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 771, Sat Dec 30 13:06:09 2000 UTC revision 842, Tue Jun 19 21:24:35 2001 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume
17    Date: 2001/06/19 17:25:00 EDT
18    Tag: blume-20010619-signals
19    Description:
20    
21    This should (hopefully) fix the long-standing signal handling bug.
22    (The runtime system was constructing a continuation record with an
23    incorrect descriptor which would cause the GC to drop data on the floor...)
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume
27    Date: 2001/06/15 15:05:00 EDT
28    Tag: blume-20010615-moresparc
29    Description:
30    
31    Here is a short late-hour update related to Sparc c-calls:
32    
33     -- made handling of double-word arguments a bit smarter
34    
35     -- instruction selection phase tries to collapse certain clumsily
36        constructed ML-Trees; typical example:
37    
38            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
39    
40        This currently has no further impact on SML/NJ since mlriscGen does
41        not seem to generate such patterns in the first place, and c-calls
42        (which did generate them in the beginning) has meanwhile been fixed
43        so as to avoid them as well.
44    
45    ----------------------------------------------------------------------
46    Name: Matthias Blume
47    Date: 2001/06/15 15:05:00 EDT
48    Tag: blume-20010615-sparc
49    Description:
50    
51    The purpose of this update is to provide an implementation of NLFFI
52    on Sparc machines.
53    
54    Here are the changes in detail:
55    
56       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
57       the Sparc implementation of the c-calls API.
58       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
59       for accessing the ML frame.  Thus, we have a real frame pointer and
60       can freely modify %sp without need for an omit-frame-ptr phase.
61       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
62       this case.
63       * ml-nlffigen has been taught to produce code for different architectures
64       and calling conventions.
65       * In a way similar to what was done in the x86 case, the Sparc
66       backend uses its own specific extension to mltree.  (For example,
67       it needs to be able to generate UNIMP instructions which are part
68       of the calling convention.)
69       * ml-nlffi-lib was reorganized to make it more modular (in particular,
70       to make it easier to plug in new machine- and os-dependent parts).
71    
72    There are some other fairly unrelated bug fixes and cleanups as well:
73    
74       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
75       that they properly share their libraries with existing SML/NJ libraries.
76       * I fixed a minor cosmetic bug in CM, supressing certain spurious
77       follow-up error messages.
78       * Updates to CM/CMB documentation.
79    
80    TODO items:
81    
82       * MLRISC should use a different register as its asmTemp on the Sparc.
83         (The current %o2 is a really bad choice because it is part of the
84         calling conventions, so things might interfere in unexpected ways.)
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2001/06/07
89    Tag: blume-20010607-calls
90    Description:
91    
92    A number of internal changes related to C calls and calling conventions:
93    
94    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
95       number of bytes popped implicitly (by the callee).  In most cases
96       this field is 0 but on x86/win32 it is some non-zero value.  This
97       is information provided for the benefit of the "omit-frameptr" pass.
98    2. The CALL instruction on the x86 carries a similar "pops" field.
99       The instruction selection phase copies its value from the ML-Tree
100       CALL statement.
101    3. On all other architectures, the instruction selection phase checks
102       whether "pops=0" and complains if not.
103    4. The c-calls implementation for x86 now accepts two calling conventions:
104       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
105       up after the call and pops is set to 0.  For "stdcall", the caller
106       does nothing, leaving the cleanup to the callee; pops is set to
107       the number of bytes that were pushed onto the stack.
108    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
109       distinguish between "ccall" and "stdcall".
110    6. The UNIMP instruction has been added to the supported Sparc instruction
111       set. (This is needed for implementing the official C calling convention
112       on this architecture.)
113    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
114       work with the latest CM.
115    
116    ----------------------------------------------------------------------
117    Name: Matthias Blume
118    Date: 2001/06/05 15:10:00 EDT
119    Tag: blume-20010605-cm-index
120    Description:
121    
122    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
123    
124    1. CM can now generate "index files".  These are human-readable files
125       that list on a per-.cm-file basis each toplevel symbol defined or
126       imported.  The location of the index file for
127       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
128       To enable index-file generation, set CM.Control.generate_index to true
129       or export an environment-symbol: export CM_GENERATE_INDEX=true.
130    
131       The CM manual has been updated accordingly.
132    
133    2. I made some slight modifications to the c-calls API in MLRISC.
134    
135         a) There is now a callback to support saving/restoring of
136            dedicated but caller-save registers around the actual call
137            instruction.
138         b) One can optionally specify a comment-annotation for the
139            call instruction.
140    
141    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
142       (For example, the comment annotation shows the C prototype of
143        the function being called.)
144    
145    ----------------------------------------------------------------------
146    Name: Matthias Blume
147    Date: 2001/06/01 13:30:00 EDT
148    Tag: blume-20010601-nlffi-cleanup
149    Description:
150    
151    This is mostly a cleanup of MLFFI stuff:
152    
153       - some signature files have been put into a more exposed place
154       - the ugly 'f type parameter is gone (simplifies types tremendously!)
155       - ml-nlffigen changed accordingly
156       - tutorial updated
157    
158    Other changes:
159    
160       - author's affiliation in CM manual(s) updated
161       - some more recognized keywords added to Allen's sml.sty
162    
163    ----------------------------------------------------------------------
164    Name: Matthias Blume
165    Date: 2001/05/25 15:30:00 EDT
166    Tag: blume-20010525-iptr
167    Description:
168    
169      - put the official 110.33-README (as it appears on the ftp server) under
170        CVS
171      - fixed a small bug related to incomplete pointer types in
172        ml-nlffigen
173      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
174        (it does not need the 'f type parameter)
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume
178    Date: 2001/05/23 14:30:00 EDT
179    Tag: Release_110_33
180    Description:
181    
182    New version number (110.33).  New bootfiles.
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume
186    Date: 2001/05/22 18:06:00 EDT
187    Tag: blume-20010522-targets
188    Description:
189    
190    Made install.sh use file config/targets.customized if it exists, falling
191    back to config/targets if it doesn't.  This way one can have a customized
192    version of the targets file without touching the "real thing", thus
193    eliminating the constant fear of accidentally checking something bogus
194    back into the CVS repository...  (File config/targets.customized must
195    not be added to the repository!)
196    
197    ----------------------------------------------------------------------
198    Name: Matthias Blume
199    Date: 2001/05/22 16:30:00 EDT
200    Tag: blume-20010522-minitut
201    Description:
202    
203    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
204       struct returns.
205    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
206       incomplete, preliminary documentation for NLFFI.
207    
208    ----------------------------------------------------------------------
209    Name: Matthias Blume
210    Date: 2001/05/14 11:30:00 EDT
211    Tag: blume-20010514-script
212    Description:
213    
214    Some bugs in install script fixed.
215    
216    In addition to that I also made a slight change to the NLFFI API:
217    Functors generated by ml-nlffigen now take the dynamic library as a
218    straight functor argument, not as a suspended one.  (The original
219    functor code used to force the suspension right away anyway, so there
220    was nothing gained by this complication of the interface.)
221    
222    ----------------------------------------------------------------------
223    Name: Matthias Blume
224    Date: 2001/05/11 14:35:00 EDT
225    Tag: blume-20010511-ml-nlffi
226    Description:
227    
228    I finally took the plunge and added my new FFI code to the main
229    repository.  For x86-linux it is now ready for prime-time.
230    
231    There are two new subdirectories of "src":
232    
233      - ml-nlffi-lib:
234           The utility library for programs using the FFI interface.
235           Here is the implementation of $/c.cm and its associated low-level
236           partners $/c-int.cm and $/memory.cm.
237      - ml-nlffigen:
238           A stand-alone program for generating ML glue code from C source
239           code.
240    
241    Building ml-nlffigen requires $/ckit-lib.cm.
242    
243    The config/install.sh script has been updates to do the Right Thing
244    (hopefully).
245    
246    Notice that the source tree for the C-Kit will not be put under "src"
247    but directly under the installation root directory.  (This is the
248    structure that currently exists on the CVS server when you check out
249    module "sml".)  Fortunately, config/install.sh knows about this oddity.
250    
251    Bugs: No documentation yet.
252    
253    ----------------------------------------------------------------------
254    Name: Matthias Blume
255    Date: 2001/05/09 16:35:00 EDT
256    Tag: blume-20010509-cpscontract
257    Description:
258    
259    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
260    wrapper/unwrapper elimination did not decrement usage counts and some
261    dead variables got overlooked by the dead-up logic.)
262    
263    ----------------------------------------------------------------------
264    Name: Lal George
265    Date: 2001/05/08  17:26:09 EDT
266    Tag: george-20010508-omit-frameptr
267    Description:
268    
269    Changes to implement the omit-frame-pointer optimization to support
270    raw C calls. For now, there is only support on the Intel x86, but
271    other architectures will follow as more experience is gained with this.
272    
273    
274    ----------------------------------------------------------------------
275    Name: Matthias Blume
276    Date: 2001/05/07 14:40:00 EDT
277    Tag: blume-20010507-proxies
278    Description:
279    
280    I made into "proxy libraries" all libraries that qualify for such a
281    change.  (A qualifying library is a library that has another library or
282    groups as its sole member and repeats that member's export list
283    verbatim.  A proxy library avoids this repetition by omitting its export
284    list, effectively inheriting the list that its (only) member exports.
285    See the CM manual for more explanation.)
286    The main effect is that explicit export lists for these libraries
287    do not have to be kepts in sync, making maintenance a bit easier.
288    
289    I also added copyright notices to many .cm-files.
290    
291    Last but not least, I made a new set of bootfiles.
292    
293    ----------------------------------------------------------------------
294    Name: Matthias Blume
295    Date: 2001/05/04 17:00:00 EDT
296    Tag: blume-20010504-cm-lsplit
297    Description:
298    
299    0. John merged pending changes to $/smlnj-lib.cm
300    
301    1. Allen's previous change accidentally backed out of one of Lal's
302       earlier changes.  I undid this mistake (re-introducing Lal's change).
303    
304    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
305       within the compiler where applicable.  There is some code simplification
306       because of that.
307    
308    3. The "split" phase (in FLINT) is now part of the default list of phases.
309       Compiler.Control.LambdaSplitting.* can be used to globally control the
310       lambda-splitting (cross-module-inlining) engine.  In addition to that,
311       it can now also be controlled on a per-source basis: CM has been taught
312       a new tool parameter applicable to ML source files.
313    
314       - To turn lambda-splitting off completely:
315            local open Compiler.Control.LambdaSplitting in
316                val _ = set Off
317            end
318       - To make "no lambda-splitting" the global default (but allow per-source
319         overriding); this is the initial setting:
320            local open Compiler.Control.LambdaSplitting in
321                val _ = set (Default NONE)
322            end
323       - To make "lambda-splitting with aggressiveness a" the global default
324         (and allow per-source overriding):
325            local open Compiler.Control.LambdaSplitting in
326                val _ = set (Default (SOME a))
327            end
328    
329       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
330         write (in the respective .cm-file):
331            a.sml (lambdasplitting:off)
332       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
333            a.sml (lambdasplitting:on)
334       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
335         <a> is a decimal non-negative integer):
336            a.sml (lambdasplitting:<a>)
337       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
338            a.sml (lambdasplitting:infinity)
339       - To use the global default for a.sml:
340            a.sml (lambdasplitting:default)
341         or simply
342            a.sml
343    
344    ----------------------------------------------------------------------
345    Name: Allen Leung
346    Date: 2001/05/04 01:57:00 EDT
347    Tag: leunga-20010504-sync
348    Description:
349    
350      MLRISC features.
351    
352      1. Fix to CMPXCHG instructions.
353      2. Changed RA interface to allow annotations in callbacks.
354      3. Added a new method to the stream interface to allow annotations updates.
355    
356    ----------------------------------------------------------------------
357    Name: Matthias Blume
358    Date: 2001/05/01 11:45:00 EDT
359    Tag: blume-20010501-pcedittmp
360    Description:
361    
362    Changed install.sh to use the current working directory instead of
363    /usr/tmp for a temporary file (pcedittmp).  The previous choice
364    of /usr/tmp caused trouble with MacOS X because of file premission
365    problems.
366    
367    ----------------------------------------------------------------------
368    Name: Matthias Blume
369    Date: 2001/04/20 11:10:00 EDT
370    Tag: blume-20010420-inMLflag
371    Description:
372    
373     - added vp_limitPtrMask to vproc-state.h
374       (for use by the raw-C-calls mechanism to implement proper interrupt
375        handling)
376     - made the ML compiler aware of various data-structure offsets so it
377       can generate code for accessing the vp_inML flag and vp_limitPtrMask
378     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
379       raw C-calls
380    
381    ----------------------------------------------------------------------
382    Name: Lal George
383    Date: 2001/04/20 09:15:28 EDT
384    Tag: george-20010420-macosX
385    Description:
386    
387     - Changes to port to Mac OS X; Darwin.
388    
389     - In the process I found that sqrt was broken on the PPC, because the
390       fsqrt instruction is not implemented.
391    
392    ----------------------------------------------------------------------
393    Name: Matthias Blume
394    Date: 2001/04/18 12:45:00 EDT
395    Tag: blume-20010418-ccalls
396    Description:
397    
398     - fixed two off-by-4 errors in the x86-specific c-calls implementation
399       (this bug prevented structure arguments containing pointers from being
400        passed correctly)
401     - changed the raw-C-call code in mlriscGen.sml in such a way that
402       structure arguments are represented as a pointer to the beginning
403       of the structure (instead of having a series of synthesized arguments,
404       one for each structure member)
405    
406     - made makeml script's verbosity level configurable via environment
407       variable (MAKEML_VERBOSITY)
408    
409     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
410       in rawmem-x86.sml; we are now using the real thing
411    
412    ----------------------------------------------------------------------
413    Name: Matthias Blume
414    Date: 2001/03/22 16:25:00 EST
415    Tag: blume-20010322-bootfiles
416    Description:
417    
418    Created a new set of bootfiles (for your automatic installation convenience).
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume
422    Date: 2001/03/22 15:10:00 EST
423    Tag: blume-20010322-rawmem-parcm
424    Description:
425    
426    1. All "raw memory access" primitives for the new FFI are implemented now
427       (at least on the x86).
428    2. Some further cleanup of CM's parallel make mechanism.
429    
430    ----------------------------------------------------------------------
431    Name: Matthias Blume
432    Date: 2001/03/19 17:53:00 EST
433    Tag: blume-20010319-parallel
434    Description:
435    
436    Parallel make (using compile servers) now works again.
437    
438    To this end, CM.stabilize and CMB.make have been modified to work in
439    two passes when compile servers are attached:
440       1. Compile everything, do not perform stabilization; this pass
441          uses compile servers
442       2. Stabilize everything; this pass does not use compile servers
443    If there are no compile servers, the two passes are combined into one
444    (as before).  Splitting the passes increases the inherent parallelism
445    in the dependency graph because the entire graph including all
446    libraries is available at the same time.  This, in turn, improves
447    server utilization.  The downside is that the master process will
448    have to do some extra work after compilation is done (because for
449    technical reasons it must re-read all the binfiles during stabilization).
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume
453    Date: 2001/03/16 12:22:00 EST
454    Tag: blume-20010316-bootfiles
455    Description:
456    
457    Created a new set of bootfiles (for your automatic installation convenience).
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume
461    Date: 2001/03/16 11:00:00 EST
462    Tag: blume-20010316-MLTREE-fixup
463    Description:
464    
465    This is a minor fixup for an (untagged) earlier commit by Allen.
466    (A file was missing).
467    
468    ----------------------------------------------------------------------
469    Name: Allen Leung
470    Date: Mon Mar  5 18:54:57 EST 2001
471    Tag: leunga-20010305-cut-support
472    
473    1. New support for alternative control-flow in MLTREE.
474       Currently we support
475    
476          FLOW_TO(CALL ...., [k1,...,kn])
477    
478       This is needed for 'cuts to' in C-- and try/handle-like constructs
479       in Moby
480    
481       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
482    
483    2. Register Allocator
484    
485       Changes in interface [from Fermin, John]
486    
487    3. Alpha 8-bit SLL support [Fermin]
488    
489    4. All architectures
490    
491       A new module (ClusterExpandCopies) for expanding parallel copies.
492    
493    ----------------------------------------------------------------------
494    Name: Allen Leung
495    Date: 2001/02/27 23:07:00 EST
496    Tag: leunga-20010227-minor-stuff
497    
498    1. Alpha bug fix for CMOVNE
499    2. Handle mltree COND(..,FCMP ...,...)
500    3. Bug fix in simplifier
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume
504    Date: 2001/01/30 17:50:00 EST
505    Tag: blume-20010130-sync
506    Description:
507    
508    This is just a minor update to sync my devel branch with the main brach.
509    The only visible change is the addition of some README files.
510    
511    ----------------------------------------------------------------------
512    Name: Matthias Blume
513    Date: 2001/01/12 23:30:00 JST
514    Tag: blume-20010112-bootfiles
515    Description:
516    
517    Made a new set of bootfiles that goes with the current state of the
518    repository.
519    
520    ----------------------------------------------------------------------
521    Name: Matthias Blume
522    Date: 2001/01/12 21:20:00 JST
523    Tag: blume-20010112-sync
524    Description:
525    
526    I am just flushing out some minor changes that had accumulated in
527    my private branch in order to sync with the main tree.  (This is
528    mainly because I had CVS trouble when trying to merge _into_ my
529    private branch.)
530    
531    Most people should be completely unaffected by this.
532    
533    ----------------------------------------------------------------------
534    Name: Allen Leung
535    Date: Thu Jan 11 21:03:00 EST 2001
536    Tag: leunga-20010111-labexp=mltree
537    Description:
538    
539    1.  Removed the type LabelExp and replace it by MLTree.
540    2.  Rewritten mltree-simplify with the pattern matcher tool.
541    3.  There were some bugs in alpha code generator which would break
542        64-bit code generation.
543    4.  Redo the tools to generate code with the
544    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
545        are now generated by perl scripts.
546    
547    ----------------------------------------------------------------------
548    Name: Matthias Blume
549    Date: 2001/01/10 21:55:00 JST
550    Tag: blume-20010110-rcc
551    Description:
552    
553    The RCC stuff now seems to work (but only on the x86).
554    This required hacking of the c-calls interface (and -implementation) in
555    MLRISC.
556    
557    Normal compiler users should be unaffected.
558    
559    ----------------------------------------------------------------------
560    Name: Matthias Blume
561    Date: 2001/01/09 01:20:00 JST
562    Tag: blume-20010109-rcc
563    Description:
564    
565    This is a fairly big patch, flushing out a large number of pending
566    changes that I made to my development copy over the last couple of days.
567    
568    Of practical relevance at this moment is a workaround for a pickling
569    bug that Allen ran into the other day.  The cause of the bug itself is
570    still unknown and it might be hard to fix it properly, but the
571    workaround has some merits of its own (namely somewhat reducing pickling
572    overhead for certain libraries).  Therefore, I think this solution should
573    be satisfactory at this time.
574    
575    The rest of the changes (i.e., the vast majority) has to do with my
576    ongoing efforts of providing direct support for C function calls from
577    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
578    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
579    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
580    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
581    handle RCC.
582    
583    The last part (the changes to mlriscGen.sml) are still known to be
584    wrong on the x86 and not implemented on all other architectures.  But
585    the infrastructure is in place. I had to change a few functor
586    signatures in the backend to be able to route the CCalls interface
587    from MLRISC there, and I had to specialize the mltree type (on the
588    x86) to include the necessary extensions. (The extensions themselves
589    were already there and redy to go in MLRISC/x86).
590    
591    Everything should be very happy as soon as someone helps me with
592    mlriscGen.sml...
593    
594    In any case, nothing of this should matter to anyone as long as the
595    new primop is not being used (which is going to be the case unless you
596    find it where I hid it :). The rest of the compiler is completely
597    unaffected.
598    
599    ----------------------------------------------------------------------
600    Name: Matthias Blume
601    Date: 2001/01/05 00:30:00 JST
602    Tag: blume-20010105-primops
603    Description:
604    
605    Added some experimental support for work that I am doing right now.
606    These changes mostly concern added primops, but there is also a new
607    experimental C library in the runtime system (but currently not enabled
608    anywhere except on Linux/X86).
609    
610    In the course of adding primops (and playing with them), I discovered that
611    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
612    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
613    To recover, I made type information in INL_PRIM mandatory and changed
614    prim.sml as well as built-in.sml accordingly.  The InLine structure now
615    has complete, correct type information (i.e., no bottom types).
616    
617    Since all these changes mean that we need new binfiles, I also bumped the
618    version number to 110.32.1.
619    
620  ----------------------------------------------------------------------  ----------------------------------------------------------------------
621  Name: Matthias Blume  Name: Matthias Blume
622  Date: 2000/12/30 22:10:00 JST  Date: 2000/12/30 22:10:00 JST

Legend:
Removed from v.771  
changed lines
  Added in v.842

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