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 575, Fri Mar 10 02:55:58 2000 UTC revision 660, Thu Jun 15 04:40:56 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: 2000/06/15 00:38:00
17    Tag: leunga-20000615-x86-peephole
18    
19    x86 Peephole fix by Fermin.  Affects c-- and moby only.
20    
21    ----------------------------------------------------------------------
22    Name: Matthias Blume
23    Date: 2000/06/12 11:40:00
24    Tag: blume-20000612-parmakefix
25    Description:
26    
27    More cleanup after changing the file naming scheme: This time I
28    repaired the parallel make mechanism for CMB.make which I broke earlier.
29    
30    ----------------------------------------------------------------------
31    Name: Allen Leung
32    Date: 2000/06/09 01:25:00
33    Tag: leunga-20000609-various
34    
35    None of these things should affect normal SML/NJ operations
36    
37    1. Peephole improvements provided by Fermin (c--)
38    2. New annotation DEFUSE for adding extra dependence (moby)
39    3. New X86 LOCK instructions (moby)
40    4. New machine description language for reservation tables (scheduling)
41    5. Fixes to various optimization/analysis modules (branch chaining, dominator
42       trees etc.)
43    6. I've changed the CM files so that they can work with versions
44       110.0.6, 110.25 and 110.28
45    
46    ----------------------------------------------------------------------
47    Name: Matthias Blume
48    Date: 2000/06/09 12:40:00
49    Tag: blume-20000609-log
50    Description:
51    
52    - Removed all(?) remaining RCS Log entries from sources.
53    
54    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
55      anchored paths).
56    
57    ----------------------------------------------------------------------
58    Name: Matthias Blume
59    Date: 2000/06/07 17:00:00 JST
60    Tag: blume-20000607-no-implicit-anchors
61    Description:
62    
63    1. This update changes the default setting for
64    CM.Control.implicit_anchors from true to false.  This means that
65    implicit anchors are no longer permitted by default.  I also tried to
66    make sure that nothing else still relies on implicit anchors.
67    (This is the next step on the schedule towards a CM that does not even
68    have the notion of implicit anchors anymore.)
69    
70    2. More CM manual updates.
71    
72    3. I managed to track down and fix the pickling bug I mentioned last
73    time.  Because of the previously existing workaround, this entails no
74    immediate practical changes.
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume
78    Date: 2000/06/06 11:15:00 JST
79    Tag: blume-20000606-lazierpickle
80    Description:
81    
82    !!!! NEW BOOT FILES !!!!
83    
84    * The main purpose of this update is to make library pickles lazier in
85    order to reduce the initial space penalty for autoloading a library.
86    As a result, it is now possible to have $smlnj/compiler.cm
87    pre-registered.  This should take care of the many complaints or
88    inquiries about missing structure Compiler.  This required changes to
89    CM's internal data structures and small tweaks to some algorithms.
90    
91    As a neat additional effect, it is no longer necessary (for the sake
92    of lean heap image files) to distinguish between a "minimal" CM and a
93    "full" CM.  Now, there is only one CM (i.e., the "full" version:
94    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
95    interactive top level. ($smlnj/cm/minimal.cm is gone.)
96    
97    To make the life of compiler-hackers easier, "makeml" now also
98    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
99    words, after you bootstrap a new sml for the first time, you will not
100    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
101    around you will still have to do it, though.)
102    
103    * A second change consists of major updates to the CM manual.  There
104    are now several appendices with summary information and also a full
105    specification of the CM description file syntax.
106    
107    * In directory src/system I added the script "allcross".  This script
108    invokes sml and cross-compiles the compiler for all supported
109    architectures.  (Useful when providing a new set of boot files.)
110    
111    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
112    added a small tweak to pickle-util.sml to work around this problem,
113    but it is not a proper fix yet.  I will investigate further.  (The
114    effect of the bug was an inflation of library pickle size.)
115    
116    * Version number increased to 110.28.1 (to avoid compatibility problems).
117    
118    ----------------------------------------------------------------------
119    Name: Allen Leung
120    Date: 2000/05/25 17:28 EDT
121    Tag: leunga-20000525-ra
122    Description:
123    
124      Fixed a bug in freezing phase of the register allocator.
125    
126    ----------------------------------------------------------------------
127    Name: Allen Leung
128    Date: 2000/05/15 22:53 EDT
129    Tag: leunga-20000515-alpha-x86-ra
130    Description:
131    
132      1. Alpha
133    
134          Slight cleanup.  Removed the instruction SGNXL
135    
136      2. X86
137    
138          Added the following instructions to the instruction set:
139    
140            ROLx, RORx,
141            BTx, BTSx, BTLx, BTRx,
142            XCHGx, and variants with the LOCK prefix
143    
144      3. Register Allocation
145    
146          The module ra-rewrite-with-renaming has been improved.
147    
148      These have no effect on SML/NJ.
149    
150    ----------------------------------------------------------------------
151    Name: Matthias Blume
152    Date: 2000/05/15 16:20:00 JST
153    Tag: blume-20000515-lightrebuild
154    Description:
155    
156    1. I added an alternative to "-rebuild" to "makeml".  The difference is
157       that prior to calling CMB.make' the CM-variable "LIGHT" will be
158       defined.  In effect, the command will not build any cross-compiler
159       backends and therefore finish more quickly.
160    
161       The "fixpt" script also takes a "-light" switch to be able to use
162       this new facility while compiling for a fixpoint.
163    
164    2. I replaced all mentions of anchored paths in group owner specifications
165       with simple relative paths (usually starting with "..").
166       The rationale is that a library's internal workings should not be
167       compromised by the lack of some anchor.  (An anchor is necessary
168       for someone who wants to refer to the library by an anchored path,
169       but it should not be necessary to build the same library in the first
170       place.)
171    
172    3. I changed the way CM's tool mechanism determines the shell command
173       string used for things like ml-yacc etc. so that it does not break
174       when CM.Control.implicit_anchors is turned off.
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume
178    Date: 2000/05/12 18:20:00 JST
179    Tag: blume-20000512-ml-build
180    Description:
181    
182    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
183    getting installed properly (by config/install.sh).
184    
185    ----------------------------------------------------------------------
186    Name: Matthias Blume
187    Date: 2000/05/12 17:30:00 JST
188    Tag: blume-20000512-anchors
189    Description:
190    
191    !!! NEW BOOT FILES !!!
192    
193    This change is in preparation of fading out support for "implicitly
194    anchored path names".  I went through all sources and used the
195    explicit (and relatively new) $-notation.  See system/README and the
196    CM manual for more info on this.
197    
198    I also modified the anchoring scheme for some things such as "smlnj",
199    "MLRISC", "cm", etc. to take advantage of the fact that explicit
200    anchors are more expressive: anchor name and first arc do not have to
201    coincide.  This entails the following user-visible change:
202    
203    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
204    particular, when you fire up sml with a command-line argument, say,
205    e.g.:
206    
207       sml '$smlnj/cmb.cm'
208    
209    At the ML toplevel prompt:
210    
211       CM.autoload "$smlnj/cmb.cm";
212    
213    There is also a new controller in CM.Control that can be used to turn
214    off all remaining support for implicit anchors by saying:
215    
216        CM.autoload "$smlnj/
217        #set CM.Control.implicit_anchors false;
218    
219    This causes CM to reject implicitly anchored paths.  This is (for the
220    time being) less permissive than the "final" version where there will
221    be no more such implicit anchors and relative paths will be just that:
222    relative.
223    
224    The next step (version after next version?) will be to make the
225    default for CM.Control.implicit_anchors false.  After the dust has
226    settled, I can then produce the "final" version of this...
227    
228    Note: Since bootstrapping is a bit tricky, I provided new boot files.
229    
230    ----------------------------------------------------------------------
231    Name: Matthias Blume
232    Date: 2000/05/11 16:30:00 JST
233    Tag: blume-20000511-sources
234    Description:
235    
236    The main change is that I added function CM.sources as a generalized
237    version of the earlier CM.makedepend.  This entails the following
238    additional changes:
239    
240      - CM.makedepend has been dropped.
241    
242      - CM manual has been updated.
243    
244      - TOOLS signature and API have been changed.
245    
246    ----------------------------------------------------------------------
247    Name: Allen Leung
248    Date: 2000/05/10 21:17 EDT
249    Tag: leunga-20000510-moby-c--ssa
250    Description:
251    
252      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
253    None of these affect SML/NJ.
254    
255    1. Register Allocation
256    
257        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
258           This module tries to remove local (i.e. basic block level) redundancies
259           during spilling.
260    
261        b. A new framework for performing region based register allocation.
262           Not yet entirely functional.
263    
264    2. X86
265    
266       a. DefUse for POP was missing the stack pointer [found by Lal]
267       b. Reload for CALL was incorrect in X86Spill [found by John]
268       c. Various fixes in X86Spill so that it can be used correctly for
269          the new spilling module.
270    
271    3. SSA/IR
272    
273       a. New module ir/dj-dataflow.sml implements elimination based
274          data flow analysis.
275    
276    4. MLRiscGen
277    
278       a. Fix for gc type annotation
279    
280    5. MDGen
281    
282       Various fixes for machine description -> ml code translation.  For ssa
283       only.
284    
285    ----------------------------------------------------------------------
286    Name: Allen Leung
287    Date: 2000/05/08 22:17 EDT
288    Tag: leunga-20000508-labexp
289    Description:
290    
291      Fermin has found a few assembly problems with constant expressions
292      generated in LabelExp.  Mostly, the problems involve extra parentheses,
293      which choke on dumb assemblers.  This is his fix.
294    
295    ----------------------------------------------------------------------
296    Name: Dave MacQueen
297    Date: 2000/04/09 14:00 EDT
298    Tag: dbm-20000502-Version_110_28
299    Description:
300    
301    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
302    
303    2. Updated config/version to 110.28
304    
305    3. Updated config/srcarchiveurl
306    
307    3. New boot files!
308       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume
312    Date: 2000/05/01 19:05:00 JST
313    Tag: blume-20000501-noweb
314    Description:
315    
316    A new noweb tool has been added.  The existing system is entirely
317    unaffected by this, but some CM users have asked for renewed noweb
318    support.  Everything is documented in the CM manual.
319    
320    New (plugin) libraries:
321    
322       noweb-tool.cm
323       nw-ext.cm
324    
325    ----------------------------------------------------------------------
326    Name: Dave MacQueen
327    Date: 2000/04/30 12:40PM EDT
328    Tag: dbm-20000430-bug_fixes
329    Description:
330    
331    1. Fix for bug 1498
332       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
333       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
334         added toRealArray function
335       smlnj/src/compiler/MiscUtil/print/ppobj.sml
336         added check for tag Obj.RealArray to array printing case in ppObj
337    
338    2. Fix for bug 1510
339       smlnj/src/compiler/Semant/types/typesutil.sml
340         fixed definition of dummyargs (used by equalTycon) so that
341         dummy args are distinct types
342    
343    ----------------------------------------------------------------------
344    Name: Matthias Blume
345    Date: 2000/04/30 01:00:00 JST
346    Tag: blume-20000430-versions
347    Description:
348    
349    1. CM version numbering added.  This is an implementation of Lal's
350       proposal for adding version numbers and version checking to .cm
351       files.  Lal said that his proposal was just that -- a proposal.
352       For the time being I went ahead and implemented it so that people
353       can comment on it.  Everything is completely backward-compatible
354       (except for the stable library format, i.e., new bootfiles!).
355    
356       As usual, see the CM manual for details.
357    
358    2. An alternative syntax for anchored paths has been implemented.
359       Dave has recently voiced the same concerns that I had when I did
360       this, so there should be some support.  My take is that eventually
361       I will let support for the current syntax (where anchors are
362       "implicit") fade out in favor of the new, explicit syntax.
363       In order to be backward-compatible, both old and new syntax are
364       currently supported.
365    
366       Again, see the CM manual for details.
367    
368    3. Parallel make is trying to be slightly smarter:  When the master
369       process finds a "bottleneck", i.e., when there is only one
370       compilation unit that can be compiled and everybody else is
371       waiting on it, then it will simply compile it directly instead
372       of clumsily telling one of the slaves to do it.
373    
374    4. Support for "unsharing" added.  This is necessary in order to be
375       able to have two different versions of the same library running
376       at the same time (e.g., for trying out a new MLRISC while still
377       having the old MLRISC linked into the current compiler, etc.)
378       See the CM manual.
379    
380    5. Simple "makedepend" functionality added for generating Makefile
381       dependency information.  (This is rather crude at the moment.
382       Expect some changes here in the future.)
383    
384    6. ".fun" added as a recognized suffix for ML files. Also documented
385       explicitly in the manual that the fallback behavior (unknown suffix
386       -> ML file) is not an official feature!
387    
388    7. Small changes to the pickler for stable libraries.
389    
390    8. Several internal changes to CM (for cleanup/improvement).
391    
392    
393    !!!! NEW BINFILES !!!!
394    
395    ----------------------------------------------------------------------
396    Name: Matthias Blume
397    Date: 2000/04/28 17:30:00 JST
398    Tag: blume-20000428-pathconfig
399    Description:
400    
401    1. I changed config/install.sh to remove duplicate entries from the
402       lib/pathconfig file at the end.  Moreover, the final version of
403       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
404       in src/system/installml.
405    
406    2. The config/install.sh script now consistently uses relative
407       pathnames in lib/pathconfig whenever the anchor is in the lib
408       directory.  (So far this was true for the libraries that come
409       pre-compiled and bundled as part of the bootfiles but not for
410       libraries that are compiled by the script itself.)
411    
412    ----------------------------------------------------------------------
413    Name: Matthias Blume
414    Date: 2000/04/26 13:10:00 JST
415    Tag: blume-20000426-fun_suffix
416    Description:
417    
418    Added ".fun" as a recognized file name suffix (for ML code).
419    
420    ----------------------------------------------------------------------
421    Name: Allen Leung
422    Date: 2000/04/25 17:00:00 EST
423    Tag: leunga-20000425-alpha-ra
424    Description:
425    
426    1. Alpha
427    
428        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
429    in C--.
430    
431    2. RA
432    
433       Added a flag "ra-dump-size" to print out the size of the flowgraph
434       and the interference graph.
435    
436    ----------------------------------------------------------------------
437    Name: Dave MacQueen
438    Date: 2000/04/25/
439    Tag: dbm-20000425-mlyacc_doc_examples
440    Description:
441      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
442      Updated all three examples in src/ml-yacc/examples to run
443      under 110.* using CM.make.
444    
445    ----------------------------------------------------------------------
446    Name: Allen Leung
447    Date: 2000/04/20 23:04:00 EST
448    Tag: leunga-20000420-ssa-c---stuff
449    Description:
450    
451      This update synchronizes my repository with Yale's.  Most of these
452    changes, however, do not affect SML/NJ at all (the RA is an exception).
453    
454    1. Register Allocator
455    
456       a. An improvement in the interference graph construction:
457          Given a copy
458    
459                s <- t
460    
461          no interference edge between s and t is added for this definition of s.
462    
463       b. I've added two new spill heuristic modules that Fermin and I developed
464          (in the new library RA.cm). These are unused in SML/NJ but maybe
465          useful for others (Moby?)
466    
467    2. X86
468    
469       a. Various fixes in the backend provided by Fermin [C--] and Lal.
470    
471    3. Alpha
472    
473       a. Added the BSR instruction and code generation that goes with it [C--]
474       b. Other fixes too numerous to recount provided by Fermin [C--]
475    
476    4. Regmaps
477    
478       a. The regmaps are not initialized with the identity physical bindings
479          at creation time.  This is unneeded.
480    
481    5. MLRISC Optimizations
482    
483       a. The DJ-Graph module can now compute the iterated dominance frontiers
484          intersects with liveness incrementally in linear time! Woohoo!
485          This is now used in my new SSA construction algorithm.
486    
487       b. THe branch reorganization module is now smarter about linear chains of
488          basic blocks.
489    
490    
491    ----------------------------------------------------------------------
492    Name: Matthias Blume
493    Date: 2000/04/12 13:52:00 JST
494    Tag: blume_main_v110p27_1
495    Description:
496    
497    Changed install.sh script to handle archive files without version number
498    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
499    name of the boot file archive.
500    
501    ----------------------------------------------------------------------
502    Name: Dave MacQueen
503    Date: 2000/04/09 14:00 EDT
504    Tag: dbm-20000410-Version_110_27
505    Description:
506    
507    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
508    
509    2. Updated src/config/version to 110.27
510    
511    3. New boot files!
512    
513    ----------------------------------------------------------------------
514    Name: Allen Leung
515    Date: 2000/04/09 19:09:00 EST
516    Tag: leunga-20000409-misc
517    Description:
518    
519    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
520    
521    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
522    
523    ----------------------------------------------------------------------
524    Name: Stefan
525    Date: 2000/04/07 10:00:00 EDT
526    Tag: monnier-20000406-branch-handling
527    Description:
528    
529    Improved handling of branches (mostly those generated from
530    polymorphic equality), removed switchoff and changed the
531    default optimization settings (more cpsopt and less flintopt).
532    
533    ----------------------------------------------------------------------
534    Name: Allen Leung
535    Date: 2000/04/06 01:30:00 EST
536    Tag: leunga-20000406-peephole-x86-SSA-2
537    Description:
538    
539       Forgot a few files.
540    
541    ----------------------------------------------------------------------
542    Name: Allen Leung
543    Date: 2000/04/06 00:36:00 EST
544    Tag: leunga-20000406-peephole-x86-SSA
545    Description:
546    
547    1.  New Peephole code
548    
549    2.  Minor improvement to X86 instruction selection
550    
551    3.  Various fixes to SSA and machine description -> code translator
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume
555    Date: 2000/04/05 12:30:00 JST
556    Tag: blume_main_v110p26p2_3
557    Description:
558    
559    This update just merges three minor cosmetic updates to CM's sources
560    to get ready for the 110.27 code freeze on Friday.  No functionality
561    has changed.
562    
563    ----------------------------------------------------------------------
564    Name: Allen Leung
565    Date: 2000/04/04 19:39:00 EST
566    Tag: leunga-20000404-x86-asm
567    Description:
568    
569    1.  Fixed a problem in X86 assembly.
570    
571        Things like
572    
573           jmp %eax
574           jmp (%eax)
575    
576        should be output as
577    
578           jmp *%eax
579           jmp *(%eax)
580    
581    2.  Assembly output
582    
583          Added a new flag
584    
585              "asm-indent-copies" (default to false)
586    
587          When this flag is on, parallel copies will be indented an extra level.
588    
589    ----------------------------------------------------------------------
590    Name: Allen Leung
591    Date: 2000/04/04 03:18:00 EST
592    Tag: leunga-20000404-C--Moby
593    Description:
594    
595        All of these fixes are related to C--, Moby, and my own optimization
596        stuff; so they shouldn't affect SML/NJ.
597    
598    1.  X86
599    
600        Various fixes related floating point, and extensions.
601    
602    2.  Alpha
603    
604        Some extra patterns related to loads with signed/zero extension
605        provided by Fermin.
606    
607    3.  Assembly
608    
609        When generating assembly, resolve the value of client defined constants,
610        instead of generating symbolic values.  This is controlled by the
611        new flag "asm-resolve-constants", which is default to true.
612    
613    4.  Machine Descriptions
614    
615        a. The precedence parser was slightly broken when parsing infixr symbols.
616        b. The type generalizing code had the bound variables reversed, resulting
617           in a problem during arity raising.
618        c. Various fixes in machine descriptions.
619    
620    ----------------------------------------------------------------------
621    Name: Matthias Blume
622    Date: 2000/04/03 16:05:00 JST
623    Tag: blume_main_v110p26p2_2
624    Description:
625    
626    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
627    now done via the ordinary static environment that is context to each
628    compilation unit.
629    
630    To this end, I arranged that instead of "structure Core" as "structure
631    _Core" is bound in the pervasive environment.  Core access is done via
632    _Core (which can never be accidentally rebound because _Core is not a
633    legal surface-syntax symbol).
634    
635    The current solution is much cleaner because the core environment is
636    now simply part of the pervasive environment which is part of every
637    compilation unit's context anyway.  In particular, this eliminates all
638    special-case handling that was necessary until now in order to deal
639    with dynamic and symbolic parts of the core environment.
640    
641    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
642    compilation manager's bootstrap compiler (actually: in the "init group"
643    handling).  See the comments in src/system/smlnj/init/init.cmi for
644    more details.
645    
646    I also tried to track down all mentions of "Core" (as string argument
647    to Symbol.strSymbol) in the compiler and replaced them with a
648    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
649    name appears in one place only.
650    
651    Binfile and bootfile format have not changed, but the switchover from
652    the old "init.cmi" to the new one is a bit tricky, so I supplied new
653    bootfiles anyway.
654    
655    ----------------------------------------------------------------------
656    Name: Allen Leung
657    Date: 2000/04/02 21:17:00 EST
658    Tag: leunga-20000402-mltree
659    Description:
660    
661       1. Renamed the constructor CALL in MLTREE by popular demand.
662       2. Added a bunch of files from my repository.  These are currently
663          used by other non-SMLNJ backends.
664    
665    ----------------------------------------------------------------------
666    Name: Allen Leung
667    Date: 2000/03/31 21:15:00 EST
668    Tag: leunga-20000331-aliasing
669    Description:
670    
671    This update contains a rewritten (and hopefully more correct) module
672    for extracting aliasing information from CPS.
673    
674       To turn on this feature:
675    
676            Compiler.Control.CG.memDisambiguate := true
677    
678       To pretty print the region information with assembly
679    
680           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
681    
682       To control how many levels of aliasing information are printed, use:
683    
684           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
685    
686       The default of n is 3.
687    
688    ----------------------------------------------------------------------
689    Name: David MacQueen
690    Date: 2000/03/31 11:15:00 EST
691    Tag: dbm-20000331-runtime_fix
692    Description:
693    
694    This update contains:
695    
696    1. runtime/c-lib/c-libraries.c
697       includes added in revision 1.2 caused compilation errors on hppa-hpux
698    
699    2. fix for bug 1556
700       system/Basis/Implementation/NJ/internal-signals.sml
701    
702    ----------------------------------------------------------------------
703    Name: Matthias Blume
704    Date: 2000/03/31 18:00:00 JST
705    Tag: blume_main_v110p26p2_1
706    Description:
707    
708    This update contains:
709    
710    1. A small change to CM's handling of stable libraries:
711       CM now maintains one "global" modmap that is used for all stable
712       libraries.  The use of such a global modmap maximizes sharing and
713       minimizes the need for re-traversing parts of environments during
714       modmap construction.  (However, this has minor impact since modmap
715       construction seems to account for just one percent or less of total
716       compile time.)
717    
718    2. I added a "genmap" phase to the statistics.  This is where I got the
719       "one percent" number (see above).
720    
721    3. CM's new tool parameter mechanism just became _even_ better. :)
722       - The parser understands named parameters and recursive options.
723       - The "make" and "shell" tools use these new features.
724         (This makes it a lot easier to cascade these tools.)
725       - There is a small syntax change: named parameters use a
726    
727           <name> : ( <option> ... )            or
728           <name> : <string>
729    
730         syntax.  Previously, named parameters were implemented in an
731         ad-hoc fashion by each tool individually (by parsing strings)
732         and had the form
733    
734           <name>=<string>
735    
736       See the CM manual for a full description of these issues.
737    
738    ----------------------------------------------------------------------
739    Name: Matthias Blume
740    Date: 2000/03/30 18:00:00 JST
741    Tag: blume_main_v110p26p2_0
742    Description:
743    
744    !!!!! WARNING !!!!!!
745    !!  New binfiles  !!
746    !!!!!!!!!!!!!!!!!!!!
747    
748    This update contains:
749    
750    1. Moderate changes to CM:
751    
752       - Changes to CM's tools mechanism.  In particular, it is now possible
753       to have tools that accept additional "command line" parameters
754       (specified in the .cm file at each instance where the tool's class is
755       used).
756    
757       This was done to accommodate the new "make" and "shell" tools which
758       facilitate fairly seamless hookup to portions of code managed using
759       Makefiles or Shell scripts.
760    
761       There are no classes "shared" or "private" anymore.  Instead, the
762       sharing annotation is now a parameter to the "sml" class.
763    
764       There is a bit of generic machinery for implementing one's own
765       tools that accept command-line parameters.  However, I am not yet fully
766       satisfied with that part, so expect changes here in the future.
767    
768       All existing tools are described in the CM manual.
769    
770       - Slightly better error handling.  (CM now suppresses many followup
771       error messages that tended to be more annoying than helpful.)
772    
773    2. Major changes to the compiler's static environment data structures.
774    
775       - no CMStaticEnv anymore.
776            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
777              but it is called Environment), no conversions between different
778              kinds of static environments
779    
780       - There is still a notion of a "modmap", but such modmaps are generated
781         on demand at the time when they are needed.  This sounds slow, but I
782         sped up the code that generates modmaps enough for this not to lead to
783         a slowdown of the compiler (at least I didn't detect any).
784    
785       - To facilitate rapid modmap generation, static environments now
786         contain an (optional) "modtree" structure.  Modtree annotations are
787         constructed by the unpickler during unpickling.  (This means that
788         the elaborator does not have to worry about modtrees at all.)
789         Modtrees have the advantage that they are compositional in the same
790         way as the environment data structure itself is compositional.
791         As a result, modtrees never hang on to parts of an environment that
792         has already been rendered "stale" by filtering or rebinding.
793    
794       - I went through many, many trials and errors before arriving at the
795         current solution.  (The initial idea of "linkpaths" did not work.)
796         But the result of all this is that I have touched a lot of files that
797         depend on the "modules" and "types" data structures (most of the
798         elaborator). There were a lot of changes during my "linkpath" trials
799         that could have been reverted to their original state but weren't.
800         Please, don't be too harsh on me for messing with this code a bit more
801         than what was strictly necessary...  (I _did_ resist the tempation
802         of doing any "global reformatting" to avoid an untimely death at
803         Dave's hands. :)
804    
805       - One positive aspect of the previous point:  At least I made sure that
806         all files that I touched now compile without warnings (other than
807         "polyEqual").
808    
809       - compiler now tends to run "leaner" (i.e., ties up less memory in
810         redundant modmaps)
811    
812    ----------------------------------------------------------------------
813    Name: Allen Leung
814    Date: 2000/03/29 18:00:00
815    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
816    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
817    Description:
818    
819       This update contains *MAJOR* changes to the way code is generated from CPS
820    in the module mlriscGen, and in various backend modules.
821    
822    CHANGES
823    =======
824    
825    1. MLRiscGen: forward propagation fix.
826    
827       There was a bug in forward propagation introduced at about the same time
828       as the MLRISC x86 backend, which prohibits coalescing to be
829       performed effectively in loops.
830    
831       Effect: speed up of loops in RISC architectures.
832               By itself, this actually slowed down certain benchmarks on the x86.
833    
834    2. MLRiscGen:  forward propagating addresses from consing.
835    
836       I've changed the way consing code is generated.  Basically I separated
837       out the initialization part:
838    
839            store tag,   offset(allocptr)
840            store elem1, offset+4(allocptr)
841            store elem2, offset+8(allocptr)
842            ...
843            store elemn, offset+4n(allocptr)
844    
845       and the address computation part:
846    
847            celladdr <- offset+4+alloctpr
848    
849       and move the address computation part
850    
851       Effect:  register pressure is generally lower as a result.  This
852                makes compilation of certain expressions much faster, such as
853                long lists with non-trivial elements.
854    
855                 [(0,0), (0,0), .... (0,0)]
856    
857    3. MLRiscGen: base pointer elimination.
858    
859        As part of the linkage mechanism, we generate the sequence:
860    
861         L:  ...  <- start of the code fragment
862    
863         L1:
864             base pointer <- linkreg - L1 + L
865    
866         The base pointer was then used for computing relocatable addresses
867       in the code fragment.  Frequently (such as in lots of continuations)
868       this is not needed.  We now eliminate this sequence whenever possible.
869    
870         For compile time efficiency, I'm using a very stupid local heuristic.
871       But in general, this should be done as a control flow analysis.
872    
873       Effect:  Smaller code size.  Speed up of most programs.
874    
875    4. Hppa back end
876    
877         Long jumps in span dependence resolution used to depend on the existence
878      of the base pointer.
879    
880         A jump to a long label L was expanded into the following sequence:
881    
882          LDIL %hi(L-8192), %r29
883          LDO  %lo(L-8192)(%r29), %r29
884          ADD  %r29, baseptr, %r29
885          BV,n %r0(%r29)
886    
887         In the presence of change (3) above, this will not work.  I've changed
888       it so that the following sequence of instructions are generated, which
889       doesn't mention the base pointer at all:
890    
891             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
892        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
893             BV,n  %r0(%r29)          /* Jump */
894    
895    5. Alpha back end
896    
897          New alpha instructions LDB/LDW have been added, as per Fermin's
898       suggestions.   This is unrelated to all other changes.
899    
900    6. X86 back end
901    
902         I've changed andl to testl in the floating point test sequence
903         whenever appropriate.  The Intel optimization guide states that
904         testl is preferable to andl.
905    
906    7. RA (x86 only)
907    
908         I've improved the spill propagation algorithm, using an approximation
909       of maximal weighted independent sets.   This seems to be necessary to
910       alleviate the negative effect in light of the slow down in (1).
911    
912         I'll write down the algorithm one of these days.
913    
914    8. MLRiscGen: frequencies
915    
916         I've added an annotation that states that all call gc blocks have zero
917       execution frequencies.  This improves register allocation on the x86.
918    
919    BENCHMARKS
920    ==========
921    
922       I've only perform the comparison on 110.25.
923    
924       The platforms are:
925    
926        HPPA  A four processor HP machine (E9000) with 5G of memory.
927        X86   A 300Hhz Pentium II with 128M of memory, and
928        SPARC An Ultra sparc 2 with 512M of memory.
929    
930       I used the following parameters for the SML benchmarks:
931    
932                 @SMLalloc
933         HPPA    256k
934         SPARC   512k
935         X86     256k
936    
937    COMPILATION TIME
938    ----------------
939       Here are the numbers comparing the compilation times of the compilers.
940       I've only compared 110.25 compiling the new sources versus
941       a fixpoint version of the new compiler compiling the same.
942    
943                     110.25                                  New
944               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
945         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
946         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
947         X86    999s    315s       94006+130691   987s    296s    108877+141957
948    
949                   110.25         New
950                Code Size      Code Size
951         HPPA   8596736         8561421
952         SPARC  8974299         8785143
953         X86    9029180         8716783
954    
955       So in summary, things are at least as good as before.   Dramatic
956       reduction in compilation is obtained on the Sparc; I can't explain it,
957       but it is reproducible.  Perhaps someone should try to reproduce this
958       on their own machines.
959    
960    SML BENCHMARKS
961    --------------
962    
963        On the average, all benchmarks perform at least as well as before.
964    
965          HPPA         Compilation Time     Spill+Reload      Run Time
966                     110.25  New            110.25    New   110.25  New
967    
968          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
969              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
970       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
971                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
972        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
973             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
974               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
975              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
976         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
977             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
978            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
979      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
980                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
981             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
982                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
983               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
984      --------------------------------------------------------------------------
985       Average                     6.12%                                   4.09%
986    
987          SPARC        Compilation Time     Spill+Reload      Run Time
988                     110.25  New            110.25    New   110.25  New
989    
990          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
991              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
992       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
993                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
994        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
995             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
996               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
997              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
998         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
999             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1000            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
1001      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
1002                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
1003             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
1004                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
1005               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
1006      ----------------------------------------------------------------------------
1007       Average                     6.94%                                    2.64%
1008    
1009          X86          Compilation Time     Spill+Reload      Run Time
1010                     110.25  New            110.25    New   110.25  New
1011    
1012          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
1013              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
1014       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
1015                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1016        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1017             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1018               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1019              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1020         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1021             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1022            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1023      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1024                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1025             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1026                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1027               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1028      ----------------------------------------------------------------------------
1029       Average                     1.22%                                     3.36%
1030    
1031    ----------------------------------------------------------------------
1032    Name: Allen Leung
1033    Date: 2000/03/23 16:25:00
1034    Tag: leunga-20000323-fix_x86_alpha
1035    Description:
1036    
1037    1. X86 fixes/changes
1038    
1039       a.  The old code generated for SETcc was completely wrong.
1040           The Intel optimization guide is VERY misleading.
1041    
1042    2. ALPHA fixes/changes
1043    
1044       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1045       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1046       c.  Added reassociation code for address computation.
1047    
1048    ----------------------------------------------------------------------
1049    Name: Allen Leung
1050    Date: 2000/03/22 01:23:00
1051    Tag: leunga-20000322-fix_x86_hppa_ra
1052    Description:
1053    
1054    1. X86 fixes/changes
1055    
1056       a.  x86Rewrite bug with MUL3 (found by Lal)
1057       b.  Added the instructions FSTS, FSTL
1058    
1059    2. PA-RISC fixes/changes
1060    
1061       a.  B label should not be a delay slot candidate!  Why did this work?
1062       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1063           as it should be.
1064       c.  The assembly syntax for fstds and fstdd was wrong.
1065       d.  Added the composite instruction COMICLR/LDO, which is the immediate
1066           operand variant of COMCLR/LDO.
1067    
1068    3. Generic MLRISC
1069    
1070       a.  shuffle.sml rewritten to be slightly more efficient
1071       b.  DIV bug in mltree-simplify fixed (found by Fermin)
1072    
1073    4. Register Allocator
1074    
1075       a.  I now release the interference graph earlier during spilling.
1076           May improve memory usage.
1077    
1078    ----------------------------------------------------------------------
1079    Name: Matthias Blume
1080    Date: 2000/03/14 14:15:32
1081    Tag: blume_main_v110p26p1_2
1082    Description:
1083    
1084    1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1085    additional argument called "template" which is an optional string that
1086    specifies the layout of the tool command line.  See the CM manual for
1087    explanation.
1088    
1089    2. A special-purpose tool can be "registered" by simply dropping the
1090    corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1091    directory where the .cm file lives that uses this tool.  (The
1092    behavior/misfeature until now was to look for the tool description
1093    files in the current working directory.)  As before, tool description
1094    files could also be anchored -- in which case they can live anywhere
1095    they like.  Following the recent e-mail discussion, this change should
1096    make it easier to have special-purpose tools that are shipped together
1097    with the sources of the program that uses them.
1098    
1099    ----------------------------------------------------------------------
1100    Name: Matthias Blume
1101    Date: 2000/03/10 07:48:34
1102    Tag: blume_main_v110p26p1_1
1103    Description:
1104    
1105    I added a re-written version of Dave's fixpt script to src/system.
1106    Changes relative to the original version:
1107      - sh-ified (not everybody has ksh)
1108      - automatically figures out which architecture it runs on
1109      - uses ./makeml a bit more cleverly
1110      - never invokes ./installml (and, thus, does not clobber your
1111        good and working installation of sml in case something goes wrong)
1112      - accepts max iteration count using option "-iter <n>"
1113      - accepts a "base" name using option "-base <base>"
1114    
1115    It does not build any extraneous heap images but directly rebuilds
1116    bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
1117    it can incorporate existing bin- and boot- hierarchies.  For example,
1118    suppose the base is set to "sml" (which is the default).  Then it
1119    successively builds
1120    
1121            sml.bin.<arch>-unix and sml.boot.<arch>-unix
1122    then    sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
1123    then    sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
1124    ...
1125    then    sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
1126    
1127    and so on.  If any of these already exist, it will just use what's
1128    there.  In particular, many people will have the initial set of bin
1129    and boot files around, so this saves time for at least one full
1130    rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
1131    <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
1132    they might be the result of an earlier partial run of fixpt (which
1133    perhaps got accidentally killed).  In this case, fixpt will quickly
1134    move through what exists before continuing where it left off earlier,
1135    and, thus, saves a lot of time.
1136    
1137    ----------------------------------------------------------------------
1138    Name: Allen Leung
1139    Date: 00/03/10 02:20:00
1140    Tag: leunga-20000310-fix_x86_asm_ra
1141    Description:
1142    
1143    More assembly output problems involving the indexed addressing mode
1144    on the x86 have been found and corrected. Thanks to Fermin Reig for the
1145    fix.
1146    
1147    The interface and implementation of the register allocator have been changed
1148    slightly to accommodate the possibility to skip the register allocation
1149    phases completely and go directly to memory allocation.  This is needed
1150    for C-- use.
1151    
1152  ----------------------------------------------------------------------  ----------------------------------------------------------------------
1153  Name: Matthias Blume  Name: Matthias Blume
# Line 43  Line 1180 
1180    it from that remote directory.    it from that remote directory.
1181    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
1182    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
1183    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
1184    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
1185    
1186    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.575  
changed lines
  Added in v.660

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