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/NOTES/HISTORY
ViewVC logotype

Diff of /sml/trunk/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 984, Wed Nov 21 19:00:08 2001 UTC revision 1133, Tue Mar 12 03:56:23 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: 2002/03/11 11 22:38:53 EST
18    Tag: george-20020311-jump-chain-elim
19    Description:
20    
21    Tested the jump chain elimination on all architectures (except the
22    hppa).  This is on by default right now and is profitable for the
23    alpha and x86, however, it may not be profitable for the sparc and ppc
24    when compiling the compiler.
25    
26    The gc test will typically jump to a label at the end of the cluster,
27    where there is another jump to an external cluster containing the actual
28    code to invoke gc. This is to allow factoring of common gc invocation
29    sequences. That is to say, we generate:
30    
31            f:
32               testgc
33               ja   L1      % jump if above to L1
34    
35            L1:
36               jmp L2
37    
38    
39    After jump chain elimination the 'ja L1' instructions is converted to
40    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
41    up being implemented in their long form (if L2 is far away) using:
42    
43            jbe     L3      % jump if below or equal to L3
44            jmp     L2
45         L3:
46            ...
47    
48    
49    For large compilation units L2  may be far away.
50    
51    
52    ----------------------------------------------------------------------
53    Name: Matthias Blume
54    Date: 2002/03/11 13:30:00 EST
55    Tag: blume-20020311-mltreeeval
56    Description:
57    
58    A functor parameter was missing.
59    
60    ----------------------------------------------------------------------
61    Name: Allen Leung
62    Date: 2002/03/11 10:30:00 EST
63    Tag: leunga-20020310-runtime-string0
64    Description:
65    
66       The representation of the empty string now points to a
67    legal null terminated C string instead of unit.  It is now possible
68    to convert an ML string into C string with InlineT.CharVector.getData.
69    This compiles into one single machine instruction.
70    
71    ----------------------------------------------------------------------
72    Name: Allen Leung
73    Date: 2002/03/10 23:55:00 EST
74    Tag: leunga-20020310-x86-call
75    Description:
76    
77       Added machine generation for CALL instruction (relative displacement mode)
78    
79    ----------------------------------------------------------------------
80    Name: Matthias Blume
81    Date: 2002/03/08 16:05:00
82    Tag: blume-20020308-entrypoints
83    Description:
84    
85    Version number bumped to 110.39.1.  NEW BOOTFILES!
86    
87    Entrypoints: non-zero offset into a code object where execution should begin.
88    
89    - Added the notion of an entrypoint to CodeObj.
90    - Added reading/writing of entrypoint info to Binfile.
91    - Made runtime system bootloader aware of entrypoints.
92    - Use the address of the label of the first function given to mlriscGen
93      as the entrypoint.  This address is currently always 0, but it will
94      not be 0 once we turn on block placement.
95    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
96      with entry points) from mlriscGen.
97    
98    ----------------------------------------------------------------------
99    Name: Allen Leung
100    Date: 2002/03/07 20:45:00 EST
101    Tag: leunga-20020307-x86-cmov
102    Description:
103    
104       Bug fixes for CMOVcc on x86.
105    
106       1. Added machine code generation for CMOVcc
107       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
108       3. CMOVcc cannot have an immediate operand as argument.
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume
112    Date: 2002/03/07 16:15:00 EST
113    Tag: blume-20020307-controls
114    Description:
115    
116    This is a very large but mostly boring patch which makes (almost)
117    every tuneable compiler knob (i.e., pretty much everything under
118    Control.* plus a few other things) configurable via both the command
119    line and environment variables in the style CM did its configuration
120    until now.
121    
122    Try starting sml with '-h' (or, if you are brave, '-H')
123    
124    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
125    implements the underlying generic mechanism.
126    
127    The interface to some of the existing such facilities has changed somewhat.
128    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
129    (The getFoo interface is still there for backward-compatibility, but its
130    use is deprecated.)
131    
132    The ml-build script passes -Cxxx=yyy command-line arguments through so
133    that one can now twiddle the compiler settings when using this "batch"
134    compiler.
135    
136    TODO items:
137    
138    We should go through and throw out all controls that are no longer
139    connected to anything.  Moreover, we should go through and provide
140    meaningful (and correct!) documentation strings for those controls
141    that still are connected.
142    
143    Currently, multiple calls to Controls.new are accepted (only the first
144    has any effect).  Eventually we should make sure that every control
145    is being made (via Controls.new) exactly once.  Future access can then
146    be done using Controls.acc.
147    
148    Finally, it would probably be a good idea to use the getter-setter
149    interface to controls rather than ref cells.  For the time being, both
150    styles are provided by the Controls module, but getter-setter pairs are
151    better if thread-safety is of any concern because they can be wrapped.
152    
153    *****************************************
154    
155    One bug fix: The function blockPlacement in three of the MLRISC
156    backpatch files used to be hard-wired to one of two possibilities at
157    link time (according to the value of the placementFlag).  But (I
158    think) it should rather sense the flag every time.
159    
160    *****************************************
161    
162    Other assorted changes (by other people who did not supply a HISTORY entry):
163    
164    1. the cross-module inliner now works much better (Monnier)
165    2. representation of weights, frequencies, and probabilities in MLRISC
166       changed in preparation of using those for weighted block placement
167       (Reppy, George)
168    
169    ----------------------------------------------------------------------
170    Name: Lal George
171    Date: 2002/03/07 14:44:24 EST 2002
172    Tag: george-20020307-weighted-block-placement
173    
174    Tested the weighted block placement optimization on all architectures
175    (except the hppa) using AMPL to generate the block and edge frequencies.
176    Changes were required in the machine properties to correctly
177    categorize trap instructions. There is an MLRISC flag
178    "weighted-block-placement" that can be used to enable weighted block
179    placement, but this will be ineffective without block/edge
180    frequencies (coming soon).
181    
182    
183    ----------------------------------------------------------------------
184    Name: Lal George
185    Date: 2002/03/05 17:24:48 EST
186    Tag: george-20020305-linkage-cluster
187    
188    In order to support the block placement optimization, a new cluster
189    is generated as the very first cluster (called the linkage cluster).
190    It contains a single jump to the 'real' entry point for the compilation
191    unit. Block placement has no effect on the linkage cluster itself, but
192    all the other clusters  have full freedom in the manner in which they
193    reorder blocks or functions.
194    
195    On the x86 the typical linkage code that is generated is:
196       ----------------------
197            .align 2
198       L0:
199            addl    $L1-L0, 72(%esp)
200            jmp     L1
201    
202    
203            .align  2
204       L1:
205       ----------------------
206    
207    72(%esp) is the memory location for the stdlink register. This
208    must contain the address of the CPS function being called. In the
209    above example, it contains the address of  L0; before
210    calling L1 (the real entry point for the compilation unit), it
211    must contain the address for L1, and hence
212    
213            addl $L1-L0, 72(%esp)
214    
215    I have tested this on all architectures except the hppa.The increase
216    in code size is of course negligible
217    
218    ----------------------------------------------------------------------
219    Name: Allen Leung
220    Date: 2002/03/03 13:20:00 EST
221    Tag: leunga-20020303-mlrisc-tools
222    
223      Added #[ ... ] expressions to mlrisc tools
224    
225    ----------------------------------------------------------------------
226    Name: Matthias Blume
227    Date: 2002/02/27 12:29:00 EST
228    Tag: blume-20020227-cdebug
229    Description:
230    
231    - made types in structure C and C_Debug to be equal
232    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
233    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
234    
235    ----------------------------------------------------------------------
236    Name: Matthias Blume
237    Date: 2002/02/26 12:00:00 EST
238    Tag: blume-20020226-ffi
239    Description:
240    
241    1. Fixed a minor bug in CM's "noweb" tool:
242       If numbering is turned off, then truly don't number (i.e., do not
243       supply the -L option to noweb).  The previous behavior was to supply
244       -L'' -- which caused noweb to use the "default" line numbering scheme.
245       Thanks to Chris Richards for pointing this out (and supplying the fix).
246    
247    2. Once again, I reworked some aspects of the FFI:
248    
249       A. The incomplete/complete type business:
250    
251       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
252         gone!
253       - ML types representing an incomplete type are now *equal* to
254         ML types representing their corresponding complete types (just like
255         in C).  This is still safe because ml-nlffigen will not generate
256         RTTI for incomplete types, nor will it generate functions that
257         require access to such RTTI.   But when ML code generated from both
258         incomplete and complete versions of the C type meet, the ML types
259         are trivially interoperable.
260    
261         NOTE:  These changes restore the full generality of the translation
262         (which was previously lost when I eliminated functorization)!
263    
264       B. Enum types:
265    
266       - Structure C now has a type constructor "enum" that is similar to
267         how the "su" constructor works.  However, "enum" is not a phantom
268         type because each "T enum" has values (and is isomorphic to
269         MLRep.Signed.int).
270       - There are generic access operations for enum objects (using
271         MLRep.Signed.int).
272       - ml-nlffigen will generate a structure E_foo for each "enum foo".
273         * The structure contains the definition of type "mlrep" (the ML-side
274         representation type of the enum).  Normally, mlrep is the same
275         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
276         then mlrep will be defined as a datatype -- thus facilitating
277         pattern matching on mlrep values.
278         ("-ec" will be suppressed if there are duplicate values in an
279          enumeration.)
280         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
281         will be generated for each C enum constant xxx.
282         * Conversion functions m2i and i2m convert between mlrep and
283         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
284         * Coversion functions c and ml convert between mlrep and "tag enum".
285         * Access functions (get/set) fetch and store mlrep values.
286       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
287         enumerations are merged into one single enumeration represented by
288         structure E_'.
289    
290    ----------------------------------------------------------------------
291    Name: Allen Leung
292    Date: 2002/02/25 04:45:00 EST
293    Tag: leunga-20020225-cps-spill
294    
295    This is a new implementation of the CPS spill phase.
296    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
297    In case of problems, replace it with the old file spill.sml
298    
299    The current compiler runs into some serious performance problems when
300    constructing a large record.  This can happen when we try to compile a
301    structure with many items.  Even a very simple structure like the following
302    makes the compiler slow down.
303    
304        structure Foo = struct
305           val x_1 = 0w1 : Word32.int
306           val x_2 = 0w2 : Word32.int
307           val x_3 = 0w3 : Word32.int
308           ...
309           val x_N = 0wN : Word32.int
310        end
311    
312    The following table shows the compile time, from N=1000 to N=4000,
313    with the old compiler:
314    
315    N
316    1000   CPS 100 spill                           0.04u  0.00s  0.00g
317           MLRISC ra                               0.06u  0.00s  0.05g
318              (spills = 0 reloads = 0)
319           TOTAL                                   0.63u  0.07s  0.21g
320    
321    1100   CPS 100 spill                           8.25u  0.32s  0.64g
322           MLRISC ra                               5.68u  0.59s  3.93g
323              (spills = 0 reloads = 0)
324           TOTAL                                   14.71u  0.99s  4.81g
325    
326    1500   CPS 100 spill                           58.55u  2.34s  1.74g
327           MLRISC ra                               5.54u  0.65s  3.91g
328              (spills = 543 reloads = 1082)
329           TOTAL                                   65.40u  3.13s  6.00g
330    
331    2000   CPS 100 spill                           126.69u  4.84s  3.08g
332           MLRISC ra                               0.80u  0.10s  0.55g
333              (spills = 42 reloads = 84)
334           TOTAL                                   129.42u  5.10s  4.13g
335    
336    3000   CPS 100 spill                           675.59u  19.03s  11.64g
337           MLRISC ra                               2.69u  0.27s  1.38g
338              (spills = 62 reloads = 124)
339           TOTAL                                   682.48u  19.61s  13.99g
340    
341    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
342           MLRISC ra                               4.96u  0.27s  2.72g
343              (spills = 85 reloads = 170)
344           TOTAL                                   2375.26u  57.21s  48.00g
345    
346    As you can see the old cps spill module suffers from some serious
347    performance problem.  But since I cannot decipher the old code fully,
348    instead of patching the problems up, I'm reimplementing it
349    with a different algorithm.  The new code is more modular,
350    smaller when compiled, and substantially faster
351    (O(n log n) time and O(n) space).  Timing of the new spill module:
352    
353    4000  CPS 100 spill                           0.02u  0.00s  0.00g
354          MLRISC ra                               0.25u  0.02s  0.15g
355             (spills=1 reloads=3)
356          TOTAL                                   7.74u  0.34s  1.62g
357    
358    Implementation details:
359    
360    As far as I can tell, the purpose of the CPS spill module is to make sure the
361    number of live variables at any program point (the bandwidth)
362    does not exceed a certain limit, which is determined by the
363    size of the spill area.
364    
365    When the bandwidth is too large, we decrease the register pressure by
366    packing live variables into spill records.  How we achieve this is
367    completely different than what we did in the old code.
368    
369    First, there is something about the MLRiscGen code generator
370    that we should be aware of:
371    
372    o MLRiscGen performs code motion!
373    
374       In particular, it will move floating point computations and
375       address computations involving only the heap pointer to
376       their use sites (if there is only a single use).
377       What this means is that if we have a CPS record construction
378       statement
379    
380           RECORD(k,vl,w,e)
381    
382       we should never count the new record address w as live if w
383       has only one use (which is often the case).
384    
385       We should do something similar to floating point, but the transformation
386       there is much more complex, so I won't deal with that.
387    
388    Secondly, there are now two new cps primops at our disposal:
389    
390     1. rawrecord of record_kind option
391        This pure operator allocates some uninitialized storage from the heap.
392        There are two forms:
393    
394         rawrecord NONE [INT n]  allocates a tagless record of length n
395         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
396                                     and initializes the tag.
397    
398     2. rawupdate of cty
399          rawupdate cty (v,i,x)
400          Assigns to x to the ith component of record v.
401          The storelist is not updated.
402    
403    We use these new primops for both spilling and increment record construction.
404    
405     1. Spilling.
406    
407        This is implemented with a linear scan algorithm (but generalized
408        to trees).  The algorithm will create a single spill record at the
409        beginning of the cps function and use rawupdate to spill to it,
410        and SELECT or SELp to reload from it.  So both spills and reloads
411        are fine-grain operations.  In contrast, in the old algorithm
412        "spills" have to be bundled together in records.
413    
414        Ideally, we should sink the spill record construction to where
415        it is needed.  We can even split the spill record into multiple ones
416        at the places where they are needed.  But CPS is not a good
417        representation for global code motion, so I'll keep it simple and
418        am not attempting this.
419    
420     2. Incremental record construction (aka record splitting).
421    
422        Long records with many component values which are simulatenously live
423        (recall that single use record addresses are not considered to
424         be live) are constructed with rawrecord and rawupdate.
425        We allocate space on the heap with rawrecord first, then gradually
426        fill it in with rawupdate.  This is the technique suggested to me
427        by Matthias.
428    
429        Some restrictions on when this is applicable:
430        1. It is not a VECTOR record.  The code generator currently does not handle
431           this case. VECTOR record uses double indirection like arrays.
432        2. All the record component values are defined in the same "basic block"
433           as the record constructor.  This is to prevent speculative
434           record construction.
435    
436    ----------------------------------------------------------------------
437    Name: Allen Leung
438    Date: 2002/02/22 01:02:00 EST
439    Tag: leunga-20020222-mlrisc-tools
440    
441    Minor bug fixes in the parser and rewriter
442    
443    ----------------------------------------------------------------------
444    Name: Allen Leung
445    Date: 2002/02/21 20:20:00 EST
446    Tag: leunga-20020221-peephole
447    
448    Regenerated the peephole files.  Some contained typos in the specification
449    and some didn't compile because of pretty printing bugs in the old version
450    of 'nowhere'.
451    
452    ----------------------------------------------------------------------
453    Name: Allen Leung
454    Date: 2002/02/19 20:20:00 EST
455    Tag: leunga-20020219-mlrisc-tools
456    Description:
457    
458       Minor bug fixes to the mlrisc-tools library:
459    
460       1.  Fixed up parsing colon suffixed keywords
461       2.  Added the ability to shut the error messages up
462       3.  Reimplemented the pretty printer and fixed up/improved
463           the pretty printing of handle and -> types.
464       4.  Fixed up generation of literal symbols in the nowhere tool.
465       5.  Added some SML keywords to to sml.sty
466    
467    ----------------------------------------------------------------------
468    Name: Matthias Blume
469    Date: 2002/02/19 16:20:00 EST
470    Tag: blume-20020219-cmffi
471    Description:
472    
473    A wild mix of changes, some minor, some major:
474    
475    * All C FFI-related libraries are now anchored under $c:
476        $/c.cm      --> $c/c.cm
477        $/c-int.cm  --> $c/internals/c-int.cm
478        $/memory.cm --> $c/memory/memory.cm
479    
480    * "make" tool (in CM) now treats its argument pathname slightly
481      differently:
482        1. If the native expansion is an absolute name, then before invoking
483           the "make" command on it, CM will apply OS.Path.mkRelative
484           (with relativeTo = OS.FileSys.getDir()) to it.
485        2. The argument will be passed through to subsequent phases of CM
486           processing without "going native".  In particular, if the argument
487           was an anchored path, then "make" will not lose track of that anchor.
488    
489    * Compiler backends now "know" their respective C calling conventions
490      instead of having to be told about it by ml-nlffigen.  This relieves
491      ml-nlffigen from one of its burdens.
492    
493    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
494    
495    * Export C_DEBUG and C_Debug from $c/c.cm.
496    
497    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
498      subtyping relationship between incomplete pointers and their complete
499      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
500      with the convention of instantiating 'o with ('t, 'c) obj whenever
501      the pointer target type is complete.  In the incomplete case, 'o
502      will be instantiated with some "'c iobj" -- a type obtained by
503      using one of the functors PointerToIncompleteType or PointerToCompleteType.
504    
505      Operations that work on both incomplete and complete pointer types are
506      typed as taking an 'o ptr while operations that require the target to
507      be known are typed as taking some ('t, 'c) obj ptr.
508    
509      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
510      where void is an eqtype without any values.  This makes it possible
511      to work on voidptr values using functions meant to operate on light
512      incomplete pointers.
513    
514    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
515      been vastly simplified.
516    
517    ----------------------------------------------------------------------
518    Name: Matthias Blume
519    Date: 2002/02/19 10:48:00 EST
520    Tag: blume-20020219-pqfix
521    Description:
522    
523    Applied Chris Okasaki's bug fix for priority queues.
524    
525    ----------------------------------------------------------------------
526    Name: Matthias Blume
527    Date: 2002/02/15 17:05:00
528    Tag: Release_110_39
529    Description:
530    
531    Last-minute retagging is becoming a tradition... :-(
532    
533    This is the working release 110.39.
534    
535    ----------------------------------------------------------------------
536    Name: Matthias Blume
537    Date: 2002/02/15 16:00:00 EST
538    Tag: Release_110_39-orig
539    Description:
540    
541    Working release 110.39.  New bootfiles.
542    
543    (Update: There was a small bug in the installer so it wouldn't work
544    with all shells.  So I retagged. -Matthias)
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blume
548    Date: 2002/02/15 14:17:00 EST
549    Tag: blume-20020215-showbindings
550    Description:
551    
552    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
553    the latter can be useful for exploring what bindings are available at
554    the interactive prompt.  (The first function returns only the list
555    of symbols that are really bound, the second prints those but also the
556    ones that CM's autoloading mechanism knows about.)
557    
558    ----------------------------------------------------------------------
559    Name: Matthias Blume
560    Date: 2002/02/15 12:08:00 EST
561    Tag: blume-20020215-iptrs
562    Description:
563    
564    Two improvements to ml-nlffigen:
565    
566      1. Write files only if they do not exist or if their current contents
567         do not coincide with what's being written.  (That is, avoid messing
568         with the time stamps unless absolutely necessary.)
569    
570      2. Implement a "repository" mechanism for generated files related
571         to "incomplete pointer types".   See the README file for details.
572    
573    ----------------------------------------------------------------------
574    Name: Matthias Blume
575    Date: 2002/02/14 11:50:00 EST
576    Tag: blume-20020214-quote
577    Description:
578    
579    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
580    because of the new and improved tag generation scheme.  (Thanks to Allen
581    Leung for pointing it out.)
582    
583    ----------------------------------------------------------------------
584    Name: Lal George
585    Date: 2002/02/14 09:55:27 EST 2002
586    Tag: george-20020214-isabelle-bug
587    Description:
588    
589    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
590    of Isabelle on the x86.
591    
592    From Allen:
593    -----------
594     I've found the problem:
595    
596         in ra-core.sml, I use the counter "blocked" to keep track of the
597         true number of elements in the freeze queue.  When the counter goes
598         to zero, I skip examining the queue.  But I've messed up the
599         bookkeeping in combine():
600    
601             else ();
602             case !ucol of
603               PSEUDO => (if !cntv > 0 then
604                     (if !cntu > 0 then blocked := !blocked - 1 else ();
605                                        ^^^^^^^^^^^^^^^^^^^^^^^
606                      moveu := mergeMoveList(!movev, !moveu)
607                     )
608                  else ();
609    
610         combine() is called to coalesce two nodes u and v.
611         I think I was thinking that if the move counts of u and v are both
612         greater than zero then after they are coalesced then one node is
613         removed from the freeze queue.  Apparently I was thinking that
614         both u and v are of low degree, but that's clearly not necessarily true.
615    
616    
617    02/12/2002:
618        Here's the patch.  HOL now compiles.
619    
620        I don't know how this impact on performance (compile
621        time or runtime).  This bug caused the RA (especially on the x86)
622        to go thru the potential spill phase when there are still nodes on the
623        freeze queue.
624    
625    
626    
627    
628    ----------------------------------------------------------------------
629    Name: Matthias Blume
630    Date: 2002/02/13 22:40:00 EST
631    Tag: blume-20020213-fptr-rtti
632    Description:
633    
634    Fixed a bug in ml-nlffigen that was introduced with one of the previous
635    updates.
636    
637    ----------------------------------------------------------------------
638    Name: Matthias Blume
639    Date: 2002/02/13 16:41:00 EST
640    Tag: blume-20020213-cmlpq
641    Description:
642    
643    Added new priority queue export symbols (which have just been added to
644    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
645    not compile and the installer would choke.)
646    
647    ----------------------------------------------------------------------
648    Name: Matthias Blume
649    Date: 2002/02/13 16:15:00 EST
650    Tag: blume-20020213-various
651    Description:
652    
653    1. More tweaks to ml-nlffigen:
654    
655       - better internal datastructures (resulting in slight speedup)
656       - "-match" option requires exact match
657       - "localized" gensym counters (untagged structs/unions nested within
658         other structs/unions or within typedefs get a fresh counter; their
659         tag will be prefixed by a concatenation of their parents' tags)
660       - bug fixes (related to calculation of transitive closure of types
661         to be included in the output)
662    
663    2. Minor Basis updates:
664    
665       - added implementations for List.collate and Option.app
666    
667    ----------------------------------------------------------------------
668    Name: Matthias Blume
669    Date: 2002/02/11 15:55:00 EST
670    Tag: blume-20020211-gensym
671    Description:
672    
673    Added a "-gensym" option to command line of ml-nlffigen.  This can be
674    used to specify a "stem" -- a string that is inserted in all "gensym'd"
675    names (ML structure names that correspond to unnamed C structs, unions,
676    and enums), so that separate runs of ml-nlffigen do not clash.
677    
678    ----------------------------------------------------------------------
679    Name: Matthias Blume
680    Date: 2002/02/11 12:05:00 EST
681    Tag: blume-20020211-gensml
682    Description:
683    
684    A quick fix for a problem with GenSML (in the pgraph-util library):
685    Make generation of toplevel "local" optional.  (Strictly speaking,
686    signature definitions within "local" are not legal SML.)
687    
688    Other than that: updates to INSTALL and cm/TODO.
689    
690    ----------------------------------------------------------------------
691    Name: Matthias Blume
692    Date: 2002/02/08 15:00:00 EST
693    Tag: blume-20020208-uniquepid
694    Description:
695    
696    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
697    
698    1. The installer (config/install.sh) has gotten smarter:
699    
700         - Configuration options are a bit easier to specify now
701           (in config/targets).
702         - Bug in recognizing .tar.bz2 files fixed.
703         - Installer automatically resolves dependencies between
704           configuration options (e.g., if you ask for eXene, you will
705           also get cml -- regardless whether you asked for it or not).
706         - Installer can run in "quieter mode" by setting the environment
707           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
708           "completely silent", though.
709         - Build HashCons library as part of smlnj-lib.
710    
711    2. A new scheme for assigning persistent identifiers to compilation
712       units (and, by extension, to types etc.) has been put into place.
713       This fixes a long-standing bug where types and even dynamic values
714       can get internally confused, thereby compromising type safety
715       (abstraction) and dynamic correctness.  See
716    
717         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
718    
719       for an example of how things could go wrong until now.
720    
721       The downside of the new scheme is that pids are not quite as
722       persistent as they used to be: CM will generate a fresh pid
723       for every compilation unit that it thinks it sees for the first
724       time.  That means that if you compile starting from a clean, fresh
725       source tree at two different times, you end up with different
726       binaries.
727    
728       Cutoff recompilation, however, has not been compromised because
729       CM keeps pid information in special caches between runs.
730    
731    ----------------------------------------------------------------------
732    Name: Lal George
733    Date: 2002/02/07 15:34:13 EST 2002
734    Tag: <none>
735    Description:
736    
737    Compilers that generate assembly code may produce  global labels
738    whose value is resolved at link time. The various peephole optimization
739    modules did not take this in account.
740    
741    TODO. The Labels.addrOf function should really return an option
742    type so that clients are forced to deal with this issue, rather
743    than an exception being raised.
744    
745    ----------------------------------------------------------------------
746    Name: Lal George
747    Date: 2002/02/06 13:55:02 EST
748    Tag: george-20020206-ra-breakup
749    Description:
750    
751    1. A bug fix from Allen.
752    
753        A typo causes extra fstp %st(0)'s to be generated at compensation
754        edges, which might cause stack underflow traps at runtime.  This
755        occurs in fft where there are extraneous fstps right before the 'into'
756        trap instruction (in this case they are harmless since none of the
757        integers overflow.)
758    
759    2. Pulled out various utility modules that were embedded in the modules
760       of the register allocator. I need these modules for other purposes, but
761       they are not complete enough to put into a library (just yet).
762    ----------------------------------------------------------------------
763    Name: Matthias Blume
764    Date: 2002/01/31 16:05:00 EST
765    Tag: blume-20020131-sparc-ccalls
766    Description:
767    
768    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
769       of extra stack space by mistake.  Fixed.
770    
771    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
772    
773    ----------------------------------------------------------------------
774    Name: Allen Leung
775    Date: 2002/01/30
776    Tag: leunga-20020130-nowhere-bug-fix
777    Description:
778    
779       MLRISC bug fixes:
780       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
781       2. MachineInt.fromString was negating its value.
782    
783    ----------------------------------------------------------------------
784    Name: Matthias Blume
785    Date: 2002/01/29
786    Tag: blume-20020129-INSTALL
787    Description:
788    
789    - Added somewhat detailed installation instructions (file INSTALL).
790    - Fixed curl-detection bug in config/install.sh.
791    - It is now possible to select the URL getter using the URLGETTER
792      environment variable:
793    
794          not set / "unknown"      --> automatic detection (script tries wget,
795                                       curl, and lynx)
796          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
797                                       how to properly invoke them)
798          other                    --> use $URLGETTER directly, it must take
799                                       precisely two command-line arguments
800                                       (source URL and destination file name)
801    
802    ----------------------------------------------------------------------
803    Name: Matthias Blume
804    Date: 2002/01/28
805    Tag: blume-20020128-sparc-ccalls
806    Description:
807    
808    - Fixed problem with calculation of "used" registers in sparc-c-calls.
809    - Make use of the allocParam argument in sparc-c-calls.
810    
811    ----------------------------------------------------------------------
812    Name: Matthias Blume
813    Date: 2002/01/28
814    Tag: blume-20020128-allocParam
815    Description:
816    
817    John Reppy:  Changes c-calls API to accept client-callback for
818    allocating extra stack space.
819    me: Corresponding changes to mlriscGen (using a dummy argument that
820        does not change the current behavior).
821    
822    ----------------------------------------------------------------------
823    Name: Matthias Blume
824    Date: 2002/01/28 12:00:00
825    Tag: Release_110_38
826    Description:
827    
828    This time for real!!!
829    
830    ----------------------------------------------------------------------
831    Name: Matthias Blume
832    Date: 2002/01/28 10:56:00 EST
833    Tag: blume-20020128-retraction
834    Description:
835    
836    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
837       with blume-Release_110_38-retracted.)
838    
839    1. Fixed a problem with incorrect rounding modes in real64.sml.
840       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
841    
842    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
843       and enums fixed.  The naming of corresponding ML identifiers should
844       now be consistent again.
845    
846    ----------------------------------------------------------------------
847    Name: Allen Leung
848    Date: 2002/01/27
849    Tag: leunga-20020127-nowhere
850    Description:
851    
852       Added a target called nowhere in the configuration scripts.
853       Enabling this will build the MLRISC 'nowhere' tool (for translating
854       programs with where-clauses into legal SML code) during installation.
855    
856    ----------------------------------------------------------------------
857    Name: Matthias Blume
858    Date: 2002/01/25 21:27:00 EST
859    Tag: blume-Release_110_38-retracted
860    Description:
861    
862    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
863    
864    README will be added later.
865    
866    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
867    (Release_110_38) removed.  Reason: Last-minute bug fixes.
868    
869    ----------------------------------------------------------------------
870    Name: Matthias Blume
871    Date: 2002/01/25
872    Tag: blume-20020125-ffi
873    Description:
874    
875    A large number of tweaks and improvements to ml-nlffi-lib and
876    ml-nlffigen:
877    
878       - ML represenation types have been streamlined
879       - getter and setter functions work with concrete values, not abstract
880         ones where possible
881       - ml-nlffigen command line more flexible (see README file there)
882       - some bugs have been fixed (hopefully)
883    
884    ----------------------------------------------------------------------
885    Name: Lal George
886    Date: 2002/01/24
887    Tag: george-20020124-risc-ra-interface
888    Description:
889    
890       There is a dramatic simplification in the interface to the
891       register allocator for RISC architectures as a result of making
892       parallel copy instructions explicit.
893    
894    ----------------------------------------------------------------------
895    Name: Matthias Blume
896    Date: 2002/01/22
897    Tag: blume-20020122-x86-ccalls
898    Description:
899    
900    Bug fix for c-calls on x86 (having to do with how char- and
901    short-arguments are being handled).
902    
903    ----------------------------------------------------------------------
904    Name: Matthias Blume
905    Date: 2002/01/21
906    Tag: blume-20020121-ff
907    Description:
908    
909    Another day of fiddling with the FFI...
910    
911    1. Bug fix/workaround:  CKIT does not complain about negative array
912       dimensions, so ml-nlffigen has to guard itself against this possibility.
913       (Otherwise a negative dimension would send it into an infinite loop.)
914    
915    2. Some of the abstract types (light objects, light pointers, most "base"
916       types) in structure C are now eqtypes.
917    
918    3. Added constructors and test functions for NULL function pointers.
919    
920    ----------------------------------------------------------------------
921    Name: Matthias Blume
922    Date: 2002/01/18
923    Tag: blume-20020118-ready-for-new-release
924    Description:
925    
926    Made config/srcarchiveurl point to a new place.  (Will provide boot
927    files shortly.)
928    
929    Maybe we christen this to be 110.38?
930    
931    ----------------------------------------------------------------------
932    Name: Matthias Blume
933    Date: 2002/01/18
934    Tag: blume-20020118-more-ffifiddle
935    Description:
936    
937    Today's FFI fiddling:
938    
939      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
940        C.Set.* that use concrete (MLRep.*) arguments and results instead
941        of abstract ones.
942    
943      - Provided word-style bit operations etc. for "int" representation
944        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
945        and Long.
946    
947    ----------------------------------------------------------------------
948    Name: Matthias Blume
949    Date: 2002/01/18
950    Tag: blume-20020118-use-x86-fp
951    Description:
952    
953    Now that x86-fast-fp seems to be working, I turned it back on again
954    by default.  (Seems to work fine now, even with the FFI.)
955    
956    Other than that, I added some documentation about the FFI to
957    src/ml-nlffigen/README and updated the FFI test examples in
958    src/ml-nlffi-lib/Tests/*.
959    
960    ----------------------------------------------------------------------
961    Name: Allen Leung
962    Date: 2002/01/17
963    Tag: leunga-20020117-x86-fast-fp-call
964    Description:
965    
966       1. Fixed a problem with handling return fp values when x86's fast fp
967          mode is turned on.
968    
969       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
970          of %f32.
971    
972       3. Added a constructor INT32lit to the ast of MLRISC tools.
973    
974    ----------------------------------------------------------------------
975    Name: Matthias Blume
976    Date: 2002/01/16
977    Tag: blume-20020116-ffifiddle
978    Description:
979    
980    More fiddling with the FFI interface:
981    
982     - Make constness 'c instead of rw wherever possible.  This eliminates
983       the need for certain explicit coercions.  (However, due to ML's
984       value polymorphism, there will still be many cases where explicit
985       coercions are necessary.  Phantom types are not the whole answer
986       to modeling a subtyping relationship in ML.)
987    
988     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
989    
990     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
991       clumsy workaround for varargs problem.)
992    
993    ----------------------------------------------------------------------
994    Name: Lal George
995    Date: 2002/01/15
996    Tag: <none>
997    Description:
998    
999    1. Since COPY instructions are no longer native to the architecture,
1000       a generic functor can be used to implement the expandCopies function.
1001    
1002    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1003       TEXT segment.
1004    
1005    ----------------------------------------------------------------------
1006    Name: Matthias Blume
1007    Date: 2002/01/15
1008    Tag: blume-20020115-ffiupdates
1009    Description:
1010    
1011    1. Fix for bug resulting in single-precision float values being returned
1012       incorrectly from FFI calls.
1013    
1014    2. Small modifications to C FFI API:
1015    
1016        - memory-allocation routines return straight objects (no options)
1017          and raise an exception in out-of-memory situations
1018        - unsafe extensions to cast between function pointers and pointers
1019          from/to ints
1020        - added structure C_Debug as an alternative to structure C where
1021          pointer-dereferencing (|*| and |*!) always check for null-pointers
1022        - added open_lib' to DynLinkage;  open_lib' works like open_lib
1023          but also takes a (possibly empty) list of existing library handles
1024          that the current library depends on
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2002/01/10
1029    Tag: blume-20020110-newffigen
1030    Description:
1031    
1032    1. Updates to portable graph code.
1033    
1034    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
1035       more scalable now so that even huge interfaces such as the one
1036       for GTK compile in finite time and space. :-)
1037       See src/ml-nlffigen/README for details on what's new.
1038    
1039    ----------------------------------------------------------------------
1040    Name: Lal George
1041    Date: 2001/01/09 14:31:35 EST 2002
1042    Tag: george-20011206-rm-native-copy
1043    Description:
1044    
1045            Removed the native COPY and FCOPY instructions
1046            from all the architectures and replaced it with the
1047            explicit COPY instruction from the previous commit.
1048    
1049            It is now possible to simplify many of the optimizations
1050            modules that manipulate copies. This has not been
1051            done in this change.
1052    
1053    ----------------------------------------------------------------------
1054    Name: Lal George
1055    Date: 2001/12/06 16:50:13 EST 2001
1056    Tag: george-20011206-mlrisc-instruction
1057    Description:
1058    
1059    Changed the representation of instructions from being fully abstract
1060    to being partially concrete. That is to say:
1061    
1062      from
1063            type instruction
1064    
1065      to
1066            type instr                              (* machine instruction *)
1067    
1068            datatype instruction =
1069                LIVE of {regs: C.cellset, spilled: C.cellset}
1070              | KILL of {regs: C.cellset, spilled: C.cellset}
1071              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1072              | ANNOTATION of {i: instruction, a: Annotations.annotation}
1073              | INSTR of instr
1074    
1075    This makes the handling of certain special instructions that appear on
1076    all architectures easier and uniform.
1077    
1078    LIVE and KILL say that a list of registers are live or killed at the
1079    program point where they appear. No spill code is generated when an
1080    element of the 'regs' field is spilled, but the register is moved to
1081    the 'spilled' (which is present, more for debugging than anything else).
1082    
1083    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1084    We used to generate:
1085    
1086            DEFFREG f1
1087            f1 := f2 + f3
1088            trapb
1089    
1090    but now generate:
1091    
1092            f1 := f2 + f3
1093            trapb
1094            LIVE {regs=[f1,f2,f3], spilled=[]}
1095    
1096    Furthermore, the DEFFREG (hack) required that all floating point instruction
1097    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1098    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1099    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1100    this architecture).
1101    
1102    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
1103    all the architectures. This will result in further simplification of the
1104    register allocator that must be aware of them for coalescing purposes, and
1105    will also simplify certain aspects of the machine description that provides
1106    callbacks related to parallel copies.
1107    
1108    ANNOTATION should be obvious, and now INSTR represents the honest to God
1109    machine instruction set!
1110    
1111    The <arch>/instructions/<arch>Instr.sml files define certain utility
1112    functions for making porting easier -- essentially converting upper case
1113    to lower case. All machine instructions (of type instr) are in upper case,
1114    and the lower case form generates an MLRISC instruction. For example on
1115    the alpha we have:
1116    
1117      datatype instr =
1118         LDA of {r:cell, b:cell, d:operand}
1119       | ...
1120    
1121      val lda : {r:cell, b:cell, d:operand} -> instruction
1122        ...
1123    
1124    where lda is just (INSTR o LDA), etc.
1125    
1126    ----------------------------------------------------------------------
1127    Name: Matthias Blume
1128    Date: 2001/11/22 21:40:00 EST
1129    Tag: Release_110_37
1130    Description:
1131    
1132    Release 110.37.  This time for real.
1133    
1134    ----------------------------------------------------------------------
1135    Name: Matthias Blume
1136    Date: 2001/11/21 16:35:00 EST
1137    Tag: blume-20011121-foot-in-mouth
1138    Description:
1139    
1140    Removed the "Release_110_37" tag because of a serious bug.
1141    This will be re-tagged once the bug is fixed.
1142    
1143    ----------------------------------------------------------------------
1144    Name: Matthias Blume
1145    Date: 2001/11/21 16:14:00 EST
1146    Tag: blume-20011121-forgottenfile
1147    Description:
1148    
1149    Forgot to add a file.  (Just a .tex-file -- part of
1150    the CM manual source.)
1151    
1152    ----------------------------------------------------------------------
1153    Name: Matthias Blume
1154    Date: 2001/11/21 16:10:00 EST
1155    Tag: blume-20011121-invalid_110_37
1156    Description:
1157    
1158    Note: I removed the original tag "Release_110_37" from this commit
1159          because we found a serious bug in all non-x86 backends.
1160          - Matthias
1161    
1162    1. Modifications to the SML/NJ code generator and to the runtime system
1163       so that code object name strings are directly inserted into code
1164       objects at code generation time.  The only business the runtime system
1165       has with this is now to read the name strings on occasions.
1166       (The encoding of the name string has also changed somewhat.)
1167    
1168    2. CM now implements a simple "set calculus" for specifying export lists.
1169       In particular, it is now possible to refer to the export lists of
1170       other libraries/groups/sources and form unions as well as differences.
1171       See the latest CM manual for details.
1172    
1173    3. An separate notion of "proxy" libraries has again be eliminated from
1174       CM's model.  (Proxy libraries are now simply a special case of using
1175       the export list calculus.)
1176    
1177    4. Some of the existing libraries now take advantage of the new set
1178       calculus.
1179       (Notice that not all libraries have been converted because some
1180       of the existing .cm-files are supposed to be backward compatible
1181       with 110.0.x.)
1182    
1183    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
1184       "General.exnMessage"!  The former relies on a certain hook to be
1185       initialized, and that often does not happen in the stand-alone case.)
1186    
1187    ----------------------------------------------------------------------
1188    Name: Lal George
1189  Date: 2001/11/21  13:56:18 EST  Date: 2001/11/21  13:56:18 EST
1190  Tag: george-2001121-pseudo-ops  Tag: george-2001121-pseudo-ops
1191  Description:  Description:
# Line 208  Line 1380 
1380    
1381  ----------------------------------------------------------------------  ----------------------------------------------------------------------
1382  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
1383  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
1384  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
1385  Description:  Description:

Legend:
Removed from v.984  
changed lines
  Added in v.1133

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