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 833, Tue May 22 22:06:12 2001 UTC revision 898, Mon Aug 13 17:44:08 2001 UTC
# Line 14  Line 14 
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name: Matthias Blume  Name: Matthias Blume
17    Date: 2001/08/13
18    Tag: blume-20010813-closures
19    Description:
20    
21    This fix was sent to us by Zhong Shao.  It is supposed to improve the
22    performance of certain loops by avoiding needless closure allocation.
23    
24    ----------------------------------------------------------------------
25    Name: Lal George
26    Date: 2001/07/31 10:03:23 EDT 2001
27    Tag: george-20010731-x86-fmalloc
28    Description: Fixed bug in x86 calls
29    
30        There was a bug where call instructions would mysteriously
31        vanish. The call instruction had to be one that returned
32        a floating point value.
33    
34    ----------------------------------------------------------------------
35    Name: Lal George
36    Date: 2001/07/19 16:36:29 EDT 2001
37    Tag: george-20010719-simple-cells
38    Description:
39    
40    I have dramatically simplified the interface for CELLS in MLRISC.
41    
42    In summary, the cells interface is broken up into three parts:
43    
44      1. CellsBasis : CELLS_BASIS
45    
46            CellsBasis is a top level structure and common for all
47            architectures.  it contains the definitions of basic datatypes
48            and utility  functions over these types.
49    
50      2. functor Cells() : CELLS
51    
52            Cells generates an interface for CELLS that incorporates the
53            specific resources on the target architecture, such as the
54            presence of special register classes, their number and size,
55            and various useful substructures.
56    
57      3. <ARCH>CELLS
58    
59            e.g. SparcCells: SPARCCELLS
60    
61            <ARCH>CELLS usually contains additional bindings for special
62            registers  on the architecture, such as:
63    
64                    val r0 : cell           (* register zero *)
65                    val y : cell            (* Y register *)
66                    val psr : cell          (* processor status register *)
67                    ...
68    
69            The structure returned by applying the Cells functor is opened
70            in this interface.
71    
72    The main implication of all this is that the datatypes for cells is
73    split between CellsBasis and CELLS -- a fairly simple change for user
74    code.
75    
76    In the old scheme the CELLS interface had a definitional binding of
77    the form:
78    
79            signature CELLS = sig
80    
81               structure CellsBasis = CellsBasis
82    
83               ...
84    
85            end
86    
87    With all the sharing constraints that goes on in MLRISC, this old
88    design  quickly leads to errors such as:
89    
90            "structure definition spec inside of sharing ... "
91    
92    
93    and appears to require an unacceptable amount of sharing and where
94    constraint hackery.
95    
96    I think this error message (the interaction of definitional specs and
97    sharing) requires more explanation on our web page.
98    
99    ----------------------------------------------------------------------
100    Name: Matthias Blume
101    Date: 2001/07/19 15:00:00 EDT
102    Tag: blume-20010719-libreorg
103    Description:
104    
105    This update puts together a fairly extensive but straightforward change
106    to the way the libraries that implement the interactive system are
107    organized:
108    
109       The biggest change is the elimination of structure Compiler.  As a
110       replacement for this structure, there is now a CM library
111       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
112       that exports all the substructures of the original structure Compiler
113       directly.  So instead of saying Compiler.Foo.bar one now simply
114       says Foo.bar.  (The CM libraries actually export a collection of
115       structures that is richer than the collection of substructures of
116       structure Compiler.)
117    
118       To make the transition smooth, there is a separate library called
119       $smlnj/compiler/compiler.cm which puts together and exports the
120       original structure Compiler (or at least something very close to it).
121    
122       There are five members of the original structure Compiler
123       that are not exported directly but which instead became members
124       of a new structure Backend (described by signature BACKEND).  These are:
125       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
126       Interact (: INTERACT), structure Machine (: MACHINE), and val
127       architecture (: string).
128    
129       Structure Compiler.Version has become structure CompilerVersion.
130    
131       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
132       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
133       or x86, respectively.
134       Each of these exports the same frontend structures that
135       $smlnj/compiler.cm exports.  But they do not have a structure Backend
136       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
137       Hppa, PPC, Sparc, or X86, respectively.
138    
139       Library $smlnj/compiler/all.cm exports the union of the exports of
140       $smlnj/compiler/<arch>.cm
141    
142       There are no structures <Arch>Compiler anymore, use
143       $smlnj/compiler/<arch>.cm instead.
144    
145       Library host-compiler-0.cm is gone.  Instead, the internal library
146       that instantiates CM is now called cm0.cm.  Selection of the host
147       compiler (backend) is no longer done here but. (Responsibility for it
148       now lies with $smlnj/compiler/current.cm.  This seems to be more
149       logical.)
150    
151       Many individual files have been moved or renamed.  Some files have
152       been split into multiple files, and some "dead" files have been deleted.
153    
154    Aside from these changes to library organization, there are also changes
155    to the way the code itself is organized:
156    
157       Structure Binfile has been re-implemented in such a way that it no
158       longer needs any knowledge of the compiler.  It exclusively deals
159       with the details of binfile layout.  It no longer invokes the
160       compiler (for the purpose of creating new prospective binfile
161       content), and it no longer has any knowledge of how to interpret
162       pickles.
163    
164       Structure Compile (: COMPILE) has been stripped down to the bare
165       essentials of compilation.  It no longer deals with linking/execution.
166       The interface has been cleaned up considerably.
167    
168       Utility routines for dealing with linking and execution have been
169       moved into their own substructures.
170    
171       (The ultimate goal of these changes is to provide a light-weight
172       binfile loader/linker (at least for, e.g., stable libraries) that
173       does not require CM or the compiler to be present.)
174    
175    CM documentation has been updated to reflect the changes to library
176    organization.
177    
178    ----------------------------------------------------------------------
179    Name: Matthias Blume
180    Date: 2001/07/10 17:30:00 EDT
181    Tag: Release_110_34
182    Description:
183    
184    Minor tweak to 110.34 (re-tagged):
185    
186      - README.html file added to CVS repository
187      - runtime compiles properly under FreeBSD 3.X and 4.X
188    
189    ----------------------------------------------------------------------
190    Name: Matthias Blume
191    Date: 2001/07/10 17:30:00 EDT
192    Tag: Release_110_34
193    Description:
194    
195    New version number (110.34). New bootfiles.
196    
197    ----------------------------------------------------------------------
198    Name: Matthias Blume
199    Date: 2001/07/09 16:00:00 EDT
200    Tag: blume-20010709-more-varargs
201    Description:
202    
203    I changed the handling of varargs in ml-nlffigen again:
204    The ellipsis ... will now simply be ignored (with an accompanying warning).
205    
206    The immediate effect is that you can actually call a varargs function
207    from ML -- but you can't actually supply any arguments beyond the ones
208    specified explicitly.  (For example, you can call printf with its format
209    string, but you cannot pass additional arguments.)
210    
211    This behavior is only marginally more useful than the one before, but
212    it has the advantage that a function or, more importantly, a function
213    type never gets dropped on the floor, thus avoiding follow-up problems with
214    other types that refer to the offending one.
215    
216    ----------------------------------------------------------------------
217    Name: Matthias Blume
218    Date: 2001/07/09 11:25:00 EDT
219    Tag: blume-20010709-varargs
220    Description:
221    
222    1. ckit-lib.cm now exports structure Error
223    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
224       with a warning accompanied by a source location.  Moreover, it
225       merely skips the offending function or type and proceeds with the
226       rest of its work.u  As a result, one can safely feed C code containing
227       "..." to ml-nlffigen.
228    3. There are some internal improvements to CM, providing slightly
229       more general string substitutions in the tools subsystem.
230    
231    ----------------------------------------------------------------------
232    Name: Matthias Blume
233    Date: 2001/06/27 15:10:00 EDT
234    Tag: blume-20010627-concur
235    Description:
236    
237    Fixed a small bug in CM's handling of parallel compilation.
238    (You could observe the bug by Control-C-interrupting an ordinary
239    CMB.make or CM.stabilize and then attaching some compile servers.
240    The result was that all of a sudden the previously interrupted
241    compilation would continue on its own.  This was because of
242    an over-optimization: CM did not bother to clean out certain queues
243    when no servers were attached "anyway", resulting in the contents
244    of these queues to grab control when new servers did get attached.)
245    
246    There is also another minor update to the CM manual.
247    
248    ----------------------------------------------------------------------
249    Name: Matthias Blume
250    Date: 2001/06/26 16:15:00 EDT
251    Tag: blume-20010626-cmdoc
252    Description:
253    
254    Minor typo fixed in CM manual (syntax diagram for libraries).
255    
256    ----------------------------------------------------------------------
257    Name: Matthias Blume
258    Date: 2001/06/25 22:55:00 EDT
259    Tag: blume-20010625-x86pc
260    Description:
261    
262    Fixed a nasty bug in the X86 assembly code that caused signal
263    handlers to fail (crash) randomly.
264    
265    ----------------------------------------------------------------------
266    Name: Matthias Blume
267    Date: 2001/06/25 12:05:00 EDT
268    Tag: blume-20010625-nlffigen
269    Description:
270    
271    This update fixes a number of minor bugs in ml-nlffigen as reported by
272    Nick Carter <nbc@andrew.cmu.edu>.
273    
274      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
275      2. Default names for generated files are now derived from the name of
276         the C file *without its directory*.  In particular, this causes generated
277         files to be placed locally even if the C file is in some system directory.
278      3. Default names for generated signatures and structures are also derived
279         from the C file name without its directory.  This avoids silly things
280         like "structure GL/GL".
281         (Other silly names are still possible because ml-nlffigen does not do
282          a thorough check of whether generated names are legal ML identifiers.
283          When in doubt, use command line arguments to force particular names.)
284    
285    ----------------------------------------------------------------------
286    Name: Matthias Blume
287    Date: 2001/06/21 12:25:00 EDT
288    Tag: blume-20010621-eXene
289    Description:
290    
291    eXene now compiles and (sort of) works again.
292    
293    The library name (for version > 110.33) is $/eXene.cm.
294    
295    I also added an new example in src/eXene/examples/nbody.  See the
296    README file there for details.
297    
298    ----------------------------------------------------------------------
299    Name: Matthias Blume
300    Date: 2001/06/20 16:40:00 EDT
301    Tag: blume-20010620-cml
302    Description:
303    
304    CML now compiles and works again.
305    
306    Libraries (for version > 110.33):
307    
308      $cml/cml.cm            Main CML library.
309      $cml/basis.cm          CML's version of $/basis.cm.
310      $cml/cml-internal.cm   Internal helper library.
311      $cml/core-cml.cm       Internal helper library.
312      $cml-lib/trace-cml.cm  Tracing facility.
313      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
314    
315    The installer (config/install.sh) has been taught how to properly
316    install this stuff.
317    
318    ----------------------------------------------------------------------
319    Name: Matthias Blume
320    Date: 2001/06/19 17:55:00 EDT
321    Tag: blume-20010619-instantiate
322    Description:
323    
324    This un-breaks the fix for bug 1432.
325    (The bug was originally fixed in 110.9 but I broke it again some
326    time after that.)
327    
328    ----------------------------------------------------------------------
329    Name: Matthias Blume
330    Date: 2001/06/19 17:25:00 EDT
331    Tag: blume-20010619-signals
332    Description:
333    
334    This should (hopefully) fix the long-standing signal handling bug.
335    (The runtime system was constructing a continuation record with an
336    incorrect descriptor which would cause the GC to drop data on the floor...)
337    
338    ----------------------------------------------------------------------
339    Name: Matthias Blume
340    Date: 2001/06/15 15:05:00 EDT
341    Tag: blume-20010615-moresparc
342    Description:
343    
344    Here is a short late-hour update related to Sparc c-calls:
345    
346     -- made handling of double-word arguments a bit smarter
347    
348     -- instruction selection phase tries to collapse certain clumsily
349        constructed ML-Trees; typical example:
350    
351            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
352    
353        This currently has no further impact on SML/NJ since mlriscGen does
354        not seem to generate such patterns in the first place, and c-calls
355        (which did generate them in the beginning) has meanwhile been fixed
356        so as to avoid them as well.
357    
358    ----------------------------------------------------------------------
359    Name: Matthias Blume
360    Date: 2001/06/15 15:05:00 EDT
361    Tag: blume-20010615-sparc
362    Description:
363    
364    The purpose of this update is to provide an implementation of NLFFI
365    on Sparc machines.
366    
367    Here are the changes in detail:
368    
369       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
370       the Sparc implementation of the c-calls API.
371       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
372       for accessing the ML frame.  Thus, we have a real frame pointer and
373       can freely modify %sp without need for an omit-frame-ptr phase.
374       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
375       this case.
376       * ml-nlffigen has been taught to produce code for different architectures
377       and calling conventions.
378       * In a way similar to what was done in the x86 case, the Sparc
379       backend uses its own specific extension to mltree.  (For example,
380       it needs to be able to generate UNIMP instructions which are part
381       of the calling convention.)
382       * ml-nlffi-lib was reorganized to make it more modular (in particular,
383       to make it easier to plug in new machine- and os-dependent parts).
384    
385    There are some other fairly unrelated bug fixes and cleanups as well:
386    
387       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
388       that they properly share their libraries with existing SML/NJ libraries.
389       * I fixed a minor cosmetic bug in CM, supressing certain spurious
390       follow-up error messages.
391       * Updates to CM/CMB documentation.
392    
393    TODO items:
394    
395       * MLRISC should use a different register as its asmTemp on the Sparc.
396         (The current %o2 is a really bad choice because it is part of the
397         calling conventions, so things might interfere in unexpected ways.)
398    
399    ----------------------------------------------------------------------
400    Name: Matthias Blume
401    Date: 2001/06/07
402    Tag: blume-20010607-calls
403    Description:
404    
405    A number of internal changes related to C calls and calling conventions:
406    
407    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
408       number of bytes popped implicitly (by the callee).  In most cases
409       this field is 0 but on x86/win32 it is some non-zero value.  This
410       is information provided for the benefit of the "omit-frameptr" pass.
411    2. The CALL instruction on the x86 carries a similar "pops" field.
412       The instruction selection phase copies its value from the ML-Tree
413       CALL statement.
414    3. On all other architectures, the instruction selection phase checks
415       whether "pops=0" and complains if not.
416    4. The c-calls implementation for x86 now accepts two calling conventions:
417       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
418       up after the call and pops is set to 0.  For "stdcall", the caller
419       does nothing, leaving the cleanup to the callee; pops is set to
420       the number of bytes that were pushed onto the stack.
421    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
422       distinguish between "ccall" and "stdcall".
423    6. The UNIMP instruction has been added to the supported Sparc instruction
424       set. (This is needed for implementing the official C calling convention
425       on this architecture.)
426    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
427       work with the latest CM.
428    
429    ----------------------------------------------------------------------
430    Name: Matthias Blume
431    Date: 2001/06/05 15:10:00 EDT
432    Tag: blume-20010605-cm-index
433    Description:
434    
435    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
436    
437    1. CM can now generate "index files".  These are human-readable files
438       that list on a per-.cm-file basis each toplevel symbol defined or
439       imported.  The location of the index file for
440       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
441       To enable index-file generation, set CM.Control.generate_index to true
442       or export an environment-symbol: export CM_GENERATE_INDEX=true.
443    
444       The CM manual has been updated accordingly.
445    
446    2. I made some slight modifications to the c-calls API in MLRISC.
447    
448         a) There is now a callback to support saving/restoring of
449            dedicated but caller-save registers around the actual call
450            instruction.
451         b) One can optionally specify a comment-annotation for the
452            call instruction.
453    
454    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
455       (For example, the comment annotation shows the C prototype of
456        the function being called.)
457    
458    ----------------------------------------------------------------------
459    Name: Matthias Blume
460    Date: 2001/06/01 13:30:00 EDT
461    Tag: blume-20010601-nlffi-cleanup
462    Description:
463    
464    This is mostly a cleanup of MLFFI stuff:
465    
466       - some signature files have been put into a more exposed place
467       - the ugly 'f type parameter is gone (simplifies types tremendously!)
468       - ml-nlffigen changed accordingly
469       - tutorial updated
470    
471    Other changes:
472    
473       - author's affiliation in CM manual(s) updated
474       - some more recognized keywords added to Allen's sml.sty
475    
476    ----------------------------------------------------------------------
477    Name: Matthias Blume
478    Date: 2001/05/25 15:30:00 EDT
479    Tag: blume-20010525-iptr
480    Description:
481    
482      - put the official 110.33-README (as it appears on the ftp server) under
483        CVS
484      - fixed a small bug related to incomplete pointer types in
485        ml-nlffigen
486      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
487        (it does not need the 'f type parameter)
488    
489    ----------------------------------------------------------------------
490    Name: Matthias Blume
491    Date: 2001/05/23 14:30:00 EDT
492    Tag: Release_110_33
493    Description:
494    
495    New version number (110.33).  New bootfiles.
496    
497    ----------------------------------------------------------------------
498    Name: Matthias Blume
499  Date: 2001/05/22 18:06:00 EDT  Date: 2001/05/22 18:06:00 EDT
500  Tag: blume-20010522-targets  Tag: blume-20010522-targets
501  Description:  Description:

Legend:
Removed from v.833  
changed lines
  Added in v.898

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