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 580, Wed Mar 22 06:33:52 2000 UTC revision 875, Tue Jul 10 21:19:41 2001 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume
17    Date: 2001/07/10 17:30:00 EDT
18    Tag: Release_110_34
19    Description:
20    
21    New version number (110.34). New bootfiles.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume
25    Date: 2001/07/09 16:00:00 EDT
26    Tag: blume-20010709-more-varargs
27    Description:
28    
29    I changed the handling of varargs in ml-nlffigen again:
30    The ellipsis ... will now simply be ignored (with an accompanying warning).
31    
32    The immediate effect is that you can actually call a varargs function
33    from ML -- but you can't actually supply any arguments beyond the ones
34    specified explicitly.  (For example, you can call printf with its format
35    string, but you cannot pass additional arguments.)
36    
37    This behavior is only marginally more useful than the one before, but
38    it has the advantage that a function or, more importantly, a function
39    type never gets dropped on the floor, thus avoiding follow-up problems with
40    other types that refer to the offending one.
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume
44    Date: 2001/07/09 11:25:00 EDT
45    Tag: blume-20010709-varargs
46    Description:
47    
48    1. ckit-lib.cm now exports structure Error
49    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
50       with a warning accompanied by a source location.  Moreover, it
51       merely skips the offending function or type and proceeds with the
52       rest of its work.u  As a result, one can safely feed C code containing
53       "..." to ml-nlffigen.
54    3. There are some internal improvements to CM, providing slightly
55       more general string substitutions in the tools subsystem.
56    
57    ----------------------------------------------------------------------
58    Name: Matthias Blume
59    Date: 2001/06/27 15:10:00 EDT
60    Tag: blume-20010627-concur
61    Description:
62    
63    Fixed a small bug in CM's handling of parallel compilation.
64    (You could observe the bug by Control-C-interrupting an ordinary
65    CMB.make or CM.stabilize and then attaching some compile servers.
66    The result was that all of a sudden the previously interrupted
67    compilation would continue on its own.  This was because of
68    an over-optimization: CM did not bother to clean out certain queues
69    when no servers were attached "anyway", resulting in the contents
70    of these queues to grab control when new servers did get attached.)
71    
72    There is also another minor update to the CM manual.
73    
74    ----------------------------------------------------------------------
75    Name: Matthias Blume
76    Date: 2001/06/26 16:15:00 EDT
77    Tag: blume-20010626-cmdoc
78    Description:
79    
80    Minor typo fixed in CM manual (syntax diagram for libraries).
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume
84    Date: 2001/06/25 22:55:00 EDT
85    Tag: blume-20010625-x86pc
86    Description:
87    
88    Fixed a nasty bug in the X86 assembly code that caused signal
89    handlers to fail (crash) randomly.
90    
91    ----------------------------------------------------------------------
92    Name: Matthias Blume
93    Date: 2001/06/25 12:05:00 EDT
94    Tag: blume-20010625-nlffigen
95    Description:
96    
97    This update fixes a number of minor bugs in ml-nlffigen as reported by
98    Nick Carter <nbc@andrew.cmu.edu>.
99    
100      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
101      2. Default names for generated files are now derived from the name of
102         the C file *without its directory*.  In particular, this causes generated
103         files to be placed locally even if the C file is in some system directory.
104      3. Default names for generated signatures and structures are also derived
105         from the C file name without its directory.  This avoids silly things
106         like "structure GL/GL".
107         (Other silly names are still possible because ml-nlffigen does not do
108          a thorough check of whether generated names are legal ML identifiers.
109          When in doubt, use command line arguments to force particular names.)
110    
111    ----------------------------------------------------------------------
112    Name: Matthias Blume
113    Date: 2001/06/21 12:25:00 EDT
114    Tag: blume-20010621-eXene
115    Description:
116    
117    eXene now compiles and (sort of) works again.
118    
119    The library name (for version > 110.33) is $/eXene.cm.
120    
121    I also added an new example in src/eXene/examples/nbody.  See the
122    README file there for details.
123    
124    ----------------------------------------------------------------------
125    Name: Matthias Blume
126    Date: 2001/06/20 16:40:00 EDT
127    Tag: blume-20010620-cml
128    Description:
129    
130    CML now compiles and works again.
131    
132    Libraries (for version > 110.33):
133    
134      $cml/cml.cm            Main CML library.
135      $cml/basis.cm          CML's version of $/basis.cm.
136      $cml/cml-internal.cm   Internal helper library.
137      $cml/core-cml.cm       Internal helper library.
138      $cml-lib/trace-cml.cm  Tracing facility.
139      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
140    
141    The installer (config/install.sh) has been taught how to properly
142    install this stuff.
143    
144    ----------------------------------------------------------------------
145    Name: Matthias Blume
146    Date: 2001/06/19 17:55:00 EDT
147    Tag: blume-20010619-instantiate
148    Description:
149    
150    This un-breaks the fix for bug 1432.
151    (The bug was originally fixed in 110.9 but I broke it again some
152    time after that.)
153    
154    ----------------------------------------------------------------------
155    Name: Matthias Blume
156    Date: 2001/06/19 17:25:00 EDT
157    Tag: blume-20010619-signals
158    Description:
159    
160    This should (hopefully) fix the long-standing signal handling bug.
161    (The runtime system was constructing a continuation record with an
162    incorrect descriptor which would cause the GC to drop data on the floor...)
163    
164    ----------------------------------------------------------------------
165    Name: Matthias Blume
166    Date: 2001/06/15 15:05:00 EDT
167    Tag: blume-20010615-moresparc
168    Description:
169    
170    Here is a short late-hour update related to Sparc c-calls:
171    
172     -- made handling of double-word arguments a bit smarter
173    
174     -- instruction selection phase tries to collapse certain clumsily
175        constructed ML-Trees; typical example:
176    
177            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
178    
179        This currently has no further impact on SML/NJ since mlriscGen does
180        not seem to generate such patterns in the first place, and c-calls
181        (which did generate them in the beginning) has meanwhile been fixed
182        so as to avoid them as well.
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume
186    Date: 2001/06/15 15:05:00 EDT
187    Tag: blume-20010615-sparc
188    Description:
189    
190    The purpose of this update is to provide an implementation of NLFFI
191    on Sparc machines.
192    
193    Here are the changes in detail:
194    
195       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
196       the Sparc implementation of the c-calls API.
197       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
198       for accessing the ML frame.  Thus, we have a real frame pointer and
199       can freely modify %sp without need for an omit-frame-ptr phase.
200       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
201       this case.
202       * ml-nlffigen has been taught to produce code for different architectures
203       and calling conventions.
204       * In a way similar to what was done in the x86 case, the Sparc
205       backend uses its own specific extension to mltree.  (For example,
206       it needs to be able to generate UNIMP instructions which are part
207       of the calling convention.)
208       * ml-nlffi-lib was reorganized to make it more modular (in particular,
209       to make it easier to plug in new machine- and os-dependent parts).
210    
211    There are some other fairly unrelated bug fixes and cleanups as well:
212    
213       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
214       that they properly share their libraries with existing SML/NJ libraries.
215       * I fixed a minor cosmetic bug in CM, supressing certain spurious
216       follow-up error messages.
217       * Updates to CM/CMB documentation.
218    
219    TODO items:
220    
221       * MLRISC should use a different register as its asmTemp on the Sparc.
222         (The current %o2 is a really bad choice because it is part of the
223         calling conventions, so things might interfere in unexpected ways.)
224    
225    ----------------------------------------------------------------------
226    Name: Matthias Blume
227    Date: 2001/06/07
228    Tag: blume-20010607-calls
229    Description:
230    
231    A number of internal changes related to C calls and calling conventions:
232    
233    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
234       number of bytes popped implicitly (by the callee).  In most cases
235       this field is 0 but on x86/win32 it is some non-zero value.  This
236       is information provided for the benefit of the "omit-frameptr" pass.
237    2. The CALL instruction on the x86 carries a similar "pops" field.
238       The instruction selection phase copies its value from the ML-Tree
239       CALL statement.
240    3. On all other architectures, the instruction selection phase checks
241       whether "pops=0" and complains if not.
242    4. The c-calls implementation for x86 now accepts two calling conventions:
243       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
244       up after the call and pops is set to 0.  For "stdcall", the caller
245       does nothing, leaving the cleanup to the callee; pops is set to
246       the number of bytes that were pushed onto the stack.
247    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
248       distinguish between "ccall" and "stdcall".
249    6. The UNIMP instruction has been added to the supported Sparc instruction
250       set. (This is needed for implementing the official C calling convention
251       on this architecture.)
252    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
253       work with the latest CM.
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume
257    Date: 2001/06/05 15:10:00 EDT
258    Tag: blume-20010605-cm-index
259    Description:
260    
261    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
262    
263    1. CM can now generate "index files".  These are human-readable files
264       that list on a per-.cm-file basis each toplevel symbol defined or
265       imported.  The location of the index file for
266       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
267       To enable index-file generation, set CM.Control.generate_index to true
268       or export an environment-symbol: export CM_GENERATE_INDEX=true.
269    
270       The CM manual has been updated accordingly.
271    
272    2. I made some slight modifications to the c-calls API in MLRISC.
273    
274         a) There is now a callback to support saving/restoring of
275            dedicated but caller-save registers around the actual call
276            instruction.
277         b) One can optionally specify a comment-annotation for the
278            call instruction.
279    
280    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
281       (For example, the comment annotation shows the C prototype of
282        the function being called.)
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume
286    Date: 2001/06/01 13:30:00 EDT
287    Tag: blume-20010601-nlffi-cleanup
288    Description:
289    
290    This is mostly a cleanup of MLFFI stuff:
291    
292       - some signature files have been put into a more exposed place
293       - the ugly 'f type parameter is gone (simplifies types tremendously!)
294       - ml-nlffigen changed accordingly
295       - tutorial updated
296    
297    Other changes:
298    
299       - author's affiliation in CM manual(s) updated
300       - some more recognized keywords added to Allen's sml.sty
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume
304    Date: 2001/05/25 15:30:00 EDT
305    Tag: blume-20010525-iptr
306    Description:
307    
308      - put the official 110.33-README (as it appears on the ftp server) under
309        CVS
310      - fixed a small bug related to incomplete pointer types in
311        ml-nlffigen
312      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
313        (it does not need the 'f type parameter)
314    
315    ----------------------------------------------------------------------
316    Name: Matthias Blume
317    Date: 2001/05/23 14:30:00 EDT
318    Tag: Release_110_33
319    Description:
320    
321    New version number (110.33).  New bootfiles.
322    
323    ----------------------------------------------------------------------
324    Name: Matthias Blume
325    Date: 2001/05/22 18:06:00 EDT
326    Tag: blume-20010522-targets
327    Description:
328    
329    Made install.sh use file config/targets.customized if it exists, falling
330    back to config/targets if it doesn't.  This way one can have a customized
331    version of the targets file without touching the "real thing", thus
332    eliminating the constant fear of accidentally checking something bogus
333    back into the CVS repository...  (File config/targets.customized must
334    not be added to the repository!)
335    
336    ----------------------------------------------------------------------
337    Name: Matthias Blume
338    Date: 2001/05/22 16:30:00 EDT
339    Tag: blume-20010522-minitut
340    Description:
341    
342    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
343       struct returns.
344    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
345       incomplete, preliminary documentation for NLFFI.
346    
347    ----------------------------------------------------------------------
348    Name: Matthias Blume
349    Date: 2001/05/14 11:30:00 EDT
350    Tag: blume-20010514-script
351    Description:
352    
353    Some bugs in install script fixed.
354    
355    In addition to that I also made a slight change to the NLFFI API:
356    Functors generated by ml-nlffigen now take the dynamic library as a
357    straight functor argument, not as a suspended one.  (The original
358    functor code used to force the suspension right away anyway, so there
359    was nothing gained by this complication of the interface.)
360    
361    ----------------------------------------------------------------------
362    Name: Matthias Blume
363    Date: 2001/05/11 14:35:00 EDT
364    Tag: blume-20010511-ml-nlffi
365    Description:
366    
367    I finally took the plunge and added my new FFI code to the main
368    repository.  For x86-linux it is now ready for prime-time.
369    
370    There are two new subdirectories of "src":
371    
372      - ml-nlffi-lib:
373           The utility library for programs using the FFI interface.
374           Here is the implementation of $/c.cm and its associated low-level
375           partners $/c-int.cm and $/memory.cm.
376      - ml-nlffigen:
377           A stand-alone program for generating ML glue code from C source
378           code.
379    
380    Building ml-nlffigen requires $/ckit-lib.cm.
381    
382    The config/install.sh script has been updates to do the Right Thing
383    (hopefully).
384    
385    Notice that the source tree for the C-Kit will not be put under "src"
386    but directly under the installation root directory.  (This is the
387    structure that currently exists on the CVS server when you check out
388    module "sml".)  Fortunately, config/install.sh knows about this oddity.
389    
390    Bugs: No documentation yet.
391    
392    ----------------------------------------------------------------------
393    Name: Matthias Blume
394    Date: 2001/05/09 16:35:00 EDT
395    Tag: blume-20010509-cpscontract
396    Description:
397    
398    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
399    wrapper/unwrapper elimination did not decrement usage counts and some
400    dead variables got overlooked by the dead-up logic.)
401    
402    ----------------------------------------------------------------------
403    Name: Lal George
404    Date: 2001/05/08  17:26:09 EDT
405    Tag: george-20010508-omit-frameptr
406    Description:
407    
408    Changes to implement the omit-frame-pointer optimization to support
409    raw C calls. For now, there is only support on the Intel x86, but
410    other architectures will follow as more experience is gained with this.
411    
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume
415    Date: 2001/05/07 14:40:00 EDT
416    Tag: blume-20010507-proxies
417    Description:
418    
419    I made into "proxy libraries" all libraries that qualify for such a
420    change.  (A qualifying library is a library that has another library or
421    groups as its sole member and repeats that member's export list
422    verbatim.  A proxy library avoids this repetition by omitting its export
423    list, effectively inheriting the list that its (only) member exports.
424    See the CM manual for more explanation.)
425    The main effect is that explicit export lists for these libraries
426    do not have to be kepts in sync, making maintenance a bit easier.
427    
428    I also added copyright notices to many .cm-files.
429    
430    Last but not least, I made a new set of bootfiles.
431    
432    ----------------------------------------------------------------------
433    Name: Matthias Blume
434    Date: 2001/05/04 17:00:00 EDT
435    Tag: blume-20010504-cm-lsplit
436    Description:
437    
438    0. John merged pending changes to $/smlnj-lib.cm
439    
440    1. Allen's previous change accidentally backed out of one of Lal's
441       earlier changes.  I undid this mistake (re-introducing Lal's change).
442    
443    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
444       within the compiler where applicable.  There is some code simplification
445       because of that.
446    
447    3. The "split" phase (in FLINT) is now part of the default list of phases.
448       Compiler.Control.LambdaSplitting.* can be used to globally control the
449       lambda-splitting (cross-module-inlining) engine.  In addition to that,
450       it can now also be controlled on a per-source basis: CM has been taught
451       a new tool parameter applicable to ML source files.
452    
453       - To turn lambda-splitting off completely:
454            local open Compiler.Control.LambdaSplitting in
455                val _ = set Off
456            end
457       - To make "no lambda-splitting" the global default (but allow per-source
458         overriding); this is the initial setting:
459            local open Compiler.Control.LambdaSplitting in
460                val _ = set (Default NONE)
461            end
462       - To make "lambda-splitting with aggressiveness a" the global default
463         (and allow per-source overriding):
464            local open Compiler.Control.LambdaSplitting in
465                val _ = set (Default (SOME a))
466            end
467    
468       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
469         write (in the respective .cm-file):
470            a.sml (lambdasplitting:off)
471       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
472            a.sml (lambdasplitting:on)
473       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
474         <a> is a decimal non-negative integer):
475            a.sml (lambdasplitting:<a>)
476       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
477            a.sml (lambdasplitting:infinity)
478       - To use the global default for a.sml:
479            a.sml (lambdasplitting:default)
480         or simply
481            a.sml
482    
483    ----------------------------------------------------------------------
484    Name: Allen Leung
485    Date: 2001/05/04 01:57:00 EDT
486    Tag: leunga-20010504-sync
487    Description:
488    
489      MLRISC features.
490    
491      1. Fix to CMPXCHG instructions.
492      2. Changed RA interface to allow annotations in callbacks.
493      3. Added a new method to the stream interface to allow annotations updates.
494    
495    ----------------------------------------------------------------------
496    Name: Matthias Blume
497    Date: 2001/05/01 11:45:00 EDT
498    Tag: blume-20010501-pcedittmp
499    Description:
500    
501    Changed install.sh to use the current working directory instead of
502    /usr/tmp for a temporary file (pcedittmp).  The previous choice
503    of /usr/tmp caused trouble with MacOS X because of file premission
504    problems.
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume
508    Date: 2001/04/20 11:10:00 EDT
509    Tag: blume-20010420-inMLflag
510    Description:
511    
512     - added vp_limitPtrMask to vproc-state.h
513       (for use by the raw-C-calls mechanism to implement proper interrupt
514        handling)
515     - made the ML compiler aware of various data-structure offsets so it
516       can generate code for accessing the vp_inML flag and vp_limitPtrMask
517     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
518       raw C-calls
519    
520    ----------------------------------------------------------------------
521    Name: Lal George
522    Date: 2001/04/20 09:15:28 EDT
523    Tag: george-20010420-macosX
524    Description:
525    
526     - Changes to port to Mac OS X; Darwin.
527    
528     - In the process I found that sqrt was broken on the PPC, because the
529       fsqrt instruction is not implemented.
530    
531    ----------------------------------------------------------------------
532    Name: Matthias Blume
533    Date: 2001/04/18 12:45:00 EDT
534    Tag: blume-20010418-ccalls
535    Description:
536    
537     - fixed two off-by-4 errors in the x86-specific c-calls implementation
538       (this bug prevented structure arguments containing pointers from being
539        passed correctly)
540     - changed the raw-C-call code in mlriscGen.sml in such a way that
541       structure arguments are represented as a pointer to the beginning
542       of the structure (instead of having a series of synthesized arguments,
543       one for each structure member)
544    
545     - made makeml script's verbosity level configurable via environment
546       variable (MAKEML_VERBOSITY)
547    
548     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
549       in rawmem-x86.sml; we are now using the real thing
550    
551    ----------------------------------------------------------------------
552    Name: Matthias Blume
553    Date: 2001/03/22 16:25:00 EST
554    Tag: blume-20010322-bootfiles
555    Description:
556    
557    Created a new set of bootfiles (for your automatic installation convenience).
558    
559    ----------------------------------------------------------------------
560    Name: Matthias Blume
561    Date: 2001/03/22 15:10:00 EST
562    Tag: blume-20010322-rawmem-parcm
563    Description:
564    
565    1. All "raw memory access" primitives for the new FFI are implemented now
566       (at least on the x86).
567    2. Some further cleanup of CM's parallel make mechanism.
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume
571    Date: 2001/03/19 17:53:00 EST
572    Tag: blume-20010319-parallel
573    Description:
574    
575    Parallel make (using compile servers) now works again.
576    
577    To this end, CM.stabilize and CMB.make have been modified to work in
578    two passes when compile servers are attached:
579       1. Compile everything, do not perform stabilization; this pass
580          uses compile servers
581       2. Stabilize everything; this pass does not use compile servers
582    If there are no compile servers, the two passes are combined into one
583    (as before).  Splitting the passes increases the inherent parallelism
584    in the dependency graph because the entire graph including all
585    libraries is available at the same time.  This, in turn, improves
586    server utilization.  The downside is that the master process will
587    have to do some extra work after compilation is done (because for
588    technical reasons it must re-read all the binfiles during stabilization).
589    
590    ----------------------------------------------------------------------
591    Name: Matthias Blume
592    Date: 2001/03/16 12:22:00 EST
593    Tag: blume-20010316-bootfiles
594    Description:
595    
596    Created a new set of bootfiles (for your automatic installation convenience).
597    
598    ----------------------------------------------------------------------
599    Name: Matthias Blume
600    Date: 2001/03/16 11:00:00 EST
601    Tag: blume-20010316-MLTREE-fixup
602    Description:
603    
604    This is a minor fixup for an (untagged) earlier commit by Allen.
605    (A file was missing).
606    
607    ----------------------------------------------------------------------
608    Name: Allen Leung
609    Date: Mon Mar  5 18:54:57 EST 2001
610    Tag: leunga-20010305-cut-support
611    
612    1. New support for alternative control-flow in MLTREE.
613       Currently we support
614    
615          FLOW_TO(CALL ...., [k1,...,kn])
616    
617       This is needed for 'cuts to' in C-- and try/handle-like constructs
618       in Moby
619    
620       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
621    
622    2. Register Allocator
623    
624       Changes in interface [from Fermin, John]
625    
626    3. Alpha 8-bit SLL support [Fermin]
627    
628    4. All architectures
629    
630       A new module (ClusterExpandCopies) for expanding parallel copies.
631    
632    ----------------------------------------------------------------------
633    Name: Allen Leung
634    Date: 2001/02/27 23:07:00 EST
635    Tag: leunga-20010227-minor-stuff
636    
637    1. Alpha bug fix for CMOVNE
638    2. Handle mltree COND(..,FCMP ...,...)
639    3. Bug fix in simplifier
640    
641    ----------------------------------------------------------------------
642    Name: Matthias Blume
643    Date: 2001/01/30 17:50:00 EST
644    Tag: blume-20010130-sync
645    Description:
646    
647    This is just a minor update to sync my devel branch with the main brach.
648    The only visible change is the addition of some README files.
649    
650    ----------------------------------------------------------------------
651    Name: Matthias Blume
652    Date: 2001/01/12 23:30:00 JST
653    Tag: blume-20010112-bootfiles
654    Description:
655    
656    Made a new set of bootfiles that goes with the current state of the
657    repository.
658    
659    ----------------------------------------------------------------------
660    Name: Matthias Blume
661    Date: 2001/01/12 21:20:00 JST
662    Tag: blume-20010112-sync
663    Description:
664    
665    I am just flushing out some minor changes that had accumulated in
666    my private branch in order to sync with the main tree.  (This is
667    mainly because I had CVS trouble when trying to merge _into_ my
668    private branch.)
669    
670    Most people should be completely unaffected by this.
671    
672    ----------------------------------------------------------------------
673    Name: Allen Leung
674    Date: Thu Jan 11 21:03:00 EST 2001
675    Tag: leunga-20010111-labexp=mltree
676    Description:
677    
678    1.  Removed the type LabelExp and replace it by MLTree.
679    2.  Rewritten mltree-simplify with the pattern matcher tool.
680    3.  There were some bugs in alpha code generator which would break
681        64-bit code generation.
682    4.  Redo the tools to generate code with the
683    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
684        are now generated by perl scripts.
685    
686    ----------------------------------------------------------------------
687    Name: Matthias Blume
688    Date: 2001/01/10 21:55:00 JST
689    Tag: blume-20010110-rcc
690    Description:
691    
692    The RCC stuff now seems to work (but only on the x86).
693    This required hacking of the c-calls interface (and -implementation) in
694    MLRISC.
695    
696    Normal compiler users should be unaffected.
697    
698    ----------------------------------------------------------------------
699    Name: Matthias Blume
700    Date: 2001/01/09 01:20:00 JST
701    Tag: blume-20010109-rcc
702    Description:
703    
704    This is a fairly big patch, flushing out a large number of pending
705    changes that I made to my development copy over the last couple of days.
706    
707    Of practical relevance at this moment is a workaround for a pickling
708    bug that Allen ran into the other day.  The cause of the bug itself is
709    still unknown and it might be hard to fix it properly, but the
710    workaround has some merits of its own (namely somewhat reducing pickling
711    overhead for certain libraries).  Therefore, I think this solution should
712    be satisfactory at this time.
713    
714    The rest of the changes (i.e., the vast majority) has to do with my
715    ongoing efforts of providing direct support for C function calls from
716    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
717    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
718    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
719    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
720    handle RCC.
721    
722    The last part (the changes to mlriscGen.sml) are still known to be
723    wrong on the x86 and not implemented on all other architectures.  But
724    the infrastructure is in place. I had to change a few functor
725    signatures in the backend to be able to route the CCalls interface
726    from MLRISC there, and I had to specialize the mltree type (on the
727    x86) to include the necessary extensions. (The extensions themselves
728    were already there and redy to go in MLRISC/x86).
729    
730    Everything should be very happy as soon as someone helps me with
731    mlriscGen.sml...
732    
733    In any case, nothing of this should matter to anyone as long as the
734    new primop is not being used (which is going to be the case unless you
735    find it where I hid it :). The rest of the compiler is completely
736    unaffected.
737    
738    ----------------------------------------------------------------------
739    Name: Matthias Blume
740    Date: 2001/01/05 00:30:00 JST
741    Tag: blume-20010105-primops
742    Description:
743    
744    Added some experimental support for work that I am doing right now.
745    These changes mostly concern added primops, but there is also a new
746    experimental C library in the runtime system (but currently not enabled
747    anywhere except on Linux/X86).
748    
749    In the course of adding primops (and playing with them), I discovered that
750    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
751    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
752    To recover, I made type information in INL_PRIM mandatory and changed
753    prim.sml as well as built-in.sml accordingly.  The InLine structure now
754    has complete, correct type information (i.e., no bottom types).
755    
756    Since all these changes mean that we need new binfiles, I also bumped the
757    version number to 110.32.1.
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume
761    Date: 2000/12/30 22:10:00 JST
762    Tag: blume-20001230-various
763    Description:
764    
765    Added proxy libraries for MLRISC and let MLRISC libraries refer
766    to each other using path anchors.  (See CM manual for explanation.)
767    
768    Updated CM documentation.
769    
770    Fixed some bugs in CM.
771    
772    Implemented "proxy" libraries (= syntactic sugar for CM).
773    
774    Added "-quiet" option to makeml and changed runtime system accordingly.
775    
776    Added cleanup handler for exportML to reset timers and compiler stats.
777    
778    ----------------------------------------------------------------------
779    Name: Lal George
780    Date: 2000/12/22 22:22:58 EST 2000
781    Tag: Release_110_32
782    Description:
783    
784            Infinite precision used throughout MLRISC.
785            see MLRISC/mltree/machine-int.sig
786    
787    ----------------------------------------------------------------------
788    Name: Matthias Blume
789    Date: 2000/12/22 23:16:00 JST
790    Tag: blume-20001222-warn
791    Description:
792    
793    Corrected wording and formatting of some CM warning message which I
794    broke in my previous patch.
795    
796    ----------------------------------------------------------------------
797    Name: Matthias Blume
798    Date: 2000/12/22 21:20:00 JST
799    Tag: blume-20001222-anchorenv
800    Description:
801    
802    Fixed CM's handling of anchor environments in connection with CMB.make.
803    
804    ----------------------------------------------------------------------
805    Name: Matthias Blume
806    Date: 2000/12/22 13:15:00 JST
807    Tag: blume-20001222-cleanup
808    Description:
809    
810    Removed src/cm/ffi which does not (and did not) belong here.
811    
812    ----------------------------------------------------------------------
813    Name: Matthias Blume
814    Date: 2000/12/21 23:55:00 JST
815    Tag: blume-20001221-exn
816    Description:
817    
818    Probably most important: CM no longer silently swallows all exceptions
819    in the compiler.
820    Plus: some other minor CM changes.  For example, CM now reports some
821    sizes for generated binfiles (code, data, envpickle, lambdapickle).
822    
823    ----------------------------------------------------------------------
824    Name: Matthias Blume
825    Date: 2000/12/15 00:01:05 JST
826    Tag: blume-20001215-dirtool
827    Description:
828    
829    - "dir" tool added.
830    - improvements and cleanup to Tools structure
831    - documentation updates
832    
833    ----------------------------------------------------------------------
834    Name: Allen Leung
835    Date: Thu Dec 14 03:45:24 EST 2000
836    Description:
837    Tag:  leunga-20001214-int-inf
838    Description:
839    
840       In IntInf, added these standard functions, which are missing from our
841    implementation:
842    
843        andb : int * int -> int
844        xorb : int * int -> int
845        orb  : int * int -> int
846        notb : int -> int
847         <<   : int * word -> int
848        ~>>  : int * word -> int
849    
850       Not tested, I hope they are correct.
851    
852    ----------------------------------------------------------------------
853    Name: Allen Leung
854    Date: Fri Dec  8 19:23:26 EST 2000
855    Description:
856    Tag:  leunga-20001208-nowhere
857    Description:
858    
859      Slight improvements to the 'nowhere' tool to handle OR-patterns,
860    to generate better error messages etc.  Plus a brief manual.
861    
862    ----------------------------------------------------------------------
863    Name: Lal George
864    Date: 2000/12/08 09:54:02 EST 2000
865    Tag: Release_110_31
866    Description:
867    
868    - Version 110.31
869    ----------------------------------------------------------------------
870    Name: Allen Leung
871    Date: Thu Dec  7 22:01:04 EST 2000
872    Tag:  leunga-20001207-cell-monster-hack
873    Description:
874    
875    Major MLRISC internal changes.  Affect all clients.
876    Summary:
877    
878    1.  Type CELLS.cell = int is now replaced by a datatype.
879        As a result, the old regmap is now gone.  Almost all interfaces
880        in MLRISC change as a consequence.
881    
882    2.  A new brand version of machine description tool (v3.0) that generates
883        modules expecting the new interface.  The old version is removed.
884    
885    3.  The RA interface has been further abstracted into two new functors.
886        RISC_RA and X86RA.  These functors have much simpler interfaces.
887        [See also directory MLRISC/demo.]
888    
889    4.  Some other new source->source code generation tools are available:
890    
891        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
892        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
893           I use this tool to generate the peephole optimizers---with the new
894           cell type changes, peephole rules are becoming difficult to write
895           without conditional pattern matching.
896    
897    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
898        cover the entire MLRISC source tree so many things broke.
899    
900    6.  CM files have been moved to the subdirectory MLRISC/cm.
901        They are moved because there are a lot of them and they clutter up the
902        root dir.
903    
904    7.  More detailed documentation to come...
905    
906        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
907        a makeml -rebuild first.  This is because of other other
908        changes that Matthias has made (see below).
909    
910    
911    ----------------------------------------------------------------------
912    Name: Matthias Blume
913    Date: 2000/11/30 23:12:00 JST
914    Tag: blume-20001130-filereorg
915    Description:
916    
917    Some manual updates and some file reorganizations in CM.
918    
919    ----------------------------------------------------------------------
920    Name: Matthias Blume
921    Date: 2000/11/24 17:45:00 JST
922    Tag: blume-20001124-link
923    Description:
924    
925    Drastically improved link traversal code for the case that the dynamic
926    value was already loaded at bootstrap time.  As a result, CM and CMB
927    now both load blazingly fast -- even on a very slow machine.  Also,
928    memory consumption has been further reduced by this.
929    
930    Warning: The format of the PIDMAP file has changed.  THerefore, to
931    bootstrap you have to do this:
932    
933    1. Run CMB.make
934    2. Make a symbolic link for the boot directory:
935         ln -s sml.boot.ARCH-OS xxx
936    3. "Rebuild" the boot directory:
937         ./makeml -boot xxx -rebuild sml ; rm xxx
938    4. Boot normally:
939          ./makeml
940    
941    ----------------------------------------------------------------------
942    Name: Matthias Blume
943    Date: 2000/11/21 21:20:00 JST
944    Tag: blume-20001121-tools
945    Description:
946    
947    Continued hacking on autoloading problem -- with success this time.
948    Also changed tool-plugin mechanism.  See new CM manual.
949    
950    ----------------------------------------------------------------------
951    Name: Matthias Blume
952    Date: 2000/11/19 14:30:00 JST
953    Tag:  blume-20001119-autoload
954    Description:
955    
956    Some hacking to make autoloading faster.  Success for CMB, no success
957    so far for CM.  There is a reduced structure CM' that autoloads faster.
958    (This is a temporary, non-documented hack to be eliminated again when
959    the general problem is solved.)
960    
961    ----------------------------------------------------------------------
962    Name: Matthias Blume
963    Date: 2000/11/17 14:10:00 JST
964    Tag: blume-20001117-pickle-lib
965    Description:
966    
967    1. Eliminated comp-lib.cm
968    2. Made pickle-lib.cm
969    3. Eliminated all uses of intset.sml (from comp-lib.cm)
970    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
971       equivalent constructs from smlnj-lib.cm (INtHashTable).
972    5. Point 4. also goes for those uses of intmap.* in MLRISC.
973       Duplicated intmap modules thrown out.
974    6. Hunted down all duplicated SCC code and replaced it with
975       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
976    7. Rewrote Feedback module.
977    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
978       should be thrown out and equivalent modules from smlnj-lib.cm
979       should be used (IntRedBlackSet, IntListSet, ...).
980    
981    Confirmed that compiler compiles to fixpoint.
982    
983    ----------------------------------------------------------------------
984    Name: Allen Leung
985    Date: 2000/11/10 18:00:00
986    Tag: leunga-20001110-new-x86-fp
987    
988    A new x86 floating point code generator has been added.
989    By default this is turned off.  To turn this on, do:
990    
991        CM.autoload "$smlnj/compiler.cm";
992        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
993    
994    Changes:
995    
996    1.  Changed FTAN to FPTAN so that the assembly output is correct.
997    2.  Changed the extension callback for FTANGENT to generate:
998    
999              fptan
1000              fstp  %st(0)
1001        instead of
1002              fptan
1003              fstpl ftempmem
1004    
1005    3.  Numerous assembly fixes for x86.
1006    
1007    5.  Cleaned up the machine code output module x86/x86MC.sml and added
1008        support for a whole bunch of instructions and addressing modes:
1009    
1010          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
1011          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
1012          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
1013          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1014          fxch %st(n)
1015          fld %st(n)
1016          fst %st(n)
1017          fst mem
1018          fstp %st(n)
1019          fucom %st(n)
1020          fucomp %st(n)
1021    
1022        All these are now generated when the fast fp mode is turned on.
1023    
1024    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2000/11/09 11:20:00 JST
1029    Tag: blume-20001109-scc
1030    Description:
1031    
1032    Eliminated some code duplication:
1033    
1034    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1035       (Otherwise the functor is useless.)
1036    2. Used GraphSCCFn where SCCUtilFun was used previously.
1037    3. Got rid of SCCUtilFun (in comp-lib.cm).
1038    
1039    ----------------------------------------------------------------------
1040    Name: Lal George
1041    Date: 2000/11/06 09:02:21 EST 2000
1042    Tag: Release_110_30
1043    Description:
1044    
1045    - Version 110.30
1046    ----------------------------------------------------------------------
1047    Name: Matthias Blume
1048    Date: 2000/11/04 14:45:00
1049    Tag: blume-20001104-mlbuild
1050    Description:
1051    
1052    - Made ml-build faster on startup.
1053    - Documentation fixes.
1054    
1055    ----------------------------------------------------------------------
1056    Name: Matthias Blume
1057    Date: 2000/11/02 17:00:00 JST
1058    Tag: blume-20001102-condcomp
1059    Description:
1060    
1061    - Small tweaks to pickler -- new BOOTFILES!
1062    - Version bumped to 110.29.2.
1063    - Added conditional compilation facility to init.cmi (see comment there).
1064    ----------------------------------------------------------------------
1065    Name: Allen Leung
1066    Date: 2000/10/23 19:31:00
1067    Tag: leunga-20001023-demo-ra
1068    
1069    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1070    2. Test programs for the graph library updated
1071    3. Some new MLRISC demo programs added
1072    
1073    ----------------------------------------------------------------------
1074    Name: Matthias Blume
1075    Date: 2000/08/31 22:15:00 JST
1076    Tag: blume-20001017-errmsg
1077    Description:
1078    
1079    More error message grief: Where there used to be no messages, there
1080    now were some that had bogus error regions.  Fixed.
1081    
1082    ----------------------------------------------------------------------
1083    Name: Matthias Blume
1084    Date: 2000/08/31 17:30:00 JST
1085    Tag: blume-20001017-v110p29p1
1086    Description:
1087    
1088    I made a version 110.29.1 with new bootfiles.
1089    
1090    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1091              CM documentation changes and a small bugfix in CM's error reporting.
1092    
1093    ----------------------------------------------------------------------
1094    Name: Lal George
1095    Date: 2000/09/27 14:42:35 EDT
1096    Tag: george-20000927-nodestatus
1097    Description:
1098    
1099    Changed the type of the nodestatus, so that:
1100    
1101            SPILLED(~1)             is now SPILLED
1102            SPILLED(m) where m>=0   is now MEMREG(m)
1103            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1104    
1105    ----------------------------------------------------------------------
1106    Name: Matthias Blume
1107    Date: 2000/09/07 14:45:00 JST
1108    Tag: blume-20000907-cmerrmsg
1109    Description:
1110    
1111    Small tweak to CM to avoid getting ML syntax error messages twice.
1112    
1113    ----------------------------------------------------------------------
1114    Name: Matthias Blume
1115    Date: 2000/08/31 18:00:00 JST
1116    Tag: blume-20000831-cvsbootfiles
1117    Description:
1118    
1119    New URL for boot files (because the 110.29 files on the BL server do
1120    now work correctly with my updated install scripts for yacc and lex).
1121    
1122    ----------------------------------------------------------------------
1123    Name: Matthias Blume
1124    Date: 2000/08/08 12:33:00 JST
1125    Tag: blume-20000808-manual
1126    Description:
1127    
1128    Tiny update to CM manual.
1129    
1130    ----------------------------------------------------------------------
1131    Name: Allen Leung
1132    Date: 2000/08/7 19:31:00
1133    Tag: leunga-20000807-a-whole-bunch-of-stuff
1134    
1135      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1136    and Mobdy.
1137    
1138    1.  x86
1139    
1140       a.  Fixes to peephole module by John and Dan.
1141       b.  Assembly fix to SETcc by Allen.
1142       c.  Fix to c-call by John.
1143       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1144       e.  Instruction selection optimization to SETcc as suggested by John.
1145    
1146           For example,
1147    
1148            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1149    
1150           should generate:
1151    
1152            MOVL a, x
1153            SUBL b, x
1154            SHRL 31, x
1155    
1156    2.  IR stuff
1157    
1158         A bunch of new DJ-graph related algorithms added.  These
1159         speed up SSA construction.
1160    
1161    3.  SSA + Scheduling
1162    
1163         Added code for SSA and scheduling to the repository
1164    
1165    ----------------------------------------------------------------------
1166    Name: Lal George
1167    Date: 2000/07/27 11:53:14 EDT
1168    
1169    Tag: lal-20000727-linux-ppc
1170    Description:
1171    
1172     Made changes to support Linux PPC.
1173     p.s. I have confirmation that the 110.29 boot files work fine.
1174    
1175    ----------------------------------------------------------------------
1176    Name: Matthias Blume
1177    Date: 2000/07/27 17:40:00 JST
1178    Tag: blume-20000727-scripts
1179    Description:
1180    
1181    !!!! WARNING !!!!
1182    You must recompile the runtime system!
1183    !!!! WARNING !!!!
1184    
1185    This is basically another round of script-enhancements:
1186    
1187    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1188       and undefine CM preprocessor symbols.
1189    
1190    2. ml-build avoids generating a new heap image if it finds that the
1191       existing one is still ok.  (The condition is that no ML file had to
1192       be recompiled and all ML files are found to be older that the heap
1193       file.)
1194    
1195       To make this work smoothly, I also hacked the runtime system as
1196       well as SMLofNJ.SysInfo to get access to the heap image suffix
1197       (.sparc-solaris, ...) that is currently being used.
1198    
1199       Moreover, the signature of CM.mk_standalone has changed.  See the
1200       CM manual.
1201    
1202    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1203       CM manual for details.)
1204    
1205    4. More CM manual updates:
1206        - all of the above has been documented.
1207        - there is now a section describing the (CM-related) command line
1208          arguments that are accepted by the "sml" command
1209    
1210    ----------------------------------------------------------------------
1211    Name: Matthias Blume
1212    Date: 2000/07/25 16:20:00 JST
1213    Tag: blume-20000725-makedepend
1214    Description:
1215    
1216    Added a script called ml-makedepend.  This can be used in makefiles
1217    for Unix' make in a way very similar to the "makedepend" command for
1218    C.
1219    
1220    The script internally uses function CM.sources.
1221    
1222    Synopsis:
1223    
1224        ml-makedepend [-f makefile] cmfile targetname
1225    
1226    The default for the makefile is "makefile" (or "Makefile" should
1227    "makefile" not exist).
1228    
1229    ml-makedepend adds a cmfile/targetname-specific section to this
1230    makefile (after removing the previous version of this section).  The
1231    section contains a single dependency specification with targetname on
1232    the LHS (targetname is an arbitrary name), and a list of files derived
1233    from the cmfile on the RHS.  Some of the files on the RHS are
1234    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1235    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1236    path names.  The makefile writer is responsible for making sure that
1237    these variables have correct at the time "make" is invoked.
1238    
1239    ----------------------------------------------------------------------
1240    Name: Matthias Blume
1241    Date: 2000/07/22 23:30:00 JST
1242    Tag: blume-20000722-urlupdate
1243    Description:
1244    
1245    Changed BOOT and config/srcarchiveurl to point to BL server:
1246    
1247        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1248    
1249    ----------------------------------------------------------------------
1250    Name: Matthias Blume
1251    Date: 2000/07/18 18:00:00 JST
1252    Tag: blume-20000718-Version_110_29
1253    Description:
1254    
1255    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1256    
1257    2. Updated config/version to 110.29
1258    
1259    3. Updated config/srcarchiveurl
1260    
1261    3. New boot files!
1262       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1263    
1264    ----------------------------------------------------------------------
1265    Name: Matthias Blume
1266    Date: 2000/07/11 13:58:00 JST
1267    Tag: blume-20000711-doctypo
1268    Description:
1269    
1270    Fixed a few typos in CM manual.
1271    
1272    ----------------------------------------------------------------------
1273    Name: Allen Leung
1274    Date: 2000/06/15 00:38:00
1275    Tag: leunga-20000704-sparc-x86
1276    
1277    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1278    2. fix to x86 RET bug [found by Dan Grossman]
1279    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1280    
1281       Affects c-- and moby only
1282    
1283    ----------------------------------------------------------------------
1284    Name: Matthias Blume
1285    Date: 2000/07/04 15:26:00
1286    Tag: blume-20000704-trigger
1287    Description:
1288    
1289    1. Improvements to CM manual.
1290    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1291       of getting a back-trace.  The function, when called, raises an
1292       internal exception which explicitly carries the full back-trace history,
1293       so it is unaffected by any intervening handle-raise pairs ("trivial"
1294       or not).  The interactive loop will print that history once it arrives
1295       at top level.
1296       Short of having all exceptions implicitly carry the full history, the
1297       recommended way of using this facility is:
1298         - compile your program with instrumentation "on"
1299         - run it, when it raises an exception, look at the history
1300         - if the history is "cut off" because of some handler, go and modify
1301           your program so that it explicitly calls BTrace.trigger
1302         - recompile (still instrumented), and rerun; look at the full history
1303    
1304    ----------------------------------------------------------------------
1305    Name: Matthias Blume
1306    Date: 2000/07/03 15:36:00 JST
1307    Tag: blume-20000702-manual
1308    Description:
1309    
1310    Small corrections and updates to CM manual.
1311    
1312    ----------------------------------------------------------------------
1313    Name: Matthias Blume
1314    Date: 2000/06/29 16:04:00 JST
1315    Tag: blume-20000629-yacctool
1316    Description:
1317    
1318    Changes:
1319    
1320    1. Class "mlyacc" now takes separate arguments to pass options to
1321       generated .sml- and .sig-files independently.
1322    2. Corresponding CM manual updates.
1323    3. BTrace module now also reports call sites.  (However, for loop clusters
1324       it only shows from where the cluster was entered.)  There are associated
1325       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1326    
1327    ----------------------------------------------------------------------
1328    Name: Matthias Blume
1329    Date: 2000/06/27 16:51:00 JST
1330    Tag: blume-20000627-noweb
1331    Description:
1332    
1333    Changes:
1334    
1335     1. Implemented "subdir" and "witness" options for noweb tool.
1336        This caused some slight internal changes in CM's tool implementation.
1337     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1338        some remaining issues from earlier path anchor changes.
1339     3. Updated CM manual accordingly.
1340    
1341     4. Changed implementation of back-tracing so that I now consider it
1342        ready for prime-time.
1343    
1344        In particular, you don't have to explicitly trigger the back-trace
1345        anymore.  Instead, if you are running BTrace-instrumented code and
1346        there is an uncaught exception (regardless of whether or not it was
1347        raised in instrumented code), the top-level evalloop will print
1348        the back-trace.
1349    
1350        Features:
1351    
1352          - Instrumented and uninstrumented code work together seemlessly.
1353            (Of course, uninstrumented code is never mentioned in actual
1354             back-traces.)
1355    
1356          - Asymptotic time- and space-complexity of instrumented code is
1357            equal to that of uninstrumented code.  (This means that
1358            tail-recursion is preserved by the instrumentation phase.)
1359    
1360          - Modules whose code has been instrumented in different sessions
1361            work together without problem.
1362    
1363          - There is no penalty whatsoever on uninstrumented code.
1364    
1365          - There is no penalty on "raise" expressions, even in
1366            instrumented code.
1367    
1368        A potential bug (or perhaps it is a feature, too):
1369    
1370          A back-trace reaches no further than the outermost instrumented
1371          non-trivial "raise".  Here, a "trivial" raise is one that is the
1372          sole RHS of a "handle" rule.  Thus, back-traces reach trough
1373    
1374               <exp> handle e => raise e
1375    
1376          and even
1377    
1378               <exp> handle Foo => raise Bar
1379    
1380          and, of course, through
1381    
1382               <exp> handle Foo => ...
1383    
1384         if the exception was not Foo.
1385    
1386         Back-traces always reach right through any un-instrumented code
1387         including any of its "handle" expressions, trivial or not.
1388    
1389       To try this out, do the following:
1390    
1391         - Erase all existing binfiles for your program.
1392           (You may keep binfiles for those modules where you think you
1393            definitely don't need back-tracing.)
1394         - Turn on back-trace instrumentation:
1395              SMLofNJ.Internals.BTrace.mode (SOME true);
1396         - Recompile your program.  (I.e., run "CM.make" or "use".)
1397         - You may now turn instrumentation off again (if you want):
1398              SMLofNJ.Internals.BTrace.mode (SOME false);
1399         - Run your program as usual.  If it raises an exception that
1400           reaches the interactive toplevel, then a back-trace will
1401           automatically be printed.  After that, the toplevel loop
1402           will print the exception history as usual.
1403    
1404    ----------------------------------------------------------------------
1405    Name: Matthias Blume
1406    Date: 2000/06/26 09:56:46 JST
1407    Tag: blume-20000626-setup
1408    Description:
1409    
1410    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1411          ML code before and after compiling a file (e.g., to set compiler
1412          flags)
1413    
1414    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1415              - associated changes to btrace.sml (BTrace instrumentation pass)
1416              - cleaner implementation of btimp.sml (BTrace tracing and report
1417                module)
1418    
1419    CM manual: * new path encoding documented
1420               * description of setup-parameter to "sml" added
1421    
1422    The biggest user-visible change to back-tracing is that it is no
1423    longer necessary to compile all traced modules within the same
1424    session.  (This was a real limitation.)
1425    
1426    ----------------------------------------------------------------------
1427    Name: Matthias Blume
1428    Date: 2000/06/24 12:40:00 JST
1429    Tag: blume-20000624-startup
1430    Description:
1431    
1432    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1433    bit too often -- to put it mildly. :)
1434    
1435    ----------------------------------------------------------------------
1436    Name: Matthias Blume
1437    Date: 2000/06/23 18:20:00 JST
1438    Tag: blume-20000623-btrace
1439    Description:
1440    
1441    This updates adds a backtrace facility to aid programmers in debugging
1442    their programs.  This involves the following changes:
1443    
1444    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1445       keeping track of the current call stack.  When programs are compiled
1446       in a special mode, the compiler will insert calls to these hooks
1447       into the user program.
1448       "Hook" means that it is possible for different implementations of
1449       back-tracing to register themselves (at different times).
1450    
1451    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1452       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1453       It can be turned on using this call:
1454         SMLofNJ.Internals.BTrace.mode (SOME true);
1455       Turning it off again:
1456         SMLofNJ.Internals.BTrace.mode (SOME false);
1457       Querying the current status:
1458         SMLofNJ.Internals.BTrace.mode NONE;
1459       Annotated programs are about twice as big as normal ones, and they
1460       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1461       where all hooks do nothing).  The slowdown with a plugin that is
1462       actually useful (such as the one supplied by default) is even greater,
1463       but in the case of the default plugin it is still only an constant
1464       factor (amortized).
1465    
1466    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1467       with a sub-structure BTrace for controlling back-tracing.  In particular,
1468       the above-mentioned function "mode" controls whether the annotation
1469       phase is invoked by the compiler.  Another important function is
1470       "trigger": when called it aborts the current execution and causes
1471       the top-level loop to print a full back-trace.
1472    
1473    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1474       for back-tracing.  It keeps track of the dynamic call stack and in
1475       addition to that it keeps a partial history at each "level" of that
1476       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1477       d and b (at separate times, dynamically), then the report will show:
1478    
1479       GOTO   d
1480             /c
1481       GOTO  \b
1482       CALL   a
1483    
1484       This shows that there was an initial non-tail call of a, then a
1485       tail-call to b or c, looping behavior in a cluster of functions that
1486       consist of b and c, and then a goto from that cluster (i.e., either from
1487       b or from c) to d.
1488    
1489       Note that (depending on the user program) the amount of information
1490       that the back-trace module has to keep track of at each level is bounded
1491       by a constant.  Thus, the whole implementation has the same asymptotical
1492       complexity as the original program (both in space and in time).
1493    
1494    5. compiler/TopLevel/interact/evalloop.sml has been modified to
1495       handle the special exception SMLofNJ.Internals.BTrace.BTrace
1496       which is raised by the "trigger" function mentioned above.
1497    
1498    Notes on usage:
1499    
1500    - Annotated code works well together with unannotated code:
1501    Unannotated calls simply do not show up at all in the backtrace.
1502    
1503    - It is not a good idea to let modules that were annotated during
1504    different sessions run at the same time.  This is because the compiler
1505    chooses small integers to identify individual functions, and there
1506    will be clashes if different modules were compiled in separate sessions.
1507    (Nothing will crash, and you will even be told about the clashes, but
1508    back-trace information will in general not be useful.)
1509    
1510    - Back-tracing can be confused by callcc and capture.
1511    
1512    - The only way of getting a back-trace right now is to explicitly
1513    invoke the "trigger" function from your user program.  Eventually, we
1514    should make every exception carry back-trace information (if
1515    available).  But since this creates more overhead at "raise"-time
1516    (similar to the current exnHistory overhead), I have not yet
1517    implemented this.  (The implementation will be rather easy.)  With
1518    exceptions carrying back-trace information, this facility will be even
1519    more useful because users don't need to modify their programs...
1520    
1521    - While it is possible to compile the compiler with back-trace
1522    annotations turned on (I did it to get some confidence in
1523    correctness), you must make absolutely sure that core.sml and
1524    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1525    be compiled with annotation because there is no core access yet, but
1526    if you compile btimp.sml with annotation, then the system will go into
1527    an infinite recursion and crash.)
1528    Since CM currently does not know about BTrace, the only way to turn
1529    annotations on and off for different modules of the compiler is to
1530    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1531    this is awkward and clumsy.
1532    
1533    Sample sessions:
1534    
1535    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1536    - SMLofNJ.Internals.BTrace.mode (SOME true);
1537    [autoloading]
1538    [autoloading done]
1539    val it = false : bool
1540    - structure X = struct
1541    -     fun main n = let
1542    -         fun a (x, 0) = d x
1543    -           | a (x, n) = b (x, n - 1)
1544    -         and b (x, n) = c (x, n)
1545    -         and c (x, n) = a (x, n)
1546    -         and d x = e (x, 3)
1547    -         and e (x, 0) = f x
1548    -           | e (x, n) = e (x, n - 1)
1549    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1550    -           | f n = n * g (n - 1)
1551    -         and g n = a (n, 3)
1552    -     in
1553    -         f n
1554    -     end
1555    - end;
1556    structure X : sig val main : int -> int end
1557    - X.main 3;
1558    *** BACK-TRACE ***
1559    GOTO   stdIn:4.2-13.20: X.main[2].f
1560    GOTO-( stdIn:4.2-13.20: X.main[2].e
1561    GOTO   stdIn:4.2-13.20: X.main[2].d
1562         / stdIn:4.2-13.20: X.main[2].a
1563         | stdIn:4.2-13.20: X.main[2].b
1564    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1565    CALL   stdIn:4.2-13.20: X.main[2].g
1566    GOTO   stdIn:4.2-13.20: X.main[2].f
1567    GOTO-( stdIn:4.2-13.20: X.main[2].e
1568    GOTO   stdIn:4.2-13.20: X.main[2].d
1569         / stdIn:4.2-13.20: X.main[2].a
1570         | stdIn:4.2-13.20: X.main[2].b
1571    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1572    CALL   stdIn:4.2-13.20: X.main[2].g
1573    GOTO   stdIn:4.2-13.20: X.main[2].f
1574    GOTO-( stdIn:4.2-13.20: X.main[2].e
1575    GOTO   stdIn:4.2-13.20: X.main[2].d
1576         / stdIn:4.2-13.20: X.main[2].a
1577         | stdIn:4.2-13.20: X.main[2].b
1578    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1579    CALL   stdIn:4.2-13.20: X.main[2].g
1580    GOTO   stdIn:4.2-13.20: X.main[2].f
1581    CALL   stdIn:2.15-17.4: X.main[2]
1582    -
1583    
1584    (Note that because of a FLINt bug the above code currently does not
1585    compile without BTrace turned on.)
1586    
1587    Here is another example, using my modified Tiger compiler:
1588    
1589    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1590    - SMLofNJ.Internals.BTrace.mode (SOME true);
1591    [autoloading]
1592    [autoloading done]
1593    val it = false : bool
1594    - CM.make "sources.cm";
1595    [autoloading]
1596    ...
1597    [autoloading done]
1598    [scanning sources.cm]
1599    [parsing (sources.cm):parse.sml]
1600    [creating directory CM/SKEL ...]
1601    [parsing (sources.cm):tiger.lex.sml]
1602    ...
1603    [wrote CM/sparc-unix/semant.sml]
1604    [compiling (sources.cm):main.sml]
1605    [wrote CM/sparc-unix/main.sml]
1606    [New bindings added.]
1607    val it = true : bool
1608    - Main.compile ("../testcases/merge.tig", "foo.out");
1609    *** BACK-TRACE ***
1610    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1611    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1612    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1613    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1614    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1615    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1616    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1617         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1618    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1619    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1620    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1621    CALL   main.sml:1.18-118.4: Main.compile[2]
1622    -
1623    
1624    ----------------------------------------------------------------------
1625    Name: Matthias Blumen
1626    Date: 2000/06/21 18:00:00 JST
1627    Tag: blume-20000621-manual
1628    Description:
1629    
1630    CM manual update: Path environments documented.
1631    
1632    ----------------------------------------------------------------------
1633    Name: Matthias Blume
1634    Date: 2000/06/19 13:40:00
1635    Tag: blume-20000619-manual
1636    Description:
1637    
1638    CM manual and system/README update.  This only covers the fact that
1639    there are no more implicit anchors.  (Path environments and the "bind"
1640    option to "cm" have yet to be documented.)
1641    
1642    ----------------------------------------------------------------------
1643    Name: Matthias Blume
1644    Date: 2000/06/19 11:05:00 JST
1645    Tag: blume-20000619-chdir-bugfix
1646    Description:
1647    
1648    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1649    
1650    ----------------------------------------------------------------------
1651    Name: Matthias Blume
1652    Date: 2000/06/18 22:00:10 JST
1653    Tag: blume-20000618-implicit-anchors-really-gone
1654    Description:
1655    
1656    I updates the previous HISTORY entry where I forgot to mention that
1657    implicit anchors are no longer with us.
1658    
1659    The current update also gets rid of the (now useless) controller
1660    CM.Control.implicit_anchors.
1661    
1662    ----------------------------------------------------------------------
1663    Name: Matthias Blume
1664    Date: 2000/06/16 17:30:00 JST
1665    Tag: blume-20000616-anchorenv
1666    Description:
1667    
1668    This patch implements the long anticipated (just kidding :) "anchor
1669    environment" mechanism.  In the course of doing this, I also
1670    re-implemented CM's internal "SrcPath" module from scratch.  The new
1671    one should be more robust in certain boundary cases.  In any case, it
1672    is a lot cleaner than its predecessor (IMHO).
1673    
1674    This time, although there is yet another boot file format change, I
1675    kept the unpickler backward-compatible.  As a result, no new bootfiles
1676    are necessary and bootstrapping is straightforward.  (You cannot read
1677    new bootfiles into an old system, but the other way around is no
1678    problem.)
1679    
1680    Visible changes:
1681    
1682    ** 0. Implicit path anchors (without the leading $-symbol) are no
1683    longer recognized at all. This means that such path names are not
1684    illegal either.  For example, the name basis.cm simply refers to a
1685    local file called "basis.cm" (i.e, the name is an ordinary path
1686    relative to .cm-files directory).  Or, to put it differently, only
1687    names that start with $ are anchored paths.
1688    
1689    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1690    vanished.
1691    
1692    John (Reppy) had critizised this as soon as I originally proposed and
1693    implemented it, but at that time I did not really deeply believe
1694    him. :) Now I came full-circle because I need the $<singlearc> syntax
1695    in another place where it cannot be seen as an abbreviation for
1696    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1697    seems to mean (i.e., it "expands" into the corresponding anchor
1698    value).
1699    
1700    However, when paths are used as members in CM description files, it
1701    continues to be true that there must be at least another arc after the
1702    anchor.  This is now enforced separately during semantic analysis
1703    (i.e., from a lexical/syntactical point of view, the notation is ok.)
1704    
1705    ** 2. The "cm" class now accepts an option "bind".  The option's value
1706    is a sub-option list of precisely two items -- one labeled "anchor"
1707    and the other one labeled "value".  As you might expect, "anchor" is
1708    used to specify an anchor name to be bound, and "value" specifies what
1709    the anchor is being bound to.
1710    
1711    The value must be a directory name and can be given in either standard
1712    syntax (including the possibility that it is itself an anchored path)
1713    or native syntax.
1714    
1715    Examples:
1716    
1717       foo.cm (bind:(anchor:bar value:$mystuff/bar))
1718       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1719    
1720    and so on.
1721    
1722    The meaning of this is that the .cm-file will be processed with an
1723    augmented anchor environment where the given anchor(s) is/are bound to
1724    the given values(s).
1725    
1726    The rationale for having this feature is this: Suppose you are trying
1727    to use two different (already stable) libraries a.cm and b.cm (that
1728    you perhaps didn't write yourself).  Further, suppose each of these
1729    two libraries internally uses its own auxiliary library $aux/lib.cm.
1730    Normally you would now have a problem because the anchor "lib" can not
1731    be bound to more than one value globally.  Therefore, the project that
1732    uses both a.cm and b.cm must locally redirect the anchor to some other
1733    place:
1734    
1735       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1736       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1737    
1738    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1739    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1740    
1741    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1742    you when you do so at the time of CM.stabilize).  Therefore, you can
1743    also use an anchored path as the value:
1744    
1745      a.cm (bind:(anchor:lib value:$a-lib))
1746      b.cm (bind:(anchor:lib value:$b-lib))
1747    
1748    Now you can globally configure (using the usual CM.Anchor.anchor or
1749    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1750    itself is always locally bound, setting it globally is no longer
1751    meaningful or necessary (but it does not hurt either).  In fact, "lib"
1752    can still be used as a global anchor for separate purposes.  As a
1753    matter of fact, one can locally define "lib" in terms of a global
1754    "lib":
1755    
1756      a.cm (bind:(anchor:lib value:$lib/a))
1757      b.cm (bind:(anchor:lib value:$lib/b))
1758    
1759    ** 3: The encoding of path names has changed.  This affects the way
1760    path names are shown in CM's progress report and also the internal
1761    protocol encoding used for parallel make.
1762    
1763    The encoding now uses one or more ':'-separated segments.  Each
1764    segments corresponds to a file that has been specified relative to the
1765    file given by its preceding segment.  The first segment is either
1766    relative to the CWD, absolute, or anchored.  Each segment itself is
1767    basically a Unix pathname; all segments but the first are relative.
1768    
1769    Example:
1770    
1771       $foo/bar/baz.cm:a/b/c.sml
1772    
1773    This path denotes the file bar/a/b/c.sml relative to the directory
1774    denoted by anchor "foo".  Notice that the encoding also includes
1775    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1776    paths are resolved relative to the .cm-files directory, so baz.cm must
1777    be ignored to get the "real" pathname.
1778    
1779    To make this fact more obvious, CM puts the names of such "virtual
1780    arcs" into parentheses when they appear in progress reports. (No
1781    parentheses will appear in the internal protocol encoding.)  Thus,
1782    what you really see is:
1783    
1784      $foo/bar/(baz.cm):a/b/c.sml
1785    
1786    I find this notation to be much more informative than before.
1787    
1788    Another new feature of the encoding is that special characters
1789    including parentheses, colons, (back)slashes, and white space are
1790    written as \ddd (where ddd is the decimal encoding of the character).
1791    
1792    *** The CM manual still needs to be updated.
1793    
1794    ----------------------------------------------------------------------
1795    Name: Allen Leung
1796    Date: 2000/06/15 00:38:00
1797    Tag: leunga-20000615-x86-peephole
1798    
1799    x86 Peephole fix by Fermin.  Affects c-- and moby only.
1800    
1801    ----------------------------------------------------------------------
1802    Name: Matthias Blume
1803    Date: 2000/06/12 11:40:00
1804    Tag: blume-20000612-parmakefix
1805    Description:
1806    
1807    More cleanup after changing the file naming scheme: This time I
1808    repaired the parallel make mechanism for CMB.make which I broke earlier.
1809    
1810    ----------------------------------------------------------------------
1811    Name: Allen Leung
1812    Date: 2000/06/09 01:25:00
1813    Tag: leunga-20000609-various
1814    
1815    None of these things should affect normal SML/NJ operations
1816    
1817    1. Peephole improvements provided by Fermin (c--)
1818    2. New annotation DEFUSE for adding extra dependence (moby)
1819    3. New X86 LOCK instructions (moby)
1820    4. New machine description language for reservation tables (scheduling)
1821    5. Fixes to various optimization/analysis modules (branch chaining, dominator
1822       trees etc.)
1823    6. I've changed the CM files so that they can work with versions
1824       110.0.6, 110.25 and 110.28
1825    
1826    ----------------------------------------------------------------------
1827    Name: Matthias Blume
1828    Date: 2000/06/09 12:40:00
1829    Tag: blume-20000609-log
1830    Description:
1831    
1832    - Removed all(?) remaining RCS Log entries from sources.
1833    
1834    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1835      anchored paths).
1836    
1837    ----------------------------------------------------------------------
1838    Name: Matthias Blume
1839    Date: 2000/06/07 17:00:00 JST
1840    Tag: blume-20000607-no-implicit-anchors
1841    Description:
1842    
1843    1. This update changes the default setting for
1844    CM.Control.implicit_anchors from true to false.  This means that
1845    implicit anchors are no longer permitted by default.  I also tried to
1846    make sure that nothing else still relies on implicit anchors.
1847    (This is the next step on the schedule towards a CM that does not even
1848    have the notion of implicit anchors anymore.)
1849    
1850    2. More CM manual updates.
1851    
1852    3. I managed to track down and fix the pickling bug I mentioned last
1853    time.  Because of the previously existing workaround, this entails no
1854    immediate practical changes.
1855    
1856    ----------------------------------------------------------------------
1857    Name: Matthias Blume
1858    Date: 2000/06/06 11:15:00 JST
1859    Tag: blume-20000606-lazierpickle
1860    Description:
1861    
1862    !!!! NEW BOOT FILES !!!!
1863    
1864    * The main purpose of this update is to make library pickles lazier in
1865    order to reduce the initial space penalty for autoloading a library.
1866    As a result, it is now possible to have $smlnj/compiler.cm
1867    pre-registered.  This should take care of the many complaints or
1868    inquiries about missing structure Compiler.  This required changes to
1869    CM's internal data structures and small tweaks to some algorithms.
1870    
1871    As a neat additional effect, it is no longer necessary (for the sake
1872    of lean heap image files) to distinguish between a "minimal" CM and a
1873    "full" CM.  Now, there is only one CM (i.e., the "full" version:
1874    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1875    interactive top level. ($smlnj/cm/minimal.cm is gone.)
1876    
1877    To make the life of compiler-hackers easier, "makeml" now also
1878    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1879    words, after you bootstrap a new sml for the first time, you will not
1880    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1881    around you will still have to do it, though.)
1882    
1883    * A second change consists of major updates to the CM manual.  There
1884    are now several appendices with summary information and also a full
1885    specification of the CM description file syntax.
1886    
1887    * In directory src/system I added the script "allcross".  This script
1888    invokes sml and cross-compiles the compiler for all supported
1889    architectures.  (Useful when providing a new set of boot files.)
1890    
1891    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
1892    added a small tweak to pickle-util.sml to work around this problem,
1893    but it is not a proper fix yet.  I will investigate further.  (The
1894    effect of the bug was an inflation of library pickle size.)
1895    
1896    * Version number increased to 110.28.1 (to avoid compatibility problems).
1897    
1898    ----------------------------------------------------------------------
1899    Name: Allen Leung
1900    Date: 2000/05/25 17:28 EDT
1901    Tag: leunga-20000525-ra
1902    Description:
1903    
1904      Fixed a bug in freezing phase of the register allocator.
1905    
1906    ----------------------------------------------------------------------
1907    Name: Allen Leung
1908    Date: 2000/05/15 22:53 EDT
1909    Tag: leunga-20000515-alpha-x86-ra
1910    Description:
1911    
1912      1. Alpha
1913    
1914          Slight cleanup.  Removed the instruction SGNXL
1915    
1916      2. X86
1917    
1918          Added the following instructions to the instruction set:
1919    
1920            ROLx, RORx,
1921            BTx, BTSx, BTLx, BTRx,
1922            XCHGx, and variants with the LOCK prefix
1923    
1924      3. Register Allocation
1925    
1926          The module ra-rewrite-with-renaming has been improved.
1927    
1928      These have no effect on SML/NJ.
1929    
1930    ----------------------------------------------------------------------
1931    Name: Matthias Blume
1932    Date: 2000/05/15 16:20:00 JST
1933    Tag: blume-20000515-lightrebuild
1934    Description:
1935    
1936    1. I added an alternative to "-rebuild" to "makeml".  The difference is
1937       that prior to calling CMB.make' the CM-variable "LIGHT" will be
1938       defined.  In effect, the command will not build any cross-compiler
1939       backends and therefore finish more quickly.
1940    
1941       The "fixpt" script also takes a "-light" switch to be able to use
1942       this new facility while compiling for a fixpoint.
1943    
1944    2. I replaced all mentions of anchored paths in group owner specifications
1945       with simple relative paths (usually starting with "..").
1946       The rationale is that a library's internal workings should not be
1947       compromised by the lack of some anchor.  (An anchor is necessary
1948       for someone who wants to refer to the library by an anchored path,
1949       but it should not be necessary to build the same library in the first
1950       place.)
1951    
1952    3. I changed the way CM's tool mechanism determines the shell command
1953       string used for things like ml-yacc etc. so that it does not break
1954       when CM.Control.implicit_anchors is turned off.
1955    
1956    ----------------------------------------------------------------------
1957    Name: Matthias Blume
1958    Date: 2000/05/12 18:20:00 JST
1959    Tag: blume-20000512-ml-build
1960    Description:
1961    
1962    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1963    getting installed properly (by config/install.sh).
1964    
1965    ----------------------------------------------------------------------
1966    Name: Matthias Blume
1967    Date: 2000/05/12 17:30:00 JST
1968    Tag: blume-20000512-anchors
1969    Description:
1970    
1971    !!! NEW BOOT FILES !!!
1972    
1973    This change is in preparation of fading out support for "implicitly
1974    anchored path names".  I went through all sources and used the
1975    explicit (and relatively new) $-notation.  See system/README and the
1976    CM manual for more info on this.
1977    
1978    I also modified the anchoring scheme for some things such as "smlnj",
1979    "MLRISC", "cm", etc. to take advantage of the fact that explicit
1980    anchors are more expressive: anchor name and first arc do not have to
1981    coincide.  This entails the following user-visible change:
1982    
1983    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1984    particular, when you fire up sml with a command-line argument, say,
1985    e.g.:
1986    
1987       sml '$smlnj/cmb.cm'
1988    
1989    At the ML toplevel prompt:
1990    
1991       CM.autoload "$smlnj/cmb.cm";
1992    
1993    There is also a new controller in CM.Control that can be used to turn
1994    off all remaining support for implicit anchors by saying:
1995    
1996        CM.autoload "$smlnj/
1997        #set CM.Control.implicit_anchors false;
1998    
1999    This causes CM to reject implicitly anchored paths.  This is (for the
2000    time being) less permissive than the "final" version where there will
2001    be no more such implicit anchors and relative paths will be just that:
2002    relative.
2003    
2004    The next step (version after next version?) will be to make the
2005    default for CM.Control.implicit_anchors false.  After the dust has
2006    settled, I can then produce the "final" version of this...
2007    
2008    Note: Since bootstrapping is a bit tricky, I provided new boot files.
2009    
2010    ----------------------------------------------------------------------
2011    Name: Matthias Blume
2012    Date: 2000/05/11 16:30:00 JST
2013    Tag: blume-20000511-sources
2014    Description:
2015    
2016    The main change is that I added function CM.sources as a generalized
2017    version of the earlier CM.makedepend.  This entails the following
2018    additional changes:
2019    
2020      - CM.makedepend has been dropped.
2021    
2022      - CM manual has been updated.
2023    
2024      - TOOLS signature and API have been changed.
2025    
2026    ----------------------------------------------------------------------
2027    Name: Allen Leung
2028    Date: 2000/05/10 21:17 EDT
2029    Tag: leunga-20000510-moby-c--ssa
2030    Description:
2031    
2032      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2033    None of these affect SML/NJ.
2034    
2035    1. Register Allocation
2036    
2037        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2038           This module tries to remove local (i.e. basic block level) redundancies
2039           during spilling.
2040    
2041        b. A new framework for performing region based register allocation.
2042           Not yet entirely functional.
2043    
2044    2. X86
2045    
2046       a. DefUse for POP was missing the stack pointer [found by Lal]
2047       b. Reload for CALL was incorrect in X86Spill [found by John]
2048       c. Various fixes in X86Spill so that it can be used correctly for
2049          the new spilling module.
2050    
2051    3. SSA/IR
2052    
2053       a. New module ir/dj-dataflow.sml implements elimination based
2054          data flow analysis.
2055    
2056    4. MLRiscGen
2057    
2058       a. Fix for gc type annotation
2059    
2060    5. MDGen
2061    
2062       Various fixes for machine description -> ml code translation.  For ssa
2063       only.
2064    
2065    ----------------------------------------------------------------------
2066    Name: Allen Leung
2067    Date: 2000/05/08 22:17 EDT
2068    Tag: leunga-20000508-labexp
2069    Description:
2070    
2071      Fermin has found a few assembly problems with constant expressions
2072      generated in LabelExp.  Mostly, the problems involve extra parentheses,
2073      which choke on dumb assemblers.  This is his fix.
2074    
2075    ----------------------------------------------------------------------
2076    Name: Dave MacQueen
2077    Date: 2000/04/09 14:00 EDT
2078    Tag: dbm-20000502-Version_110_28
2079    Description:
2080    
2081    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2082    
2083    2. Updated config/version to 110.28
2084    
2085    3. Updated config/srcarchiveurl
2086    
2087    3. New boot files!
2088       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2089    
2090    ----------------------------------------------------------------------
2091    Name: Matthias Blume
2092    Date: 2000/05/01 19:05:00 JST
2093    Tag: blume-20000501-noweb
2094    Description:
2095    
2096    A new noweb tool has been added.  The existing system is entirely
2097    unaffected by this, but some CM users have asked for renewed noweb
2098    support.  Everything is documented in the CM manual.
2099    
2100    New (plugin) libraries:
2101    
2102       noweb-tool.cm
2103       nw-ext.cm
2104    
2105    ----------------------------------------------------------------------
2106    Name: Dave MacQueen
2107    Date: 2000/04/30 12:40PM EDT
2108    Tag: dbm-20000430-bug_fixes
2109    Description:
2110    
2111    1. Fix for bug 1498
2112       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2113       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2114         added toRealArray function
2115       smlnj/src/compiler/MiscUtil/print/ppobj.sml
2116         added check for tag Obj.RealArray to array printing case in ppObj
2117    
2118    2. Fix for bug 1510
2119       smlnj/src/compiler/Semant/types/typesutil.sml
2120         fixed definition of dummyargs (used by equalTycon) so that
2121         dummy args are distinct types
2122    
2123    ----------------------------------------------------------------------
2124    Name: Matthias Blume
2125    Date: 2000/04/30 01:00:00 JST
2126    Tag: blume-20000430-versions
2127    Description:
2128    
2129    1. CM version numbering added.  This is an implementation of Lal's
2130       proposal for adding version numbers and version checking to .cm
2131       files.  Lal said that his proposal was just that -- a proposal.
2132       For the time being I went ahead and implemented it so that people
2133       can comment on it.  Everything is completely backward-compatible
2134       (except for the stable library format, i.e., new bootfiles!).
2135    
2136       As usual, see the CM manual for details.
2137    
2138    2. An alternative syntax for anchored paths has been implemented.
2139       Dave has recently voiced the same concerns that I had when I did
2140       this, so there should be some support.  My take is that eventually
2141       I will let support for the current syntax (where anchors are
2142       "implicit") fade out in favor of the new, explicit syntax.
2143       In order to be backward-compatible, both old and new syntax are
2144       currently supported.
2145    
2146       Again, see the CM manual for details.
2147    
2148    3. Parallel make is trying to be slightly smarter:  When the master
2149       process finds a "bottleneck", i.e., when there is only one
2150       compilation unit that can be compiled and everybody else is
2151       waiting on it, then it will simply compile it directly instead
2152       of clumsily telling one of the slaves to do it.
2153    
2154    4. Support for "unsharing" added.  This is necessary in order to be
2155       able to have two different versions of the same library running
2156       at the same time (e.g., for trying out a new MLRISC while still
2157       having the old MLRISC linked into the current compiler, etc.)
2158       See the CM manual.
2159    
2160    5. Simple "makedepend" functionality added for generating Makefile
2161       dependency information.  (This is rather crude at the moment.
2162       Expect some changes here in the future.)
2163    
2164    6. ".fun" added as a recognized suffix for ML files. Also documented
2165       explicitly in the manual that the fallback behavior (unknown suffix
2166       -> ML file) is not an official feature!
2167    
2168    7. Small changes to the pickler for stable libraries.
2169    
2170    8. Several internal changes to CM (for cleanup/improvement).
2171    
2172    
2173    !!!! NEW BINFILES !!!!
2174    
2175    ----------------------------------------------------------------------
2176    Name: Matthias Blume
2177    Date: 2000/04/28 17:30:00 JST
2178    Tag: blume-20000428-pathconfig
2179    Description:
2180    
2181    1. I changed config/install.sh to remove duplicate entries from the
2182       lib/pathconfig file at the end.  Moreover, the final version of
2183       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
2184       in src/system/installml.
2185    
2186    2. The config/install.sh script now consistently uses relative
2187       pathnames in lib/pathconfig whenever the anchor is in the lib
2188       directory.  (So far this was true for the libraries that come
2189       pre-compiled and bundled as part of the bootfiles but not for
2190       libraries that are compiled by the script itself.)
2191    
2192    ----------------------------------------------------------------------
2193    Name: Matthias Blume
2194    Date: 2000/04/26 13:10:00 JST
2195    Tag: blume-20000426-fun_suffix
2196    Description:
2197    
2198    Added ".fun" as a recognized file name suffix (for ML code).
2199    
2200    ----------------------------------------------------------------------
2201    Name: Allen Leung
2202    Date: 2000/04/25 17:00:00 EST
2203    Tag: leunga-20000425-alpha-ra
2204    Description:
2205    
2206    1. Alpha
2207    
2208        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
2209    in C--.
2210    
2211    2. RA
2212    
2213       Added a flag "ra-dump-size" to print out the size of the flowgraph
2214       and the interference graph.
2215    
2216    ----------------------------------------------------------------------
2217    Name: Dave MacQueen
2218    Date: 2000/04/25/
2219    Tag: dbm-20000425-mlyacc_doc_examples
2220    Description:
2221      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2222      Updated all three examples in src/ml-yacc/examples to run
2223      under 110.* using CM.make.
2224    
2225    ----------------------------------------------------------------------
2226    Name: Allen Leung
2227    Date: 2000/04/20 23:04:00 EST
2228    Tag: leunga-20000420-ssa-c---stuff
2229    Description:
2230    
2231      This update synchronizes my repository with Yale's.  Most of these
2232    changes, however, do not affect SML/NJ at all (the RA is an exception).
2233    
2234    1. Register Allocator
2235    
2236       a. An improvement in the interference graph construction:
2237          Given a copy
2238    
2239                s <- t
2240    
2241          no interference edge between s and t is added for this definition of s.
2242    
2243       b. I've added two new spill heuristic modules that Fermin and I developed
2244          (in the new library RA.cm). These are unused in SML/NJ but maybe
2245          useful for others (Moby?)
2246    
2247    2. X86
2248    
2249       a. Various fixes in the backend provided by Fermin [C--] and Lal.
2250    
2251    3. Alpha
2252    
2253       a. Added the BSR instruction and code generation that goes with it [C--]
2254       b. Other fixes too numerous to recount provided by Fermin [C--]
2255    
2256    4. Regmaps
2257    
2258       a. The regmaps are not initialized with the identity physical bindings
2259          at creation time.  This is unneeded.
2260    
2261    5. MLRISC Optimizations
2262    
2263       a. The DJ-Graph module can now compute the iterated dominance frontiers
2264          intersects with liveness incrementally in linear time! Woohoo!
2265          This is now used in my new SSA construction algorithm.
2266    
2267       b. THe branch reorganization module is now smarter about linear chains of
2268          basic blocks.
2269    
2270    
2271    ----------------------------------------------------------------------
2272    Name: Matthias Blume
2273    Date: 2000/04/12 13:52:00 JST
2274    Tag: blume_main_v110p27_1
2275    Description:
2276    
2277    Changed install.sh script to handle archive files without version number
2278    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2279    name of the boot file archive.
2280    
2281    ----------------------------------------------------------------------
2282    Name: Dave MacQueen
2283    Date: 2000/04/09 14:00 EDT
2284    Tag: dbm-20000410-Version_110_27
2285    Description:
2286    
2287    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2288    
2289    2. Updated src/config/version to 110.27
2290    
2291    3. New boot files!
2292    
2293    ----------------------------------------------------------------------
2294    Name: Allen Leung
2295    Date: 2000/04/09 19:09:00 EST
2296    Tag: leunga-20000409-misc
2297    Description:
2298    
2299    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
2300    
2301    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
2302    
2303    ----------------------------------------------------------------------
2304    Name: Stefan
2305    Date: 2000/04/07 10:00:00 EDT
2306    Tag: monnier-20000406-branch-handling
2307    Description:
2308    
2309    Improved handling of branches (mostly those generated from
2310    polymorphic equality), removed switchoff and changed the
2311    default optimization settings (more cpsopt and less flintopt).
2312    
2313    ----------------------------------------------------------------------
2314    Name: Allen Leung
2315    Date: 2000/04/06 01:30:00 EST
2316    Tag: leunga-20000406-peephole-x86-SSA-2
2317    Description:
2318    
2319       Forgot a few files.
2320    
2321    ----------------------------------------------------------------------
2322    Name: Allen Leung
2323    Date: 2000/04/06 00:36:00 EST
2324    Tag: leunga-20000406-peephole-x86-SSA
2325    Description:
2326    
2327    1.  New Peephole code
2328    
2329    2.  Minor improvement to X86 instruction selection
2330    
2331    3.  Various fixes to SSA and machine description -> code translator
2332    
2333    ----------------------------------------------------------------------
2334    Name: Matthias Blume
2335    Date: 2000/04/05 12:30:00 JST
2336    Tag: blume_main_v110p26p2_3
2337    Description:
2338    
2339    This update just merges three minor cosmetic updates to CM's sources
2340    to get ready for the 110.27 code freeze on Friday.  No functionality
2341    has changed.
2342    
2343    ----------------------------------------------------------------------
2344    Name: Allen Leung
2345    Date: 2000/04/04 19:39:00 EST
2346    Tag: leunga-20000404-x86-asm
2347    Description:
2348    
2349    1.  Fixed a problem in X86 assembly.
2350    
2351        Things like
2352    
2353           jmp %eax
2354           jmp (%eax)
2355    
2356        should be output as
2357    
2358           jmp *%eax
2359           jmp *(%eax)
2360    
2361    2.  Assembly output
2362    
2363          Added a new flag
2364    
2365              "asm-indent-copies" (default to false)
2366    
2367          When this flag is on, parallel copies will be indented an extra level.
2368    
2369    ----------------------------------------------------------------------
2370    Name: Allen Leung
2371    Date: 2000/04/04 03:18:00 EST
2372    Tag: leunga-20000404-C--Moby
2373    Description:
2374    
2375        All of these fixes are related to C--, Moby, and my own optimization
2376        stuff; so they shouldn't affect SML/NJ.
2377    
2378    1.  X86
2379    
2380        Various fixes related floating point, and extensions.
2381    
2382    2.  Alpha
2383    
2384        Some extra patterns related to loads with signed/zero extension
2385        provided by Fermin.
2386    
2387    3.  Assembly
2388    
2389        When generating assembly, resolve the value of client defined constants,
2390        instead of generating symbolic values.  This is controlled by the
2391        new flag "asm-resolve-constants", which is default to true.
2392    
2393    4.  Machine Descriptions
2394    
2395        a. The precedence parser was slightly broken when parsing infixr symbols.
2396        b. The type generalizing code had the bound variables reversed, resulting
2397           in a problem during arity raising.
2398        c. Various fixes in machine descriptions.
2399    
2400    ----------------------------------------------------------------------
2401    Name: Matthias Blume
2402    Date: 2000/04/03 16:05:00 JST
2403    Tag: blume_main_v110p26p2_2
2404    Description:
2405    
2406    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
2407    now done via the ordinary static environment that is context to each
2408    compilation unit.
2409    
2410    To this end, I arranged that instead of "structure Core" as "structure
2411    _Core" is bound in the pervasive environment.  Core access is done via
2412    _Core (which can never be accidentally rebound because _Core is not a
2413    legal surface-syntax symbol).
2414    
2415    The current solution is much cleaner because the core environment is
2416    now simply part of the pervasive environment which is part of every
2417    compilation unit's context anyway.  In particular, this eliminates all
2418    special-case handling that was necessary until now in order to deal
2419    with dynamic and symbolic parts of the core environment.
2420    
2421    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2422    compilation manager's bootstrap compiler (actually: in the "init group"
2423    handling).  See the comments in src/system/smlnj/init/init.cmi for
2424    more details.
2425    
2426    I also tried to track down all mentions of "Core" (as string argument
2427    to Symbol.strSymbol) in the compiler and replaced them with a
2428    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
2429    name appears in one place only.
2430    
2431    Binfile and bootfile format have not changed, but the switchover from
2432    the old "init.cmi" to the new one is a bit tricky, so I supplied new
2433    bootfiles anyway.
2434    
2435    ----------------------------------------------------------------------
2436    Name: Allen Leung
2437    Date: 2000/04/02 21:17:00 EST
2438    Tag: leunga-20000402-mltree
2439    Description:
2440    
2441       1. Renamed the constructor CALL in MLTREE by popular demand.
2442       2. Added a bunch of files from my repository.  These are currently
2443          used by other non-SMLNJ backends.
2444    
2445    ----------------------------------------------------------------------
2446    Name: Allen Leung
2447    Date: 2000/03/31 21:15:00 EST
2448    Tag: leunga-20000331-aliasing
2449    Description:
2450    
2451    This update contains a rewritten (and hopefully more correct) module
2452    for extracting aliasing information from CPS.
2453    
2454       To turn on this feature:
2455    
2456            Compiler.Control.CG.memDisambiguate := true
2457    
2458       To pretty print the region information with assembly
2459    
2460           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
2461    
2462       To control how many levels of aliasing information are printed, use:
2463    
2464           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
2465    
2466       The default of n is 3.
2467    
2468    ----------------------------------------------------------------------
2469    Name: David MacQueen
2470    Date: 2000/03/31 11:15:00 EST
2471    Tag: dbm-20000331-runtime_fix
2472    Description:
2473    
2474    This update contains:
2475    
2476    1. runtime/c-lib/c-libraries.c
2477       includes added in revision 1.2 caused compilation errors on hppa-hpux
2478    
2479    2. fix for bug 1556
2480       system/Basis/Implementation/NJ/internal-signals.sml
2481    
2482    ----------------------------------------------------------------------
2483    Name: Matthias Blume
2484    Date: 2000/03/31 18:00:00 JST
2485    Tag: blume_main_v110p26p2_1
2486    Description:
2487    
2488    This update contains:
2489    
2490    1. A small change to CM's handling of stable libraries:
2491       CM now maintains one "global" modmap that is used for all stable
2492       libraries.  The use of such a global modmap maximizes sharing and
2493       minimizes the need for re-traversing parts of environments during
2494       modmap construction.  (However, this has minor impact since modmap
2495       construction seems to account for just one percent or less of total
2496       compile time.)
2497    
2498    2. I added a "genmap" phase to the statistics.  This is where I got the
2499       "one percent" number (see above).
2500    
2501    3. CM's new tool parameter mechanism just became _even_ better. :)
2502       - The parser understands named parameters and recursive options.
2503       - The "make" and "shell" tools use these new features.
2504         (This makes it a lot easier to cascade these tools.)
2505       - There is a small syntax change: named parameters use a
2506    
2507           <name> : ( <option> ... )            or
2508           <name> : <string>
2509    
2510         syntax.  Previously, named parameters were implemented in an
2511         ad-hoc fashion by each tool individually (by parsing strings)
2512         and had the form
2513    
2514           <name>=<string>
2515    
2516       See the CM manual for a full description of these issues.
2517    
2518    ----------------------------------------------------------------------
2519    Name: Matthias Blume
2520    Date: 2000/03/30 18:00:00 JST
2521    Tag: blume_main_v110p26p2_0
2522    Description:
2523    
2524    !!!!! WARNING !!!!!!
2525    !!  New binfiles  !!
2526    !!!!!!!!!!!!!!!!!!!!
2527    
2528    This update contains:
2529    
2530    1. Moderate changes to CM:
2531    
2532       - Changes to CM's tools mechanism.  In particular, it is now possible
2533       to have tools that accept additional "command line" parameters
2534       (specified in the .cm file at each instance where the tool's class is
2535       used).
2536    
2537       This was done to accommodate the new "make" and "shell" tools which
2538       facilitate fairly seamless hookup to portions of code managed using
2539       Makefiles or Shell scripts.
2540    
2541       There are no classes "shared" or "private" anymore.  Instead, the
2542       sharing annotation is now a parameter to the "sml" class.
2543    
2544       There is a bit of generic machinery for implementing one's own
2545       tools that accept command-line parameters.  However, I am not yet fully
2546       satisfied with that part, so expect changes here in the future.
2547    
2548       All existing tools are described in the CM manual.
2549    
2550       - Slightly better error handling.  (CM now suppresses many followup
2551       error messages that tended to be more annoying than helpful.)
2552    
2553    2. Major changes to the compiler's static environment data structures.
2554    
2555       - no CMStaticEnv anymore.
2556            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
2557              but it is called Environment), no conversions between different
2558              kinds of static environments
2559    
2560       - There is still a notion of a "modmap", but such modmaps are generated
2561         on demand at the time when they are needed.  This sounds slow, but I
2562         sped up the code that generates modmaps enough for this not to lead to
2563         a slowdown of the compiler (at least I didn't detect any).
2564    
2565       - To facilitate rapid modmap generation, static environments now
2566         contain an (optional) "modtree" structure.  Modtree annotations are
2567         constructed by the unpickler during unpickling.  (This means that
2568         the elaborator does not have to worry about modtrees at all.)
2569         Modtrees have the advantage that they are compositional in the same
2570         way as the environment data structure itself is compositional.
2571         As a result, modtrees never hang on to parts of an environment that
2572         has already been rendered "stale" by filtering or rebinding.
2573    
2574       - I went through many, many trials and errors before arriving at the
2575         current solution.  (The initial idea of "linkpaths" did not work.)
2576         But the result of all this is that I have touched a lot of files that
2577         depend on the "modules" and "types" data structures (most of the
2578         elaborator). There were a lot of changes during my "linkpath" trials
2579         that could have been reverted to their original state but weren't.
2580         Please, don't be too harsh on me for messing with this code a bit more
2581         than what was strictly necessary...  (I _did_ resist the tempation
2582         of doing any "global reformatting" to avoid an untimely death at
2583         Dave's hands. :)
2584    
2585       - One positive aspect of the previous point:  At least I made sure that
2586         all files that I touched now compile without warnings (other than
2587         "polyEqual").
2588    
2589       - compiler now tends to run "leaner" (i.e., ties up less memory in
2590         redundant modmaps)
2591    
2592    ----------------------------------------------------------------------
2593    Name: Allen Leung
2594    Date: 2000/03/29 18:00:00
2595    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
2596    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
2597    Description:
2598    
2599       This update contains *MAJOR* changes to the way code is generated from CPS
2600    in the module mlriscGen, and in various backend modules.
2601    
2602    CHANGES
2603    =======
2604    
2605    1. MLRiscGen: forward propagation fix.
2606    
2607       There was a bug in forward propagation introduced at about the same time
2608       as the MLRISC x86 backend, which prohibits coalescing to be
2609       performed effectively in loops.
2610    
2611       Effect: speed up of loops in RISC architectures.
2612               By itself, this actually slowed down certain benchmarks on the x86.
2613    
2614    2. MLRiscGen:  forward propagating addresses from consing.
2615    
2616       I've changed the way consing code is generated.  Basically I separated
2617       out the initialization part:
2618    
2619            store tag,   offset(allocptr)
2620            store elem1, offset+4(allocptr)
2621            store elem2, offset+8(allocptr)
2622            ...
2623            store elemn, offset+4n(allocptr)
2624    
2625       and the address computation part:
2626    
2627            celladdr <- offset+4+alloctpr
2628    
2629       and move the address computation part
2630    
2631       Effect:  register pressure is generally lower as a result.  This
2632                makes compilation of certain expressions much faster, such as
2633                long lists with non-trivial elements.
2634    
2635                 [(0,0), (0,0), .... (0,0)]
2636    
2637    3. MLRiscGen: base pointer elimination.
2638    
2639        As part of the linkage mechanism, we generate the sequence:
2640    
2641         L:  ...  <- start of the code fragment
2642    
2643         L1:
2644             base pointer <- linkreg - L1 + L
2645    
2646         The base pointer was then used for computing relocatable addresses
2647       in the code fragment.  Frequently (such as in lots of continuations)
2648       this is not needed.  We now eliminate this sequence whenever possible.
2649    
2650         For compile time efficiency, I'm using a very stupid local heuristic.
2651       But in general, this should be done as a control flow analysis.
2652    
2653       Effect:  Smaller code size.  Speed up of most programs.
2654    
2655    4. Hppa back end
2656    
2657         Long jumps in span dependence resolution used to depend on the existence
2658      of the base pointer.
2659    
2660         A jump to a long label L was expanded into the following sequence:
2661    
2662          LDIL %hi(L-8192), %r29
2663          LDO  %lo(L-8192)(%r29), %r29
2664          ADD  %r29, baseptr, %r29
2665          BV,n %r0(%r29)
2666    
2667         In the presence of change (3) above, this will not work.  I've changed
2668       it so that the following sequence of instructions are generated, which
2669       doesn't mention the base pointer at all:
2670    
2671             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
2672        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
2673             BV,n  %r0(%r29)          /* Jump */
2674    
2675    5. Alpha back end
2676    
2677          New alpha instructions LDB/LDW have been added, as per Fermin's
2678       suggestions.   This is unrelated to all other changes.
2679    
2680    6. X86 back end
2681    
2682         I've changed andl to testl in the floating point test sequence
2683         whenever appropriate.  The Intel optimization guide states that
2684         testl is preferable to andl.
2685    
2686    7. RA (x86 only)
2687    
2688         I've improved the spill propagation algorithm, using an approximation
2689       of maximal weighted independent sets.   This seems to be necessary to
2690       alleviate the negative effect in light of the slow down in (1).
2691    
2692         I'll write down the algorithm one of these days.
2693    
2694    8. MLRiscGen: frequencies
2695    
2696         I've added an annotation that states that all call gc blocks have zero
2697       execution frequencies.  This improves register allocation on the x86.
2698    
2699    BENCHMARKS
2700    ==========
2701    
2702       I've only perform the comparison on 110.25.
2703    
2704       The platforms are:
2705    
2706        HPPA  A four processor HP machine (E9000) with 5G of memory.
2707        X86   A 300Hhz Pentium II with 128M of memory, and
2708        SPARC An Ultra sparc 2 with 512M of memory.
2709    
2710       I used the following parameters for the SML benchmarks:
2711    
2712                 @SMLalloc
2713         HPPA    256k
2714         SPARC   512k
2715         X86     256k
2716    
2717    COMPILATION TIME
2718    ----------------
2719       Here are the numbers comparing the compilation times of the compilers.
2720       I've only compared 110.25 compiling the new sources versus
2721       a fixpoint version of the new compiler compiling the same.
2722    
2723                     110.25                                  New
2724               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
2725         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
2726         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
2727         X86    999s    315s       94006+130691   987s    296s    108877+141957
2728    
2729                   110.25         New
2730                Code Size      Code Size
2731         HPPA   8596736         8561421
2732         SPARC  8974299         8785143
2733         X86    9029180         8716783
2734    
2735       So in summary, things are at least as good as before.   Dramatic
2736       reduction in compilation is obtained on the Sparc; I can't explain it,
2737       but it is reproducible.  Perhaps someone should try to reproduce this
2738       on their own machines.
2739    
2740    SML BENCHMARKS
2741    --------------
2742    
2743        On the average, all benchmarks perform at least as well as before.
2744    
2745          HPPA         Compilation Time     Spill+Reload      Run Time
2746                     110.25  New            110.25    New   110.25  New
2747    
2748          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
2749              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
2750       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
2751                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
2752        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
2753             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
2754               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
2755              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
2756         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
2757             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
2758            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
2759      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
2760                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
2761             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
2762                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
2763               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
2764      --------------------------------------------------------------------------
2765       Average                     6.12%                                   4.09%
2766    
2767          SPARC        Compilation Time     Spill+Reload      Run Time
2768                     110.25  New            110.25    New   110.25  New
2769    
2770          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
2771              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
2772       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
2773                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
2774        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
2775             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
2776               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
2777              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
2778         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
2779             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
2780            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
2781      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
2782                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
2783             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
2784                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
2785               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
2786      ----------------------------------------------------------------------------
2787       Average                     6.94%                                    2.64%
2788    
2789          X86          Compilation Time     Spill+Reload      Run Time
2790                     110.25  New            110.25    New   110.25  New
2791    
2792          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
2793              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
2794       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
2795                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
2796        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
2797             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
2798               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
2799              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
2800         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
2801             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
2802            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
2803      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
2804                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
2805             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
2806                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
2807               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
2808      ----------------------------------------------------------------------------
2809       Average                     1.22%                                     3.36%
2810    
2811    ----------------------------------------------------------------------
2812    Name: Allen Leung
2813    Date: 2000/03/23 16:25:00
2814    Tag: leunga-20000323-fix_x86_alpha
2815    Description:
2816    
2817    1. X86 fixes/changes
2818    
2819       a.  The old code generated for SETcc was completely wrong.
2820           The Intel optimization guide is VERY misleading.
2821    
2822    2. ALPHA fixes/changes
2823    
2824       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2825       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2826       c.  Added reassociation code for address computation.
2827    
2828  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2829  Name: Allen Leung  Name: Allen Leung
2830  Date: 2000/03/22 01:23:00  Date: 2000/03/22 01:23:00
# Line 49  Line 2863 
2863    
2864  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2865  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
2866  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
2867  explanation.  explanation.
2868    
2869  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
2870  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2871  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
2872  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 96  Line 2910 
2910  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
2911  <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,
2912  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
2913  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
2914  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
2915  and, thus, saves a lot of time.  and, thus, saves a lot of time.
2916    
# Line 146  Line 2960 
2960    it from that remote directory.    it from that remote directory.
2961    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
2962    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
2963    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
2964    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
2965    
2966    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.580  
changed lines
  Added in v.875

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