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 570, Wed Mar 8 17:30:13 2000 UTC revision 657, Fri Jun 9 05:20:54 2000 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    ----------------------------------------------------------------------
15    Name: Allen Leung
16    Date: 2000/06/09 01:25:00
17    Tag: leunga-20000609-various
18    
19    None of these things should affect normal SML/NJ operations
20    
21    1. Peephole improvements provided by Fermin (c--)
22    2. New annotation DEFUSE for adding extra dependence (moby)
23    3. New X86 LOCK instructions (moby)
24    4. New machine description language for reservation tables (scheduling)
25    5. Fixes to various optimization/analysis modules (branch chaining, dominator
26       trees etc.)
27    6. I've changed the CM files so that they can work with versions
28       110.0.6, 110.25 and 110.28
29    
30  ----------------------------------------------------------------------  ----------------------------------------------------------------------
31  Name:  Name: Matthias Blume
32  Date:  Date: 2000/06/09 12:40:00
33  Tag:  Tag: blume-20000609-log
34    Description:
35    
36    - Removed all(?) remaining RCS Log entries from sources.
37    
38    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
39      anchored paths).
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume
43    Date: 2000/06/07 17:00:00 JST
44    Tag: blume-20000607-no-implicit-anchors
45    Description:
46    
47    1. This update changes the default setting for
48    CM.Control.implicit_anchors from true to false.  This means that
49    implicit anchors are no longer permitted by default.  I also tried to
50    make sure that nothing else still relies on implicit anchors.
51    (This is the next step on the schedule towards a CM that does not even
52    have the notion of implicit anchors anymore.)
53    
54    2. More CM manual updates.
55    
56    3. I managed to track down and fix the pickling bug I mentioned last
57    time.  Because of the previously existing workaround, this entails no
58    immediate practical changes.
59    
60    ----------------------------------------------------------------------
61    Name: Matthias Blume
62    Date: 2000/06/06 11:15:00 JST
63    Tag: blume-20000606-lazierpickle
64    Description:
65    
66    !!!! NEW BOOT FILES !!!!
67    
68    * The main purpose of this update is to make library pickles lazier in
69    order to reduce the initial space penalty for autoloading a library.
70    As a result, it is now possible to have $smlnj/compiler.cm
71    pre-registered.  This should take care of the many complaints or
72    inquiries about missing structure Compiler.  This required changes to
73    CM's internal data structures and small tweaks to some algorithms.
74    
75    As a neat additional effect, it is no longer necessary (for the sake
76    of lean heap image files) to distinguish between a "minimal" CM and a
77    "full" CM.  Now, there is only one CM (i.e., the "full" version:
78    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
79    interactive top level. ($smlnj/cm/minimal.cm is gone.)
80    
81    To make the life of compiler-hackers easier, "makeml" now also
82    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
83    words, after you bootstrap a new sml for the first time, you will not
84    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
85    around you will still have to do it, though.)
86    
87    * A second change consists of major updates to the CM manual.  There
88    are now several appendices with summary information and also a full
89    specification of the CM description file syntax.
90    
91    * In directory src/system I added the script "allcross".  This script
92    invokes sml and cross-compiles the compiler for all supported
93    architectures.  (Useful when providing a new set of boot files.)
94    
95    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
96    added a small tweak to pickle-util.sml to work around this problem,
97    but it is not a proper fix yet.  I will investigate further.  (The
98    effect of the bug was an inflation of library pickle size.)
99    
100    * Version number increased to 110.28.1 (to avoid compatibility problems).
101    
102    ----------------------------------------------------------------------
103    Name: Allen Leung
104    Date: 2000/05/25 17:28 EDT
105    Tag: leunga-20000525-ra
106    Description:
107    
108      Fixed a bug in freezing phase of the register allocator.
109    
110    ----------------------------------------------------------------------
111    Name: Allen Leung
112    Date: 2000/05/15 22:53 EDT
113    Tag: leunga-20000515-alpha-x86-ra
114    Description:
115    
116      1. Alpha
117    
118          Slight cleanup.  Removed the instruction SGNXL
119    
120      2. X86
121    
122          Added the following instructions to the instruction set:
123    
124            ROLx, RORx,
125            BTx, BTSx, BTLx, BTRx,
126            XCHGx, and variants with the LOCK prefix
127    
128      3. Register Allocation
129    
130          The module ra-rewrite-with-renaming has been improved.
131    
132      These have no effect on SML/NJ.
133    
134    ----------------------------------------------------------------------
135    Name: Matthias Blume
136    Date: 2000/05/15 16:20:00 JST
137    Tag: blume-20000515-lightrebuild
138    Description:
139    
140    1. I added an alternative to "-rebuild" to "makeml".  The difference is
141       that prior to calling CMB.make' the CM-variable "LIGHT" will be
142       defined.  In effect, the command will not build any cross-compiler
143       backends and therefore finish more quickly.
144    
145       The "fixpt" script also takes a "-light" switch to be able to use
146       this new facility while compiling for a fixpoint.
147    
148    2. I replaced all mentions of anchored paths in group owner specifications
149       with simple relative paths (usually starting with "..").
150       The rationale is that a library's internal workings should not be
151       compromised by the lack of some anchor.  (An anchor is necessary
152       for someone who wants to refer to the library by an anchored path,
153       but it should not be necessary to build the same library in the first
154       place.)
155    
156    3. I changed the way CM's tool mechanism determines the shell command
157       string used for things like ml-yacc etc. so that it does not break
158       when CM.Control.implicit_anchors is turned off.
159    
160    ----------------------------------------------------------------------
161    Name: Matthias Blume
162    Date: 2000/05/12 18:20:00 JST
163    Tag: blume-20000512-ml-build
164    Description:
165    
166    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
167    getting installed properly (by config/install.sh).
168    
169    ----------------------------------------------------------------------
170    Name: Matthias Blume
171    Date: 2000/05/12 17:30:00 JST
172    Tag: blume-20000512-anchors
173    Description:
174    
175    !!! NEW BOOT FILES !!!
176    
177    This change is in preparation of fading out support for "implicitly
178    anchored path names".  I went through all sources and used the
179    explicit (and relatively new) $-notation.  See system/README and the
180    CM manual for more info on this.
181    
182    I also modified the anchoring scheme for some things such as "smlnj",
183    "MLRISC", "cm", etc. to take advantage of the fact that explicit
184    anchors are more expressive: anchor name and first arc do not have to
185    coincide.  This entails the following user-visible change:
186    
187    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
188    particular, when you fire up sml with a command-line argument, say,
189    e.g.:
190    
191       sml '$smlnj/cmb.cm'
192    
193    At the ML toplevel prompt:
194    
195       CM.autoload "$smlnj/cmb.cm";
196    
197    There is also a new controller in CM.Control that can be used to turn
198    off all remaining support for implicit anchors by saying:
199    
200        CM.autoload "$smlnj/
201        #set CM.Control.implicit_anchors false;
202    
203    This causes CM to reject implicitly anchored paths.  This is (for the
204    time being) less permissive than the "final" version where there will
205    be no more such implicit anchors and relative paths will be just that:
206    relative.
207    
208    The next step (version after next version?) will be to make the
209    default for CM.Control.implicit_anchors false.  After the dust has
210    settled, I can then produce the "final" version of this...
211    
212    Note: Since bootstrapping is a bit tricky, I provided new boot files.
213    
214    ----------------------------------------------------------------------
215    Name: Matthias Blume
216    Date: 2000/05/11 16:30:00 JST
217    Tag: blume-20000511-sources
218    Description:
219    
220    The main change is that I added function CM.sources as a generalized
221    version of the earlier CM.makedepend.  This entails the following
222    additional changes:
223    
224      - CM.makedepend has been dropped.
225    
226      - CM manual has been updated.
227    
228      - TOOLS signature and API have been changed.
229    
230    ----------------------------------------------------------------------
231    Name: Allen Leung
232    Date: 2000/05/10 21:17 EDT
233    Tag: leunga-20000510-moby-c--ssa
234    Description:
235    
236      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
237    None of these affect SML/NJ.
238    
239    1. Register Allocation
240    
241        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
242           This module tries to remove local (i.e. basic block level) redundancies
243           during spilling.
244    
245        b. A new framework for performing region based register allocation.
246           Not yet entirely functional.
247    
248    2. X86
249    
250       a. DefUse for POP was missing the stack pointer [found by Lal]
251       b. Reload for CALL was incorrect in X86Spill [found by John]
252       c. Various fixes in X86Spill so that it can be used correctly for
253          the new spilling module.
254    
255    3. SSA/IR
256    
257       a. New module ir/dj-dataflow.sml implements elimination based
258          data flow analysis.
259    
260    4. MLRiscGen
261    
262       a. Fix for gc type annotation
263    
264    5. MDGen
265    
266       Various fixes for machine description -> ml code translation.  For ssa
267       only.
268    
269    ----------------------------------------------------------------------
270    Name: Allen Leung
271    Date: 2000/05/08 22:17 EDT
272    Tag: leunga-20000508-labexp
273    Description:
274    
275      Fermin has found a few assembly problems with constant expressions
276      generated in LabelExp.  Mostly, the problems involve extra parentheses,
277      which choke on dumb assemblers.  This is his fix.
278    
279    ----------------------------------------------------------------------
280    Name: Dave MacQueen
281    Date: 2000/04/09 14:00 EDT
282    Tag: dbm-20000502-Version_110_28
283    Description:
284    
285    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
286    
287    2. Updated config/version to 110.28
288    
289    3. Updated config/srcarchiveurl
290    
291    3. New boot files!
292       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume
296    Date: 2000/05/01 19:05:00 JST
297    Tag: blume-20000501-noweb
298    Description:
299    
300    A new noweb tool has been added.  The existing system is entirely
301    unaffected by this, but some CM users have asked for renewed noweb
302    support.  Everything is documented in the CM manual.
303    
304    New (plugin) libraries:
305    
306       noweb-tool.cm
307       nw-ext.cm
308    
309    ----------------------------------------------------------------------
310    Name: Dave MacQueen
311    Date: 2000/04/30 12:40PM EDT
312    Tag: dbm-20000430-bug_fixes
313    Description:
314    
315    1. Fix for bug 1498
316       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
317       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
318         added toRealArray function
319       smlnj/src/compiler/MiscUtil/print/ppobj.sml
320         added check for tag Obj.RealArray to array printing case in ppObj
321    
322    2. Fix for bug 1510
323       smlnj/src/compiler/Semant/types/typesutil.sml
324         fixed definition of dummyargs (used by equalTycon) so that
325         dummy args are distinct types
326    
327    ----------------------------------------------------------------------
328    Name: Matthias Blume
329    Date: 2000/04/30 01:00:00 JST
330    Tag: blume-20000430-versions
331    Description:
332    
333    1. CM version numbering added.  This is an implementation of Lal's
334       proposal for adding version numbers and version checking to .cm
335       files.  Lal said that his proposal was just that -- a proposal.
336       For the time being I went ahead and implemented it so that people
337       can comment on it.  Everything is completely backward-compatible
338       (except for the stable library format, i.e., new bootfiles!).
339    
340       As usual, see the CM manual for details.
341    
342    2. An alternative syntax for anchored paths has been implemented.
343       Dave has recently voiced the same concerns that I had when I did
344       this, so there should be some support.  My take is that eventually
345       I will let support for the current syntax (where anchors are
346       "implicit") fade out in favor of the new, explicit syntax.
347       In order to be backward-compatible, both old and new syntax are
348       currently supported.
349    
350       Again, see the CM manual for details.
351    
352    3. Parallel make is trying to be slightly smarter:  When the master
353       process finds a "bottleneck", i.e., when there is only one
354       compilation unit that can be compiled and everybody else is
355       waiting on it, then it will simply compile it directly instead
356       of clumsily telling one of the slaves to do it.
357    
358    4. Support for "unsharing" added.  This is necessary in order to be
359       able to have two different versions of the same library running
360       at the same time (e.g., for trying out a new MLRISC while still
361       having the old MLRISC linked into the current compiler, etc.)
362       See the CM manual.
363    
364    5. Simple "makedepend" functionality added for generating Makefile
365       dependency information.  (This is rather crude at the moment.
366       Expect some changes here in the future.)
367    
368    6. ".fun" added as a recognized suffix for ML files. Also documented
369       explicitly in the manual that the fallback behavior (unknown suffix
370       -> ML file) is not an official feature!
371    
372    7. Small changes to the pickler for stable libraries.
373    
374    8. Several internal changes to CM (for cleanup/improvement).
375    
376    
377    !!!! NEW BINFILES !!!!
378    
379    ----------------------------------------------------------------------
380    Name: Matthias Blume
381    Date: 2000/04/28 17:30:00 JST
382    Tag: blume-20000428-pathconfig
383    Description:
384    
385    1. I changed config/install.sh to remove duplicate entries from the
386       lib/pathconfig file at the end.  Moreover, the final version of
387       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
388       in src/system/installml.
389    
390    2. The config/install.sh script now consistently uses relative
391       pathnames in lib/pathconfig whenever the anchor is in the lib
392       directory.  (So far this was true for the libraries that come
393       pre-compiled and bundled as part of the bootfiles but not for
394       libraries that are compiled by the script itself.)
395    
396    ----------------------------------------------------------------------
397    Name: Matthias Blume
398    Date: 2000/04/26 13:10:00 JST
399    Tag: blume-20000426-fun_suffix
400    Description:
401    
402    Added ".fun" as a recognized file name suffix (for ML code).
403    
404    ----------------------------------------------------------------------
405    Name: Allen Leung
406    Date: 2000/04/25 17:00:00 EST
407    Tag: leunga-20000425-alpha-ra
408    Description:
409    
410    1. Alpha
411    
412        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
413    in C--.
414    
415    2. RA
416    
417       Added a flag "ra-dump-size" to print out the size of the flowgraph
418       and the interference graph.
419    
420    ----------------------------------------------------------------------
421    Name: Dave MacQueen
422    Date: 2000/04/25/
423    Tag: dbm-20000425-mlyacc_doc_examples
424    Description:
425      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
426      Updated all three examples in src/ml-yacc/examples to run
427      under 110.* using CM.make.
428    
429    ----------------------------------------------------------------------
430    Name: Allen Leung
431    Date: 2000/04/20 23:04:00 EST
432    Tag: leunga-20000420-ssa-c---stuff
433    Description:
434    
435      This update synchronizes my repository with Yale's.  Most of these
436    changes, however, do not affect SML/NJ at all (the RA is an exception).
437    
438    1. Register Allocator
439    
440       a. An improvement in the interference graph construction:
441          Given a copy
442    
443                s <- t
444    
445          no interference edge between s and t is added for this definition of s.
446    
447       b. I've added two new spill heuristic modules that Fermin and I developed
448          (in the new library RA.cm). These are unused in SML/NJ but maybe
449          useful for others (Moby?)
450    
451    2. X86
452    
453       a. Various fixes in the backend provided by Fermin [C--] and Lal.
454    
455    3. Alpha
456    
457       a. Added the BSR instruction and code generation that goes with it [C--]
458       b. Other fixes too numerous to recount provided by Fermin [C--]
459    
460    4. Regmaps
461    
462       a. The regmaps are not initialized with the identity physical bindings
463          at creation time.  This is unneeded.
464    
465    5. MLRISC Optimizations
466    
467       a. The DJ-Graph module can now compute the iterated dominance frontiers
468          intersects with liveness incrementally in linear time! Woohoo!
469          This is now used in my new SSA construction algorithm.
470    
471       b. THe branch reorganization module is now smarter about linear chains of
472          basic blocks.
473    
474    
475    ----------------------------------------------------------------------
476    Name: Matthias Blume
477    Date: 2000/04/12 13:52:00 JST
478    Tag: blume_main_v110p27_1
479    Description:
480    
481    Changed install.sh script to handle archive files without version number
482    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
483    name of the boot file archive.
484    
485    ----------------------------------------------------------------------
486    Name: Dave MacQueen
487    Date: 2000/04/09 14:00 EDT
488    Tag: dbm-20000410-Version_110_27
489    Description:
490    
491    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
492    
493    2. Updated src/config/version to 110.27
494    
495    3. New boot files!
496    
497    ----------------------------------------------------------------------
498    Name: Allen Leung
499    Date: 2000/04/09 19:09:00 EST
500    Tag: leunga-20000409-misc
501    Description:
502    
503    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
504    
505    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
506    
507    ----------------------------------------------------------------------
508    Name: Stefan
509    Date: 2000/04/07 10:00:00 EDT
510    Tag: monnier-20000406-branch-handling
511    Description:
512    
513    Improved handling of branches (mostly those generated from
514    polymorphic equality), removed switchoff and changed the
515    default optimization settings (more cpsopt and less flintopt).
516    
517    ----------------------------------------------------------------------
518    Name: Allen Leung
519    Date: 2000/04/06 01:30:00 EST
520    Tag: leunga-20000406-peephole-x86-SSA-2
521    Description:
522    
523       Forgot a few files.
524    
525    ----------------------------------------------------------------------
526    Name: Allen Leung
527    Date: 2000/04/06 00:36:00 EST
528    Tag: leunga-20000406-peephole-x86-SSA
529    Description:
530    
531    1.  New Peephole code
532    
533    2.  Minor improvement to X86 instruction selection
534    
535    3.  Various fixes to SSA and machine description -> code translator
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume
539    Date: 2000/04/05 12:30:00 JST
540    Tag: blume_main_v110p26p2_3
541    Description:
542    
543    This update just merges three minor cosmetic updates to CM's sources
544    to get ready for the 110.27 code freeze on Friday.  No functionality
545    has changed.
546    
547    ----------------------------------------------------------------------
548    Name: Allen Leung
549    Date: 2000/04/04 19:39:00 EST
550    Tag: leunga-20000404-x86-asm
551    Description:
552    
553    1.  Fixed a problem in X86 assembly.
554    
555        Things like
556    
557           jmp %eax
558           jmp (%eax)
559    
560        should be output as
561    
562           jmp *%eax
563           jmp *(%eax)
564    
565    2.  Assembly output
566    
567          Added a new flag
568    
569              "asm-indent-copies" (default to false)
570    
571          When this flag is on, parallel copies will be indented an extra level.
572    
573    ----------------------------------------------------------------------
574    Name: Allen Leung
575    Date: 2000/04/04 03:18:00 EST
576    Tag: leunga-20000404-C--Moby
577    Description:
578    
579        All of these fixes are related to C--, Moby, and my own optimization
580        stuff; so they shouldn't affect SML/NJ.
581    
582    1.  X86
583    
584        Various fixes related floating point, and extensions.
585    
586    2.  Alpha
587    
588        Some extra patterns related to loads with signed/zero extension
589        provided by Fermin.
590    
591    3.  Assembly
592    
593        When generating assemby, resolve the value of client defined constants,
594        instead of generating symbolic values.  This is controlled by the
595        new flag "asm-resolve-constants", which is default to true.
596    
597    4.  Machine Descriptions
598    
599        a. The precedence parser was slightly broken when parsing infixr symbols.
600        b. The type generalizing code had the bound variables reversed, resulting
601           in a problem during arity raising.
602        c. Various fixes in machine descriptions.
603    
604    ----------------------------------------------------------------------
605    Name: Matthias Blume
606    Date: 2000/04/03 16:05:00 JST
607    Tag: blume_main_v110p26p2_2
608    Description:
609    
610    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
611    now done via the ordinary static environment that is context to each
612    compilation unit.
613    
614    To this end, I arranged that instead of "structure Core" as "structure
615    _Core" is bound in the pervasive environment.  Core access is done via
616    _Core (which can never be accidentially rebound because _Core is not a
617    legal surface-syntax symbol).
618    
619    The current solution is much cleaner because the core environment is
620    now simply part of the pervasive environment which is part of every
621    compilation unit's context anyway.  In particular, this eliminates all
622    special-case handling that was necessary until now in order to deal
623    with dynamic and symbolic parts of the core environment.
624    
625    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
626    compilation mananger's bootstrap compiler (actually: in the "init group"
627    handling).  See the comments in src/system/smlnj/init/init.cmi for
628    more details.
629    
630    I also tried to track down all mentions of "Core" (as string argument
631    to Symbol.strSymbol) in the compiler and replaced them with a
632    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
633    name appears in one place only.
634    
635    Binfile and bootfile format have not changed, but the switchover from
636    the old "init.cmi" to the new one is a bit tricky, so I supplied new
637    bootfiles anyway.
638    
639    ----------------------------------------------------------------------
640    Name: Allen Leung
641    Date: 2000/04/02 21:17:00 EST
642    Tag: leunga-20000402-mltree
643    Description:
644    
645       1. Renamed the constructor CALL in MLTREE by popular demand.
646       2. Added a bunch of files from my repository.  These are currently
647          used by other non-SMLNJ backends.
648    
649    ----------------------------------------------------------------------
650    Name: Allen Leung
651    Date: 2000/03/31 21:15:00 EST
652    Tag: leunga-20000331-aliasing
653    Description:
654    
655    This update contains a rewritten (and hopefully more correct) module
656    for extracting aliasing information from CPS.
657    
658       To turn on this feature:
659    
660            Compiler.Control.CG.memDisambiguate := true
661    
662       To pretty print the region information with assembly
663    
664           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
665    
666       To control how many levels of aliasing information are printed, use:
667    
668           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
669    
670       The default of n is 3.
671    
672    ----------------------------------------------------------------------
673    Name: David MacQueen
674    Date: 2000/03/31 11:15:00 EST
675    Tag: dbm-20000331-runtime_fix
676    Description:
677    
678    This update contains:
679    
680    1. runtime/c-lib/c-libraries.c
681       includes added in revision 1.2 caused compilation errors on hppa-hpux
682    
683    2. fix for bug 1556
684       system/Basis/Implementation/NJ/internal-signals.sml
685    
686    ----------------------------------------------------------------------
687    Name: Matthias Blume
688    Date: 2000/03/31 18:00:00 JST
689    Tag: blume_main_v110p26p2_1
690    Description:
691    
692    This update contains:
693    
694    1. A small change to CM's handling of stable libraries:
695       CM now maintains one "global" modmap that is used for all stable
696       libraries.  The use of such a global modmap maximizes sharing and
697       minimizes the need for re-traversing parts of environments during
698       modmap construction.  (However, this has minor impact since modmap
699       construction seems to account for just one percent or less of total
700       compile time.)
701    
702    2. I added a "genmap" phase to the statistics.  This is where I got the
703       "one percent" number (see above).
704    
705    3. CM's new tool parameter mechanism just became _even_ better. :)
706       - The parser understands named parameters and recursive options.
707       - The "make" and "shell" tools use these new features.
708         (This makes it a lot easier to cascade these tools.)
709       - There is a small syntax change: named parameters use a
710    
711           <name> : ( <option> ... )            or
712           <name> : <string>
713    
714         syntax.  Previously, named parameters were implemented in an
715         ad-hoc fashion by each tool individually (by parsing strings)
716         and had the form
717    
718           <name>=<string>
719    
720       See the CM manual for a full description of these issues.
721    
722    ----------------------------------------------------------------------
723    Name: Matthias Blume
724    Date: 2000/03/30 18:00:00 JST
725    Tag: blume_main_v110p26p2_0
726    Description:
727    
728    !!!!! WARNING !!!!!!
729    !!  New binfiles  !!
730    !!!!!!!!!!!!!!!!!!!!
731    
732    This update contains:
733    
734    1. Moderate changes to CM:
735    
736       - Changes to CM's tools mechanism.  In particular, it is now possible
737       to have tools that accept additional "command line" parameters
738       (specified in the .cm file at each instance where the tool's class is
739       used).
740    
741       This was done to accomodate the new "make" and "shell" tools which
742       facilitate fairly seemless hookup to portions of code managed using
743       Makefiles or Shell scripts.
744    
745       There are no classes "shared" or "private" anymore.  Instead, the
746       sharing annotation is now a parameter to the "sml" class.
747    
748       There is a bit of generic machinery for implementing one's own
749       tools that accept command-line parameters.  However, I am not yet fully
750       satisfied with that part, so expect changes here in the future.
751    
752       All existing tools are described in the CM manual.
753    
754       - Slightly better error handling.  (CM now surpresses many followup
755       error messages that tended to be more annoying than helpful.)
756    
757    2. Major changes to the compiler's static environment data structures.
758    
759       - no CMStaticEnv anymore.
760            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
761              but it is called Environment), no conversions between different
762              kinds of static environments
763    
764       - There is still a notion of a "modmap", but such modmaps are generated
765         on demand at the time when they are needed.  This sounds slow, but I
766         sped up the code that generates modmaps enough for this not to lead to
767         a slowdown of the compiler (at least I didn't detect any).
768    
769       - To facilitate rapid modmap generation, static environments now
770         contain an (optional) "modtree" structure.  Modtree annotations are
771         constructed by the unpickler during unpickling.  (This means that
772         the elaborator does not have to worry about modtrees at all.)
773         Modtrees have the advantage that they are compositional in the same
774         way as the environment data structure itself is compositional.
775         As a result, modtrees never hang on to parts of an environment that
776         has already been rendered "stale" by filtering or rebinding.
777    
778       - I went through many, many trials and errors before arriving at the
779         current solution.  (The initial idea of "linkpaths" did not work.)
780         But the result of all this is that I have touched a lot of files that
781         depend on the "modules" and "types" data structures (most of the
782         elaborator). There were a lot of changes during my "linkpath" trials
783         that could have been reverted to their original state but weren't.
784         Please, don't be too harsh on me for messing with this code a bit more
785         than what was strictly necessary...  (I _did_ resist the tempation
786         of doing any "global reformatting" to avoid an untimely death at
787         Dave's hands. :)
788    
789       - One positive aspect of the previous point:  At least I made sure that
790         all files that I touched now compile without warnings (other than
791         "polyEqual").
792    
793       - compiler now tends to run "leaner" (i.e., ties up less memory in
794         redundant modmaps)
795    
796    ----------------------------------------------------------------------
797    Name: Allen Leung
798    Date: 2000/03/29 18:00:00
799    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
800    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
801    Description:
802    
803       This update contains *MAJOR* changes to the way code is generated from CPS
804    in the module mlriscGen, and in various backend modules.
805    
806    CHANGES
807    =======
808    
809    1. MLRiscGen: forward propagation fix.
810    
811       There was a bug in forward propagation introduced at about the same time
812       as the MLRISC x86 backend, which prohibits coalescing to be
813       performed effectively in loops.
814    
815       Effect: speed up of loops in RISC architectures.
816               By itself, this actually slowed down certain benchmarks on the x86.
817    
818    2. MLRiscGen:  forward propagating addresses from consing.
819    
820       I've changed the way consing code is generated.  Basically I separated
821       out the initialization part:
822    
823            store tag,   offset(allocptr)
824            store elem1, offset+4(allocptr)
825            store elem2, offset+8(allocptr)
826            ...
827            store elemn, offset+4n(allocptr)
828    
829       and the address computation part:
830    
831            celladdr <- offset+4+alloctpr
832    
833       and move the address computation part
834    
835       Effect:  register pressure is generally lower as a result.  This
836                makes compilation of certain expressions much faster, such as
837                long lists with non-trivial elements.
838    
839                 [(0,0), (0,0), .... (0,0)]
840    
841    3. MLRiscGen: base pointer elimination.
842    
843        As part of the linkage mechanism, we generate the sequence:
844    
845         L:  ...  <- start of the code fragment
846    
847         L1:
848             base pointer <- linkreg - L1 + L
849    
850         The base pointer was then used for computing relocatable addresses
851       in the code fragment.  Frequently (such as in lots of continuations)
852       this is not needed.  We now eliminate this sequence whenever possible.
853    
854         For compile time efficiency, I'm using a very stupid local heuristic.
855       But in general, this should be done as a control flow analysis.
856    
857       Effect:  Smaller code size.  Speed up of most programs.
858    
859    4. Hppa back end
860    
861         Long jumps in span dependence resolution used to depend on the existence
862      of the base pointer.
863    
864         A jump to a long label L was expanded into the following sequence:
865    
866          LDIL %hi(L-8192), %r29
867          LDO  %lo(L-8192)(%r29), %r29
868          ADD  %r29, baseptr, %r29
869          BV,n %r0(%r29)
870    
871         In the presence of change (3) above, this will not work.  I've changed
872       it so that the following sequence of instructions are generated, which
873       doesn't mention the base pointer at all:
874    
875             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
876        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
877             BV,n  %r0(%r29)          /* Jump */
878    
879    5. Alpha back end
880    
881          New alpha instructions LDB/LDW have been added, as per Fermin's
882       suggestions.   This is unrelated to all other changes.
883    
884    6. X86 back end
885    
886         I've changed andl to testl in the floating point test sequence
887         whenever appropriate.  The Intel optimization guide states that
888         testl is perferable to andl.
889    
890    7. RA (x86 only)
891    
892         I've improved the spill propagation algorithm, using an approximation
893       of maximal weighted independent sets.   This seems to be necessary to
894       alleviate the negative effect in light of the slow down in (1).
895    
896         I'll write down the algorithm one of these days.
897    
898    8. MLRiscGen: frequencies
899    
900         I've added an annotation that states that all call gc blocks have zero
901       execution frequencies.  This improves register allocation on the x86.
902    
903    BENCHMARKS
904    ==========
905    
906       I've only perform the comparison on 110.25.
907    
908       The platforms are:
909    
910        HPPA  A four processor HP machine (E9000) with 5G of memory.
911        X86   A 300Hhz Pentium II with 128M of memory, and
912        SPARC An Ultra sparc 2 with 512M of memory.
913    
914       I used the following parameters for the SML benchmarks:
915    
916                 @SMLalloc
917         HPPA    256k
918         SPARC   512k
919         X86     256k
920    
921    COMPILATION TIME
922    ----------------
923       Here are the numbers comparing the compilation times of the compilers.
924       I've only compared 110.25 compiling the new sources versus
925       a fixpoint version of the new compiler compiling the same.
926    
927                     110.25                                  New
928               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
929         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
930         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
931         X86    999s    315s       94006+130691   987s    296s    108877+141957
932    
933                   110.25         New
934                Code Size      Code Size
935         HPPA   8596736         8561421
936         SPARC  8974299         8785143
937         X86    9029180         8716783
938    
939       So in summary, things are at least as good as before.   Dramatic
940       reduction in compilation is obtained on the Sparc; I can't explain it,
941       but it is reproducible.  Perhaps someone should try to reproduce this
942       on their own machines.
943    
944    SML BENCHMARKS
945    --------------
946    
947        On the average, all benchmarks perform at least as well as before.
948    
949          HPPA         Compilation Time     Spill+Reload      Run Time
950                     110.25  New            110.25    New   110.25  New
951    
952          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
953              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
954       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
955                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
956        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
957             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
958               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
959              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
960         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
961             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
962            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
963      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
964                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
965             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
966                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
967               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
968      --------------------------------------------------------------------------
969       Average                     6.12%                                   4.09%
970    
971          SPARC        Compilation Time     Spill+Reload      Run Time
972                     110.25  New            110.25    New   110.25  New
973    
974          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
975              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
976       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
977                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
978        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
979             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
980               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
981              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
982         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
983             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
984            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
985      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
986                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
987             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
988                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
989               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
990      ----------------------------------------------------------------------------
991       Average                     6.94%                                    2.64%
992    
993          X86          Compilation Time     Spill+Reload      Run Time
994                     110.25  New            110.25    New   110.25  New
995    
996          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
997              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
998       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
999                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1000        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1001             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1002               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1003              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1004         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1005             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1006            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1007      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1008                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1009             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1010                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1011               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1012      ----------------------------------------------------------------------------
1013       Average                     1.22%                                     3.36%
1014    
1015    ----------------------------------------------------------------------
1016    Name: Allen Leung
1017    Date: 2000/03/23 16:25:00
1018    Tag: leunga-20000323-fix_x86_alpha
1019    Description:
1020    
1021    1. X86 fixes/changes
1022    
1023       a.  The old code generated for SETcc was completely wrong.
1024           The Intel optimization guide is VERY misleading.
1025    
1026    2. ALPHA fixes/changes
1027    
1028       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1029       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1030       c.  Added reassociation code for address computation.
1031    
1032    ----------------------------------------------------------------------
1033    Name: Allen Leung
1034    Date: 2000/03/22 01:23:00
1035    Tag: leunga-20000322-fix_x86_hppa_ra
1036    Description:
1037    
1038    1. X86 fixes/changes
1039    
1040       a.  x86Rewrite bug with MUL3 (found by Lal)
1041       b.  Added the instructions FSTS, FSTL
1042    
1043    2. PA-RISC fixes/changes
1044    
1045       a.  B label should not be a delay slot candidate!  Why did this work?
1046       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1047           as it should be.
1048       c.  The assembly syntax for fstds and fstdd was wrong.
1049       d.  Added the composite instruction COMICLR/LDO, which is the immediate
1050           operand variant of COMCLR/LDO.
1051    
1052    3. Generic MLRISC
1053    
1054       a.  shuffle.sml rewritten to be slightly more efficient
1055       b.  DIV bug in mltree-simplify fixed (found by Fermin)
1056    
1057    4. Register Allocator
1058    
1059       a.  I now release the interference graph earlier during spilling.
1060           May improve memory usage.
1061    
1062    ----------------------------------------------------------------------
1063    Name: Matthias Blume
1064    Date: 2000/03/14 14:15:32
1065    Tag: blume_main_v110p26p1_2
1066    Description:
1067    
1068    1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1069    additional argument called "template" which is an optional string that
1070    specifiel the layout of the tool command line.  See the CM manual for
1071    explanation.
1072    
1073    2. A special-purpose tool can be "regisitered" by simply dropping the
1074    corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1075    directory where the .cm file lives that uses this tool.  (The
1076    behavior/misfeature until now was to look for the tool description
1077    files in the current working directory.)  As before, tool description
1078    files could also be anchored -- in which case they can live anywhere
1079    they like.  Following the recent e-mail discussion, this change should
1080    make it easier to have special-purpose tools that are shipped together
1081    with the sources of the program that uses them.
1082    
1083    ----------------------------------------------------------------------
1084    Name: Matthias Blume
1085    Date: 2000/03/10 07:48:34
1086    Tag: blume_main_v110p26p1_1
1087    Description:
1088    
1089    I added a re-written version of Dave's fixpt script to src/system.
1090    Changes relative to the original version:
1091      - sh-ified (not everybody has ksh)
1092      - automatically figures out which architecture it runs on
1093      - uses ./makeml a bit more cleverly
1094      - never invokes ./installml (and, thus, does not clobber your
1095        good and working installation of sml in case something goes wrong)
1096      - accepts max iteration count using option "-iter <n>"
1097      - accepts a "base" name using option "-base <base>"
1098    
1099    It does not build any extraneous heap images but directly rebuilds
1100    bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
1101    it can incorporate existing bin- and boot- hierarchies.  For example,
1102    suppose the base is set to "sml" (which is the default).  Then it
1103    successively builds
1104    
1105            sml.bin.<arch>-unix and sml.boot.<arch>-unix
1106    then    sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
1107    then    sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
1108    ...
1109    then    sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
1110    
1111    and so on.  If any of these already exist, it will just use what's
1112    there.  In particular, many people will have the initial set of bin
1113    and boot files around, so this saves time for at least one full
1114    rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
1115    <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
1116    they might be the result of an earlier partial run of fixpt (which
1117    perhaps got accidentially killed).  In this case, fixpt will quickly
1118    move through what exists before continuing where it left off earlier,
1119    and, thus, saves a lot of time.
1120    
1121    ----------------------------------------------------------------------
1122    Name: Allen Leung
1123    Date: 00/03/10 02:20:00
1124    Tag: leunga-20000310-fix_x86_asm_ra
1125    Description:
1126    
1127    More assembly output problems involving the indexed addressing mode
1128    on the x86 have been found and corrected. Thanks to Fermin Reig for the
1129    fix.
1130    
1131    The interface and implementation of the register allocator have been changed
1132    slightly to accommodate the possibility to skip the register allocation
1133    phases completely and go directly to memory allocation.  This is needed
1134    for C-- use.
1135    
1136    ----------------------------------------------------------------------
1137    Name: Matthias Blume
1138    Date: 00/03/09 10:23:53
1139    Tag: blume_main_v110p26p1_0
1140    Description:
1141    
1142    * Complete re-organization of library names.  Many libraries have been
1143    consolidated so that they share the same path anchor.  For example,
1144    all MLRISC-related libraries are anchored at MLRISC, most libraries that
1145    are SML/NJ-specific are under "smlnj".  Notice that names like
1146    host-cmb.cm or host-compiler.cm no longer exist.  See system/README
1147    for a complete description of the new naming scheme.  Quick reference:
1148    
1149       host-cmb.cm        -> smlnj/cmb.cm
1150       host-compiler.cm   -> smlnj/compiler.cm
1151       full-cm.cm         -> smlnj/cm.cm
1152       <arch>-<os>.cm     -> smlnj/cmb/<arch>-<os>.cm
1153       <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
1154    
1155    * Bug fixes in CM.
1156        - exceptions in user code are being passed through (i.e., reach top level)
1157        - more bugs in paranoia mode fixed
1158        - bug related to checking group owners fixed
1159    
1160    * New install.sh script that automagically fetches archive files:
1161      The new file config/srcarchiveurl must contain the URL of the
1162      (remote) directory that contains bin files (or other source archives).
1163      If install.sh does not find the archive locally, it tries to get
1164      it from that remote directory.
1165      This should simplify installation further:  For machines that have
1166      access to the internet, just fetch <version>-config.tgz, unpack it,
1167      edit config/targets, and go (run config/install.sh).  The scipt will
1168      fetch everything else that it might need all by itself.
1169    
1170      For CVS users, this mechanism is not relevant for source archives, but
1171      it is convenient for getting new sets of binfiles.
1172    
1173      Archives should be tar files compressed with either gzip, compress, or
1174      bzip2.  The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
1175    
1176    ----------------------------------------------------------------------
1177    Name: Matthias Blume
1178    Date: 2000/03/07 04:01:04
1179    Tag: blume_main_v110_26_2
1180  Description:  Description:
1181    - size info in BOOTLIST
1182         * no fixed upper limits for number of bootfiles or length of
1183           bootfile names in runtime
1184         * falling back to old behavior if no BOOTLIST size info found
1185    - allocation size heuristics in .run-sml
1186         * tries to read cache size from /proc/cpuinfo (this is important for
1187            small-cache Celeron systems!)
1188    - install.sh robustified
1189    - CM manual updates
1190    - paranoid mode
1191         * no more CMB.deliver() (i.e., all done by CMB.make())
1192         * can re-use existing sml.boot.* files
1193         * init.cmi now treated as library
1194         * library stamps for consistency checks
1195    - sml.boot.<arch>-<os>/PIDMAP file
1196         * This file is read by the CM startup code.  This is used to minimize
1197           the amount of dynamic state that needs to be stowed away for the
1198           purpose of sharing between interactive system and user code.
1199    - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
1200         * Upon request by Elsa.  Anchors now controlled by get-set-pair
1201           like most other CM state variables.
1202    - Compiler.CMSA eliminated
1203         * No longer supported by CM anyway.
1204    - fixed bugs in pickler that kept biting Stefan
1205         * past refs to past refs (was caused by the possibility that
1206           ad-hoc sharing is more discriminating than hash-cons sharing)
1207         * integer overflow on LargeInt.minInt
1208    - ml-{lex,yacc} build scripts now use new mechanism
1209      for building standalone programs
1210    - fixed several gcc -Wall warnings that were caused by missing header
1211      files, missing initializations, etc., in runtime (not all warnings
1212      eliminated, though)

Legend:
Removed from v.570  
changed lines
  Added in v.657

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