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 1032, Tue Jan 22 16:59:16 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: Matthias Blume
17    Date: 2002/01/22
18    Tag: blume-20020122-x86-ccalls
19    Description:
20    
21    Bug fix for c-calls on x86 (having to do with how char- and
22    short-arguments are being handled).
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume
26    Date: 2002/01/21
27    Tag: blume-20020121-ff
28    Description:
29    
30    Another day of fiddling with the FFI...
31    
32    1. Bug fix/workaround:  CKIT does not complain about negative array
33       dimensions, so ml-nlffigen has to guard itself against this possibility.
34       (Otherwise a negative dimension would send it into an infinite loop.)
35    
36    2. Some of the abstract types (light objects, light pointers, most "base"
37       types) in structure C are now eqtypes.
38    
39    3. Added constructors and test functions for NULL function pointers.
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume
43    Date: 2002/01/18
44    Tag: blume-20020118-ready-for-new-release
45    Description:
46    
47    Made config/srcarchiveurl point to a new place.  (Will provide boot
48    files shortly.)
49    
50    Maybe we christen this to be 110.38?
51    
52    ----------------------------------------------------------------------
53    Name: Matthias Blume
54    Date: 2002/01/18
55    Tag: blume-20020118-more-ffifiddle
56    Description:
57    
58    Today's FFI fiddling:
59    
60      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
61        C.Set.* that use concrete (MLRep.*) arguments and results instead
62        of abstract ones.
63    
64      - Provided word-style bit operations etc. for "int" representation
65        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
66        and Long.
67    
68    ----------------------------------------------------------------------
69    Name: Matthias Blume
70    Date: 2002/01/18
71    Tag: blume-20020118-use-x86-fp
72    Description:
73    
74    Now that x86-fast-fp seems to be working, I turned it back on again
75    by default.  (Seems to work fine now, even with the FFI.)
76    
77    Other than that, I added some documentation about the FFI to
78    src/ml-nlffigen/README and updated the FFI test examples in
79    src/ml-nlffi-lib/Tests/*.
80    
81    ----------------------------------------------------------------------
82    Name: Allen Leung
83    Date: 2002/01/17
84    Tag: leunga-20020117-x86-fast-fp-call
85    Description:
86    
87       1. Fixed a problem with handling return fp values when x86's fast fp
88          mode is turned on.
89    
90       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
91          of %f32.
92    
93       3. Added a constructor INT32lit to the ast of MLRISC tools.
94    
95    ----------------------------------------------------------------------
96    Name: Matthias Blume
97    Date: 2002/01/16
98    Tag: blume-20020116-ffifiddle
99    Description:
100    
101    More fiddling with the FFI interface:
102    
103     - Make constness 'c instead of rw wherever possible.  This eliminates
104       the need for certain explicit coercions.  (However, due to ML's
105       value polymorphism, there will still be many cases where explicit
106       coercions are necessary.  Phantom types are not the whole answer
107       to modeling a subtyping relationship in ML.)
108    
109     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
110    
111     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
112       clumsy workaround for varargs problem.)
113    
114    ----------------------------------------------------------------------
115    Name: Lal George
116    Date: 2002/01/15
117    Tag: <none>
118    Description:
119    
120    1. Since COPY instructions are no longer native to the architecture,
121       a generic functor can be used to implement the expandCopies function.
122    
123    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
124       TEXT segment.
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume
128    Date: 2002/01/15
129    Tag: blume-20020115-ffiupdates
130    Description:
131    
132    1. Fix for bug resulting in single-precision float values being returned
133       incorrectly from FFI calls.
134    
135    2. Small modifications to C FFI API:
136    
137        - memory-allocation routines return straight objects (no options)
138          and raise an exception in out-of-memory situations
139        - unsafe extensions to cast between function pointers and pointers
140          from/to ints
141        - added structure C_Debug as an alternative to structure C where
142          pointer-dereferencing (|*| and |*!) always check for null-pointers
143        - added open_lib' to DynLinkage;  open_lib' works like open_lib
144          but also takes a (possibly empty) list of existing library handles
145          that the current library depends on
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume
149    Date: 2002/01/10
150    Tag: blume-20020110-newffigen
151    Description:
152    
153    1. Updates to portable graph code.
154    
155    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
156       more scalable now so that even huge interfaces such as the one
157       for GTK compile in finite time and space. :-)
158       See src/ml-nlffigen/README for details on what's new.
159    
160    ----------------------------------------------------------------------
161    Name: Lal George
162    Date: 2001/01/09 14:31:35 EST 2002
163    Tag: george-20011206-rm-native-copy
164    Description:
165    
166            Removed the native COPY and FCOPY instructions
167            from all the architectures and replaced it with the
168            explicit COPY instruction from the previous commit.
169    
170            It is now possible to simplify many of the optimizations
171            modules that manipulate copies. This has not been
172            done in this change.
173    
174    ----------------------------------------------------------------------
175    Name: Lal George
176    Date: 2001/12/06 16:50:13 EST 2001
177    Tag: george-20011206-mlrisc-instruction
178    Description:
179    
180    Changed the representation of instructions from being fully abstract
181    to being partially concrete. That is to say:
182    
183      from
184            type instruction
185    
186      to
187            type instr                              (* machine instruction *)
188    
189            datatype instruction =
190                LIVE of {regs: C.cellset, spilled: C.cellset}
191              | KILL of {regs: C.cellset, spilled: C.cellset}
192              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
193              | ANNOTATION of {i: instruction, a: Annotations.annotation}
194              | INSTR of instr
195    
196    This makes the handling of certain special instructions that appear on
197    all architectures easier and uniform.
198    
199    LIVE and KILL say that a list of registers are live or killed at the
200    program point where they appear. No spill code is generated when an
201    element of the 'regs' field is spilled, but the register is moved to
202    the 'spilled' (which is present, more for debugging than anything else).
203    
204    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
205    We used to generate:
206    
207            DEFFREG f1
208            f1 := f2 + f3
209            trapb
210    
211    but now generate:
212    
213            f1 := f2 + f3
214            trapb
215            LIVE {regs=[f1,f2,f3], spilled=[]}
216    
217    Furthermore, the DEFFREG (hack) required that all floating point instruction
218    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
219    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
220    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
221    this architecture).
222    
223    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
224    all the architectures. This will result in further simplification of the
225    register allocator that must be aware of them for coalescing purposes, and
226    will also simplify certain aspects of the machine description that provides
227    callbacks related to parallel copies.
228    
229    ANNOTATION should be obvious, and now INSTR represents the honest to God
230    machine instruction set!
231    
232    The <arch>/instructions/<arch>Instr.sml files define certain utility
233    functions for making porting easier -- essentially converting upper case
234    to lower case. All machine instructions (of type instr) are in upper case,
235    and the lower case form generates an MLRISC instruction. For example on
236    the alpha we have:
237    
238      datatype instr =
239         LDA of {r:cell, b:cell, d:operand}
240       | ...
241    
242      val lda : {r:cell, b:cell, d:operand} -> instruction
243        ...
244    
245    where lda is just (INSTR o LDA), etc.
246    
247    ----------------------------------------------------------------------
248    Name: Matthias Blume
249    Date: 2001/11/22 21:40:00 EST
250    Tag: Release_110_37
251    Description:
252    
253    Release 110.37.  This time for real.
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume
257    Date: 2001/11/21 16:35:00 EST
258    Tag: blume-20011121-foot-in-mouth
259    Description:
260    
261    Removed the "Release_110_37" tag because of a serious bug.
262    This will be re-tagged once the bug is fixed.
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume
266    Date: 2001/11/21 16:14:00 EST
267    Tag: blume-20011121-forgottenfile
268    Description:
269    
270    Forgot to add a file.  (Just a .tex-file -- part of
271    the CM manual source.)
272    
273    ----------------------------------------------------------------------
274    Name: Matthias Blume
275    Date: 2001/11/21 16:10:00 EST
276    Tag: blume-20011121-invalid_110_37
277    Description:
278    
279    Note: I removed the original tag "Release_110_37" from this commit
280          because we found a serious bug in all non-x86 backends.
281          - Matthias
282    
283    1. Modifications to the SML/NJ code generator and to the runtime system
284       so that code object name strings are directly inserted into code
285       objects at code generation time.  The only business the runtime system
286       has with this is now to read the name strings on occasions.
287       (The encoding of the name string has also changed somewhat.)
288    
289    2. CM now implements a simple "set calculus" for specifying export lists.
290       In particular, it is now possible to refer to the export lists of
291       other libraries/groups/sources and form unions as well as differences.
292       See the latest CM manual for details.
293    
294    3. An separate notion of "proxy" libraries has again be eliminated from
295       CM's model.  (Proxy libraries are now simply a special case of using
296       the export list calculus.)
297    
298    4. Some of the existing libraries now take advantage of the new set
299       calculus.
300       (Notice that not all libraries have been converted because some
301       of the existing .cm-files are supposed to be backward compatible
302       with 110.0.x.)
303    
304    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
305       "General.exnMessage"!  The former relies on a certain hook to be
306       initialized, and that often does not happen in the stand-alone case.)
307    
308    ----------------------------------------------------------------------
309    Name: Lal George
310    Date: 2001/11/21  13:56:18 EST
311    Tag: george-2001121-pseudo-ops
312    Description:
313    
314      Implemented a complete redesign of MLRISC pseudo-ops. Now there
315      ought to never be any question of incompatabilities with
316      pseudo-op syntax expected by host assemblers.
317    
318      For now, only modules supporting GAS syntax are implemented
319      but more should follow, such as MASM, and vendor assembler
320      syntax, e.g. IBM as, Sun as, etc.
321    
322    ----------------------------------------------------------------------
323    Name: Matthias Blume
324    Date: 2001/11/14 11:52:00 EST
325    Tag: blume-20011114-srcname
326    Description:
327    
328    1. Routed the name of the current source file to mlriscgen where it
329       should be directly emitted into the code object.  (This last part
330       is yet to be done.)
331    
332    2. Some cleanup of the pgraph code to make it match the proposal that
333       I put out the other day.  (The proposal notwithstanding, things are
334       still in flux here.)
335    
336    ----------------------------------------------------------------------
337    Name: Lal George
338    Date: 2001/11/14 09:44:04 EST
339    Tag:
340    Description:
341    
342      Fix for a backpatching bug reported by Allen.
343    
344      Because the boundary between short and long span-dependent
345      instructions is +/- 128, there are an astounding number of
346      span-dependent instructions whose size is over estimated.
347    
348      Allen came up with the idea of letting the size of span
349      dependent instructions be non-monotonic, for a maxIter
350      number of times, after which the size must be monotonically
351      increasing.
352    
353      This table shows the number of span-dependent instructions
354      whose size was over-estimated as a function of maxIter, for the
355      file Parse/parse/ml.grm.sml:
356    
357         maxIter            # of instructions:
358            10                      687
359            20                      438
360            30                      198
361            40                        0
362    
363      In compiling the compiler, there is no significant difference in
364      compilation speed between maxIter=10 and maxIter=40. Actually,
365      my measurements showed that maxIter=40 was a tad faster than
366      maxIter=10! Also 96% of the  files in the compiler reach a fix
367      point within 13 iterations, so fixing maxIter at 40, while high,
368      is okay.
369    
370    ----------------------------------------------------------------------
371    Name: Matthias Blume
372    Date: 2001/10/31 15:25:00 EST
373    Tag: blume-20011031-pgraph
374    Description:
375    
376    CKIT:
377    * Changed the "Function" constructor of type Ast.ctype to carry optional
378      argument identifiers.
379    * Changed the return type of TypeUtil.getFunction accordingly.
380    * Type equality ignores the argument names.
381    * TypeUtil.composite tries to preserve argument names but gives up quickly
382      if there is a mismatch.
383    
384    installation script:
385    * attempts to use "curl" if available (unless "wget" is available as well)
386    
387    CM:
388    * has an experimental implementation of "portable graphs" which I will
389      soon propose as an implementation-independent library format
390    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
391    
392    NLFFI-LIB:
393    * some cleanup (all cosmetic)
394    
395    NLFFIGEN:
396    * temporarily disabled the mechanism that suppresses ML output for
397      C definitions whose identifiers start with an underscore character
398    * generate val bindings for enum constants
399    * user can request that only one style (light or heavy) is being used;
400      default is to use both (command-line arguments: -heavy and -light)
401    * fixed bug in handling of function types involving incomplete pointers
402    * generate ML entry points that take record arguments (i.e., using
403      named arguments) for C functions that have a prototype with named
404      arguments
405      (see changes to CKIT)
406    
407    ----------------------------------------------------------------------
408    Name: Allen Leung
409    Date: 2001/10/27 20:34:00 EDT
410    Tag: leunga-20011027-x86-fast-fp-call
411    Description:
412    
413       Fixed the bug described in blume-20010920-slowfp.
414    
415       The fix involves
416          1. generating FCOPYs in FSTP in ia32-svid
417          2. marking a CALL with the appropriate annotation
418    
419    ----------------------------------------------------------------------
420    Name: Matthias Blume
421    Date: 2001/10/16 11:32:00 EDT
422    Tag: blume-20011016-netbsd
423    Description:
424    
425    Underscore patch from Chris Richards (fixing problem with compiling
426    runtime system under recent NetBSD).
427    
428    ----------------------------------------------------------------------
429    Name: Allen Leung
430    Date: 2001/10/12 17:18:32 EDT 2001
431    Tag: leung-20011012-x86-printflowgraph
432    Description:
433    
434    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
435    
436    ----------------------------------------------------------------------
437    Name: Lal George
438    Date: 2001/10/11 23:51:34 EDT
439    Tag: george-20011011-too-many-instrs
440    Description:
441    
442    The representation of a program point never expected to see more
443    than 65536 instructions in a basic block!
444    
445    ----------------------------------------------------------------------
446  Name: Lal George  Name: Lal George
447  Date: 2001/10/09 09:41:37 EDT  Date: 2001/10/09 09:41:37 EDT
448  Tag: george-20011008-mlrisc-labels  Tag: george-20011008-mlrisc-labels

Legend:
Removed from v.952  
changed lines
  Added in v.1032

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