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 583, Thu Mar 23 21:52:30 2000 UTC revision 670, Sun Jun 18 13:10:57 2000 UTC
# Line 12  Line 12 
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14  ----------------------------------------------------------------------  ----------------------------------------------------------------------
15    Name: Matthias Blume
16    Date: 2000/06/18 22:00:10 JST
17    Tag: blume-20000618-implicit-anchors-really-gone
18    Description:
19    
20    I updates the previous HISTORY entry where I forgot to mention that
21    implicit anchors are no longer with us.
22    
23    The current update also gets rid of the (now useless) controller
24    CM.Control.implicit_anchors.
25    
26    ----------------------------------------------------------------------
27    Name: Matthias Blume
28    Date: 2000/06/16 17:30:00 JST
29    Tag: blume-20000616-anchorenv
30    Description:
31    
32    This patch implements the long anticipated (just kidding :) "anchor
33    environment" mechanism.  In the course of doing this, I also
34    re-implemented CM's internal "SrcPath" module from scratch.  The new
35    one should be more robust in certain boundary cases.  In any case, it
36    is a lot cleaner than its predecessor (IMHO).
37    
38    This time, although there is yet another boot file format change, I
39    kept the unpickler backward-compatible.  As a result, no new bootfiles
40    are necessary and bootstrapping is straightforward.  (You cannot read
41    new bootfiles into an old system, but the other way around is no
42    problem.)
43    
44    Visible changes:
45    
46    ** 0. Implicit path anchors (without the leading $-symbol) are no
47    longer recognized at all. This means that such path names are not
48    illegal either.  For example, the name basis.cm simply refers to a
49    local file called "basis.cm" (i.e, the name is an ordinary path
50    relative to .cm-files directory).  Or, to put it differently, only
51    names that start with $ are anchored paths.
52    
53    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
54    vanished.
55    
56    John (Reppy) had critizised this as soon as I originally proposed and
57    implemented it, but at that time I did not really deeply believe
58    him. :) Now I came full-circle because I need the $<singlearc> syntax
59    in another place where it cannot be seen as an abbreviation for
60    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
61    seems to mean (i.e., it "expands" into the corresponding anchor
62    value).
63    
64    However, when paths are used as members in CM description files, it
65    continues to be true that there must be at least another arc after the
66    anchor.  This is now enforced separately during semantic analysis
67    (i.e., from a lexical/syntactical point of view, the notation is ok.)
68    
69    ** 2. The "cm" class now accepts an option "bind".  The option's value
70    is a sub-option list of precisely two items -- one labeled "anchor"
71    and the other one labeled "value".  As you might expect, "anchor" is
72    used to specify an anchor name to be bound, and "value" specifies what
73    the anchor is being bound to.
74    
75    The value must be a directory name and can be given in either standard
76    syntax (including the possibility that it is itself an anchored path)
77    or native syntax.
78    
79    Examples:
80    
81       foo.cm (bind:(anchor:bar value:$mystuff/bar))
82       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
83    
84    and so on.
85    
86    The meaning of this is that the .cm-file will be processed with an
87    augmented anchor environment where the given anchor(s) is/are bound to
88    the given values(s).
89    
90    The rationale for having this feature is this: Suppose you are trying
91    to use two different (already stable) libraries a.cm and b.cm (that
92    you perhaps didn't write yourself).  Further, suppose each of these
93    two libraries internally uses its own auxiliary library $aux/lib.cm.
94    Normally you would now have a problem because the anchor "lib" can not
95    be bound to more than one value globally.  Therefore, the project that
96    uses both a.cm and b.cm must locally redirect the anchor to some other
97    place:
98    
99       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
100       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
101    
102    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
103    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
104    
105    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
106    you when you do so at the time of CM.stabilize).  Therefore, you can
107    also use an anchored path as the value:
108    
109      a.cm (bind:(anchor:lib value:$a-lib))
110      b.cm (bind:(anchor:lib value:$b-lib))
111    
112    Now you can globally configure (using the usual CM.Anchor.anchor or
113    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
114    itself is always locally bound, setting it globally is no longer
115    meaningful or necessary (but it does not hurt either).  In fact, "lib"
116    can still be used as a global anchor for separate purposes.  As a
117    matter of fact, one can locally define "lib" in terms of a global
118    "lib":
119    
120      a.cm (bind:(anchor:lib value:$lib/a))
121      b.cm (bind:(anchor:lib value:$lib/b))
122    
123    ** 3: The encoding of path names has changed.  This affects the way
124    path names are shown in CM's progress report and also the internal
125    protocol encoding used for parallel make.
126    
127    The encoding now uses one or more ':'-separated segments.  Each
128    segments corresponds to a file that has been specified relative to the
129    file given by its preceding segment.  The first segment is either
130    relative to the CWD, absolute, or anchored.  Each segment itself is
131    basically a Unix pathname; all segments but the first are relative.
132    
133    Example:
134    
135       $foo/bar/baz.cm:a/b/c.sml
136    
137    This path denotes the file bar/a/b/c.sml relative to the directory
138    denoted by anchor "foo".  Notice that the encoding also includes
139    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
140    paths are resolved relative to the .cm-files directory, so baz.cm must
141    be ignored to get the "real" pathname.
142    
143    To make this fact more obvious, CM puts the names of such "virtual
144    arcs" into parentheses when they appear in progress reports. (No
145    parentheses will appear in the internal protocol encoding.)  Thus,
146    what you really see is:
147    
148      $foo/bar/(baz.cm):a/b/c.sml
149    
150    I find this notation to be much more informative than before.
151    
152    Another new feature of the encoding is that special characters
153    including parentheses, colons, (back)slashes, and white space are
154    written as \ddd (where ddd is the decimal encoding of the character).
155    
156    *** The CM manual still needs to be updated.
157    
158    ----------------------------------------------------------------------
159    Name: Allen Leung
160    Date: 2000/06/15 00:38:00
161    Tag: leunga-20000615-x86-peephole
162    
163    x86 Peephole fix by Fermin.  Affects c-- and moby only.
164    
165    ----------------------------------------------------------------------
166    Name: Matthias Blume
167    Date: 2000/06/12 11:40:00
168    Tag: blume-20000612-parmakefix
169    Description:
170    
171    More cleanup after changing the file naming scheme: This time I
172    repaired the parallel make mechanism for CMB.make which I broke earlier.
173    
174    ----------------------------------------------------------------------
175    Name: Allen Leung
176    Date: 2000/06/09 01:25:00
177    Tag: leunga-20000609-various
178    
179    None of these things should affect normal SML/NJ operations
180    
181    1. Peephole improvements provided by Fermin (c--)
182    2. New annotation DEFUSE for adding extra dependence (moby)
183    3. New X86 LOCK instructions (moby)
184    4. New machine description language for reservation tables (scheduling)
185    5. Fixes to various optimization/analysis modules (branch chaining, dominator
186       trees etc.)
187    6. I've changed the CM files so that they can work with versions
188       110.0.6, 110.25 and 110.28
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume
192    Date: 2000/06/09 12:40:00
193    Tag: blume-20000609-log
194    Description:
195    
196    - Removed all(?) remaining RCS Log entries from sources.
197    
198    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
199      anchored paths).
200    
201    ----------------------------------------------------------------------
202    Name: Matthias Blume
203    Date: 2000/06/07 17:00:00 JST
204    Tag: blume-20000607-no-implicit-anchors
205    Description:
206    
207    1. This update changes the default setting for
208    CM.Control.implicit_anchors from true to false.  This means that
209    implicit anchors are no longer permitted by default.  I also tried to
210    make sure that nothing else still relies on implicit anchors.
211    (This is the next step on the schedule towards a CM that does not even
212    have the notion of implicit anchors anymore.)
213    
214    2. More CM manual updates.
215    
216    3. I managed to track down and fix the pickling bug I mentioned last
217    time.  Because of the previously existing workaround, this entails no
218    immediate practical changes.
219    
220    ----------------------------------------------------------------------
221    Name: Matthias Blume
222    Date: 2000/06/06 11:15:00 JST
223    Tag: blume-20000606-lazierpickle
224    Description:
225    
226    !!!! NEW BOOT FILES !!!!
227    
228    * The main purpose of this update is to make library pickles lazier in
229    order to reduce the initial space penalty for autoloading a library.
230    As a result, it is now possible to have $smlnj/compiler.cm
231    pre-registered.  This should take care of the many complaints or
232    inquiries about missing structure Compiler.  This required changes to
233    CM's internal data structures and small tweaks to some algorithms.
234    
235    As a neat additional effect, it is no longer necessary (for the sake
236    of lean heap image files) to distinguish between a "minimal" CM and a
237    "full" CM.  Now, there is only one CM (i.e., the "full" version:
238    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
239    interactive top level. ($smlnj/cm/minimal.cm is gone.)
240    
241    To make the life of compiler-hackers easier, "makeml" now also
242    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
243    words, after you bootstrap a new sml for the first time, you will not
244    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
245    around you will still have to do it, though.)
246    
247    * A second change consists of major updates to the CM manual.  There
248    are now several appendices with summary information and also a full
249    specification of the CM description file syntax.
250    
251    * In directory src/system I added the script "allcross".  This script
252    invokes sml and cross-compiles the compiler for all supported
253    architectures.  (Useful when providing a new set of boot files.)
254    
255    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
256    added a small tweak to pickle-util.sml to work around this problem,
257    but it is not a proper fix yet.  I will investigate further.  (The
258    effect of the bug was an inflation of library pickle size.)
259    
260    * Version number increased to 110.28.1 (to avoid compatibility problems).
261    
262    ----------------------------------------------------------------------
263    Name: Allen Leung
264    Date: 2000/05/25 17:28 EDT
265    Tag: leunga-20000525-ra
266    Description:
267    
268      Fixed a bug in freezing phase of the register allocator.
269    
270    ----------------------------------------------------------------------
271    Name: Allen Leung
272    Date: 2000/05/15 22:53 EDT
273    Tag: leunga-20000515-alpha-x86-ra
274    Description:
275    
276      1. Alpha
277    
278          Slight cleanup.  Removed the instruction SGNXL
279    
280      2. X86
281    
282          Added the following instructions to the instruction set:
283    
284            ROLx, RORx,
285            BTx, BTSx, BTLx, BTRx,
286            XCHGx, and variants with the LOCK prefix
287    
288      3. Register Allocation
289    
290          The module ra-rewrite-with-renaming has been improved.
291    
292      These have no effect on SML/NJ.
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume
296    Date: 2000/05/15 16:20:00 JST
297    Tag: blume-20000515-lightrebuild
298    Description:
299    
300    1. I added an alternative to "-rebuild" to "makeml".  The difference is
301       that prior to calling CMB.make' the CM-variable "LIGHT" will be
302       defined.  In effect, the command will not build any cross-compiler
303       backends and therefore finish more quickly.
304    
305       The "fixpt" script also takes a "-light" switch to be able to use
306       this new facility while compiling for a fixpoint.
307    
308    2. I replaced all mentions of anchored paths in group owner specifications
309       with simple relative paths (usually starting with "..").
310       The rationale is that a library's internal workings should not be
311       compromised by the lack of some anchor.  (An anchor is necessary
312       for someone who wants to refer to the library by an anchored path,
313       but it should not be necessary to build the same library in the first
314       place.)
315    
316    3. I changed the way CM's tool mechanism determines the shell command
317       string used for things like ml-yacc etc. so that it does not break
318       when CM.Control.implicit_anchors is turned off.
319    
320    ----------------------------------------------------------------------
321    Name: Matthias Blume
322    Date: 2000/05/12 18:20:00 JST
323    Tag: blume-20000512-ml-build
324    Description:
325    
326    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
327    getting installed properly (by config/install.sh).
328    
329    ----------------------------------------------------------------------
330    Name: Matthias Blume
331    Date: 2000/05/12 17:30:00 JST
332    Tag: blume-20000512-anchors
333    Description:
334    
335    !!! NEW BOOT FILES !!!
336    
337    This change is in preparation of fading out support for "implicitly
338    anchored path names".  I went through all sources and used the
339    explicit (and relatively new) $-notation.  See system/README and the
340    CM manual for more info on this.
341    
342    I also modified the anchoring scheme for some things such as "smlnj",
343    "MLRISC", "cm", etc. to take advantage of the fact that explicit
344    anchors are more expressive: anchor name and first arc do not have to
345    coincide.  This entails the following user-visible change:
346    
347    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
348    particular, when you fire up sml with a command-line argument, say,
349    e.g.:
350    
351       sml '$smlnj/cmb.cm'
352    
353    At the ML toplevel prompt:
354    
355       CM.autoload "$smlnj/cmb.cm";
356    
357    There is also a new controller in CM.Control that can be used to turn
358    off all remaining support for implicit anchors by saying:
359    
360        CM.autoload "$smlnj/
361        #set CM.Control.implicit_anchors false;
362    
363    This causes CM to reject implicitly anchored paths.  This is (for the
364    time being) less permissive than the "final" version where there will
365    be no more such implicit anchors and relative paths will be just that:
366    relative.
367    
368    The next step (version after next version?) will be to make the
369    default for CM.Control.implicit_anchors false.  After the dust has
370    settled, I can then produce the "final" version of this...
371    
372    Note: Since bootstrapping is a bit tricky, I provided new boot files.
373    
374    ----------------------------------------------------------------------
375    Name: Matthias Blume
376    Date: 2000/05/11 16:30:00 JST
377    Tag: blume-20000511-sources
378    Description:
379    
380    The main change is that I added function CM.sources as a generalized
381    version of the earlier CM.makedepend.  This entails the following
382    additional changes:
383    
384      - CM.makedepend has been dropped.
385    
386      - CM manual has been updated.
387    
388      - TOOLS signature and API have been changed.
389    
390    ----------------------------------------------------------------------
391    Name: Allen Leung
392    Date: 2000/05/10 21:17 EDT
393    Tag: leunga-20000510-moby-c--ssa
394    Description:
395    
396      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
397    None of these affect SML/NJ.
398    
399    1. Register Allocation
400    
401        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
402           This module tries to remove local (i.e. basic block level) redundancies
403           during spilling.
404    
405        b. A new framework for performing region based register allocation.
406           Not yet entirely functional.
407    
408    2. X86
409    
410       a. DefUse for POP was missing the stack pointer [found by Lal]
411       b. Reload for CALL was incorrect in X86Spill [found by John]
412       c. Various fixes in X86Spill so that it can be used correctly for
413          the new spilling module.
414    
415    3. SSA/IR
416    
417       a. New module ir/dj-dataflow.sml implements elimination based
418          data flow analysis.
419    
420    4. MLRiscGen
421    
422       a. Fix for gc type annotation
423    
424    5. MDGen
425    
426       Various fixes for machine description -> ml code translation.  For ssa
427       only.
428    
429    ----------------------------------------------------------------------
430    Name: Allen Leung
431    Date: 2000/05/08 22:17 EDT
432    Tag: leunga-20000508-labexp
433    Description:
434    
435      Fermin has found a few assembly problems with constant expressions
436      generated in LabelExp.  Mostly, the problems involve extra parentheses,
437      which choke on dumb assemblers.  This is his fix.
438    
439    ----------------------------------------------------------------------
440    Name: Dave MacQueen
441    Date: 2000/04/09 14:00 EDT
442    Tag: dbm-20000502-Version_110_28
443    Description:
444    
445    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
446    
447    2. Updated config/version to 110.28
448    
449    3. Updated config/srcarchiveurl
450    
451    3. New boot files!
452       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
453    
454    ----------------------------------------------------------------------
455    Name: Matthias Blume
456    Date: 2000/05/01 19:05:00 JST
457    Tag: blume-20000501-noweb
458    Description:
459    
460    A new noweb tool has been added.  The existing system is entirely
461    unaffected by this, but some CM users have asked for renewed noweb
462    support.  Everything is documented in the CM manual.
463    
464    New (plugin) libraries:
465    
466       noweb-tool.cm
467       nw-ext.cm
468    
469    ----------------------------------------------------------------------
470    Name: Dave MacQueen
471    Date: 2000/04/30 12:40PM EDT
472    Tag: dbm-20000430-bug_fixes
473    Description:
474    
475    1. Fix for bug 1498
476       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
477       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
478         added toRealArray function
479       smlnj/src/compiler/MiscUtil/print/ppobj.sml
480         added check for tag Obj.RealArray to array printing case in ppObj
481    
482    2. Fix for bug 1510
483       smlnj/src/compiler/Semant/types/typesutil.sml
484         fixed definition of dummyargs (used by equalTycon) so that
485         dummy args are distinct types
486    
487    ----------------------------------------------------------------------
488    Name: Matthias Blume
489    Date: 2000/04/30 01:00:00 JST
490    Tag: blume-20000430-versions
491    Description:
492    
493    1. CM version numbering added.  This is an implementation of Lal's
494       proposal for adding version numbers and version checking to .cm
495       files.  Lal said that his proposal was just that -- a proposal.
496       For the time being I went ahead and implemented it so that people
497       can comment on it.  Everything is completely backward-compatible
498       (except for the stable library format, i.e., new bootfiles!).
499    
500       As usual, see the CM manual for details.
501    
502    2. An alternative syntax for anchored paths has been implemented.
503       Dave has recently voiced the same concerns that I had when I did
504       this, so there should be some support.  My take is that eventually
505       I will let support for the current syntax (where anchors are
506       "implicit") fade out in favor of the new, explicit syntax.
507       In order to be backward-compatible, both old and new syntax are
508       currently supported.
509    
510       Again, see the CM manual for details.
511    
512    3. Parallel make is trying to be slightly smarter:  When the master
513       process finds a "bottleneck", i.e., when there is only one
514       compilation unit that can be compiled and everybody else is
515       waiting on it, then it will simply compile it directly instead
516       of clumsily telling one of the slaves to do it.
517    
518    4. Support for "unsharing" added.  This is necessary in order to be
519       able to have two different versions of the same library running
520       at the same time (e.g., for trying out a new MLRISC while still
521       having the old MLRISC linked into the current compiler, etc.)
522       See the CM manual.
523    
524    5. Simple "makedepend" functionality added for generating Makefile
525       dependency information.  (This is rather crude at the moment.
526       Expect some changes here in the future.)
527    
528    6. ".fun" added as a recognized suffix for ML files. Also documented
529       explicitly in the manual that the fallback behavior (unknown suffix
530       -> ML file) is not an official feature!
531    
532    7. Small changes to the pickler for stable libraries.
533    
534    8. Several internal changes to CM (for cleanup/improvement).
535    
536    
537    !!!! NEW BINFILES !!!!
538    
539    ----------------------------------------------------------------------
540    Name: Matthias Blume
541    Date: 2000/04/28 17:30:00 JST
542    Tag: blume-20000428-pathconfig
543    Description:
544    
545    1. I changed config/install.sh to remove duplicate entries from the
546       lib/pathconfig file at the end.  Moreover, the final version of
547       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
548       in src/system/installml.
549    
550    2. The config/install.sh script now consistently uses relative
551       pathnames in lib/pathconfig whenever the anchor is in the lib
552       directory.  (So far this was true for the libraries that come
553       pre-compiled and bundled as part of the bootfiles but not for
554       libraries that are compiled by the script itself.)
555    
556    ----------------------------------------------------------------------
557    Name: Matthias Blume
558    Date: 2000/04/26 13:10:00 JST
559    Tag: blume-20000426-fun_suffix
560    Description:
561    
562    Added ".fun" as a recognized file name suffix (for ML code).
563    
564    ----------------------------------------------------------------------
565    Name: Allen Leung
566    Date: 2000/04/25 17:00:00 EST
567    Tag: leunga-20000425-alpha-ra
568    Description:
569    
570    1. Alpha
571    
572        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
573    in C--.
574    
575    2. RA
576    
577       Added a flag "ra-dump-size" to print out the size of the flowgraph
578       and the interference graph.
579    
580    ----------------------------------------------------------------------
581    Name: Dave MacQueen
582    Date: 2000/04/25/
583    Tag: dbm-20000425-mlyacc_doc_examples
584    Description:
585      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
586      Updated all three examples in src/ml-yacc/examples to run
587      under 110.* using CM.make.
588    
589    ----------------------------------------------------------------------
590    Name: Allen Leung
591    Date: 2000/04/20 23:04:00 EST
592    Tag: leunga-20000420-ssa-c---stuff
593    Description:
594    
595      This update synchronizes my repository with Yale's.  Most of these
596    changes, however, do not affect SML/NJ at all (the RA is an exception).
597    
598    1. Register Allocator
599    
600       a. An improvement in the interference graph construction:
601          Given a copy
602    
603                s <- t
604    
605          no interference edge between s and t is added for this definition of s.
606    
607       b. I've added two new spill heuristic modules that Fermin and I developed
608          (in the new library RA.cm). These are unused in SML/NJ but maybe
609          useful for others (Moby?)
610    
611    2. X86
612    
613       a. Various fixes in the backend provided by Fermin [C--] and Lal.
614    
615    3. Alpha
616    
617       a. Added the BSR instruction and code generation that goes with it [C--]
618       b. Other fixes too numerous to recount provided by Fermin [C--]
619    
620    4. Regmaps
621    
622       a. The regmaps are not initialized with the identity physical bindings
623          at creation time.  This is unneeded.
624    
625    5. MLRISC Optimizations
626    
627       a. The DJ-Graph module can now compute the iterated dominance frontiers
628          intersects with liveness incrementally in linear time! Woohoo!
629          This is now used in my new SSA construction algorithm.
630    
631       b. THe branch reorganization module is now smarter about linear chains of
632          basic blocks.
633    
634    
635    ----------------------------------------------------------------------
636    Name: Matthias Blume
637    Date: 2000/04/12 13:52:00 JST
638    Tag: blume_main_v110p27_1
639    Description:
640    
641    Changed install.sh script to handle archive files without version number
642    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
643    name of the boot file archive.
644    
645    ----------------------------------------------------------------------
646    Name: Dave MacQueen
647    Date: 2000/04/09 14:00 EDT
648    Tag: dbm-20000410-Version_110_27
649    Description:
650    
651    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
652    
653    2. Updated src/config/version to 110.27
654    
655    3. New boot files!
656    
657    ----------------------------------------------------------------------
658    Name: Allen Leung
659    Date: 2000/04/09 19:09:00 EST
660    Tag: leunga-20000409-misc
661    Description:
662    
663    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
664    
665    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
666    
667    ----------------------------------------------------------------------
668    Name: Stefan
669    Date: 2000/04/07 10:00:00 EDT
670    Tag: monnier-20000406-branch-handling
671    Description:
672    
673    Improved handling of branches (mostly those generated from
674    polymorphic equality), removed switchoff and changed the
675    default optimization settings (more cpsopt and less flintopt).
676    
677    ----------------------------------------------------------------------
678    Name: Allen Leung
679    Date: 2000/04/06 01:30:00 EST
680    Tag: leunga-20000406-peephole-x86-SSA-2
681    Description:
682    
683       Forgot a few files.
684    
685    ----------------------------------------------------------------------
686    Name: Allen Leung
687    Date: 2000/04/06 00:36:00 EST
688    Tag: leunga-20000406-peephole-x86-SSA
689    Description:
690    
691    1.  New Peephole code
692    
693    2.  Minor improvement to X86 instruction selection
694    
695    3.  Various fixes to SSA and machine description -> code translator
696    
697    ----------------------------------------------------------------------
698    Name: Matthias Blume
699    Date: 2000/04/05 12:30:00 JST
700    Tag: blume_main_v110p26p2_3
701    Description:
702    
703    This update just merges three minor cosmetic updates to CM's sources
704    to get ready for the 110.27 code freeze on Friday.  No functionality
705    has changed.
706    
707    ----------------------------------------------------------------------
708    Name: Allen Leung
709    Date: 2000/04/04 19:39:00 EST
710    Tag: leunga-20000404-x86-asm
711    Description:
712    
713    1.  Fixed a problem in X86 assembly.
714    
715        Things like
716    
717           jmp %eax
718           jmp (%eax)
719    
720        should be output as
721    
722           jmp *%eax
723           jmp *(%eax)
724    
725    2.  Assembly output
726    
727          Added a new flag
728    
729              "asm-indent-copies" (default to false)
730    
731          When this flag is on, parallel copies will be indented an extra level.
732    
733    ----------------------------------------------------------------------
734    Name: Allen Leung
735    Date: 2000/04/04 03:18:00 EST
736    Tag: leunga-20000404-C--Moby
737    Description:
738    
739        All of these fixes are related to C--, Moby, and my own optimization
740        stuff; so they shouldn't affect SML/NJ.
741    
742    1.  X86
743    
744        Various fixes related floating point, and extensions.
745    
746    2.  Alpha
747    
748        Some extra patterns related to loads with signed/zero extension
749        provided by Fermin.
750    
751    3.  Assembly
752    
753        When generating assembly, resolve the value of client defined constants,
754        instead of generating symbolic values.  This is controlled by the
755        new flag "asm-resolve-constants", which is default to true.
756    
757    4.  Machine Descriptions
758    
759        a. The precedence parser was slightly broken when parsing infixr symbols.
760        b. The type generalizing code had the bound variables reversed, resulting
761           in a problem during arity raising.
762        c. Various fixes in machine descriptions.
763    
764    ----------------------------------------------------------------------
765    Name: Matthias Blume
766    Date: 2000/04/03 16:05:00 JST
767    Tag: blume_main_v110p26p2_2
768    Description:
769    
770    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
771    now done via the ordinary static environment that is context to each
772    compilation unit.
773    
774    To this end, I arranged that instead of "structure Core" as "structure
775    _Core" is bound in the pervasive environment.  Core access is done via
776    _Core (which can never be accidentally rebound because _Core is not a
777    legal surface-syntax symbol).
778    
779    The current solution is much cleaner because the core environment is
780    now simply part of the pervasive environment which is part of every
781    compilation unit's context anyway.  In particular, this eliminates all
782    special-case handling that was necessary until now in order to deal
783    with dynamic and symbolic parts of the core environment.
784    
785    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
786    compilation manager's bootstrap compiler (actually: in the "init group"
787    handling).  See the comments in src/system/smlnj/init/init.cmi for
788    more details.
789    
790    I also tried to track down all mentions of "Core" (as string argument
791    to Symbol.strSymbol) in the compiler and replaced them with a
792    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
793    name appears in one place only.
794    
795    Binfile and bootfile format have not changed, but the switchover from
796    the old "init.cmi" to the new one is a bit tricky, so I supplied new
797    bootfiles anyway.
798    
799    ----------------------------------------------------------------------
800    Name: Allen Leung
801    Date: 2000/04/02 21:17:00 EST
802    Tag: leunga-20000402-mltree
803    Description:
804    
805       1. Renamed the constructor CALL in MLTREE by popular demand.
806       2. Added a bunch of files from my repository.  These are currently
807          used by other non-SMLNJ backends.
808    
809    ----------------------------------------------------------------------
810    Name: Allen Leung
811    Date: 2000/03/31 21:15:00 EST
812    Tag: leunga-20000331-aliasing
813    Description:
814    
815    This update contains a rewritten (and hopefully more correct) module
816    for extracting aliasing information from CPS.
817    
818       To turn on this feature:
819    
820            Compiler.Control.CG.memDisambiguate := true
821    
822       To pretty print the region information with assembly
823    
824           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
825    
826       To control how many levels of aliasing information are printed, use:
827    
828           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
829    
830       The default of n is 3.
831    
832    ----------------------------------------------------------------------
833    Name: David MacQueen
834    Date: 2000/03/31 11:15:00 EST
835    Tag: dbm-20000331-runtime_fix
836    Description:
837    
838    This update contains:
839    
840    1. runtime/c-lib/c-libraries.c
841       includes added in revision 1.2 caused compilation errors on hppa-hpux
842    
843    2. fix for bug 1556
844       system/Basis/Implementation/NJ/internal-signals.sml
845    
846    ----------------------------------------------------------------------
847    Name: Matthias Blume
848    Date: 2000/03/31 18:00:00 JST
849    Tag: blume_main_v110p26p2_1
850    Description:
851    
852    This update contains:
853    
854    1. A small change to CM's handling of stable libraries:
855       CM now maintains one "global" modmap that is used for all stable
856       libraries.  The use of such a global modmap maximizes sharing and
857       minimizes the need for re-traversing parts of environments during
858       modmap construction.  (However, this has minor impact since modmap
859       construction seems to account for just one percent or less of total
860       compile time.)
861    
862    2. I added a "genmap" phase to the statistics.  This is where I got the
863       "one percent" number (see above).
864    
865    3. CM's new tool parameter mechanism just became _even_ better. :)
866       - The parser understands named parameters and recursive options.
867       - The "make" and "shell" tools use these new features.
868         (This makes it a lot easier to cascade these tools.)
869       - There is a small syntax change: named parameters use a
870    
871           <name> : ( <option> ... )            or
872           <name> : <string>
873    
874         syntax.  Previously, named parameters were implemented in an
875         ad-hoc fashion by each tool individually (by parsing strings)
876         and had the form
877    
878           <name>=<string>
879    
880       See the CM manual for a full description of these issues.
881    
882    ----------------------------------------------------------------------
883    Name: Matthias Blume
884    Date: 2000/03/30 18:00:00 JST
885    Tag: blume_main_v110p26p2_0
886    Description:
887    
888    !!!!! WARNING !!!!!!
889    !!  New binfiles  !!
890    !!!!!!!!!!!!!!!!!!!!
891    
892    This update contains:
893    
894    1. Moderate changes to CM:
895    
896       - Changes to CM's tools mechanism.  In particular, it is now possible
897       to have tools that accept additional "command line" parameters
898       (specified in the .cm file at each instance where the tool's class is
899       used).
900    
901       This was done to accommodate the new "make" and "shell" tools which
902       facilitate fairly seamless hookup to portions of code managed using
903       Makefiles or Shell scripts.
904    
905       There are no classes "shared" or "private" anymore.  Instead, the
906       sharing annotation is now a parameter to the "sml" class.
907    
908       There is a bit of generic machinery for implementing one's own
909       tools that accept command-line parameters.  However, I am not yet fully
910       satisfied with that part, so expect changes here in the future.
911    
912       All existing tools are described in the CM manual.
913    
914       - Slightly better error handling.  (CM now suppresses many followup
915       error messages that tended to be more annoying than helpful.)
916    
917    2. Major changes to the compiler's static environment data structures.
918    
919       - no CMStaticEnv anymore.
920            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
921              but it is called Environment), no conversions between different
922              kinds of static environments
923    
924       - There is still a notion of a "modmap", but such modmaps are generated
925         on demand at the time when they are needed.  This sounds slow, but I
926         sped up the code that generates modmaps enough for this not to lead to
927         a slowdown of the compiler (at least I didn't detect any).
928    
929       - To facilitate rapid modmap generation, static environments now
930         contain an (optional) "modtree" structure.  Modtree annotations are
931         constructed by the unpickler during unpickling.  (This means that
932         the elaborator does not have to worry about modtrees at all.)
933         Modtrees have the advantage that they are compositional in the same
934         way as the environment data structure itself is compositional.
935         As a result, modtrees never hang on to parts of an environment that
936         has already been rendered "stale" by filtering or rebinding.
937    
938       - I went through many, many trials and errors before arriving at the
939         current solution.  (The initial idea of "linkpaths" did not work.)
940         But the result of all this is that I have touched a lot of files that
941         depend on the "modules" and "types" data structures (most of the
942         elaborator). There were a lot of changes during my "linkpath" trials
943         that could have been reverted to their original state but weren't.
944         Please, don't be too harsh on me for messing with this code a bit more
945         than what was strictly necessary...  (I _did_ resist the tempation
946         of doing any "global reformatting" to avoid an untimely death at
947         Dave's hands. :)
948    
949       - One positive aspect of the previous point:  At least I made sure that
950         all files that I touched now compile without warnings (other than
951         "polyEqual").
952    
953       - compiler now tends to run "leaner" (i.e., ties up less memory in
954         redundant modmaps)
955    
956    ----------------------------------------------------------------------
957    Name: Allen Leung
958    Date: 2000/03/29 18:00:00
959    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
960    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
961    Description:
962    
963       This update contains *MAJOR* changes to the way code is generated from CPS
964    in the module mlriscGen, and in various backend modules.
965    
966    CHANGES
967    =======
968    
969    1. MLRiscGen: forward propagation fix.
970    
971       There was a bug in forward propagation introduced at about the same time
972       as the MLRISC x86 backend, which prohibits coalescing to be
973       performed effectively in loops.
974    
975       Effect: speed up of loops in RISC architectures.
976               By itself, this actually slowed down certain benchmarks on the x86.
977    
978    2. MLRiscGen:  forward propagating addresses from consing.
979    
980       I've changed the way consing code is generated.  Basically I separated
981       out the initialization part:
982    
983            store tag,   offset(allocptr)
984            store elem1, offset+4(allocptr)
985            store elem2, offset+8(allocptr)
986            ...
987            store elemn, offset+4n(allocptr)
988    
989       and the address computation part:
990    
991            celladdr <- offset+4+alloctpr
992    
993       and move the address computation part
994    
995       Effect:  register pressure is generally lower as a result.  This
996                makes compilation of certain expressions much faster, such as
997                long lists with non-trivial elements.
998    
999                 [(0,0), (0,0), .... (0,0)]
1000    
1001    3. MLRiscGen: base pointer elimination.
1002    
1003        As part of the linkage mechanism, we generate the sequence:
1004    
1005         L:  ...  <- start of the code fragment
1006    
1007         L1:
1008             base pointer <- linkreg - L1 + L
1009    
1010         The base pointer was then used for computing relocatable addresses
1011       in the code fragment.  Frequently (such as in lots of continuations)
1012       this is not needed.  We now eliminate this sequence whenever possible.
1013    
1014         For compile time efficiency, I'm using a very stupid local heuristic.
1015       But in general, this should be done as a control flow analysis.
1016    
1017       Effect:  Smaller code size.  Speed up of most programs.
1018    
1019    4. Hppa back end
1020    
1021         Long jumps in span dependence resolution used to depend on the existence
1022      of the base pointer.
1023    
1024         A jump to a long label L was expanded into the following sequence:
1025    
1026          LDIL %hi(L-8192), %r29
1027          LDO  %lo(L-8192)(%r29), %r29
1028          ADD  %r29, baseptr, %r29
1029          BV,n %r0(%r29)
1030    
1031         In the presence of change (3) above, this will not work.  I've changed
1032       it so that the following sequence of instructions are generated, which
1033       doesn't mention the base pointer at all:
1034    
1035             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
1036        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
1037             BV,n  %r0(%r29)          /* Jump */
1038    
1039    5. Alpha back end
1040    
1041          New alpha instructions LDB/LDW have been added, as per Fermin's
1042       suggestions.   This is unrelated to all other changes.
1043    
1044    6. X86 back end
1045    
1046         I've changed andl to testl in the floating point test sequence
1047         whenever appropriate.  The Intel optimization guide states that
1048         testl is preferable to andl.
1049    
1050    7. RA (x86 only)
1051    
1052         I've improved the spill propagation algorithm, using an approximation
1053       of maximal weighted independent sets.   This seems to be necessary to
1054       alleviate the negative effect in light of the slow down in (1).
1055    
1056         I'll write down the algorithm one of these days.
1057    
1058    8. MLRiscGen: frequencies
1059    
1060         I've added an annotation that states that all call gc blocks have zero
1061       execution frequencies.  This improves register allocation on the x86.
1062    
1063    BENCHMARKS
1064    ==========
1065    
1066       I've only perform the comparison on 110.25.
1067    
1068       The platforms are:
1069    
1070        HPPA  A four processor HP machine (E9000) with 5G of memory.
1071        X86   A 300Hhz Pentium II with 128M of memory, and
1072        SPARC An Ultra sparc 2 with 512M of memory.
1073    
1074       I used the following parameters for the SML benchmarks:
1075    
1076                 @SMLalloc
1077         HPPA    256k
1078         SPARC   512k
1079         X86     256k
1080    
1081    COMPILATION TIME
1082    ----------------
1083       Here are the numbers comparing the compilation times of the compilers.
1084       I've only compared 110.25 compiling the new sources versus
1085       a fixpoint version of the new compiler compiling the same.
1086    
1087                     110.25                                  New
1088               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
1089         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
1090         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
1091         X86    999s    315s       94006+130691   987s    296s    108877+141957
1092    
1093                   110.25         New
1094                Code Size      Code Size
1095         HPPA   8596736         8561421
1096         SPARC  8974299         8785143
1097         X86    9029180         8716783
1098    
1099       So in summary, things are at least as good as before.   Dramatic
1100       reduction in compilation is obtained on the Sparc; I can't explain it,
1101       but it is reproducible.  Perhaps someone should try to reproduce this
1102       on their own machines.
1103    
1104    SML BENCHMARKS
1105    --------------
1106    
1107        On the average, all benchmarks perform at least as well as before.
1108    
1109          HPPA         Compilation Time     Spill+Reload      Run Time
1110                     110.25  New            110.25    New   110.25  New
1111    
1112          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
1113              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
1114       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
1115                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
1116        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
1117             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
1118               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
1119              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
1120         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
1121             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
1122            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
1123      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
1124                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
1125             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
1126                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
1127               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
1128      --------------------------------------------------------------------------
1129       Average                     6.12%                                   4.09%
1130    
1131          SPARC        Compilation Time     Spill+Reload      Run Time
1132                     110.25  New            110.25    New   110.25  New
1133    
1134          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
1135              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
1136       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
1137                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
1138        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
1139             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
1140               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
1141              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
1142         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
1143             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1144            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
1145      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
1146                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
1147             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
1148                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
1149               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
1150      ----------------------------------------------------------------------------
1151       Average                     6.94%                                    2.64%
1152    
1153          X86          Compilation Time     Spill+Reload      Run Time
1154                     110.25  New            110.25    New   110.25  New
1155    
1156          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
1157              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
1158       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
1159                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1160        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1161             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1162               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1163              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1164         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1165             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1166            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1167      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1168                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1169             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1170                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1171               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1172      ----------------------------------------------------------------------------
1173       Average                     1.22%                                     3.36%
1174    
1175    ----------------------------------------------------------------------
1176  Name: Allen Leung  Name: Allen Leung
1177  Date: 2000/03/23 16:25:00  Date: 2000/03/23 16:25:00
1178  Tag: leunga-20000323-fix_x86_alpha  Tag: leunga-20000323-fix_x86_alpha
# Line 66  Line 1227 
1227    
1228  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1229  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
1230  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
1231  explanation.  explanation.
1232    
1233  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
1234  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1235  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
1236  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 113  Line 1274 
1274  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
1275  <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,
1276  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
1277  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
1278  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
1279  and, thus, saves a lot of time.  and, thus, saves a lot of time.
1280    
# Line 163  Line 1324 
1324    it from that remote directory.    it from that remote directory.
1325    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
1326    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
1327    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
1328    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
1329    
1330    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.583  
changed lines
  Added in v.670

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