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 1034, Thu Jan 24 05:53:53 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: 2002/01/24
18    Tag: george-20020124-risc-ra-interface
19    Description:
20    
21       There is a dramatic simplification in the interface to the
22       register allocator for RISC architectures as a result of making
23       parallel copy instructions explicit.
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume
27    Date: 2002/01/22
28    Tag: blume-20020122-x86-ccalls
29    Description:
30    
31    Bug fix for c-calls on x86 (having to do with how char- and
32    short-arguments are being handled).
33    
34    ----------------------------------------------------------------------
35    Name: Matthias Blume
36    Date: 2002/01/21
37    Tag: blume-20020121-ff
38    Description:
39    
40    Another day of fiddling with the FFI...
41    
42    1. Bug fix/workaround:  CKIT does not complain about negative array
43       dimensions, so ml-nlffigen has to guard itself against this possibility.
44       (Otherwise a negative dimension would send it into an infinite loop.)
45    
46    2. Some of the abstract types (light objects, light pointers, most "base"
47       types) in structure C are now eqtypes.
48    
49    3. Added constructors and test functions for NULL function pointers.
50    
51    ----------------------------------------------------------------------
52    Name: Matthias Blume
53    Date: 2002/01/18
54    Tag: blume-20020118-ready-for-new-release
55    Description:
56    
57    Made config/srcarchiveurl point to a new place.  (Will provide boot
58    files shortly.)
59    
60    Maybe we christen this to be 110.38?
61    
62    ----------------------------------------------------------------------
63    Name: Matthias Blume
64    Date: 2002/01/18
65    Tag: blume-20020118-more-ffifiddle
66    Description:
67    
68    Today's FFI fiddling:
69    
70      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
71        C.Set.* that use concrete (MLRep.*) arguments and results instead
72        of abstract ones.
73    
74      - Provided word-style bit operations etc. for "int" representation
75        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
76        and Long.
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume
80    Date: 2002/01/18
81    Tag: blume-20020118-use-x86-fp
82    Description:
83    
84    Now that x86-fast-fp seems to be working, I turned it back on again
85    by default.  (Seems to work fine now, even with the FFI.)
86    
87    Other than that, I added some documentation about the FFI to
88    src/ml-nlffigen/README and updated the FFI test examples in
89    src/ml-nlffi-lib/Tests/*.
90    
91    ----------------------------------------------------------------------
92    Name: Allen Leung
93    Date: 2002/01/17
94    Tag: leunga-20020117-x86-fast-fp-call
95    Description:
96    
97       1. Fixed a problem with handling return fp values when x86's fast fp
98          mode is turned on.
99    
100       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
101          of %f32.
102    
103       3. Added a constructor INT32lit to the ast of MLRISC tools.
104    
105    ----------------------------------------------------------------------
106    Name: Matthias Blume
107    Date: 2002/01/16
108    Tag: blume-20020116-ffifiddle
109    Description:
110    
111    More fiddling with the FFI interface:
112    
113     - Make constness 'c instead of rw wherever possible.  This eliminates
114       the need for certain explicit coercions.  (However, due to ML's
115       value polymorphism, there will still be many cases where explicit
116       coercions are necessary.  Phantom types are not the whole answer
117       to modeling a subtyping relationship in ML.)
118    
119     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
120    
121     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
122       clumsy workaround for varargs problem.)
123    
124    ----------------------------------------------------------------------
125    Name: Lal George
126    Date: 2002/01/15
127    Tag: <none>
128    Description:
129    
130    1. Since COPY instructions are no longer native to the architecture,
131       a generic functor can be used to implement the expandCopies function.
132    
133    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
134       TEXT segment.
135    
136    ----------------------------------------------------------------------
137    Name: Matthias Blume
138    Date: 2002/01/15
139    Tag: blume-20020115-ffiupdates
140    Description:
141    
142    1. Fix for bug resulting in single-precision float values being returned
143       incorrectly from FFI calls.
144    
145    2. Small modifications to C FFI API:
146    
147        - memory-allocation routines return straight objects (no options)
148          and raise an exception in out-of-memory situations
149        - unsafe extensions to cast between function pointers and pointers
150          from/to ints
151        - added structure C_Debug as an alternative to structure C where
152          pointer-dereferencing (|*| and |*!) always check for null-pointers
153        - added open_lib' to DynLinkage;  open_lib' works like open_lib
154          but also takes a (possibly empty) list of existing library handles
155          that the current library depends on
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume
159    Date: 2002/01/10
160    Tag: blume-20020110-newffigen
161    Description:
162    
163    1. Updates to portable graph code.
164    
165    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
166       more scalable now so that even huge interfaces such as the one
167       for GTK compile in finite time and space. :-)
168       See src/ml-nlffigen/README for details on what's new.
169    
170    ----------------------------------------------------------------------
171    Name: Lal George
172    Date: 2001/01/09 14:31:35 EST 2002
173    Tag: george-20011206-rm-native-copy
174    Description:
175    
176            Removed the native COPY and FCOPY instructions
177            from all the architectures and replaced it with the
178            explicit COPY instruction from the previous commit.
179    
180            It is now possible to simplify many of the optimizations
181            modules that manipulate copies. This has not been
182            done in this change.
183    
184    ----------------------------------------------------------------------
185    Name: Lal George
186    Date: 2001/12/06 16:50:13 EST 2001
187    Tag: george-20011206-mlrisc-instruction
188    Description:
189    
190    Changed the representation of instructions from being fully abstract
191    to being partially concrete. That is to say:
192    
193      from
194            type instruction
195    
196      to
197            type instr                              (* machine instruction *)
198    
199            datatype instruction =
200                LIVE of {regs: C.cellset, spilled: C.cellset}
201              | KILL of {regs: C.cellset, spilled: C.cellset}
202              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
203              | ANNOTATION of {i: instruction, a: Annotations.annotation}
204              | INSTR of instr
205    
206    This makes the handling of certain special instructions that appear on
207    all architectures easier and uniform.
208    
209    LIVE and KILL say that a list of registers are live or killed at the
210    program point where they appear. No spill code is generated when an
211    element of the 'regs' field is spilled, but the register is moved to
212    the 'spilled' (which is present, more for debugging than anything else).
213    
214    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
215    We used to generate:
216    
217            DEFFREG f1
218            f1 := f2 + f3
219            trapb
220    
221    but now generate:
222    
223            f1 := f2 + f3
224            trapb
225            LIVE {regs=[f1,f2,f3], spilled=[]}
226    
227    Furthermore, the DEFFREG (hack) required that all floating point instruction
228    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
229    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
230    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
231    this architecture).
232    
233    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
234    all the architectures. This will result in further simplification of the
235    register allocator that must be aware of them for coalescing purposes, and
236    will also simplify certain aspects of the machine description that provides
237    callbacks related to parallel copies.
238    
239    ANNOTATION should be obvious, and now INSTR represents the honest to God
240    machine instruction set!
241    
242    The <arch>/instructions/<arch>Instr.sml files define certain utility
243    functions for making porting easier -- essentially converting upper case
244    to lower case. All machine instructions (of type instr) are in upper case,
245    and the lower case form generates an MLRISC instruction. For example on
246    the alpha we have:
247    
248      datatype instr =
249         LDA of {r:cell, b:cell, d:operand}
250       | ...
251    
252      val lda : {r:cell, b:cell, d:operand} -> instruction
253        ...
254    
255    where lda is just (INSTR o LDA), etc.
256    
257    ----------------------------------------------------------------------
258    Name: Matthias Blume
259    Date: 2001/11/22 21:40:00 EST
260    Tag: Release_110_37
261    Description:
262    
263    Release 110.37.  This time for real.
264    
265    ----------------------------------------------------------------------
266    Name: Matthias Blume
267    Date: 2001/11/21 16:35:00 EST
268    Tag: blume-20011121-foot-in-mouth
269    Description:
270    
271    Removed the "Release_110_37" tag because of a serious bug.
272    This will be re-tagged once the bug is fixed.
273    
274    ----------------------------------------------------------------------
275    Name: Matthias Blume
276    Date: 2001/11/21 16:14:00 EST
277    Tag: blume-20011121-forgottenfile
278    Description:
279    
280    Forgot to add a file.  (Just a .tex-file -- part of
281    the CM manual source.)
282    
283    ----------------------------------------------------------------------
284    Name: Matthias Blume
285    Date: 2001/11/21 16:10:00 EST
286    Tag: blume-20011121-invalid_110_37
287    Description:
288    
289    Note: I removed the original tag "Release_110_37" from this commit
290          because we found a serious bug in all non-x86 backends.
291          - Matthias
292    
293    1. Modifications to the SML/NJ code generator and to the runtime system
294       so that code object name strings are directly inserted into code
295       objects at code generation time.  The only business the runtime system
296       has with this is now to read the name strings on occasions.
297       (The encoding of the name string has also changed somewhat.)
298    
299    2. CM now implements a simple "set calculus" for specifying export lists.
300       In particular, it is now possible to refer to the export lists of
301       other libraries/groups/sources and form unions as well as differences.
302       See the latest CM manual for details.
303    
304    3. An separate notion of "proxy" libraries has again be eliminated from
305       CM's model.  (Proxy libraries are now simply a special case of using
306       the export list calculus.)
307    
308    4. Some of the existing libraries now take advantage of the new set
309       calculus.
310       (Notice that not all libraries have been converted because some
311       of the existing .cm-files are supposed to be backward compatible
312       with 110.0.x.)
313    
314    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
315       "General.exnMessage"!  The former relies on a certain hook to be
316       initialized, and that often does not happen in the stand-alone case.)
317    
318    ----------------------------------------------------------------------
319    Name: Lal George
320    Date: 2001/11/21  13:56:18 EST
321    Tag: george-2001121-pseudo-ops
322    Description:
323    
324      Implemented a complete redesign of MLRISC pseudo-ops. Now there
325      ought to never be any question of incompatabilities with
326      pseudo-op syntax expected by host assemblers.
327    
328      For now, only modules supporting GAS syntax are implemented
329      but more should follow, such as MASM, and vendor assembler
330      syntax, e.g. IBM as, Sun as, etc.
331    
332    ----------------------------------------------------------------------
333    Name: Matthias Blume
334    Date: 2001/11/14 11:52:00 EST
335    Tag: blume-20011114-srcname
336    Description:
337    
338    1. Routed the name of the current source file to mlriscgen where it
339       should be directly emitted into the code object.  (This last part
340       is yet to be done.)
341    
342    2. Some cleanup of the pgraph code to make it match the proposal that
343       I put out the other day.  (The proposal notwithstanding, things are
344       still in flux here.)
345    
346    ----------------------------------------------------------------------
347    Name: Lal George
348    Date: 2001/11/14 09:44:04 EST
349    Tag:
350    Description:
351    
352      Fix for a backpatching bug reported by Allen.
353    
354      Because the boundary between short and long span-dependent
355      instructions is +/- 128, there are an astounding number of
356      span-dependent instructions whose size is over estimated.
357    
358      Allen came up with the idea of letting the size of span
359      dependent instructions be non-monotonic, for a maxIter
360      number of times, after which the size must be monotonically
361      increasing.
362    
363      This table shows the number of span-dependent instructions
364      whose size was over-estimated as a function of maxIter, for the
365      file Parse/parse/ml.grm.sml:
366    
367         maxIter            # of instructions:
368            10                      687
369            20                      438
370            30                      198
371            40                        0
372    
373      In compiling the compiler, there is no significant difference in
374      compilation speed between maxIter=10 and maxIter=40. Actually,
375      my measurements showed that maxIter=40 was a tad faster than
376      maxIter=10! Also 96% of the  files in the compiler reach a fix
377      point within 13 iterations, so fixing maxIter at 40, while high,
378      is okay.
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume
382    Date: 2001/10/31 15:25:00 EST
383    Tag: blume-20011031-pgraph
384    Description:
385    
386    CKIT:
387    * Changed the "Function" constructor of type Ast.ctype to carry optional
388      argument identifiers.
389    * Changed the return type of TypeUtil.getFunction accordingly.
390    * Type equality ignores the argument names.
391    * TypeUtil.composite tries to preserve argument names but gives up quickly
392      if there is a mismatch.
393    
394    installation script:
395    * attempts to use "curl" if available (unless "wget" is available as well)
396    
397    CM:
398    * has an experimental implementation of "portable graphs" which I will
399      soon propose as an implementation-independent library format
400    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
401    
402    NLFFI-LIB:
403    * some cleanup (all cosmetic)
404    
405    NLFFIGEN:
406    * temporarily disabled the mechanism that suppresses ML output for
407      C definitions whose identifiers start with an underscore character
408    * generate val bindings for enum constants
409    * user can request that only one style (light or heavy) is being used;
410      default is to use both (command-line arguments: -heavy and -light)
411    * fixed bug in handling of function types involving incomplete pointers
412    * generate ML entry points that take record arguments (i.e., using
413      named arguments) for C functions that have a prototype with named
414      arguments
415      (see changes to CKIT)
416    
417    ----------------------------------------------------------------------
418    Name: Allen Leung
419    Date: 2001/10/27 20:34:00 EDT
420    Tag: leunga-20011027-x86-fast-fp-call
421    Description:
422    
423       Fixed the bug described in blume-20010920-slowfp.
424    
425       The fix involves
426          1. generating FCOPYs in FSTP in ia32-svid
427          2. marking a CALL with the appropriate annotation
428    
429    ----------------------------------------------------------------------
430    Name: Matthias Blume
431    Date: 2001/10/16 11:32:00 EDT
432    Tag: blume-20011016-netbsd
433    Description:
434    
435    Underscore patch from Chris Richards (fixing problem with compiling
436    runtime system under recent NetBSD).
437    
438    ----------------------------------------------------------------------
439    Name: Allen Leung
440    Date: 2001/10/12 17:18:32 EDT 2001
441    Tag: leung-20011012-x86-printflowgraph
442    Description:
443    
444    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
445    
446    ----------------------------------------------------------------------
447    Name: Lal George
448    Date: 2001/10/11 23:51:34 EDT
449    Tag: george-20011011-too-many-instrs
450    Description:
451    
452    The representation of a program point never expected to see more
453    than 65536 instructions in a basic block!
454    
455    ----------------------------------------------------------------------
456    Name: Lal George
457    Date: 2001/10/09 09:41:37 EDT
458    Tag: george-20011008-mlrisc-labels
459    Description:
460    
461    Changed the machine description files to support printing of
462    local and global labels in assembly code, based on host assembler
463    conventions.
464    
465    ----------------------------------------------------------------------
466    Name: Matthias Blume
467    Date: 2001/09/25 15:25:00 EDT
468    Tag: blume-20010925-exninfo
469    Description:
470    
471    I provided a non-hook implementation of exnName (at the toplevel) and
472    made the "dummy" implementation of exnMessage (at the toplevel) more
473    useful: if nothing gets "hooked in", then at least you are going to
474    see the exception name and a message indicating why you don't see more.
475    
476    [For the time being, programs that need exnMessage and want to use
477    ml-build should either use General.exnMessage (strongly recommended) or
478    refer to structure General at some other point so that CM sees a
479    static dependency.]
480    
481    [Similar remarks go for "print" and "use":  If you want to use their
482    functionality in stand-alone programs generated by ml-build, then use
483    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
484    
485    ----------------------------------------------------------------------
486    Name: Matthias Blume
487    Date: 2001/09/20 17:28:00 EDT
488    Tag: blume-20010920-slowfp
489    Description:
490    
491    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
492    
493    ----------------------------------------------------------------------
494  Name: Matthias Blume  Name: Matthias Blume
495  Date: 2001/09/20 17:20:00 EDT  Date: 2001/09/20 17:20:00 EDT
496  Tag: blume-20010920-canonicalpaths  Tag: blume-20010920-canonicalpaths
# Line 33  Line 511 
511    
512  ----------------------------------------------------------------------  ----------------------------------------------------------------------
513  Name: Matthias Blume  Name: Matthias Blume
514    >>>>>>> 1.169
515  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
516  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
517  Description:  Description:

Legend:
Removed from v.937  
changed lines
  Added in v.1034

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