Changes to CM's tools mechanism.  In particular, it is now possible
1752       to have tools that accept additional "command line" parameters
1753       (specified in the .cm file at each instance where the tool's class is
1754       used).
1755    
1756       This was done to accommodate the new "make" and "shell" tools which
1757       facilitate fairly seamless hookup to portions of code managed using
1758       Makefiles or Shell scripts.
1759    
1760       There are no classes "shared" or "private" anymore.  Instead, the
1761       sharing annotation is now a parameter to the "sml" class.
1762    
1763       There is a bit of generic machinery for implementing one's own
1764       tools that accept command-line parameters.  However, I am not yet fully
1765       satisfied with that part, so expect changes here in the future.
1766    
1767       All existing tools are described in the CM manual.
1768    
1769       - Slightly better error handling.  (CM now suppresses many followup
1770       error messages that tended to be more annoying than helpful.)
1771    
1772    2. Major changes to the compiler's static environment data structures.
1773    
1774       - no CMStaticEnv anymore.
1775            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
1776              but it is called Environment), no conversions between different
1777              kinds of static environments
1778    
1779       - There is still a notion of a "modmap", but such modmaps are generated
1780         on demand at the time when they are needed.  This sounds slow, but I
1781         sped up the code that generates modmaps enough for this not to lead to
1782         a slowdown of the compiler (at least I didn't detect any).
1783    
1784       - To facilitate rapid modmap generation, static environments now
1785         contain an (optional) "modtree" structure.  Modtree annotations are
1786         constructed by the unpickler during unpickling.  (This means that
1787         the elaborator does not have to worry about modtrees at all.)
1788         Modtrees have the advantage that they are compositional in the same
1789         way as the environment data structure itself is compositional.
1790         As a result, modtrees never hang on to parts of an environment that
1791         has already been rendered "stale" by filtering or rebinding.
1792    
1793       - I went through many, many trials and errors before arriving at the
1794         current solution.  (The initial idea of "linkpaths" did not work.)
1795         But the result of all this is that I have touched a lot of files that
1796         depend on the "modules" and "types" data structures (most of the
1797         elaborator). There were a lot of changes during my "linkpath" trials
1798         that could have been reverted to their original state but weren't.
1799         Please, don't be too harsh on me for messing with this code a bit more
1800         than what was strictly necessary...  (I _did_ resist the tempation
1801         of doing any "global reformatting" to avoid an untimely death at
1802         Dave's hands. :)
1803    
1804       - One positive aspect of the previous point:  At least I made sure that
1805         all files that I touched now compile without warnings (other than
1806         "polyEqual").
1807    
1808       - compiler now tends to run "leaner" (i.e., ties up less memory in
1809         redundant modmaps)
1810    
1811    ----------------------------------------------------------------------
1812    Name: Allen Leung
1813    Date: 2000/03/29 18:00:00
1814    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
1815    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
1816    Description:
1817    
1818       This update contains *MAJOR* changes to the way code is generated from CPS
1819    in the module mlriscGen, and in various backend modules.
1820    
1821    CHANGES
1822    =======
1823    
1824    1. MLRiscGen: forward propagation fix.
1825    
1826       There was a bug in forward propagation introduced at about the same time
1827       as the MLRISC x86 backend, which prohibits coalescing to be
1828       performed effectively in loops.
1829    
1830       Effect: speed up of loops in RISC architectures.
1831               By itself, this actually slowed down certain benchmarks on the x86.
1832    
1833    2. MLRiscGen:  forward propagating addresses from consing.
1834    
1835       I've changed the way consing code is generated.  Basically I separated
1836       out the initialization part:
1837    
1838            store tag,   offset(allocptr)
1839            store elem1, offset+4(allocptr)
1840            store elem2, offset+8(allocptr)
1841            ...
1842            store elemn, offset+4n(allocptr)
1843    
1844       and the address computation part:
1845    
1846            celladdr <- offset+4+alloctpr
1847    
1848       and move the address computation part
1849    
1850       Effect:  register pressure is generally lower as a result.  This
1851                makes compilation of certain expressions much faster, such as
1852                long lists with non-trivial elements.
1853    
1854                 [(0,0), (0,0), .... (0,0)]
1855    
1856    3. MLRiscGen: base pointer elimination.
1857    
1858        As part of the linkage mechanism, we generate the sequence:
1859    
1860         L:  ...  <- start of the code fragment
1861    
1862         L1:
1863             base pointer <- linkreg - L1 + L
1864    
1865         The base pointer was then used for computing relocatable addresses
1866       in the code fragment.  Frequently (such as in lots of continuations)
1867       this is not needed.  We now eliminate this sequence whenever possible.
1868    
1869         For compile time efficiency, I'm using a very stupid local heuristic.
1870       But in general, this should be done as a control flow analysis.
1871    
1872       Effect:  Smaller code size.  Speed up of most programs.
1873    
1874    4. Hppa back end
1875    
1876         Long jumps in span dependence resolution used to depend on the existence
1877      of the base pointer.
1878    
1879         A jump to a long label L was expanded into the following sequence:
1880    
1881          LDIL %hi(L-8192), %r29
1882          LDO  %lo(L-8192)(%r29), %r29
1883          ADD  %r29, baseptr, %r29
1884          BV,n %r0(%r29)
1885    
1886         In the presence of change (3) above, this will not work.  I've changed
1887       it so that the following sequence of instructions are generated, which
1888       doesn't mention the base pointer at all:
1889    
1890             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
1891        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
1892             BV,n  %r0(%r29)          /* Jump */
1893    
1894    5. Alpha back end
1895    
1896          New alpha instructions LDB/LDW have been added, as per Fermin's
1897       suggestions.   This is unrelated to all other changes.
1898    
1899    6. X86 back end
1900    
1901         I've changed andl to testl in the floating point test sequence
1902         whenever appropriate.  The Intel optimization guide states that
1903         testl is preferable to andl.
1904    
1905    7. RA (x86 only)
1906    
1907         I've improved the spill propagation algorithm, using an approximation
1908       of maximal weighted independent sets.   This seems to be necessary to
1909       alleviate the negative effect in light of the slow down in (1).
1910    
1911         I'll write down the algorithm one of these days.
1912    
1913    8. MLRiscGen: frequencies
1914    
1915         I've added an annotation that states that all call gc blocks have zero
1916       execution frequencies.  This improves register allocation on the x86.
1917    
1918    BENCHMARKS
1919    ==========
1920    
1921       I've only perform the comparison on 110.25.
1922    
1923       The platforms are:
1924    
1925        HPPA  A four processor HP machine (E9000) with 5G of memory.
1926        X86   A 300Hhz Pentium II with 128M of memory, and
1927        SPARC An Ultra sparc 2 with 512M of memory.
1928    
1929       I used the following parameters for the SML benchmarks:
1930    
1931                 @SMLalloc
1932         HPPA    256k
1933         SPARC   512k
1934         X86     256k
1935    
1936    COMPILATION TIME
1937    ----------------
1938       Here are the numbers comparing the compilation times of the compilers.
1939       I've only compared 110.25 compiling the new sources versus
1940       a fixpoint version of the new compiler compiling the same.
1941    
1942                     110.25                                  New
1943               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
1944         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
1945         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
1946         X86    999s    315s       94006+130691   987s    296s    108877+141957
1947    
1948                   110.25         New
1949                Code Size      Code Size
1950         HPPA   8596736         8561421
1951         SPARC  8974299         8785143
1952         X86    9029180         8716783
1953    
1954       So in summary, things are at least as good as before.   Dramatic
1955       reduction in compilation is obtained on the Sparc; I can't explain it,
1956       but it is reproducible.  Perhaps someone should try to reproduce this
1957       on their own machines.
1958    
1959    SML BENCHMARKS
1960    --------------
1961    
1962        On the average, all benchmarks perform at least as well as before.
1963    
1964          HPPA         Compilation Time     Spill+Reload      Run Time
1965                     110.25  New            110.25    New   110.25  New
1966    
1967          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
1968              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
1969       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
1970                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
1971        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
1972             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
1973               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
1974              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
1975         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
1976             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
1977            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
1978      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
1979                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
1980             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
1981                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
1982               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
1983      --------------------------------------------------------------------------
1984       Average                     6.12%                                   4.09%
1985    
1986          SPARC        Compilation Time     Spill+Reload      Run Time
1987                     110.25  New            110.25    New   110.25  New
1988    
1989          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
1990              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
1991       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
1992                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
1993        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
1994             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
1995               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
1996              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
1997         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
1998             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1999            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
2000      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
2001                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
2002             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
2003                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
2004               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
2005      ----------------------------------------------------------------------------
2006       Average                     6.94%                                    2.64%
2007    
2008          X86          Compilation Time     Spill+Reload      Run Time
2009                     110.25  New            110.25    New   110.25  New
2010    
2011          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
2012              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
2013       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
2014                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
2015        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
2016             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
2017               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
2018              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
2019         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
2020             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
2021            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
2022      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
2023                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
2024             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
2025                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
2026               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
2027      ----------------------------------------------------------------------------
2028       Average                     1.22%                                     3.36%
2029    
2030    ----------------------------------------------------------------------
2031    Name: Allen Leung
2032    Date: 2000/03/23 16:25:00
2033    Tag: leunga-20000323-fix_x86_alpha
2034    Description:
2035    
2036    1. X86 fixes/changes
2037    
2038       a.  The old code generated for SETcc was completely wrong.
2039           The Intel optimization guide is VERY misleading.
2040    
2041    2. ALPHA fixes/changes
2042    
2043       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2044       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2045       c.  Added reassociation code for address computation.
2046    
2047    ----------------------------------------------------------------------
2048    Name: Allen Leung
2049    Date: 2000/03/22 01:23:00
2050    Tag: leunga-20000322-fix_x86_hppa_ra
2051    Description:
2052    
2053    1. X86 fixes/changes
2054    
2055       a.  x86Rewrite bug with MUL3 (found by Lal)
2056       b.  Added the instructions FSTS, FSTL
2057    
2058    2. PA-RISC fixes/changes
2059    
2060       a.  B label should not be a delay slot candidate!  Why did this work?
2061       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
2062           as it should be.
2063       c.  The assembly syntax for fstds and fstdd was wrong.
2064       d.  Added the composite instruction COMICLR/LDO, which is the immediate
2065           operand variant of COMCLR/LDO.
2066    
2067    3. Generic MLRISC
2068    
2069       a.  shuffle.sml rewritten to be slightly more efficient
2070       b.  DIV bug in mltree-simplify fixed (found by Fermin)
2071    
2072    4. Register Allocator
2073    
2074       a.  I now release the interference graph earlier during spilling.
2075           May improve memory usage.
2076    
2077    ----------------------------------------------------------------------
2078    Name: Matthias Blume
2079  Date: 2000/03/14 14:15:32  Date: 2000/03/14 14:15:32
2080  Tag: blume_main_v110p26p1_2  Tag: blume_main_v110p26p1_2
2081  Description:  Description:
2082    
2083  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2084  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
2085  specifiel the layout of the tool command line.  See the CM manual for  specifies the layout of the tool command line.  See the CM manual for
2086  explanation.  explanation.
2087    
2088  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
2089  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2090  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
2091  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 66  Line 2129 
2129  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2130  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
2131  they might be the result of an earlier partial run of fixpt (which  they might be the result of an earlier partial run of fixpt (which
2132  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
2133  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
2134  and, thus, saves a lot of time.  and, thus, saves a lot of time.
2135    
# Line 116  Line 2179 
2179    it from that remote directory.    it from that remote directory.
2180    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
2181    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
2182    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
2183    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
2184    
2185    For CVS users, this mechanism is not relevant for source archives, but    For CVS users, this mechanism is not relevant for source archives, but

Legend:
Removed from v.578  
changed lines
  Added in v.759

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