Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 578, Tue Mar 14 05:16:29 2000 UTC revision 675, Fri Jun 23 09:18:18 2000 UTC
# Line 13  Line 13 
13  Description:  Description:
14  ----------------------------------------------------------------------  ----------------------------------------------------------------------
15  Name: Matthias Blume  Name: Matthias Blume
16    Date: 2000/06/23 18:20:00 JST
17    Tag: blume-20000623-btrace
18    Description:
19    
20    This updates adds a backtrace facility to aid programmers in debugging
21    their programs.  This involves the following changes:
22    
23    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
24       keeping track of the current call stack.  When programs are compiled
25       in a special mode, the compiler will insert calls to these hooks
26       into the user program.
27       "Hook" means that it is possible for different implementations of
28       back-tracing to register themselves (at different times).
29    
30    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
31       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
32       It can be turned on using this call:
33         SMLofNJ.Internals.BTrace.mode (SOME true);
34       Turning it off again:
35         SMLofNJ.Internals.BTrace.mode (SOME false);
36       Querying the current status:
37         SMLofNJ.Internals.BTrace.mode NONE;
38       Annotated programs are about twice as big as normal ones, and they
39       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
40       where all hooks do nothing).  The slowdown with a plugin that is
41       actually useful (such as the one supplied by default) is even greater,
42       but in the case of the default plugin it is still only an constant
43       factor (amortized).
44    
45    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
46       with a sub-structure BTrace for controlling back-tracing.  In particular,
47       the above-mentioned function "mode" controls whether the annotation
48       phase is invoked by the compiler.  Another important function is
49       "trigger": when called it aborts the current execution and causes
50       the top-level loop to print a full back-trace.
51    
52    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
53       for back-tracing.  It keeps track of the dynamic call stack and in
54       addition to that it keeps a partial history at each "level" of that
55       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
56       d and b (at separate times, dynamically), then the report will show:
57    
58       GOTO   d
59             /c
60       GOTO  \b
61       CALL   a
62    
63       This shows that there was an initial non-tail call of a, then a
64       tail-call to b or c, looping behavior in a cluster of functions that
65       consist of b and c, and then a goto from that cluster (i.e., either from
66       b or from c) to d.
67    
68       Note that (depending on the user program) the amount of information
69       that the back-trace module has to keep track of at each level is bounded
70       by a constant.  Thus, the whole implementation has the same asymptotical
71       complexity as the original program (both in space and in time).
72    
73    5. compiler/TopLevel/interact/evalloop.sml has been modified to
74       handle the special exception SMLofNJ.Internals.BTrace.BTrace
75       which is raised by the "trigger" function mentioned above.
76    
77    Notes on usage:
78    
79    - Annotated code works well together with unannotated code:
80    Unannotated calls simply do not show up at all in the backtrace.
81    
82    - It is not a good idea to let modules that were annotated during
83    different sessions run at the same time.  This is because the compiler
84    chooses small integers to identify individual functions, and there
85    will be clashes if different modules were compiled in separate sessions.
86    (Nothing will crash, and you will even be told about the clashes, but
87    back-trace information will in general not be useful.)
88    
89    - Back-tracing can be confused by callcc and capture.
90    
91    - The only way of getting a back-trace right now is to explicitly
92    invoke the "trigger" function from your user program.  Eventually, we
93    should make every exception carry back-trace information (if
94    available).  But since this creates more overhead at "raise"-time
95    (similar to the current exnHistory overhead), I have not yet
96    implemented this.  (The implementation will be rather easy.)  With
97    exceptions carrying back-trace information, this facility will be even
98    more useful because users don't need to modify their programs...
99    
100    - While it is possible to compile the compiler with back-trace
101    annotations turned on (I did it to get some confidence in
102    correctness), you must make absolutely sure that core.sml and
103    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
104    be compiled with annotation because there is no core access yet, but
105    if you compile btimp.sml with annotation, then the system will go into
106    an infinite recursion and crash.)
107    Since CM currently does not know about BTrace, the only way to turn
108    annotations on and off for different modules of the compiler is to
109    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
110    this is awkward and clumsy.
111    
112    Sample sessions:
113    
114    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
115    - SMLofNJ.Internals.BTrace.mode (SOME true);
116    [autoloading]
117    [autoloading done]
118    val it = false : bool
119    - structure X = struct
120    -     fun main n = let
121    -         fun a (x, 0) = d x
122    -           | a (x, n) = b (x, n - 1)
123    -         and b (x, n) = c (x, n)
124    -         and c (x, n) = a (x, n)
125    -         and d x = e (x, 3)
126    -         and e (x, 0) = f x
127    -           | e (x, n) = e (x, n - 1)
128    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
129    -           | f n = n * g (n - 1)
130    -         and g n = a (n, 3)
131    -     in
132    -         f n
133    -     end
134    - end;
135    structure X : sig val main : int -> int end
136    - X.main 3;
137    *** BACK-TRACE ***
138    GOTO   stdIn:4.2-13.20: X.main[2].f
139    GOTO-( stdIn:4.2-13.20: X.main[2].e
140    GOTO   stdIn:4.2-13.20: X.main[2].d
141         / stdIn:4.2-13.20: X.main[2].a
142         | stdIn:4.2-13.20: X.main[2].b
143    GOTO-\ stdIn:4.2-13.20: X.main[2].c
144    CALL   stdIn:4.2-13.20: X.main[2].g
145    GOTO   stdIn:4.2-13.20: X.main[2].f
146    GOTO-( stdIn:4.2-13.20: X.main[2].e
147    GOTO   stdIn:4.2-13.20: X.main[2].d
148         / stdIn:4.2-13.20: X.main[2].a
149         | stdIn:4.2-13.20: X.main[2].b
150    GOTO-\ stdIn:4.2-13.20: X.main[2].c
151    CALL   stdIn:4.2-13.20: X.main[2].g
152    GOTO   stdIn:4.2-13.20: X.main[2].f
153    GOTO-( stdIn:4.2-13.20: X.main[2].e
154    GOTO   stdIn:4.2-13.20: X.main[2].d
155         / stdIn:4.2-13.20: X.main[2].a
156         | stdIn:4.2-13.20: X.main[2].b
157    GOTO-\ stdIn:4.2-13.20: X.main[2].c
158    CALL   stdIn:4.2-13.20: X.main[2].g
159    GOTO   stdIn:4.2-13.20: X.main[2].f
160    CALL   stdIn:2.15-17.4: X.main[2]
161    -
162    
163    (Note that because of a FLINt bug the above code currently does not
164    compile without BTrace turned on.)
165    
166    Here is another example, using my modified Tiger compiler:
167    
168    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
169    - SMLofNJ.Internals.BTrace.mode (SOME true);
170    [autoloading]
171    [autoloading done]
172    val it = false : bool
173    - CM.make "sources.cm";
174    [autoloading]
175    ...
176    [autoloading done]
177    [scanning sources.cm]
178    [parsing (sources.cm):parse.sml]
179    [creating directory CM/SKEL ...]
180    [parsing (sources.cm):tiger.lex.sml]
181    ...
182    [wrote CM/sparc-unix/semant.sml]
183    [compiling (sources.cm):main.sml]
184    [wrote CM/sparc-unix/main.sml]
185    [New bindings added.]
186    val it = true : bool
187    - Main.compile ("../testcases/merge.tig", "foo.out");
188    *** BACK-TRACE ***
189    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
190    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
191    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
192    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
193    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
194    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
195    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
196         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
197    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
198    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
199    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
200    CALL   main.sml:1.18-118.4: Main.compile[2]
201    -
202    
203    ----------------------------------------------------------------------
204    Name: Matthias Blumen
205    Date: 2000/06/21 18:00:00 JST
206    Tag: blume-20000621-manual
207    Description:
208    
209    CM manual update: Path environments documented.
210    
211    ----------------------------------------------------------------------
212    Name: Matthias Blume
213    Date: 2000/06/19 13:40:00
214    Tag: blume-20000619-manual
215    Description:
216    
217    CM manual and system/README update.  This only covers the fact that
218    there are no more implicit anchors.  (Path environments and the "bind"
219    option to "cm" have yet to be documented.)
220    
221    ----------------------------------------------------------------------
222    Name: Matthias Blume
223    Date: 2000/06/19 11:05:00 JST
224    Tag: blume-20000619-chdir-bugfix
225    Description:
226    
227    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
228    
229    ----------------------------------------------------------------------
230    Name: Matthias Blume
231    Date: 2000/06/18 22:00:10 JST
232    Tag: blume-20000618-implicit-anchors-really-gone
233    Description:
234    
235    I updates the previous HISTORY entry where I forgot to mention that
236    implicit anchors are no longer with us.
237    
238    The current update also gets rid of the (now useless) controller
239    CM.Control.implicit_anchors.
240    
241    ----------------------------------------------------------------------
242    Name: Matthias Blume
243    Date: 2000/06/16 17:30:00 JST
244    Tag: blume-20000616-anchorenv
245    Description:
246    
247    This patch implements the long anticipated (just kidding :) "anchor
248    environment" mechanism.  In the course of doing this, I also
249    re-implemented CM's internal "SrcPath" module from scratch.  The new
250    one should be more robust in certain boundary cases.  In any case, it
251    is a lot cleaner than its predecessor (IMHO).
252    
253    This time, although there is yet another boot file format change, I
254    kept the unpickler backward-compatible.  As a result, no new bootfiles
255    are necessary and bootstrapping is straightforward.  (You cannot read
256    new bootfiles into an old system, but the other way around is no
257    problem.)
258    
259    Visible changes:
260    
261    ** 0. Implicit path anchors (without the leading $-symbol) are no
262    longer recognized at all. This means that such path names are not
263    illegal either.  For example, the name basis.cm simply refers to a
264    local file called "basis.cm" (i.e, the name is an ordinary path
265    relative to .cm-files directory).  Or, to put it differently, only
266    names that start with $ are anchored paths.
267    
268    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
269    vanished.
270    
271    John (Reppy) had critizised this as soon as I originally proposed and
272    implemented it, but at that time I did not really deeply believe
273    him. :) Now I came full-circle because I need the $<singlearc> syntax
274    in another place where it cannot be seen as an abbreviation for
275    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
276    seems to mean (i.e., it "expands" into the corresponding anchor
277    value).
278    
279    However, when paths are used as members in CM description files, it
280    continues to be true that there must be at least another arc after the
281    anchor.  This is now enforced separately during semantic analysis
282    (i.e., from a lexical/syntactical point of view, the notation is ok.)
283    
284    ** 2. The "cm" class now accepts an option "bind".  The option's value
285    is a sub-option list of precisely two items -- one labeled "anchor"
286    and the other one labeled "value".  As you might expect, "anchor" is
287    used to specify an anchor name to be bound, and "value" specifies what
288    the anchor is being bound to.
289    
290    The value must be a directory name and can be given in either standard
291    syntax (including the possibility that it is itself an anchored path)
292    or native syntax.
293    
294    Examples:
295    
296       foo.cm (bind:(anchor:bar value:$mystuff/bar))
297       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
298    
299    and so on.
300    
301    The meaning of this is that the .cm-file will be processed with an
302    augmented anchor environment where the given anchor(s) is/are bound to
303    the given values(s).
304    
305    The rationale for having this feature is this: Suppose you are trying
306    to use two different (already stable) libraries a.cm and b.cm (that
307    you perhaps didn't write yourself).  Further, suppose each of these
308    two libraries internally uses its own auxiliary library $aux/lib.cm.
309    Normally you would now have a problem because the anchor "lib" can not
310    be bound to more than one value globally.  Therefore, the project that
311    uses both a.cm and b.cm must locally redirect the anchor to some other
312    place:
313    
314       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
315       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
316    
317    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
318    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
319    
320    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
321    you when you do so at the time of CM.stabilize).  Therefore, you can
322    also use an anchored path as the value:
323    
324      a.cm (bind:(anchor:lib value:$a-lib))
325      b.cm (bind:(anchor:lib value:$b-lib))
326    
327    Now you can globally configure (using the usual CM.Anchor.anchor or
328    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
329    itself is always locally bound, setting it globally is no longer
330    meaningful or necessary (but it does not hurt either).  In fact, "lib"
331    can still be used as a global anchor for separate purposes.  As a
332    matter of fact, one can locally define "lib" in terms of a global
333    "lib":
334    
335      a.cm (bind:(anchor:lib value:$lib/a))
336      b.cm (bind:(anchor:lib value:$lib/b))
337    
338    ** 3: The encoding of path names has changed.  This affects the way
339    path names are shown in CM's progress report and also the internal
340    protocol encoding used for parallel make.
341    
342    The encoding now uses one or more ':'-separated segments.  Each
343    segments corresponds to a file that has been specified relative to the
344    file given by its preceding segment.  The first segment is either
345    relative to the CWD, absolute, or anchored.  Each segment itself is
346    basically a Unix pathname; all segments but the first are relative.
347    
348    Example:
349    
350       $foo/bar/baz.cm:a/b/c.sml
351    
352    This path denotes the file bar/a/b/c.sml relative to the directory
353    denoted by anchor "foo".  Notice that the encoding also includes
354    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
355    paths are resolved relative to the .cm-files directory, so baz.cm must
356    be ignored to get the "real" pathname.
357    
358    To make this fact more obvious, CM puts the names of such "virtual
359    arcs" into parentheses when they appear in progress reports. (No
360    parentheses will appear in the internal protocol encoding.)  Thus,
361    what you really see is:
362    
363      $foo/bar/(baz.cm):a/b/c.sml
364    
365    I find this notation to be much more informative than before.
366    
367    Another new feature of the encoding is that special characters
368    including parentheses, colons, (back)slashes, and white space are
369    written as \ddd (where ddd is the decimal encoding of the character).
370    
371    *** The CM manual still needs to be updated.
372    
373    ----------------------------------------------------------------------
374    Name: Allen Leung
375    Date: 2000/06/15 00:38:00
376    Tag: leunga-20000615-x86-peephole
377    
378    x86 Peephole fix by Fermin.  Affects c-- and moby only.
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume
382    Date: 2000/06/12 11:40:00
383    Tag: blume-20000612-parmakefix
384    Description:
385    
386    More cleanup after changing the file naming scheme: This time I
387    repaired the parallel make mechanism for CMB.make which I broke earlier.
388    
389    ----------------------------------------------------------------------
390    Name: Allen Leung
391    Date: 2000/06/09 01:25:00
392    Tag: leunga-20000609-various
393    
394    None of these things should affect normal SML/NJ operations
395    
396    1. Peephole improvements provided by Fermin (c--)
397    2. New annotation DEFUSE for adding extra dependence (moby)
398    3. New X86 LOCK instructions (moby)
399    4. New machine description language for reservation tables (scheduling)
400    5. Fixes to various optimization/analysis modules (branch chaining, dominator
401       trees etc.)
402    6. I've changed the CM files so that they can work with versions
403       110.0.6, 110.25 and 110.28
404    
405    ----------------------------------------------------------------------
406    Name: Matthias Blume
407    Date: 2000/06/09 12:40:00
408    Tag: blume-20000609-log
409    Description:
410    
411    - Removed all(?) remaining RCS Log entries from sources.
412    
413    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
414      anchored paths).
415    
416    ----------------------------------------------------------------------
417    Name: Matthias Blume
418    Date: 2000/06/07 17:00:00 JST
419    Tag: blume-20000607-no-implicit-anchors
420    Description:
421    
422    1. This update changes the default setting for
423    CM.Control.implicit_anchors from true to false.  This means that
424    implicit anchors are no longer permitted by default.  I also tried to
425    make sure that nothing else still relies on implicit anchors.
426    (This is the next step on the schedule towards a CM that does not even
427    have the notion of implicit anchors anymore.)
428    
429    2. More CM manual updates.
430    
431    3. I managed to track down and fix the pickling bug I mentioned last
432    time.  Because of the previously existing workaround, this entails no
433    immediate practical changes.
434    
435    ----------------------------------------------------------------------
436    Name: Matthias Blume
437    Date: 2000/06/06 11:15:00 JST
438    Tag: blume-20000606-lazierpickle
439    Description:
440    
441    !!!! NEW BOOT FILES !!!!
442    
443    * The main purpose of this update is to make library pickles lazier in
444    order to reduce the initial space penalty for autoloading a library.
445    As a result, it is now possible to have $smlnj/compiler.cm
446    pre-registered.  This should take care of the many complaints or
447    inquiries about missing structure Compiler.  This required changes to
448    CM's internal data structures and small tweaks to some algorithms.
449    
450    As a neat additional effect, it is no longer necessary (for the sake
451    of lean heap image files) to distinguish between a "minimal" CM and a
452    "full" CM.  Now, there is only one CM (i.e., the "full" version:
453    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
454    interactive top level. ($smlnj/cm/minimal.cm is gone.)
455    
456    To make the life of compiler-hackers easier, "makeml" now also
457    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
458    words, after you bootstrap a new sml for the first time, you will not
459    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
460    around you will still have to do it, though.)
461    
462    * A second change consists of major updates to the CM manual.  There
463    are now several appendices with summary information and also a full
464    specification of the CM description file syntax.
465    
466    * In directory src/system I added the script "allcross".  This script
467    invokes sml and cross-compiles the compiler for all supported
468    architectures.  (Useful when providing a new set of boot files.)
469    
470    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
471    added a small tweak to pickle-util.sml to work around this problem,
472    but it is not a proper fix yet.  I will investigate further.  (The
473    effect of the bug was an inflation of library pickle size.)
474    
475    * Version number increased to 110.28.1 (to avoid compatibility problems).
476    
477    ----------------------------------------------------------------------
478    Name: Allen Leung
479    Date: 2000/05/25 17:28 EDT
480    Tag: leunga-20000525-ra
481    Description:
482    
483      Fixed a bug in freezing phase of the register allocator.
484    
485    ----------------------------------------------------------------------
486    Name: Allen Leung
487    Date: 2000/05/15 22:53 EDT
488    Tag: leunga-20000515-alpha-x86-ra
489    Description:
490    
491      1. Alpha
492    
493          Slight cleanup.  Removed the instruction SGNXL
494    
495      2. X86
496    
497          Added the following instructions to the instruction set:
498    
499            ROLx, RORx,
500            BTx, BTSx, BTLx, BTRx,
501            XCHGx, and variants with the LOCK prefix
502    
503      3. Register Allocation
504    
505          The module ra-rewrite-with-renaming has been improved.
506    
507      These have no effect on SML/NJ.
508    
509    ----------------------------------------------------------------------
510    Name: Matthias Blume
511    Date: 2000/05/15 16:20:00 JST
512    Tag: blume-20000515-lightrebuild
513    Description:
514    
515    1. I added an alternative to "-rebuild" to "makeml".  The difference is
516       that prior to calling CMB.make' the CM-variable "LIGHT" will be
517       defined.  In effect, the command will not build any cross-compiler
518       backends and therefore finish more quickly.
519    
520       The "fixpt" script also takes a "-light" switch to be able to use
521       this new facility while compiling for a fixpoint.
522    
523    2. I replaced all mentions of anchored paths in group owner specifications
524       with simple relative paths (usually starting with "..").
525       The rationale is that a library's internal workings should not be
526       compromised by the lack of some anchor.  (An anchor is necessary
527       for someone who wants to refer to the library by an anchored path,
528       but it should not be necessary to build the same library in the first
529       place.)
530    
531    3. I changed the way CM's tool mechanism determines the shell command
532       string used for things like ml-yacc etc. so that it does not break
533       when CM.Control.implicit_anchors is turned off.
534    
535    ----------------------------------------------------------------------
536    Name: Matthias Blume
537    Date: 2000/05/12 18:20:00 JST
538    Tag: blume-20000512-ml-build
539    Description:
540    
541    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
542    getting installed properly (by config/install.sh).
543    
544    ----------------------------------------------------------------------
545    Name: Matthias Blume
546    Date: 2000/05/12 17:30:00 JST
547    Tag: blume-20000512-anchors
548    Description:
549    
550    !!! NEW BOOT FILES !!!
551    
552    This change is in preparation of fading out support for "implicitly
553    anchored path names".  I went through all sources and used the
554    explicit (and relatively new) $-notation.  See system/README and the
555    CM manual for more info on this.
556    
557    I also modified the anchoring scheme for some things such as "smlnj",
558    "MLRISC", "cm", etc. to take advantage of the fact that explicit
559    anchors are more expressive: anchor name and first arc do not have to
560    coincide.  This entails the following user-visible change:
561    
562    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
563    particular, when you fire up sml with a command-line argument, say,
564    e.g.:
565    
566       sml '$smlnj/cmb.cm'
567    
568    At the ML toplevel prompt:
569    
570       CM.autoload "$smlnj/cmb.cm";
571    
572    There is also a new controller in CM.Control that can be used to turn
573    off all remaining support for implicit anchors by saying:
574    
575        CM.autoload "$smlnj/
576        #set CM.Control.implicit_anchors false;
577    
578    This causes CM to reject implicitly anchored paths.  This is (for the
579    time being) less permissive than the "final" version where there will
580    be no more such implicit anchors and relative paths will be just that:
581    relative.
582    
583    The next step (version after next version?) will be to make the
584    default for CM.Control.implicit_anchors false.  After the dust has
585    settled, I can then produce the "final" version of this...
586    
587    Note: Since bootstrapping is a bit tricky, I provided new boot files.
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume
591    Date: 2000/05/11 16:30:00 JST
592    Tag: blume-20000511-sources
593    Description:
594    
595    The main change is that I added function CM.sources as a generalized
596    version of the earlier CM.makedepend.  This entails the following
597    additional changes:
598    
599      - CM.makedepend has been dropped.
600    
601      - CM manual has been updated.
602    
603      - TOOLS signature and API have been changed.
604    
605    ----------------------------------------------------------------------
606    Name: Allen Leung
607    Date: 2000/05/10 21:17 EDT
608    Tag: leunga-20000510-moby-c--ssa
609    Description:
610    
611      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
612    None of these affect SML/NJ.
613    
614    1. Register Allocation
615    
616        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
617           This module tries to remove local (i.e. basic block level) redundancies
618           during spilling.
619    
620        b. A new framework for performing region based register allocation.
621           Not yet entirely functional.
622    
623    2. X86
624    
625       a. DefUse for POP was missing the stack pointer [found by Lal]
626       b. Reload for CALL was incorrect in X86Spill [found by John]
627       c. Various fixes in X86Spill so that it can be used correctly for
628          the new spilling module.
629    
630    3. SSA/IR
631    
632       a. New module ir/dj-dataflow.sml implements elimination based
633          data flow analysis.
634    
635    4. MLRiscGen
636    
637       a. Fix for gc type annotation
638    
639    5. MDGen
640    
641       Various fixes for machine description -> ml code translation.  For ssa
642       only.
643    
644    ----------------------------------------------------------------------
645    Name: Allen Leung
646    Date: 2000/05/08 22:17 EDT
647    Tag: leunga-20000508-labexp
648    Description:
649    
650      Fermin has found a few assembly problems with constant expressions
651      generated in LabelExp.  Mostly, the problems involve extra parentheses,
652      which choke on dumb assemblers.  This is his fix.
653    
654    ----------------------------------------------------------------------
655    Name: Dave MacQueen
656    Date: 2000/04/09 14:00 EDT
657    Tag: dbm-20000502-Version_110_28
658    Description:
659    
660    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
661    
662    2. Updated config/version to 110.28
663    
664    3. Updated config/srcarchiveurl
665    
666    3. New boot files!
667       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
668    
669    ----------------------------------------------------------------------
670    Name: Matthias Blume
671    Date: 2000/05/01 19:05:00 JST
672    Tag: blume-20000501-noweb
673    Description:
674    
675    A new noweb tool has been added.  The existing system is entirely
676    unaffected by this, but some CM users have asked for renewed noweb
677    support.  Everything is documented in the CM manual.
678    
679    New (plugin) libraries:
680    
681       noweb-tool.cm
682       nw-ext.cm
683    
684    ----------------------------------------------------------------------
685    Name: Dave MacQueen
686    Date: 2000/04/30 12:40PM EDT
687    Tag: dbm-20000430-bug_fixes
688    Description:
689    
690    1. Fix for bug 1498
691       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
692       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
693         added toRealArray function
694       smlnj/src/compiler/MiscUtil/print/ppobj.sml
695         added check for tag Obj.RealArray to array printing case in ppObj
696    
697    2. Fix for bug 1510
698       smlnj/src/compiler/Semant/types/typesutil.sml
699         fixed definition of dummyargs (used by equalTycon) so that
700         dummy args are distinct types
701    
702    ----------------------------------------------------------------------
703    Name: Matthias Blume
704    Date: 2000/04/30 01:00:00 JST
705    Tag: blume-20000430-versions
706    Description:
707    
708    1. CM version numbering added.  This is an implementation of Lal's
709       proposal for adding version numbers and version checking to .cm
710       files.  Lal said that his proposal was just that -- a proposal.
711       For the time being I went ahead and implemented it so that people
712       can comment on it.  Everything is completely backward-compatible
713       (except for the stable library format, i.e., new bootfiles!).
714    
715       As usual, see the CM manual for details.
716    
717    2. An alternative syntax for anchored paths has been implemented.
718       Dave has recently voiced the same concerns that I had when I did
719       this, so there should be some support.  My take is that eventually
720       I will let support for the current syntax (where anchors are
721       "implicit") fade out in favor of the new, explicit syntax.
722       In order to be backward-compatible, both old and new syntax are
723       currently supported.
724    
725       Again, see the CM manual for details.
726    
727    3. Parallel make is trying to be slightly smarter:  When the master
728       process finds a "bottleneck", i.e., when there is only one
729       compilation unit that can be compiled and everybody else is
730       waiting on it, then it will simply compile it directly instead
731       of clumsily telling one of the slaves to do it.
732    
733    4. Support for "unsharing" added.  This is necessary in order to be
734       able to have two different versions of the same library running
735       at the same time (e.g., for trying out a new MLRISC while still
736       having the old MLRISC linked into the current compiler, etc.)
737       See the CM manual.
738    
739    5. Simple "makedepend" functionality added for generating Makefile
740       dependency information.  (This is rather crude at the moment.
741       Expect some changes here in the future.)
742    
743    6. ".fun" added as a recognized suffix for ML files. Also documented
744       explicitly in the manual that the fallback behavior (unknown suffix
745       -> ML file) is not an official feature!
746    
747    7. Small changes to the pickler for stable libraries.
748    
749    8. Several internal changes to CM (for cleanup/improvement).
750    
751    
752    !!!! NEW BINFILES !!!!
753    
754    ----------------------------------------------------------------------
755    Name: Matthias Blume
756    Date: 2000/04/28 17:30:00 JST
757    Tag: blume-20000428-pathconfig
758    Description:
759    
760    1. I changed config/install.sh to remove duplicate entries from the
761       lib/pathconfig file at the end.  Moreover, the final version of
762       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
763       in src/system/installml.
764    
765    2. The config/install.sh script now consistently uses relative
766       pathnames in lib/pathconfig whenever the anchor is in the lib
767       directory.  (So far this was true for the libraries that come
768       pre-compiled and bundled as part of the bootfiles but not for
769       libraries that are compiled by the script itself.)
770    
771    ----------------------------------------------------------------------
772    Name: Matthias Blume
773    Date: 2000/04/26 13:10:00 JST
774    Tag: blume-20000426-fun_suffix
775    Description:
776    
777    Added ".fun" as a recognized file name suffix (for ML code).
778    
779    ----------------------------------------------------------------------
780    Name: Allen Leung
781    Date: 2000/04/25 17:00:00 EST
782    Tag: leunga-20000425-alpha-ra
783    Description:
784    
785    1. Alpha
786    
787        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
788    in C--.
789    
790    2. RA
791    
792       Added a flag "ra-dump-size" to print out the size of the flowgraph
793       and the interference graph.
794    
795    ----------------------------------------------------------------------
796    Name: Dave MacQueen
797    Date: 2000/04/25/
798    Tag: dbm-20000425-mlyacc_doc_examples
799    Description:
800      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
801      Updated all three examples in src/ml-yacc/examples to run
802      under 110.* using CM.make.
803    
804    ----------------------------------------------------------------------
805    Name: Allen Leung
806    Date: 2000/04/20 23:04:00 EST
807    Tag: leunga-20000420-ssa-c---stuff
808    Description:
809    
810      This update synchronizes my repository with Yale's.  Most of these
811    changes, however, do not affect SML/NJ at all (the RA is an exception).
812    
813    1. Register Allocator
814    
815       a. An improvement in the interference graph construction:
816          Given a copy
817    
818                s <- t
819    
820          no interference edge between s and t is added for this definition of s.
821    
822       b. I've added two new spill heuristic modules that Fermin and I developed
823          (in the new library RA.cm). These are unused in SML/NJ but maybe
824          useful for others (Moby?)
825    
826    2. X86
827    
828       a. Various fixes in the backend provided by Fermin [C--] and Lal.
829    
830    3. Alpha
831    
832       a. Added the BSR instruction and code generation that goes with it [C--]
833       b. Other fixes too numerous to recount provided by Fermin [C--]
834    
835    4. Regmaps
836    
837       a. The regmaps are not initialized with the identity physical bindings
838          at creation time.  This is unneeded.
839    
840    5. MLRISC Optimizations
841    
842       a. The DJ-Graph module can now compute the iterated dominance frontiers
843          intersects with liveness incrementally in linear time! Woohoo!
844          This is now used in my new SSA construction algorithm.
845    
846       b. THe branch reorganization module is now smarter about linear chains of
847          basic blocks.
848    
849    
850    ----------------------------------------------------------------------
851    Name: Matthias Blume
852    Date: 2000/04/12 13:52:00 JST
853    Tag: blume_main_v110p27_1
854    Description:
855    
856    Changed install.sh script to handle archive files without version number
857    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
858    name of the boot file archive.
859    
860    ----------------------------------------------------------------------
861    Name: Dave MacQueen
862    Date: 2000/04/09 14:00 EDT
863    Tag: dbm-20000410-Version_110_27
864    Description:
865    
866    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
867    
868    2. Updated src/config/version to 110.27
869    
870    3. New boot files!
871    
872    ----------------------------------------------------------------------
873    Name: Allen Leung
874    Date: 2000/04/09 19:09:00 EST
875    Tag: leunga-20000409-misc
876    Description:
877    
878    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
879    
880    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
881    
882    ----------------------------------------------------------------------
883    Name: Stefan
884    Date: 2000/04/07 10:00:00 EDT
885    Tag: monnier-20000406-branch-handling
886    Description:
887    
888    Improved handling of branches (mostly those generated from
889    polymorphic equality), removed switchoff and changed the
890    default optimization settings (more cpsopt and less flintopt).
891    
892    ----------------------------------------------------------------------
893    Name: Allen Leung
894    Date: 2000/04/06 01:30:00 EST
895    Tag: leunga-20000406-peephole-x86-SSA-2
896    Description:
897    
898       Forgot a few files.
899    
900    ----------------------------------------------------------------------
901    Name: Allen Leung
902    Date: 2000/04/06 00:36:00 EST
903    Tag: leunga-20000406-peephole-x86-SSA
904    Description:
905    
906    1.  New Peephole code
907    
908    2.  Minor improvement to X86 instruction selection
909    
910    3.  Various fixes to SSA and machine description -> code translator
911    
912    ----------------------------------------------------------------------
913    Name: Matthias Blume
914    Date: 2000/04/05 12:30:00 JST
915    Tag: blume_main_v110p26p2_3
916    Description:
917    
918    This update just merges three minor cosmetic updates to CM's sources
919    to get ready for the 110.27 code freeze on Friday.  No functionality
920    has changed.
921    
922    ----------------------------------------------------------------------
923    Name: Allen Leung
924    Date: 2000/04/04 19:39:00 EST
925    Tag: leunga-20000404-x86-asm
926    Description:
927    
928    1.  Fixed a problem in X86 assembly.
929    
930        Things like
931    
932           jmp %eax
933           jmp (%eax)
934    
935        should be output as
936    
937           jmp *%eax
938           jmp *(%eax)
939    
940    2.  Assembly output
941    
942          Added a new flag
943    
944              "asm-indent-copies" (default to false)
945    
946          When this flag is on, parallel copies will be indented an extra level.
947    
948    ----------------------------------------------------------------------
949    Name: Allen Leung
950    Date: 2000/04/04 03:18:00 EST
951    Tag: leunga-20000404-C--Moby
952    Description:
953    
954        All of these fixes are related to C--, Moby, and my own optimization
955        stuff; so they shouldn't affect SML/NJ.
956    
957    1.  X86
958    
959        Various fixes related floating point, and extensions.
960    
961    2.  Alpha
962    
963        Some extra patterns related to loads with signed/zero extension
964        provided by Fermin.
965    
966    3.  Assembly
967    
968        When generating assembly, resolve the value of client defined constants,
969        instead of generating symbolic values.  This is controlled by the
970        new flag "asm-resolve-constants", which is default to true.
971    
972    4.  Machine Descriptions
973    
974        a. The precedence parser was slightly broken when parsing infixr symbols.
975        b. The type generalizing code had the bound variables reversed, resulting
976           in a problem during arity raising.
977        c. Various fixes in machine descriptions.
978    
979    ----------------------------------------------------------------------
980    Name: Matthias Blume
981    Date: 2000/04/03 16:05:00 JST
982    Tag: blume_main_v110p26p2_2
983    Description:
984    
985    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
986    now done via the ordinary static environment that is context to each
987    compilation unit.
988    
989    To this end, I arranged that instead of "structure Core" as "structure
990    _Core" is bound in the pervasive environment.  Core access is done via
991    _Core (which can never be accidentally rebound because _Core is not a
992    legal surface-syntax symbol).
993    
994    The current solution is much cleaner because the core environment is
995    now simply part of the pervasive environment which is part of every
996    compilation unit's context anyway.  In particular, this eliminates all
997    special-case handling that was necessary until now in order to deal
998    with dynamic and symbolic parts of the core environment.
999    
1000    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1001    compilation manager's bootstrap compiler (actually: in the "init group"
1002    handling).  See the comments in src/system/smlnj/init/init.cmi for
1003    more details.
1004    
1005    I also tried to track down all mentions of "Core" (as string argument
1006    to Symbol.strSymbol) in the compiler and replaced them with a
1007    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
1008    name appears in one place only.
1009    
1010    Binfile and bootfile format have not changed, but the switchover from
1011    the old "init.cmi" to the new one is a bit tricky, so I supplied new
1012    bootfiles anyway.
1013    
1014    ----------------------------------------------------------------------
1015    Name: Allen Leung
1016    Date: 2000/04/02 21:17:00 EST
1017    Tag: leunga-20000402-mltree
1018    Description:
1019    
1020       1. Renamed the constructor CALL in MLTREE by popular demand.
1021       2. Added a bunch of files from my repository.  These are currently
1022          used by other non-SMLNJ backends.
1023    
1024    ----------------------------------------------------------------------
1025    Name: Allen Leung
1026    Date: 2000/03/31 21:15:00 EST
1027    Tag: leunga-20000331-aliasing
1028    Description:
1029    
1030    This update contains a rewritten (and hopefully more correct) module
1031    for extracting aliasing information from CPS.
1032    
1033       To turn on this feature:
1034    
1035            Compiler.Control.CG.memDisambiguate := true
1036    
1037       To pretty print the region information with assembly
1038    
1039           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
1040    
1041       To control how many levels of aliasing information are printed, use:
1042    
1043           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
1044    
1045       The default of n is 3.
1046    
1047    ----------------------------------------------------------------------
1048    Name: David MacQueen
1049    Date: 2000/03/31 11:15:00 EST
1050    Tag: dbm-20000331-runtime_fix
1051    Description:
1052    
1053    This update contains:
1054    
1055    1. runtime/c-lib/c-libraries.c
1056       includes added in revision 1.2 caused compilation errors on hppa-hpux
1057    
1058    2. fix for bug 1556
1059       system/Basis/Implementation/NJ/internal-signals.sml
1060    
1061    ----------------------------------------------------------------------
1062    Name: Matthias Blume
1063    Date: 2000/03/31 18:00:00 JST
1064    Tag: blume_main_v110p26p2_1
1065    Description:
1066    
1067    This update contains:
1068    
1069    1. A small change to CM's handling of stable libraries:
1070       CM now maintains one "global" modmap that is used for all stable
1071       libraries.  The use of such a global modmap maximizes sharing and
1072       minimizes the need for re-traversing parts of environments during
1073       modmap construction.  (However, this has minor impact since modmap
1074       construction seems to account for just one percent or less of total
1075       compile time.)
1076    
1077    2. I added a "genmap" phase to the statistics.  This is where I got the
1078       "one percent" number (see above).
1079    
1080    3. CM's new tool parameter mechanism just became _even_ better. :)
1081       - The parser understands named parameters and recursive options.
1082       - The "make" and "shell" tools use these new features.
1083         (This makes it a lot easier to cascade these tools.)
1084       - There is a small syntax change: named parameters use a
1085    
1086           <name> : ( <option> ... )            or
1087           <name> : <string>
1088    
1089         syntax.  Previously, named parameters were implemented in an
1090         ad-hoc fashion by each tool individually (by parsing strings)
1091         and had the form
1092    
1093           <name>=<string>
1094    
1095       See the CM manual for a full description of these issues.
1096    
1097    ----------------------------------------------------------------------
1098    Name: Matthias Blume
1099    Date: 2000/03/30 18:00:00 JST
1100    Tag: blume_main_v110p26p2_0
1101    Description:
1102    
1103    !!!!! WARNING !!!!!!
1104    !!  New binfiles  !!
1105    !!!!!!!!!!!!!!!!!!!!
1106    
1107    This update contains:
1108    
1109    1. Moderate changes to CM:
1110    
1111       - Changes to CM's tools mechanism.  In particular, it is now possible
1112       to have tools that accept additional "command line" parameters
1113       (specified in the .cm file at each instance where the tool's class is
1114       used).
1115    
1116       This was done to accommodate the new "make" and "shell" tools which
1117       facilitate fairly seamless hookup to portions of code managed using
1118       Makefiles or Shell scripts.
1119    
1120       There are no classes "shared" or "private" anymore.  Instead, the
1121       sharing annotation is now a parameter to the "sml" class.
1122    
1123       There is a bit of generic machinery for implementing one's own
1124       tools that accept command-line parameters.  However, I am not yet fully
1125       satisfied with that part, so expect changes here in the future.
1126    
1127       All existing tools are described in the CM manual.
1128    
1129       - Slightly better error handling.  (CM now suppresses many followup
1130       error messages that tended to be more annoying than helpful.)
1131    
1132    2. Major changes to the compiler's static environment data structures.
1133    
1134       - no CMStaticEnv anymore.
1135            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
1136              but it is called Environment), no conversions between different
1137              kinds of static environments
1138    
1139       - There is still a notion of a "modmap", but such modmaps are generated
1140         on demand at the time when they are needed.  This sounds slow, but I
1141         sped up the code that generates modmaps enough for this not to lead to
1142         a slowdown of the compiler (at least I didn't detect any).
1143    
1144       - To facilitate rapid modmap generation, static environments now
1145         contain an (optional) "modtree" structure.  Modtree annotations are
1146         constructed by the unpickler during unpickling.  (This means that
1147         the elaborator does not have to worry about modtrees at all.)
1148         Modtrees have the advantage that they are compositional in the same
1149         way as the environment data structure itself is compositional.
1150         As a result, modtrees never hang on to parts of an environment that
1151         has already been rendered "stale" by filtering or rebinding.
1152    
1153       - I went through many, many trials and errors before arriving at the
1154         current solution.  (The initial idea of "linkpaths" did not work.)
1155         But the result of all this is that I have touched a lot of files that
1156         depend on the "modules" and "types" data structures (most of the
1157         elaborator). There were a lot of changes during my "linkpath" trials
1158         that could have been reverted to their original state but weren't.
1159         Please, don't be too harsh on me for messing with this code a bit more
1160         than what was strictly necessary...  (I _did_ resist the tempation
1161         of doing any "global reformatting" to avoid an untimely death at
1162         Dave's hands. :)
1163    
1164       - One positive aspect of the previous point:  At least I made sure that
1165         all files that I touched now compile without warnings (other than
1166         "polyEqual").
1167    
1168       - compiler now tends to run "leaner" (i.e., ties up less memory in
1169         redundant modmaps)
1170    
1171    ----------------------------------------------------------------------
1172    Name: Allen Leung
1173    Date: 2000/03/29 18:00:00
1174    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
1175    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
1176    Description:
1177    
1178       This update contains *MAJOR* changes to the way code is generated from CPS
1179    in the module mlriscGen, and in various backend modules.
1180    
1181    CHANGES
1182    =======
1183    
1184    1. MLRiscGen: forward propagation fix.
1185    
1186       There was a bug in forward propagation introduced at about the same time
1187       as the MLRISC x86 backend, which prohibits coalescing to be
1188       performed effectively in loops.
1189    
1190       Effect: speed up of loops in RISC architectures.
1191               By itself, this actually slowed down certain benchmarks on the x86.
1192    
1193    2. MLRiscGen:  forward propagating addresses from consing.
1194    
1195       I've changed the way consing code is generated.  Basically I separated
1196       out the initialization part:
1197    
1198            store tag,   offset(allocptr)
1199            store elem1, offset+4(allocptr)
1200            store elem2, offset+8(allocptr)
1201            ...
1202            store elemn, offset+4n(allocptr)
1203    
1204       and the address computation part:
1205    
1206            celladdr <- offset+4+alloctpr
1207    
1208       and move the address computation part
1209    
1210       Effect:  register pressure is generally lower as a result.  This
1211                makes compilation of certain expressions much faster, such as
1212                long lists with non-trivial elements.
1213    
1214                 [(0,0), (0,0), .... (0,0)]
1215    
1216    3. MLRiscGen: base pointer elimination.
1217    
1218        As part of the linkage mechanism, we generate the sequence:
1219    
1220         L:  ...  <- start of the code fragment
1221    
1222         L1:
1223             base pointer <- linkreg - L1 + L
1224    
1225         The base pointer was then used for computing relocatable addresses
1226       in the code fragment.  Frequently (such as in lots of continuations)
1227       this is not needed.  We now eliminate this sequence whenever possible.
1228    
1229         For compile time efficiency, I'm using a very stupid local heuristic.
1230       But in general, this should be done as a control flow analysis.
1231    
1232       Effect:  Smaller code size.  Speed up of most programs.
1233    
1234    4. Hppa back end
1235    
1236         Long jumps in span dependence resolution used to depend on the existence
1237      of the base pointer.
1238    
1239         A jump to a long label L was expanded into the following sequence:
1240    
1241          LDIL %hi(L-8192), %r29
1242          LDO  %lo(L-8192)(%r29), %r29
1243          ADD  %r29, baseptr, %r29
1244          BV,n %r0(%r29)
1245    
1246         In the presence of change (3) above, this will not work.  I've changed
1247       it so that the following sequence of instructions are generated, which
1248       doesn't mention the base pointer at all:
1249    
1250             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
1251        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
1252             BV,n  %r0(%r29)          /* Jump */
1253    
1254    5. Alpha back end
1255    
1256          New alpha instructions LDB/LDW have been added, as per Fermin's
1257       suggestions.   This is unrelated to all other changes.
1258    
1259    6. X86 back end
1260    
1261         I've changed andl to testl in the floating point test sequence
1262         whenever appropriate.  The Intel optimization guide states that
1263         testl is preferable to andl.
1264    
1265    7. RA (x86 only)
1266    
1267         I've improved the spill propagation algorithm, using an approximation
1268       of maximal weighted independent sets.   This seems to be necessary to
1269       alleviate the negative effect in light of the slow down in (1).
1270    
1271         I'll write down the algorithm one of these days.
1272    
1273    8. MLRiscGen: frequencies
1274    
1275         I've added an annotation that states that all call gc blocks have zero
1276       execution frequencies.  This improves register allocation on the x86.
1277    
1278    BENCHMARKS
1279    ==========
1280    
1281       I've only perform the comparison on 110.25.
1282    
1283       The platforms are:
1284    
1285        HPPA  A four processor HP machine (E9000) with 5G of memory.
1286        X86   A 300Hhz Pentium II with 128M of memory, and
1287        SPARC An Ultra sparc 2 with 512M of memory.
1288    
1289       I used the following parameters for the SML benchmarks:
1290    
1291                 @SMLalloc
1292         HPPA    256k
1293         SPARC   512k
1294         X86     256k
1295    
1296    COMPILATION TIME
1297    ----------------
1298       Here are the numbers comparing the compilation times of the compilers.
1299       I've only compared 110.25 compiling the new sources versus
1300       a fixpoint version of the new compiler compiling the same.
1301    
1302                     110.25                                  New
1303               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
1304         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
1305         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
1306         X86    999s    315s       94006+130691   987s    296s    108877+141957
1307    
1308                   110.25         New
1309                Code Size      Code Size
1310         HPPA   8596736         8561421
1311         SPARC  8974299         8785143
1312         X86    9029180         8716783
1313    
1314       So in summary, things are at least as good as before.   Dramatic
1315       reduction in compilation is obtained on the Sparc; I can't explain it,
1316       but it is reproducible.  Perhaps someone should try to reproduce this
1317       on their own machines.
1318    
1319    SML BENCHMARKS
1320    --------------
1321    
1322        On the average, all benchmarks perform at least as well as before.
1323    
1324          HPPA         Compilation Time     Spill+Reload      Run Time
1325                     110.25  New            110.25    New   110.25  New
1326    
1327          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
1328              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
1329       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
1330                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
1331        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
1332             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
1333               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
1334              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
1335         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
1336             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
1337            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
1338      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
1339                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
1340             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
1341                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
1342               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
1343      --------------------------------------------------------------------------
1344       Average                     6.12%                                   4.09%
1345    
1346          SPARC        Compilation Time     Spill+Reload      Run Time
1347                     110.25  New            110.25    New   110.25  New
1348    
1349          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
1350              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
1351       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
1352                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
1353        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
1354             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
1355               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
1356              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
1357         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
1358             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1359            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
1360      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
1361                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
1362             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
1363                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
1364               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
1365      ----------------------------------------------------------------------------
1366       Average                     6.94%                                    2.64%
1367    
1368          X86          Compilation Time     Spill+Reload      Run Time
1369                     110.25  New            110.25    New   110.25  New
1370    
1371          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
1372              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
1373       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
1374                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1375        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1376             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1377               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1378              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1379         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1380             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1381            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1382      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1383                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1384             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1385                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1386               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1387      ----------------------------------------------------------------------------
1388       Average                     1.22%                                     3.36%
1389    
1390    ----------------------------------------------------------------------
1391    Name: Allen Leung
1392    Date: 2000/03/23 16:25:00
1393    Tag: leunga-20000323-fix_x86_alpha
1394    Description:
1395    
1396    1. X86 fixes/changes
1397    
1398       a.  The old code generated for SETcc was completely wrong.
1399           The Intel optimization guide is VERY misleading.
1400    
1401    2. ALPHA fixes/changes
1402    
1403       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1404       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1405       c.  Added reassociation code for address computation.
1406    
1407    ----------------------------------------------------------------------
1408    Name: Allen Leung
1409    Date: 2000/03/22 01:23:00
1410    Tag: leunga-20000322-fix_x86_hppa_ra
1411    Description:
1412    
1413    1. X86 fixes/changes
1414    
1415       a.  x86Rewrite bug with MUL3 (found by Lal)
1416       b.  Added the instructions FSTS, FSTL
1417    
1418    2. PA-RISC fixes/changes
1419    
1420       a.  B label should not be a delay slot candidate!  Why did this work?
1421       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1422           as it should be.
1423       c.  The assembly syntax for fstds and fstdd was wrong.
1424       d.  Added the composite instruction COMICLR/LDO, which is the immediate
1425           operand variant of COMCLR/LDO.
1426    
1427    3. Generic MLRISC
1428    
1429       a.  shuffle.sml rewritten to be slightly more efficient
1430       b.  DIV bug in mltree-simplify fixed (found by Fermin)
1431    
1432    4. Register Allocator
1433    
1434       a.  I now release the interference graph earlier during spilling.
1435           May improve memory usage.
1436    
1437    ----------------------------------------------------------------------
1438    Name: Matthias Blume
1439  Date: 2000/03/14 14:15:32  Date: 2000/03/14 14:15:32
1440  Tag: blume_main_v110p26p1_2  Tag: blume_main_v110p26p1_2
1441  Description:  Description:
1442    
1443  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1444  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
1445  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
1446  explanation.  explanation.
1447    
1448  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
1449  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1450  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
1451  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 66  Line 1489 
1489  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
1490  <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,
1491  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
1492  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
1493  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
1494  and, thus, saves a lot of time.  and, thus, saves a lot of time.
1495    
# Line 116  Line 1539 
1539    it from that remote directory.    it from that remote directory.
1540    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
1541    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
1542    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
1543    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
1544    
1545    For CVS users, this mechanism is not relevant for source archives, but    For CVS users, this mechanism is not relevant for source archives, but

Legend:
Removed from v.578  
changed lines
  Added in v.675

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