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 937, Thu Sep 20 21:20:28 2001 UTC revision 1055, Thu Feb 7 20:38:59 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
17    Date: 2002/02/07 15:34:13 EST 2002
18    Tag: <none>
19    Description:
20    
21    Compilers that generate assembly code may produce  global labels
22    whose value is resolved at link time. The various peephole optimization
23    modules did not take this in account.
24    
25    TODO. The Labels.addrOf function should really return an option
26    type so that clients are forced to deal with this issue, rather
27    than an exception being raised.
28    
29    ----------------------------------------------------------------------
30    Name: Lal George
31    Date: 2002/02/06 13:55:02 EST
32    Tag: george-20020206-ra-breakup
33    Description:
34    
35    1. A bug fix from Allen.
36    
37        A typo causes extra fstp %st(0)'s to be generated at compensation
38        edges, which might cause stack underflow traps at runtime.  This
39        occurs in fft where there are extraneous fstps right before the 'into'
40        trap instruction (in this case they are harmless since none of the
41        integers overflow.)
42    
43    2. Pulled out various utility modules that were embedded in the modules
44       of the register allocator. I need these modules for other purposes, but
45       they are not complete enough to put into a library (just yet).
46    ----------------------------------------------------------------------
47    Name: Matthias Blume
48    Date: 2002/01/31 16:05:00 EST
49    Tag: blume-20020131-sparc-ccalls
50    Description:
51    
52    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
53       of extra stack space by mistake.  Fixed.
54    
55    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
56    
57    ----------------------------------------------------------------------
58    Name: Allen Leung
59    Date: 2002/01/30
60    Tag: leunga-20020130-nowhere-bug-fix
61    Description:
62    
63       MLRISC bug fixes:
64       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
65       2. MachineInt.fromString was negating its value.
66    
67    ----------------------------------------------------------------------
68    Name: Matthias Blume
69    Date: 2002/01/29
70    Tag: blume-20020129-INSTALL
71    Description:
72    
73    - Added somewhat detailed installation instructions (file INSTALL).
74    - Fixed curl-detection bug in config/install.sh.
75    - It is now possible to select the URL getter using the URLGETTER
76      environment variable:
77    
78          not set / "unknown"      --> automatic detection (script tries wget,
79                                       curl, and lynx)
80          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
81                                       how to properly invoke them)
82          other                    --> use $URLGETTER directly, it must take
83                                       precisely two command-line arguments
84                                       (source URL and destination file name)
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2002/01/28
89    Tag: blume-20020128-sparc-ccalls
90    Description:
91    
92    - Fixed problem with calculation of "used" registers in sparc-c-calls.
93    - Make use of the allocParam argument in sparc-c-calls.
94    
95    ----------------------------------------------------------------------
96    Name: Matthias Blume
97    Date: 2002/01/28
98    Tag: blume-20020128-allocParam
99    Description:
100    
101    John Reppy:  Changes c-calls API to accept client-callback for
102    allocating extra stack space.
103    me: Corresponding changes to mlriscGen (using a dummy argument that
104        does not change the current behavior).
105    
106    ----------------------------------------------------------------------
107    Name: Matthias Blume
108    Date: 2002/01/28 12:00:00
109    Tag: Release_110_38
110    Description:
111    
112    This time for real!!!
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume
116    Date: 2002/01/28 10:56:00 EST
117    Tag: blume-20020128-retraction
118    Description:
119    
120    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
121       with blume-Release_110_38-retracted.)
122    
123    1. Fixed a problem with incorrect rounding modes in real64.sml.
124       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
125    
126    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
127       and enums fixed.  The naming of corresponding ML identifiers should
128       now be consistent again.
129    
130    ----------------------------------------------------------------------
131    Name: Allen Leung
132    Date: 2002/01/27
133    Tag: leunga-20020127-nowhere
134    Description:
135    
136       Added a target called nowhere in the configuration scripts.
137       Enabling this will build the MLRISC 'nowhere' tool (for translating
138       programs with where-clauses into legal SML code) during installation.
139    
140    ----------------------------------------------------------------------
141    Name: Matthias Blume
142    Date: 2002/01/25 21:27:00 EST
143    Tag: blume-Release_110_38-retracted
144    Description:
145    
146    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
147    
148    README will be added later.
149    
150    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
151    (Release_110_38) removed.  Reason: Last-minute bug fixes.
152    
153    ----------------------------------------------------------------------
154    Name: Matthias Blume
155    Date: 2002/01/25
156    Tag: blume-20020125-ffi
157    Description:
158    
159    A large number of tweaks and improvements to ml-nlffi-lib and
160    ml-nlffigen:
161    
162       - ML represenation types have been streamlined
163       - getter and setter functions work with concrete values, not abstract
164         ones where possible
165       - ml-nlffigen command line more flexible (see README file there)
166       - some bugs have been fixed (hopefully)
167    
168    ----------------------------------------------------------------------
169    Name: Lal George
170    Date: 2002/01/24
171    Tag: george-20020124-risc-ra-interface
172    Description:
173    
174       There is a dramatic simplification in the interface to the
175       register allocator for RISC architectures as a result of making
176       parallel copy instructions explicit.
177    
178    ----------------------------------------------------------------------
179    Name: Matthias Blume
180    Date: 2002/01/22
181    Tag: blume-20020122-x86-ccalls
182    Description:
183    
184    Bug fix for c-calls on x86 (having to do with how char- and
185    short-arguments are being handled).
186    
187    ----------------------------------------------------------------------
188    Name: Matthias Blume
189    Date: 2002/01/21
190    Tag: blume-20020121-ff
191    Description:
192    
193    Another day of fiddling with the FFI...
194    
195    1. Bug fix/workaround:  CKIT does not complain about negative array
196       dimensions, so ml-nlffigen has to guard itself against this possibility.
197       (Otherwise a negative dimension would send it into an infinite loop.)
198    
199    2. Some of the abstract types (light objects, light pointers, most "base"
200       types) in structure C are now eqtypes.
201    
202    3. Added constructors and test functions for NULL function pointers.
203    
204    ----------------------------------------------------------------------
205    Name: Matthias Blume
206    Date: 2002/01/18
207    Tag: blume-20020118-ready-for-new-release
208    Description:
209    
210    Made config/srcarchiveurl point to a new place.  (Will provide boot
211    files shortly.)
212    
213    Maybe we christen this to be 110.38?
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume
217    Date: 2002/01/18
218    Tag: blume-20020118-more-ffifiddle
219    Description:
220    
221    Today's FFI fiddling:
222    
223      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
224        C.Set.* that use concrete (MLRep.*) arguments and results instead
225        of abstract ones.
226    
227      - Provided word-style bit operations etc. for "int" representation
228        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
229        and Long.
230    
231    ----------------------------------------------------------------------
232    Name: Matthias Blume
233    Date: 2002/01/18
234    Tag: blume-20020118-use-x86-fp
235    Description:
236    
237    Now that x86-fast-fp seems to be working, I turned it back on again
238    by default.  (Seems to work fine now, even with the FFI.)
239    
240    Other than that, I added some documentation about the FFI to
241    src/ml-nlffigen/README and updated the FFI test examples in
242    src/ml-nlffi-lib/Tests/*.
243    
244    ----------------------------------------------------------------------
245    Name: Allen Leung
246    Date: 2002/01/17
247    Tag: leunga-20020117-x86-fast-fp-call
248    Description:
249    
250       1. Fixed a problem with handling return fp values when x86's fast fp
251          mode is turned on.
252    
253       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
254          of %f32.
255    
256       3. Added a constructor INT32lit to the ast of MLRISC tools.
257    
258    ----------------------------------------------------------------------
259    Name: Matthias Blume
260    Date: 2002/01/16
261    Tag: blume-20020116-ffifiddle
262    Description:
263    
264    More fiddling with the FFI interface:
265    
266     - Make constness 'c instead of rw wherever possible.  This eliminates
267       the need for certain explicit coercions.  (However, due to ML's
268       value polymorphism, there will still be many cases where explicit
269       coercions are necessary.  Phantom types are not the whole answer
270       to modeling a subtyping relationship in ML.)
271    
272     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
273    
274     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
275       clumsy workaround for varargs problem.)
276    
277    ----------------------------------------------------------------------
278    Name: Lal George
279    Date: 2002/01/15
280    Tag: <none>
281    Description:
282    
283    1. Since COPY instructions are no longer native to the architecture,
284       a generic functor can be used to implement the expandCopies function.
285    
286    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
287       TEXT segment.
288    
289    ----------------------------------------------------------------------
290    Name: Matthias Blume
291    Date: 2002/01/15
292    Tag: blume-20020115-ffiupdates
293    Description:
294    
295    1. Fix for bug resulting in single-precision float values being returned
296       incorrectly from FFI calls.
297    
298    2. Small modifications to C FFI API:
299    
300        - memory-allocation routines return straight objects (no options)
301          and raise an exception in out-of-memory situations
302        - unsafe extensions to cast between function pointers and pointers
303          from/to ints
304        - added structure C_Debug as an alternative to structure C where
305          pointer-dereferencing (|*| and |*!) always check for null-pointers
306        - added open_lib' to DynLinkage;  open_lib' works like open_lib
307          but also takes a (possibly empty) list of existing library handles
308          that the current library depends on
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume
312    Date: 2002/01/10
313    Tag: blume-20020110-newffigen
314    Description:
315    
316    1. Updates to portable graph code.
317    
318    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
319       more scalable now so that even huge interfaces such as the one
320       for GTK compile in finite time and space. :-)
321       See src/ml-nlffigen/README for details on what's new.
322    
323    ----------------------------------------------------------------------
324    Name: Lal George
325    Date: 2001/01/09 14:31:35 EST 2002
326    Tag: george-20011206-rm-native-copy
327    Description:
328    
329            Removed the native COPY and FCOPY instructions
330            from all the architectures and replaced it with the
331            explicit COPY instruction from the previous commit.
332    
333            It is now possible to simplify many of the optimizations
334            modules that manipulate copies. This has not been
335            done in this change.
336    
337    ----------------------------------------------------------------------
338    Name: Lal George
339    Date: 2001/12/06 16:50:13 EST 2001
340    Tag: george-20011206-mlrisc-instruction
341    Description:
342    
343    Changed the representation of instructions from being fully abstract
344    to being partially concrete. That is to say:
345    
346      from
347            type instruction
348    
349      to
350            type instr                              (* machine instruction *)
351    
352            datatype instruction =
353                LIVE of {regs: C.cellset, spilled: C.cellset}
354              | KILL of {regs: C.cellset, spilled: C.cellset}
355              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
356              | ANNOTATION of {i: instruction, a: Annotations.annotation}
357              | INSTR of instr
358    
359    This makes the handling of certain special instructions that appear on
360    all architectures easier and uniform.
361    
362    LIVE and KILL say that a list of registers are live or killed at the
363    program point where they appear. No spill code is generated when an
364    element of the 'regs' field is spilled, but the register is moved to
365    the 'spilled' (which is present, more for debugging than anything else).
366    
367    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
368    We used to generate:
369    
370            DEFFREG f1
371            f1 := f2 + f3
372            trapb
373    
374    but now generate:
375    
376            f1 := f2 + f3
377            trapb
378            LIVE {regs=[f1,f2,f3], spilled=[]}
379    
380    Furthermore, the DEFFREG (hack) required that all floating point instruction
381    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
382    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
383    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
384    this architecture).
385    
386    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
387    all the architectures. This will result in further simplification of the
388    register allocator that must be aware of them for coalescing purposes, and
389    will also simplify certain aspects of the machine description that provides
390    callbacks related to parallel copies.
391    
392    ANNOTATION should be obvious, and now INSTR represents the honest to God
393    machine instruction set!
394    
395    The <arch>/instructions/<arch>Instr.sml files define certain utility
396    functions for making porting easier -- essentially converting upper case
397    to lower case. All machine instructions (of type instr) are in upper case,
398    and the lower case form generates an MLRISC instruction. For example on
399    the alpha we have:
400    
401      datatype instr =
402         LDA of {r:cell, b:cell, d:operand}
403       | ...
404    
405      val lda : {r:cell, b:cell, d:operand} -> instruction
406        ...
407    
408    where lda is just (INSTR o LDA), etc.
409    
410    ----------------------------------------------------------------------
411    Name: Matthias Blume
412    Date: 2001/11/22 21:40:00 EST
413    Tag: Release_110_37
414    Description:
415    
416    Release 110.37.  This time for real.
417    
418    ----------------------------------------------------------------------
419    Name: Matthias Blume
420    Date: 2001/11/21 16:35:00 EST
421    Tag: blume-20011121-foot-in-mouth
422    Description:
423    
424    Removed the "Release_110_37" tag because of a serious bug.
425    This will be re-tagged once the bug is fixed.
426    
427    ----------------------------------------------------------------------
428    Name: Matthias Blume
429    Date: 2001/11/21 16:14:00 EST
430    Tag: blume-20011121-forgottenfile
431    Description:
432    
433    Forgot to add a file.  (Just a .tex-file -- part of
434    the CM manual source.)
435    
436    ----------------------------------------------------------------------
437    Name: Matthias Blume
438    Date: 2001/11/21 16:10:00 EST
439    Tag: blume-20011121-invalid_110_37
440    Description:
441    
442    Note: I removed the original tag "Release_110_37" from this commit
443          because we found a serious bug in all non-x86 backends.
444          - Matthias
445    
446    1. Modifications to the SML/NJ code generator and to the runtime system
447       so that code object name strings are directly inserted into code
448       objects at code generation time.  The only business the runtime system
449       has with this is now to read the name strings on occasions.
450       (The encoding of the name string has also changed somewhat.)
451    
452    2. CM now implements a simple "set calculus" for specifying export lists.
453       In particular, it is now possible to refer to the export lists of
454       other libraries/groups/sources and form unions as well as differences.
455       See the latest CM manual for details.
456    
457    3. An separate notion of "proxy" libraries has again be eliminated from
458       CM's model.  (Proxy libraries are now simply a special case of using
459       the export list calculus.)
460    
461    4. Some of the existing libraries now take advantage of the new set
462       calculus.
463       (Notice that not all libraries have been converted because some
464       of the existing .cm-files are supposed to be backward compatible
465       with 110.0.x.)
466    
467    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
468       "General.exnMessage"!  The former relies on a certain hook to be
469       initialized, and that often does not happen in the stand-alone case.)
470    
471    ----------------------------------------------------------------------
472    Name: Lal George
473    Date: 2001/11/21  13:56:18 EST
474    Tag: george-2001121-pseudo-ops
475    Description:
476    
477      Implemented a complete redesign of MLRISC pseudo-ops. Now there
478      ought to never be any question of incompatabilities with
479      pseudo-op syntax expected by host assemblers.
480    
481      For now, only modules supporting GAS syntax are implemented
482      but more should follow, such as MASM, and vendor assembler
483      syntax, e.g. IBM as, Sun as, etc.
484    
485    ----------------------------------------------------------------------
486    Name: Matthias Blume
487    Date: 2001/11/14 11:52:00 EST
488    Tag: blume-20011114-srcname
489    Description:
490    
491    1. Routed the name of the current source file to mlriscgen where it
492       should be directly emitted into the code object.  (This last part
493       is yet to be done.)
494    
495    2. Some cleanup of the pgraph code to make it match the proposal that
496       I put out the other day.  (The proposal notwithstanding, things are
497       still in flux here.)
498    
499    ----------------------------------------------------------------------
500    Name: Lal George
501    Date: 2001/11/14 09:44:04 EST
502    Tag:
503    Description:
504    
505      Fix for a backpatching bug reported by Allen.
506    
507      Because the boundary between short and long span-dependent
508      instructions is +/- 128, there are an astounding number of
509      span-dependent instructions whose size is over estimated.
510    
511      Allen came up with the idea of letting the size of span
512      dependent instructions be non-monotonic, for a maxIter
513      number of times, after which the size must be monotonically
514      increasing.
515    
516      This table shows the number of span-dependent instructions
517      whose size was over-estimated as a function of maxIter, for the
518      file Parse/parse/ml.grm.sml:
519    
520         maxIter            # of instructions:
521            10                      687
522            20                      438
523            30                      198
524            40                        0
525    
526      In compiling the compiler, there is no significant difference in
527      compilation speed between maxIter=10 and maxIter=40. Actually,
528      my measurements showed that maxIter=40 was a tad faster than
529      maxIter=10! Also 96% of the  files in the compiler reach a fix
530      point within 13 iterations, so fixing maxIter at 40, while high,
531      is okay.
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume
535    Date: 2001/10/31 15:25:00 EST
536    Tag: blume-20011031-pgraph
537    Description:
538    
539    CKIT:
540    * Changed the "Function" constructor of type Ast.ctype to carry optional
541      argument identifiers.
542    * Changed the return type of TypeUtil.getFunction accordingly.
543    * Type equality ignores the argument names.
544    * TypeUtil.composite tries to preserve argument names but gives up quickly
545      if there is a mismatch.
546    
547    installation script:
548    * attempts to use "curl" if available (unless "wget" is available as well)
549    
550    CM:
551    * has an experimental implementation of "portable graphs" which I will
552      soon propose as an implementation-independent library format
553    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
554    
555    NLFFI-LIB:
556    * some cleanup (all cosmetic)
557    
558    NLFFIGEN:
559    * temporarily disabled the mechanism that suppresses ML output for
560      C definitions whose identifiers start with an underscore character
561    * generate val bindings for enum constants
562    * user can request that only one style (light or heavy) is being used;
563      default is to use both (command-line arguments: -heavy and -light)
564    * fixed bug in handling of function types involving incomplete pointers
565    * generate ML entry points that take record arguments (i.e., using
566      named arguments) for C functions that have a prototype with named
567      arguments
568      (see changes to CKIT)
569    
570    ----------------------------------------------------------------------
571    Name: Allen Leung
572    Date: 2001/10/27 20:34:00 EDT
573    Tag: leunga-20011027-x86-fast-fp-call
574    Description:
575    
576       Fixed the bug described in blume-20010920-slowfp.
577    
578       The fix involves
579          1. generating FCOPYs in FSTP in ia32-svid
580          2. marking a CALL with the appropriate annotation
581    
582    ----------------------------------------------------------------------
583    Name: Matthias Blume
584    Date: 2001/10/16 11:32:00 EDT
585    Tag: blume-20011016-netbsd
586    Description:
587    
588    Underscore patch from Chris Richards (fixing problem with compiling
589    runtime system under recent NetBSD).
590    
591    ----------------------------------------------------------------------
592    Name: Allen Leung
593    Date: 2001/10/12 17:18:32 EDT 2001
594    Tag: leung-20011012-x86-printflowgraph
595    Description:
596    
597    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
598    
599    ----------------------------------------------------------------------
600    Name: Lal George
601    Date: 2001/10/11 23:51:34 EDT
602    Tag: george-20011011-too-many-instrs
603    Description:
604    
605    The representation of a program point never expected to see more
606    than 65536 instructions in a basic block!
607    
608    ----------------------------------------------------------------------
609    Name: Lal George
610    Date: 2001/10/09 09:41:37 EDT
611    Tag: george-20011008-mlrisc-labels
612    Description:
613    
614    Changed the machine description files to support printing of
615    local and global labels in assembly code, based on host assembler
616    conventions.
617    
618    ----------------------------------------------------------------------
619    Name: Matthias Blume
620    Date: 2001/09/25 15:25:00 EDT
621    Tag: blume-20010925-exninfo
622    Description:
623    
624    I provided a non-hook implementation of exnName (at the toplevel) and
625    made the "dummy" implementation of exnMessage (at the toplevel) more
626    useful: if nothing gets "hooked in", then at least you are going to
627    see the exception name and a message indicating why you don't see more.
628    
629    [For the time being, programs that need exnMessage and want to use
630    ml-build should either use General.exnMessage (strongly recommended) or
631    refer to structure General at some other point so that CM sees a
632    static dependency.]
633    
634    [Similar remarks go for "print" and "use":  If you want to use their
635    functionality in stand-alone programs generated by ml-build, then use
636    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
637    
638    ----------------------------------------------------------------------
639    Name: Matthias Blume
640    Date: 2001/09/20 17:28:00 EDT
641    Tag: blume-20010920-slowfp
642    Description:
643    
644    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
645    
646    ----------------------------------------------------------------------
647  Name: Matthias Blume  Name: Matthias Blume
648  Date: 2001/09/20 17:20:00 EDT  Date: 2001/09/20 17:20:00 EDT
649  Tag: blume-20010920-canonicalpaths  Tag: blume-20010920-canonicalpaths

Legend:
Removed from v.937  
changed lines
  Added in v.1055

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