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 1158, Fri Mar 22 01:21:25 2002 UTC revision 1192, Wed May 15 14:02:06 2002 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Lal George
17    Date: 2002/05/15 09:20:10 EDT
18    Tag: george-20020515-pseudo-op-decls
19    Description:
20    
21            All pseudo-ops emitted before the first segment declaration
22    such as TEXT, DATA, and BSS directives are assumed to be global
23    declarations and are emitted first in the assembly file. This is
24    useful in a number of situations where one has pseudo-ops that are not
25    specific to any segment, and also works around the constraint that one
26    cannot have client pseudo-ops in the TEXT segment.
27    
28            Because no segment is associated with these declarations it is
29    an error to allocate any space or objects before the first segment
30    directive and an exception will be raised. However, we cannot make
31    this check for client pseudo-ops.
32    
33            These top level declarations are a field in the CFG graph_info.
34    In theory you can continue to add to this field after the CFG has been
35    built -- provided you know what you are doing;-)
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume
39    Date: 2002/05/13 16:40:00 EDT
40    Tag: blume-20020513-pp-etc
41    Description:
42    
43    A few minor bugfixes:
44    
45      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
46        (Bogus printouts for redefined bindings still occur.  Compiler
47        bug should no longer occur now.  We need to redo the prettyprinter
48        from scratch.)
49    
50      - CM pathname printer now also adds escape sequences for ( and )
51    
52      - commend and docu fixes for ml-nlffi
53    
54    ----------------------------------------------------------------------
55    Name: Matthias Blume
56    Date: 2002/05/10 16:40:00 EDT
57    Tag: blume-20020510-erg-textio
58    Description:
59    
60    Applied the following bugfix provided by Emden Gansner:
61    
62        Output is corrupted when outputSubstr is used rather than output.
63    
64        The problem occurs when a substring
65    
66            ss = (s, dataStart, dataLen)
67    
68        where dataStart > 0, fills a stream buffer with avail bytes left.
69        avail bytes of s, starting at index dataStart, are copied into the
70        buffer, the buffer is flushed, and then the remaining dataLen-avail
71        bytes of ss are copied into the beginning of the buffer. Instead of
72        starting this copy at index dataStart+avail in s, the current code
73        starts the copy at index avail.
74    
75        Fix:
76        In text-io-fn.sml, change line 695 from
77             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
78        to
79             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
80    
81    ----------------------------------------------------------------------
82    Name: Matthias Blume
83    Date: 2002/04/12 13:55:00 EDT
84    Tag: blume-20020412-assyntax
85    Description:
86    
87    1. Grabbed newer assyntax.h from the XFree86 project.
88    2. Fiddled with how to compile X86.prim.asm without warnings.
89    3. (Very) Minor cleanup in CM.
90    
91    ----------------------------------------------------------------------
92    Name: Matthias Blume
93    Date: 2002/04/01 (no joke!) 17:07:00 EST
94    Tag: blume-20020401-x86div
95    Description:
96    
97    Added full support for div/mod/rem/quot on the x86, using the machine
98    instruction's two results (without clumsily recomputing the remainder)
99    directly where appropriate.
100    
101    Some more extensive power-of-two support was added to the x86 instruction
102    selector (avoiding expensive divs, mods, and muls where they can be
103    replaced with cheaper shifts and masks).  However, this sort of thing
104    ought to be done earlier, e.g., within the CPS optimizer so that
105    all architectures benefit from it.
106    
107    The compiler compiles to a fixed point, but changes might be somewhat
108    fragile nevertheless.  Please, report any strange things that you might
109    see wrt. div/mod/quot/rem...
110    
111    ----------------------------------------------------------------------
112    Name: Matthias Blume
113    Date: 2002/03/29 17:22:00
114    Tag: blume-20020329-div
115    Description:
116    
117    Fixed my broken div/mod logic.  Unfortunately, this means that the
118    inline code for div/mod now has one more comparison than before.
119    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
120    The problem was with quotient = 0, because that alone does not tell
121    us which way the rounding went.  One then has to look at whether
122    remainder and divisor have the same sign...  :(
123    
124    Anyway, I replaced the bootfiles with fresh ones...
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume
128    Date: 2002/03/29 14:10:00 EST
129    Tag: blume-20020329-inlprims
130    Description:
131    
132    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
133    
134    Primops have changed. This means that the bin/boot-file formats have
135    changed as well.
136    
137    To make sure that there is no confusion, I made a new version.
138    
139    
140    CHANGES:
141    
142    * removed REMT from mltree (remainder should never overflow).
143    
144    * added primops to deal with divisions of all flavors to the frontend
145    
146    * handled these primops all the way through so they map to their respective
147      MLRISC support
148    
149    * used these primops in the implementation of Int, Int32, Word, Word32
150    
151    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
152    
153    * parameterized INLMIN, INLMAX, and INLABS by a numkind
154    
155    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
156      floating point
157    
158    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
159      and Real (but Real.abs maps to a separate floating-point-only primop)
160    
161    
162    TODO items:
163    
164    * Hacked Alpha32 instruction selection, disabling the selection of REMx
165      instructions because the machine instruction encoder cannot handle
166      them.  (Hppa, PPC, and Sparc instruction selection did not handle
167      REM in the first place, and REM is supported by the x86 machine coder.)
168    
169    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
170      selection phase.  (The two can be streamlined because the hardware
171      delivers both quotient and remainder at the same time anyway.)
172    
173    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
174      (Currently the behavior is inconsistent both across architectures and
175      wrt. the draft Basis spec.)
176    
177    * Word8 should eventually be handled natively, too.
178    
179    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
180      as if there currently is no execution path that could trigger it in
181      SML/NJ.  (The assumptions underlying functions arith and promotable do not
182      hold for things like multiplication and division.)
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume
186    Date: 2002/03/27 16:27:00 EST
187    Tag: blume-20020327-mlrisc-divisions
188    Description:
189    
190    Added support for all four division operations (ML's div, mod, quot,
191    and rem) to MLRISC.  In the course of doing so, I also rationalized
192    the naming (no more annoying switch-around of DIV and QUOT), by
193    parameterizing the operation by div_rounding_mode (which can be either
194    DIV_TO_ZERO or DIV_TO_NEGINF).
195    
196    The generic MLTreeGen functor takes care of compiling all four
197    operations down to only round-to-zero div.
198    
199    Missing pieces:
200    
201      * Doing something smarter than relying on MLTreeGen on architectures
202        like, e.g., the x86 where hardware division delivers both quotient and
203        remainder at the same time.  With this, the implementation of the
204        round-to-neginf operations could be further streamlined.
205    
206      * Remove inlining support for div/mod/rem from the frontend and replace it
207        with primops that get carried through to the backend.  Do this for all
208        int and word types.
209    
210    ----------------------------------------------------------------------
211    Name: Matthias Blume
212    Date: 2002/03/25 17:25:00 EST
213    Tag: blume-20020325-divmod
214    Description:
215    
216    I improved (hopefully without breaking them) the implementation of Int.div,
217    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
218    advantage of the following observations:
219    
220      Let  q = x quot y      r = x rem y
221           d = x div  y      m = x mod y
222    
223    where "quot" is the round-to-zero version of integer division that
224    hardware usually provides.  Then we have:
225    
226         r = x - q * y        where neither the * nor the - will overflow
227         d = if q >= 0 orelse x = q * y then q else q - 1
228                              where neither the * nor the - will overflow
229         m = if q >= 0 orelse r = 0 then r else r + y
230                              where the + will not overflow
231    
232    This results in substantial simplification of the generated code.
233    The following table shows the number of CFG nodes and edges generated
234    for
235            fun f (x, y) = x OPER y
236            (* with OPER \in div, mod, quot, rem *)
237    
238    
239        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
240        --------------------------------------------------------
241         div |         24 |         39 |         12 |         16
242         mod |         41 |         71 |         12 |         16
243        quot |          8 |         10 |          8 |         10
244         rem |         10 |         14 |          8 |         10
245    
246    
247    ----------------------------------------------------------------------
248    Name: Matthias Blume
249    Date: 2002/03/25 22:06:00 EST
250    Tag: blume-20020325-cprotobug
251    Description:
252    
253    Fixed a bug in cproto (c prototype decoder).
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume
257    Date: 2002/03/25 16:00:00 EST
258    Tag: blume-20020325-raw-primops
259    Description:
260    
261    I did some cleanup to Allen's new primop code and
262    replaced yesterday's bootfiles with new ones.
263    (But they are stored in the same place.)
264    
265    ----------------------------------------------------------------------
266    Name: Matthias Blume
267    Date: 2002/03/24 22:40:00 EST
268    Tag: blume-20020324-bootfiles
269    Description:
270    
271    Made the bootfiles that Allen asked for.
272    
273    ----------------------------------------------------------------------
274    Name: Allen Leung
275    Date: 2002/03/23 15:50:00 EST
276    Tag: leunga-20020323-flint-cps-rcc-primops
277    Description:
278    
279      1. Changes to FLINT primops:
280    
281        (* make a call to a C-function;
282         * The primop carries C function prototype information and specifies
283         * which of its (ML-) arguments are floating point. C prototype
284         * information is for use by the backend, ML information is for
285         * use by the CPS converter. *)
286      | RAW_CCALL of { c_proto: CTypes.c_proto,
287                       ml_args: ccall_type list,
288                       ml_res_opt: ccall_type option,
289                       reentrant : bool
290                     } option
291       (* Allocate uninitialized storage on the heap.
292        * The record is meant to hold short-lived C objects, i.e., they
293        * are not ML pointers.  With the tag, the representation is
294        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
295        *)
296      | RAW_RECORD of {tag:bool,sz:int}
297      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
298    
299      2.  These CPS primops are now overloaded:
300    
301           rawload of {kind:numkind}
302           rawstore of {kind:numkind}
303    
304          The one argument form is:
305    
306             rawload {kind} address
307    
308          The two argument form is:
309    
310             rawload {kind} [ml object, byte-offset]
311    
312      3. RAW_CCALL/RCC now takes two extra arguments:
313    
314         a. The first is whether the C call is reentrant, i.e., whether
315            ML state should be saved and restored.
316         b. The second argument is a string argument specifying the name of
317            library and the C function.
318    
319         These things are currently not handled in the code generator, yet.
320    
321      4. In CProto,
322    
323         An encoding type of "bool" means "ml object" and is mapped into
324         C prototype of PTR.  Note that "bool" is different than "string",
325         even though "string" is also mapped into PTR, because "bool"
326         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
327    
328      5. Pickler/unpicker
329    
330         Changed to handle RAW_RECORD and newest RAW_CCALL
331    
332      6. MLRiscGen,
333    
334         1. Changed to handle the new rawload/rawstore/rawrecord operators.
335         2. Code for handling C Calls has been moved to a new module CPSCCalls,
336            in the file CodeGen/cpscompile/cps-c-calls.sml
337    
338      7. Added the conditional move operator
339    
340             condmove of branch
341    
342         to cps.  Generation of this is still buggy so it is currently
343         disabled.
344    
345    ----------------------------------------------------------------------
346    Name: Lal George
347    Date: 2002/03/22 14:18:25 EST
348    Tag: george-20020322-cps-branch-prob
349    Description:
350    
351    Implemented the Ball-Larus branch prediction-heuristics, and
352    incorporated graphical viewers for control flow graphs.
353    
354    Ball-Larus Heuristics:
355    ---------------------
356    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
357    
358    By design it uses the Dempster-Shafer theory for combining
359    probabilities.  For example, in the function:
360    
361        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
362    
363    the ball-larus heuristics predicts that the n=0 is unlikely
364    (OH-heuristic), and the 'then' branch is unlikely because of the
365    RH-heuristic -- giving the 'then' branch an even lower combined
366    probability using the Dempster-Shafer theory.
367    
368    Finally, John Reppy's loop analysis in MLRISC, further lowers the
369    probability of the 'then' branch because of the loop in the else
370    branch.
371    
372    
373    Graphical Viewing:
374    ------------------
375    I merely plugged in Allen's graphical viewers into the compiler. The
376    additional code is not much. At the top level, saying:
377    
378            Control.MLRISC.getFlag "cfg-graphical-view" := true;
379    
380    will display the graphical view of the control flow graph just before
381    back-patching.  daVinci must be in your path for this to work. If
382    daVinci is not available, then the default viewer can be changed
383    using:
384    
385            Control.MLRISC.getString "viewer"
386    
387    which can be set to "dot" or "vcg" for the corresponding viewers. Of
388    course, these viewers must be in your path.
389    
390    The above will display the compilation unit at the level of clusters,
391    many of which are small, boring, and un-interesting. Also setting:
392    
393            Control.MLRISC.getInt "cfg-graphical-view_size"
394    
395    will display clusters that are larger than the value set by the above.
396    
397    
398    ----------------------------------------------------------------------
399    Name: Matthias Blume
400    Date: 2002/03/21 22:20:00 EST
401    Tag: blume-20020321-kmp-bugfix
402    Description:
403    
404    Changed the interface to the KMP routine in PreString and fixed
405    a minor bug in one place where it was used.
406    
407    ----------------------------------------------------------------------
408  Name: Allen Leung  Name: Allen Leung
409  Date: 2002/03/21 20:30:00 EST  Date: 2002/03/21 20:30:00 EST
410  Tag: leunga-20020321-cfg  Tag: leunga-20020321-cfg

Legend:
Removed from v.1158  
changed lines
  Added in v.1192

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