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 972, Sun Oct 28 00:37:02 2001 UTC revision 1011, Thu Jan 10 20:22:04 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    Name: Matthias Blume
16    Date: 2002/01/10
17    Tag: blume-20020110-newffigen
18    Description:
19    
20    1. Updates to portable graph code.
21    
22    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
23       more scalable now so that even huge interfaces such as the one
24       for GTK compile in finite time and space. :-)
25       See src/ml-nlffigen/README for details on what's new.
26    
27    ----------------------------------------------------------------------
28    Name: Lal George
29    Date: 2001/01/09 14:31:35 EST 2002
30    Tag: george-20011206-rm-native-copy
31    Description:
32    
33            Removed the native COPY and FCOPY instructions
34            from all the architectures and replaced it with the
35            explicit COPY instruction from the previous commit.
36    
37            It is now possible to simplify many of the optimizations
38            modules that manipulate copies. This has not been
39            done in this change.
40    
41    ----------------------------------------------------------------------
42    Name: Lal George
43    Date: 2001/12/06 16:50:13 EST 2001
44    Tag: george-20011206-mlrisc-instruction
45    Description:
46    
47    Changed the representation of instructions from being fully abstract
48    to being partially concrete. That is to say:
49    
50      from
51            type instruction
52    
53      to
54            type instr                              (* machine instruction *)
55    
56            datatype instruction =
57                LIVE of {regs: C.cellset, spilled: C.cellset}
58              | KILL of {regs: C.cellset, spilled: C.cellset}
59              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
60              | ANNOTATION of {i: instruction, a: Annotations.annotation}
61              | INSTR of instr
62    
63    This makes the handling of certain special instructions that appear on
64    all architectures easier and uniform.
65    
66    LIVE and KILL say that a list of registers are live or killed at the
67    program point where they appear. No spill code is generated when an
68    element of the 'regs' field is spilled, but the register is moved to
69    the 'spilled' (which is present, more for debugging than anything else).
70    
71    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
72    We used to generate:
73    
74            DEFFREG f1
75            f1 := f2 + f3
76            trapb
77    
78    but now generate:
79    
80            f1 := f2 + f3
81            trapb
82            LIVE {regs=[f1,f2,f3], spilled=[]}
83    
84    Furthermore, the DEFFREG (hack) required that all floating point instruction
85    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
86    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
87    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
88    this architecture).
89    
90    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
91    all the architectures. This will result in further simplification of the
92    register allocator that must be aware of them for coalescing purposes, and
93    will also simplify certain aspects of the machine description that provides
94    callbacks related to parallel copies.
95    
96    ANNOTATION should be obvious, and now INSTR represents the honest to God
97    machine instruction set!
98    
99    The <arch>/instructions/<arch>Instr.sml files define certain utility
100    functions for making porting easier -- essentially converting upper case
101    to lower case. All machine instructions (of type instr) are in upper case,
102    and the lower case form generates an MLRISC instruction. For example on
103    the alpha we have:
104    
105      datatype instr =
106         LDA of {r:cell, b:cell, d:operand}
107       | ...
108    
109      val lda : {r:cell, b:cell, d:operand} -> instruction
110        ...
111    
112    where lda is just (INSTR o LDA), etc.
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume
116    Date: 2001/11/22 21:40:00 EST
117    Tag: Release_110_37
118    Description:
119    
120    Release 110.37.  This time for real.
121    
122    ----------------------------------------------------------------------
123    Name: Matthias Blume
124    Date: 2001/11/21 16:35:00 EST
125    Tag: blume-20011121-foot-in-mouth
126    Description:
127    
128    Removed the "Release_110_37" tag because of a serious bug.
129    This will be re-tagged once the bug is fixed.
130    
131    ----------------------------------------------------------------------
132    Name: Matthias Blume
133    Date: 2001/11/21 16:14:00 EST
134    Tag: blume-20011121-forgottenfile
135    Description:
136    
137    Forgot to add a file.  (Just a .tex-file -- part of
138    the CM manual source.)
139    
140    ----------------------------------------------------------------------
141    Name: Matthias Blume
142    Date: 2001/11/21 16:10:00 EST
143    Tag: blume-20011121-invalid_110_37
144    Description:
145    
146    Note: I removed the original tag "Release_110_37" from this commit
147          because we found a serious bug in all non-x86 backends.
148          - Matthias
149    
150    1. Modifications to the SML/NJ code generator and to the runtime system
151       so that code object name strings are directly inserted into code
152       objects at code generation time.  The only business the runtime system
153       has with this is now to read the name strings on occasions.
154       (The encoding of the name string has also changed somewhat.)
155    
156    2. CM now implements a simple "set calculus" for specifying export lists.
157       In particular, it is now possible to refer to the export lists of
158       other libraries/groups/sources and form unions as well as differences.
159       See the latest CM manual for details.
160    
161    3. An separate notion of "proxy" libraries has again be eliminated from
162       CM's model.  (Proxy libraries are now simply a special case of using
163       the export list calculus.)
164    
165    4. Some of the existing libraries now take advantage of the new set
166       calculus.
167       (Notice that not all libraries have been converted because some
168       of the existing .cm-files are supposed to be backward compatible
169       with 110.0.x.)
170    
171    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
172       "General.exnMessage"!  The former relies on a certain hook to be
173       initialized, and that often does not happen in the stand-alone case.)
174    
175    ----------------------------------------------------------------------
176    Name: Lal George
177    Date: 2001/11/21  13:56:18 EST
178    Tag: george-2001121-pseudo-ops
179    Description:
180    
181      Implemented a complete redesign of MLRISC pseudo-ops. Now there
182      ought to never be any question of incompatabilities with
183      pseudo-op syntax expected by host assemblers.
184    
185      For now, only modules supporting GAS syntax are implemented
186      but more should follow, such as MASM, and vendor assembler
187      syntax, e.g. IBM as, Sun as, etc.
188    
189    ----------------------------------------------------------------------
190    Name: Matthias Blume
191    Date: 2001/11/14 11:52:00 EST
192    Tag: blume-20011114-srcname
193    Description:
194    
195    1. Routed the name of the current source file to mlriscgen where it
196       should be directly emitted into the code object.  (This last part
197       is yet to be done.)
198    
199    2. Some cleanup of the pgraph code to make it match the proposal that
200       I put out the other day.  (The proposal notwithstanding, things are
201       still in flux here.)
202    
203    ----------------------------------------------------------------------
204    Name: Lal George
205    Date: 2001/11/14 09:44:04 EST
206    Tag:
207    Description:
208    
209      Fix for a backpatching bug reported by Allen.
210    
211      Because the boundary between short and long span-dependent
212      instructions is +/- 128, there are an astounding number of
213      span-dependent instructions whose size is over estimated.
214    
215      Allen came up with the idea of letting the size of span
216      dependent instructions be non-monotonic, for a maxIter
217      number of times, after which the size must be monotonically
218      increasing.
219    
220      This table shows the number of span-dependent instructions
221      whose size was over-estimated as a function of maxIter, for the
222      file Parse/parse/ml.grm.sml:
223    
224         maxIter            # of instructions:
225            10                      687
226            20                      438
227            30                      198
228            40                        0
229    
230      In compiling the compiler, there is no significant difference in
231      compilation speed between maxIter=10 and maxIter=40. Actually,
232      my measurements showed that maxIter=40 was a tad faster than
233      maxIter=10! Also 96% of the  files in the compiler reach a fix
234      point within 13 iterations, so fixing maxIter at 40, while high,
235      is okay.
236    
237    ----------------------------------------------------------------------
238    Name: Matthias Blume
239    Date: 2001/10/31 15:25:00 EST
240    Tag: blume-20011031-pgraph
241    Description:
242    
243    CKIT:
244    * Changed the "Function" constructor of type Ast.ctype to carry optional
245      argument identifiers.
246    * Changed the return type of TypeUtil.getFunction accordingly.
247    * Type equality ignores the argument names.
248    * TypeUtil.composite tries to preserve argument names but gives up quickly
249      if there is a mismatch.
250    
251    installation script:
252    * attempts to use "curl" if available (unless "wget" is available as well)
253    
254    CM:
255    * has an experimental implementation of "portable graphs" which I will
256      soon propose as an implementation-independent library format
257    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
258    
259    NLFFI-LIB:
260    * some cleanup (all cosmetic)
261    
262    NLFFIGEN:
263    * temporarily disabled the mechanism that suppresses ML output for
264      C definitions whose identifiers start with an underscore character
265    * generate val bindings for enum constants
266    * user can request that only one style (light or heavy) is being used;
267      default is to use both (command-line arguments: -heavy and -light)
268    * fixed bug in handling of function types involving incomplete pointers
269    * generate ML entry points that take record arguments (i.e., using
270      named arguments) for C functions that have a prototype with named
271      arguments
272      (see changes to CKIT)
273    
274  ----------------------------------------------------------------------  ----------------------------------------------------------------------
275  Name: Allen Leung  Name: Allen Leung

Legend:
Removed from v.972  
changed lines
  Added in v.1011

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