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 952, Tue Oct 9 14:00:40 2001 UTC revision 1054, Wed Feb 6 19:27:27 2002 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name: Lal George  Name: Lal George
17    Date: 2002/02/06 13:55:02 EST
18    Tag: george-20020206-ra-breakup
19    Description:
20    
21    1. A bug fix from Allen.
22    
23        A typo causes extra fstp %st(0)'s to be generated at compensation
24        edges, which might cause stack underflow traps at runtime.  This
25        occurs in fft where there are extraneous fstps right before the 'into'
26        trap instruction (in this case they are harmless since none of the
27        integers overflow.)
28    
29    2. Pulled out various utility modules that were embedded in the modules
30       of the register allocator. I need these modules for other purposes, but
31       they are not complete enough to put into a library (just yet).
32    ----------------------------------------------------------------------
33    Name: Matthias Blume
34    Date: 2002/01/31 16:05:00 EST
35    Tag: blume-20020131-sparc-ccalls
36    Description:
37    
38    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
39       of extra stack space by mistake.  Fixed.
40    
41    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
42    
43    ----------------------------------------------------------------------
44    Name: Allen Leung
45    Date: 2002/01/30
46    Tag: leunga-20020130-nowhere-bug-fix
47    Description:
48    
49       MLRISC bug fixes:
50       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
51       2. MachineInt.fromString was negating its value.
52    
53    ----------------------------------------------------------------------
54    Name: Matthias Blume
55    Date: 2002/01/29
56    Tag: blume-20020129-INSTALL
57    Description:
58    
59    - Added somewhat detailed installation instructions (file INSTALL).
60    - Fixed curl-detection bug in config/install.sh.
61    - It is now possible to select the URL getter using the URLGETTER
62      environment variable:
63    
64          not set / "unknown"      --> automatic detection (script tries wget,
65                                       curl, and lynx)
66          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
67                                       how to properly invoke them)
68          other                    --> use $URLGETTER directly, it must take
69                                       precisely two command-line arguments
70                                       (source URL and destination file name)
71    
72    ----------------------------------------------------------------------
73    Name: Matthias Blume
74    Date: 2002/01/28
75    Tag: blume-20020128-sparc-ccalls
76    Description:
77    
78    - Fixed problem with calculation of "used" registers in sparc-c-calls.
79    - Make use of the allocParam argument in sparc-c-calls.
80    
81    ----------------------------------------------------------------------
82    Name: Matthias Blume
83    Date: 2002/01/28
84    Tag: blume-20020128-allocParam
85    Description:
86    
87    John Reppy:  Changes c-calls API to accept client-callback for
88    allocating extra stack space.
89    me: Corresponding changes to mlriscGen (using a dummy argument that
90        does not change the current behavior).
91    
92    ----------------------------------------------------------------------
93    Name: Matthias Blume
94    Date: 2002/01/28 12:00:00
95    Tag: Release_110_38
96    Description:
97    
98    This time for real!!!
99    
100    ----------------------------------------------------------------------
101    Name: Matthias Blume
102    Date: 2002/01/28 10:56:00 EST
103    Tag: blume-20020128-retraction
104    Description:
105    
106    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
107       with blume-Release_110_38-retracted.)
108    
109    1. Fixed a problem with incorrect rounding modes in real64.sml.
110       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
111    
112    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
113       and enums fixed.  The naming of corresponding ML identifiers should
114       now be consistent again.
115    
116    ----------------------------------------------------------------------
117    Name: Allen Leung
118    Date: 2002/01/27
119    Tag: leunga-20020127-nowhere
120    Description:
121    
122       Added a target called nowhere in the configuration scripts.
123       Enabling this will build the MLRISC 'nowhere' tool (for translating
124       programs with where-clauses into legal SML code) during installation.
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume
128    Date: 2002/01/25 21:27:00 EST
129    Tag: blume-Release_110_38-retracted
130    Description:
131    
132    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
133    
134    README will be added later.
135    
136    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
137    (Release_110_38) removed.  Reason: Last-minute bug fixes.
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume
141    Date: 2002/01/25
142    Tag: blume-20020125-ffi
143    Description:
144    
145    A large number of tweaks and improvements to ml-nlffi-lib and
146    ml-nlffigen:
147    
148       - ML represenation types have been streamlined
149       - getter and setter functions work with concrete values, not abstract
150         ones where possible
151       - ml-nlffigen command line more flexible (see README file there)
152       - some bugs have been fixed (hopefully)
153    
154    ----------------------------------------------------------------------
155    Name: Lal George
156    Date: 2002/01/24
157    Tag: george-20020124-risc-ra-interface
158    Description:
159    
160       There is a dramatic simplification in the interface to the
161       register allocator for RISC architectures as a result of making
162       parallel copy instructions explicit.
163    
164    ----------------------------------------------------------------------
165    Name: Matthias Blume
166    Date: 2002/01/22
167    Tag: blume-20020122-x86-ccalls
168    Description:
169    
170    Bug fix for c-calls on x86 (having to do with how char- and
171    short-arguments are being handled).
172    
173    ----------------------------------------------------------------------
174    Name: Matthias Blume
175    Date: 2002/01/21
176    Tag: blume-20020121-ff
177    Description:
178    
179    Another day of fiddling with the FFI...
180    
181    1. Bug fix/workaround:  CKIT does not complain about negative array
182       dimensions, so ml-nlffigen has to guard itself against this possibility.
183       (Otherwise a negative dimension would send it into an infinite loop.)
184    
185    2. Some of the abstract types (light objects, light pointers, most "base"
186       types) in structure C are now eqtypes.
187    
188    3. Added constructors and test functions for NULL function pointers.
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume
192    Date: 2002/01/18
193    Tag: blume-20020118-ready-for-new-release
194    Description:
195    
196    Made config/srcarchiveurl point to a new place.  (Will provide boot
197    files shortly.)
198    
199    Maybe we christen this to be 110.38?
200    
201    ----------------------------------------------------------------------
202    Name: Matthias Blume
203    Date: 2002/01/18
204    Tag: blume-20020118-more-ffifiddle
205    Description:
206    
207    Today's FFI fiddling:
208    
209      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
210        C.Set.* that use concrete (MLRep.*) arguments and results instead
211        of abstract ones.
212    
213      - Provided word-style bit operations etc. for "int" representation
214        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
215        and Long.
216    
217    ----------------------------------------------------------------------
218    Name: Matthias Blume
219    Date: 2002/01/18
220    Tag: blume-20020118-use-x86-fp
221    Description:
222    
223    Now that x86-fast-fp seems to be working, I turned it back on again
224    by default.  (Seems to work fine now, even with the FFI.)
225    
226    Other than that, I added some documentation about the FFI to
227    src/ml-nlffigen/README and updated the FFI test examples in
228    src/ml-nlffi-lib/Tests/*.
229    
230    ----------------------------------------------------------------------
231    Name: Allen Leung
232    Date: 2002/01/17
233    Tag: leunga-20020117-x86-fast-fp-call
234    Description:
235    
236       1. Fixed a problem with handling return fp values when x86's fast fp
237          mode is turned on.
238    
239       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
240          of %f32.
241    
242       3. Added a constructor INT32lit to the ast of MLRISC tools.
243    
244    ----------------------------------------------------------------------
245    Name: Matthias Blume
246    Date: 2002/01/16
247    Tag: blume-20020116-ffifiddle
248    Description:
249    
250    More fiddling with the FFI interface:
251    
252     - Make constness 'c instead of rw wherever possible.  This eliminates
253       the need for certain explicit coercions.  (However, due to ML's
254       value polymorphism, there will still be many cases where explicit
255       coercions are necessary.  Phantom types are not the whole answer
256       to modeling a subtyping relationship in ML.)
257    
258     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
259    
260     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
261       clumsy workaround for varargs problem.)
262    
263    ----------------------------------------------------------------------
264    Name: Lal George
265    Date: 2002/01/15
266    Tag: <none>
267    Description:
268    
269    1. Since COPY instructions are no longer native to the architecture,
270       a generic functor can be used to implement the expandCopies function.
271    
272    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
273       TEXT segment.
274    
275    ----------------------------------------------------------------------
276    Name: Matthias Blume
277    Date: 2002/01/15
278    Tag: blume-20020115-ffiupdates
279    Description:
280    
281    1. Fix for bug resulting in single-precision float values being returned
282       incorrectly from FFI calls.
283    
284    2. Small modifications to C FFI API:
285    
286        - memory-allocation routines return straight objects (no options)
287          and raise an exception in out-of-memory situations
288        - unsafe extensions to cast between function pointers and pointers
289          from/to ints
290        - added structure C_Debug as an alternative to structure C where
291          pointer-dereferencing (|*| and |*!) always check for null-pointers
292        - added open_lib' to DynLinkage;  open_lib' works like open_lib
293          but also takes a (possibly empty) list of existing library handles
294          that the current library depends on
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume
298    Date: 2002/01/10
299    Tag: blume-20020110-newffigen
300    Description:
301    
302    1. Updates to portable graph code.
303    
304    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
305       more scalable now so that even huge interfaces such as the one
306       for GTK compile in finite time and space. :-)
307       See src/ml-nlffigen/README for details on what's new.
308    
309    ----------------------------------------------------------------------
310    Name: Lal George
311    Date: 2001/01/09 14:31:35 EST 2002
312    Tag: george-20011206-rm-native-copy
313    Description:
314    
315            Removed the native COPY and FCOPY instructions
316            from all the architectures and replaced it with the
317            explicit COPY instruction from the previous commit.
318    
319            It is now possible to simplify many of the optimizations
320            modules that manipulate copies. This has not been
321            done in this change.
322    
323    ----------------------------------------------------------------------
324    Name: Lal George
325    Date: 2001/12/06 16:50:13 EST 2001
326    Tag: george-20011206-mlrisc-instruction
327    Description:
328    
329    Changed the representation of instructions from being fully abstract
330    to being partially concrete. That is to say:
331    
332      from
333            type instruction
334    
335      to
336            type instr                              (* machine instruction *)
337    
338            datatype instruction =
339                LIVE of {regs: C.cellset, spilled: C.cellset}
340              | KILL of {regs: C.cellset, spilled: C.cellset}
341              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
342              | ANNOTATION of {i: instruction, a: Annotations.annotation}
343              | INSTR of instr
344    
345    This makes the handling of certain special instructions that appear on
346    all architectures easier and uniform.
347    
348    LIVE and KILL say that a list of registers are live or killed at the
349    program point where they appear. No spill code is generated when an
350    element of the 'regs' field is spilled, but the register is moved to
351    the 'spilled' (which is present, more for debugging than anything else).
352    
353    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
354    We used to generate:
355    
356            DEFFREG f1
357            f1 := f2 + f3
358            trapb
359    
360    but now generate:
361    
362            f1 := f2 + f3
363            trapb
364            LIVE {regs=[f1,f2,f3], spilled=[]}
365    
366    Furthermore, the DEFFREG (hack) required that all floating point instruction
367    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
368    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
369    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
370    this architecture).
371    
372    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
373    all the architectures. This will result in further simplification of the
374    register allocator that must be aware of them for coalescing purposes, and
375    will also simplify certain aspects of the machine description that provides
376    callbacks related to parallel copies.
377    
378    ANNOTATION should be obvious, and now INSTR represents the honest to God
379    machine instruction set!
380    
381    The <arch>/instructions/<arch>Instr.sml files define certain utility
382    functions for making porting easier -- essentially converting upper case
383    to lower case. All machine instructions (of type instr) are in upper case,
384    and the lower case form generates an MLRISC instruction. For example on
385    the alpha we have:
386    
387      datatype instr =
388         LDA of {r:cell, b:cell, d:operand}
389       | ...
390    
391      val lda : {r:cell, b:cell, d:operand} -> instruction
392        ...
393    
394    where lda is just (INSTR o LDA), etc.
395    
396    ----------------------------------------------------------------------
397    Name: Matthias Blume
398    Date: 2001/11/22 21:40:00 EST
399    Tag: Release_110_37
400    Description:
401    
402    Release 110.37.  This time for real.
403    
404    ----------------------------------------------------------------------
405    Name: Matthias Blume
406    Date: 2001/11/21 16:35:00 EST
407    Tag: blume-20011121-foot-in-mouth
408    Description:
409    
410    Removed the "Release_110_37" tag because of a serious bug.
411    This will be re-tagged once the bug is fixed.
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume
415    Date: 2001/11/21 16:14:00 EST
416    Tag: blume-20011121-forgottenfile
417    Description:
418    
419    Forgot to add a file.  (Just a .tex-file -- part of
420    the CM manual source.)
421    
422    ----------------------------------------------------------------------
423    Name: Matthias Blume
424    Date: 2001/11/21 16:10:00 EST
425    Tag: blume-20011121-invalid_110_37
426    Description:
427    
428    Note: I removed the original tag "Release_110_37" from this commit
429          because we found a serious bug in all non-x86 backends.
430          - Matthias
431    
432    1. Modifications to the SML/NJ code generator and to the runtime system
433       so that code object name strings are directly inserted into code
434       objects at code generation time.  The only business the runtime system
435       has with this is now to read the name strings on occasions.
436       (The encoding of the name string has also changed somewhat.)
437    
438    2. CM now implements a simple "set calculus" for specifying export lists.
439       In particular, it is now possible to refer to the export lists of
440       other libraries/groups/sources and form unions as well as differences.
441       See the latest CM manual for details.
442    
443    3. An separate notion of "proxy" libraries has again be eliminated from
444       CM's model.  (Proxy libraries are now simply a special case of using
445       the export list calculus.)
446    
447    4. Some of the existing libraries now take advantage of the new set
448       calculus.
449       (Notice that not all libraries have been converted because some
450       of the existing .cm-files are supposed to be backward compatible
451       with 110.0.x.)
452    
453    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
454       "General.exnMessage"!  The former relies on a certain hook to be
455       initialized, and that often does not happen in the stand-alone case.)
456    
457    ----------------------------------------------------------------------
458    Name: Lal George
459    Date: 2001/11/21  13:56:18 EST
460    Tag: george-2001121-pseudo-ops
461    Description:
462    
463      Implemented a complete redesign of MLRISC pseudo-ops. Now there
464      ought to never be any question of incompatabilities with
465      pseudo-op syntax expected by host assemblers.
466    
467      For now, only modules supporting GAS syntax are implemented
468      but more should follow, such as MASM, and vendor assembler
469      syntax, e.g. IBM as, Sun as, etc.
470    
471    ----------------------------------------------------------------------
472    Name: Matthias Blume
473    Date: 2001/11/14 11:52:00 EST
474    Tag: blume-20011114-srcname
475    Description:
476    
477    1. Routed the name of the current source file to mlriscgen where it
478       should be directly emitted into the code object.  (This last part
479       is yet to be done.)
480    
481    2. Some cleanup of the pgraph code to make it match the proposal that
482       I put out the other day.  (The proposal notwithstanding, things are
483       still in flux here.)
484    
485    ----------------------------------------------------------------------
486    Name: Lal George
487    Date: 2001/11/14 09:44:04 EST
488    Tag:
489    Description:
490    
491      Fix for a backpatching bug reported by Allen.
492    
493      Because the boundary between short and long span-dependent
494      instructions is +/- 128, there are an astounding number of
495      span-dependent instructions whose size is over estimated.
496    
497      Allen came up with the idea of letting the size of span
498      dependent instructions be non-monotonic, for a maxIter
499      number of times, after which the size must be monotonically
500      increasing.
501    
502      This table shows the number of span-dependent instructions
503      whose size was over-estimated as a function of maxIter, for the
504      file Parse/parse/ml.grm.sml:
505    
506         maxIter            # of instructions:
507            10                      687
508            20                      438
509            30                      198
510            40                        0
511    
512      In compiling the compiler, there is no significant difference in
513      compilation speed between maxIter=10 and maxIter=40. Actually,
514      my measurements showed that maxIter=40 was a tad faster than
515      maxIter=10! Also 96% of the  files in the compiler reach a fix
516      point within 13 iterations, so fixing maxIter at 40, while high,
517      is okay.
518    
519    ----------------------------------------------------------------------
520    Name: Matthias Blume
521    Date: 2001/10/31 15:25:00 EST
522    Tag: blume-20011031-pgraph
523    Description:
524    
525    CKIT:
526    * Changed the "Function" constructor of type Ast.ctype to carry optional
527      argument identifiers.
528    * Changed the return type of TypeUtil.getFunction accordingly.
529    * Type equality ignores the argument names.
530    * TypeUtil.composite tries to preserve argument names but gives up quickly
531      if there is a mismatch.
532    
533    installation script:
534    * attempts to use "curl" if available (unless "wget" is available as well)
535    
536    CM:
537    * has an experimental implementation of "portable graphs" which I will
538      soon propose as an implementation-independent library format
539    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
540    
541    NLFFI-LIB:
542    * some cleanup (all cosmetic)
543    
544    NLFFIGEN:
545    * temporarily disabled the mechanism that suppresses ML output for
546      C definitions whose identifiers start with an underscore character
547    * generate val bindings for enum constants
548    * user can request that only one style (light or heavy) is being used;
549      default is to use both (command-line arguments: -heavy and -light)
550    * fixed bug in handling of function types involving incomplete pointers
551    * generate ML entry points that take record arguments (i.e., using
552      named arguments) for C functions that have a prototype with named
553      arguments
554      (see changes to CKIT)
555    
556    ----------------------------------------------------------------------
557    Name: Allen Leung
558    Date: 2001/10/27 20:34:00 EDT
559    Tag: leunga-20011027-x86-fast-fp-call
560    Description:
561    
562       Fixed the bug described in blume-20010920-slowfp.
563    
564       The fix involves
565          1. generating FCOPYs in FSTP in ia32-svid
566          2. marking a CALL with the appropriate annotation
567    
568    ----------------------------------------------------------------------
569    Name: Matthias Blume
570    Date: 2001/10/16 11:32:00 EDT
571    Tag: blume-20011016-netbsd
572    Description:
573    
574    Underscore patch from Chris Richards (fixing problem with compiling
575    runtime system under recent NetBSD).
576    
577    ----------------------------------------------------------------------
578    Name: Allen Leung
579    Date: 2001/10/12 17:18:32 EDT 2001
580    Tag: leung-20011012-x86-printflowgraph
581    Description:
582    
583    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
584    
585    ----------------------------------------------------------------------
586    Name: Lal George
587    Date: 2001/10/11 23:51:34 EDT
588    Tag: george-20011011-too-many-instrs
589    Description:
590    
591    The representation of a program point never expected to see more
592    than 65536 instructions in a basic block!
593    
594    ----------------------------------------------------------------------
595    Name: Lal George
596  Date: 2001/10/09 09:41:37 EDT  Date: 2001/10/09 09:41:37 EDT
597  Tag: george-20011008-mlrisc-labels  Tag: george-20011008-mlrisc-labels
598  Description:  Description:
# Line 71  Line 650 
650    
651  ----------------------------------------------------------------------  ----------------------------------------------------------------------
652  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
653  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
654  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
655  Description:  Description:

Legend:
Removed from v.952  
changed lines
  Added in v.1054

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